A SRPP-Input Tube Amplifier For Headphones And Loudspeakers.

by Tony Frazer

frazer1.jpg

I built the prototype of this headphone amplifier in Oct’96 and remain impressed by it’s performance. Imaging is particularly good – I make much greater use of my headphones than ever before! Hum is low and unintrusive. I have used 60 ohm and 300 ohm headphones with this amp. The amp is designed also to drive an efficient pair of 8 ohm loudspeakers. If you do build it, please let me know how you get on – ideas for improvement are always welcome!

frazer2.gif
Figure 1

Figure 1 shows the amplifier schematic for one channel. The circuit topology is a SRPP (Series Regulated Push-Pull) input stage which is AC-coupled to a Parallel-Triode Cathode Bias output stage. To check the circuit gain, I injected a 1kHz sine wave at 1V p/p into the grid of the lower V1. I measured 12V p/p at the grid of V2, and 110V p/p at the anode of V2. The output stage runs at 32mA, 234V, i.e. 7.5 Watts. Class A amplifiers are typically about 20% efficient, so it would be reasonable to expect about 1.5 Watts output.

The amp was designed to work with 8 ohm loudspeakers. Nearly all loudspeaker drive units I have encountered have a nominal impedance of 8 ohms – of course this varies with frequency by a few ohms. When driving headphones, the audio output transformer must be shunted with a 10-ohm resistor (Ro) to present the correct load to the transformer secondary. The shunt allows higher impedance headphones to be used. Remove the shunt (add a switch to take it out of the circuit) when driving loudspeakers.

A wide range of headphone impedances can be used with this amplifier because of its low impedance output. A pair of 60 ohm headphones will present 1/((1/10)+(1/60))= 8.57 ohm load with 60 ohm headphones, while the amp will see a 1/((1/10+(1/300)) = 9.68 ohm load with 300 ohm headphones. I know it isn’t efficient from a dedicated headphone amplifier perspective, where the output transformer would have a secondary matched to the headphone impedance. However, with the 10-ohm shunt, there is still plenty of power.

CONSTRUCTION NOTES

WARNING: Tube circuits involve potentially LETHAL HIGH VOLTAGES and should only be tackled by experienced persons with due regard for SAFETY. Don’t fiddle with the circuit while wearing your ‘phones!

frazer3.jpg
Figure 2

General construction notes for my prototype:

The unit is built on a two-section ‘U’-type Aluminium chassis WxDxH:12x6x2″ with holes in the base section and around valveholders for ventilation.

  1. Chassis mounting B9A valveholders are used throughout.
  2. Point-to-point wiring is used – including a couple of tagstrips.
  3. Mains and output transformers are situated at opposite ends of the chassis, mounted on butyl spacers to suppress vibration which may be picked up by the input valves, with the mains transformer at 90 degrees to minimise magnetic coupling.
Chassis plate layout:
+-------------------------------+  Key:
|         [h] [h]  [V2] [o/p ]  |  tr = Mains Transformer
|[   ]    [c] [c]       [    ]  |  R = rectifier tube
|[tr ][R]              [V1][V1] |  h = heater supply capacitor
|[   ]      [C]         [    ]  |  c = 100uF HT capacitor
|                  [V2] [o/p ]  |  C = 330uF HT capacitor
+-------------------------------+  V1 = 7025; V2 = 12BH7
                                   o/p = SE output transformer

AMPLIFIER CIRCUIT NOTES:

 

  • V1 is a Sovtek 7025/12AX7WA (i.e. a high spec ECC83). I tried 6072 (12AY7) and ECC81 types (with minor bias adjustments). ECC81 sounds a bit sterile, ECC83 has warmth! Raytheon 6072 was also OK, but past experience suggests they may go microphonic.
  • V2 is 12BH7 – I use a couple of Brimars, which, oddly enough appear to have slightly different internal structures! Although they both sound the same, this is something to watch out for if symmetry is important to you!
  • VOL and BALANCE controls are dual gang pots.
  • BALANCE slider is wired to input side for one channel and VOL side for the other. In use, VOL up to half-scale with a CD player is about the limit.
  • All resistors are 0.6W Metal Film unless otherwise stated.
  • 100nF coupling capacitor is polypropylene.
  • Screened leads link Input socket-BAL-VOL-V1 and ground at the Input socket.

Output tranformer is a home-made 2k7-8R using dual isolated bobbin for safety. The ‘1’ on the schematic denotes start (inside) of the windings. Primary: 800t of 32g; Secondary: 45t of 22g. I used a 20VA transformer kit to source the laminations and bobbin. Everything you need to make all the transformers can be obtained from Maplin Electronics by mail order. From the catalogue:

Stock Number / Description
YJ61R / 20VA Transformer Kit
YJ62S / 50VA Transformer Kit

Maplin Electronics also do copper wire on 50g and 250g reels in the gauges required (I would suggest the 250g reels). Heavy items such as transformers have an additional carriage charge.

Maplin Electronics
PO Box 777
Rayleigh
Essex SS6 8LU
United Kingdom

Laminations 2 5/8″ x 2 1/4″; 7/8″ stack. I use a variable-speed hand drill and a coach-bolt/wooden block which the bobbin fits over. On slowest speed, I count revolutions for a minute, then calculate winding time from that. Secondary is wound manually. Don’t interleave ‘E’ and ‘I’ laminatons as you would with a power transformer. No extra gap spacer is required as the ampere turns are well within limits for DC saturation. I use plenty of 1-hour epoxy resin to hold it all together. (That was not intended to be a lesson in winding transformers!)

  • Heater wiring not shown – see Power Supply schematics (figures 3 and 4) for heater supply. All heaters connected in parallel: connect pin 4 to pin 5, centre is pin 9.

 

frazer4.gif
Figure 3

frazer5.gif
Figure 4

NOTES ON THE POWER SUPPLY:

There are two versions of the power supply: the original prototype’s power supply (figure 3) and an all-silicon version (figure 4). The original uses a half-wave thermionic rectifier (EZ81). The all-silicon power supply uses a different transformer HT secondary, and the rectifier heater winding is not required. I have shown a capacitor-choke-capacitor filter which should prove quieter than the original resistor-capacitor chain filter.

Basically, the PSU provides +234V DC for the amp and 6.3V DC for the heaters, but because I am using an SRPP stage where the cathode of the upper half of V1 is at quite a high voltage with respect to gnd, I use the voltage-divider network to ‘pull up’ the heater voltage (with respect to gnd) to reduce the potential difference between cathodes and heaters.

Some tubes have a specified design limit – I got into the habit of doing this with SRPP stages using the 6072 tube for which a maximum of 90 volts K/h is specified, however the ECC83/12AX7 has a maximum ‘peak’ K/h of 200 volts specified – it is arguably unnecesary for the ECC83, but for the cost of a couple of resistors and a cap. it seems a worthwhile precaution to prolong tube life, or when experimenting with different tubes.

A green L.E.D. and 1K5 resistor (not shown) are connected across the smoothed heater supply as a pilot light.

Addendum

11/9/98: Corrected resistor value from 47 to 47K ohms in figures 4 and 5.

c. 1997 Tony Frazer.

No-Compromise OTL Tube Headphone Amplifier.

by Andrea Ciuffoli

ciuff_nc1.jpg

This OTL tube headphone amplifier is a more refined design of my previous amplifiers (see A Top-Level Headphone Amplifier). The sound of this version is very “real” and so detailed that you can hear the quality of the source CD player and CD music.

Here, I allowed no compromise to get the best sound for this type of application. It can be used without any limitation as preamplifier. It can drive any kind of power amplifier (such as my Power Follower 99), because it has enough output voltage swing, enough gain (about 30db) and a very low output impedance (about 30 ohms).

ciuff_nc2.gif
Figure 1

The schematic for one channel of the headphone amplifier is shown in figure 1. Please see A Top-Level Headphone Amplifier for general information about the circuit. The improvements in this version come from the careful selection of parts, the use of tube diodes, and the elimination of the interstage MKP capacitor.

Please use only Allen-Bradley, 2W carbon resistors where specified in the schematic. The capacitors must be ELNA Cerafine without any kind of bypass, but the more expensive Black Gate capacitors could be better. I have not tested the Black Gates, but I know (because I have read so in many articles) that the Black Gates need many hours of use before they sound good. It is also necessary to switch on the amplifier every day, or the sound gets bad again.

For the class A voltage-gain stage, I tested many tubes, such as the JAN-Philips 5814, the JAN-Philips 5687, the Mullard E182CC, the JAN-Philips 6SN7 and the RCA 3A5 direct-heating twin-triode. I found that the Raytheon 5842 hi-mu triodes had the best sound (JAN-Raytheon 5842WA). It was not necessary to select the tubes. They cost about $17 US – very good for this product. I bought them at:

SELECTION COMPONENTS
via G. De Leva 13
00179 Roma
ITALY
tel. 0039 06 7840118
email: mc7455@mclink.it

SELECTION COMPONENTS will give a plot of the tubes for only $7 US.

The output stage is a paralleled-triode cathode follower. For the output tubes, I tried the E182CC and E288CC and nothing else. I used the Mullard E182CC and suggest only tubes by Mullard, JAN-Philips or Siemens in the output. The amplifier will drive headphones of 200 to 600 ohms impedance. For more output to drive 150 ohm headphones, add another 22K ohm resistor (9) to Rk to increase the bias current of the E182CC.

ciuff_nc3.gif
Figure 2

The power supply schematic is shown in figure 2. To get no-compromise performance, the rectification should be vacuum tubes only. I tested the JAN-Philips 5Y3 and the RCA 5R4. I suggest using two 5Y3s or one 5R4, as you like. Only the first capacitor after the diodes could be a normal electrolytic type or better, a paper-oil type. The inductors can have an internal resistance of up to 70 ohms max. and no more!!! I used inductors made by Trau in Italy, and these cost about $23 US.

The filament supply is AC, but I am developing an alternative supply for use in a country like Italy, where I live and where the voltage goes from 230 volts in the morning to the 196 volts in the evening. In my previous article, I presented a very good DC filament supply, but in tests, I heard better sound with the AC supply. To avoid turn-on thumps, wait about 30 seconds after switch-on before connecting the headphones. I will probably add a time-delay to the audio output, but integrated with a new regulator for the filament supply.

ciuff_nc4.jpg
Figure 3

The frame of the chassis is made of walnut and measures 53cm x 30cm x 5cm. The top is a sheet of 3mm aluminum. I kept the wood separated from the metal with gaskets placed at the screw points. All of the audio input and output jacks are gold-plated. The extra octal tube socket seen in the photograph was for testing a 6SN7 tube.

Addendum

5/17/99: Decreased anode power supply from 280VDC to 190-200VDC to increase lifetime of the tubes and for better sound, removed output capacitor after rectification in power supply to prevent current peaks on the diode and added option to drive 150 ohm headphones.

c. 1999, Andrea Ciuffoli.
For commercial use of the circuits in this article, please contact Andrea Ciuffoli.
From Andrea Ciuffoli’s Home Page. Republished with permission.

Top-Level OTL Tube Headphone Amplifier.

by Andrea Ciuffoli

This article shows my tube headphone amplifier designs for driving headphones with an impedance of 200 to 600 ohms (I am using these amplifiers with the Sennheiser 580, which has a 300-ohm impedance). The problem with many tube headphone amplifiers is the high output impedance and the low output current that decrease the sonic performance. I have analyzed many schematics, from a hybrid design with an E88CC voltage gain stage and a direct-coupled IRF610 MOSFET output stage to a design involving a 6080 or 6C33C-B as output stage in a cathode follower configuration.

The hybrid with the IRF610 source follower does not have very good sound, the amp with the 6080 output stage has pretty high output impedance (more devices per channel would lower output impedance, but then the tubes must be carefully selected) and the 6C33C-B is not very linear for a headphone amp (but it is very good in an OTL amp for driving loudspeakers).

DESIGN GOALS

The design goals for my amplifiers were:

  • no regulator tubes are used such as 6080 and 6C33C-B, because they are considered less linear
  • low output resistance (about 33 ohm)
  • the bias current of the output stage should be enough for any music peak (20Vpp on 300ohm)
  • very good sound
  • low power consumption
  • small size (30 x 40 cm)
  • some choices for the power supply

The designs have been optimized using a SPICE circuit simulator with an advanced tube model by Mithat F. Konar. This model is very close to real tube performance, and I have developed an interactive program to find the right parameters. Here are the results for the E182CC (simulation in red, actual in white):

ciuff1_1.gif
Figure 1 – E182CC SPICE Simulation

and for the 5814 (ECC82 military type):

ciuff1_2.gif
Figure 2 – 5814 SPICE Simulation

The SPICE model is so close that it is sometimes difficult to see the difference between model and real tube curves. The real tube curves were generated with Audiomatica’s Sofia vacuum tube curve tracer. The SPICE circuit simulator is available for free on Unix (Berkeley) and on Windows (WinSpice), and there are many commercial ports on Windows 95/NT. [Editor: an evaluation copy of a GUI-based SPICE simulator for Windows can be ordered from OrCAD.]

THE CIRCUIT

ciuff1_3a.gif
Figure 3a

ciuff1_3b.gif
Figure 3b

ciuff1_3c.gif
Figure 3c

The headphone amplifier circuits (figures 3a, 3b and 3c) use a single-ended triode class A gain stage and a paralleled-triode cathode follower output stage. The first version with E82CC/5814 should be considered the starting point. It has good sound but with some distortion because noval (new 9-pin tubes) tubes are not very linear. I have tested the first version with a 5814 JAN Philips. It could be that other tubes of this same type such as the E82CC/ECC82 Mullard or Jan Philips will have better sound. Each section of 5814 is biased at 4.42mA and dissipates 3.76W. To reduce the distortion of the first version with E82CC/ECC82/5814 tube, I don’t have a bypass on cathode resistor to leave some local feedback, but you can try to bypass it with 220uF, 16V ELNA Cerafine electrolytic capacitor as with the other versions.

The second version with the 3A5 (I used a RCA 3A5), a direct heating tube, gives less distortion and has a very liquid sound as only direct heating tubes can do. The 3A5 MUST have a DC filament supply, such as the 6.3VDC supply shown in figure 7. The third version with the a Jan Philips 6SN7 (that I now prefer) has a less liquid sound than the 3A5, but the distortion is very low and the bass response is fantastic! Sometimes during listening sessions with the 6SN7 I thought that the high frequencies were missing, but this is not true. Note: the 3A5 is sensitive to microphonics (if you touch the table where the amplifier is, you will hear a “gong” in the headphones). I have mounted the 3A5 on a floating socket. Tube dampeners might also reduce the microphonic effect, but I have not tried them.

In all versions, the output tube V2 is biased at 26mA per section (52mA combined) for a total dissipation of 6W (2x3W). Use only Mullard or JAN Philips tubes.

ciuff1_3d.gif
Figure 3d

All resistors should be Allen Bradley carbon types for the better sound. The 7 x 22K, 2W resistors can be replaced by one 3K ohm Caddock non-inductive type (20w) in the TO220 case. Figure 3d shows how I mounted the 7x22K resistors. The copper board templates are included in the PC board layouts below.

The inter-stage and output capacitors are the Jensen or Audio Note paper-oil types, but a MKP type (such as Solen) will also work. For the electrolytic capacitors, I prefer Black Gate or ELNA Cerafine (by Audio Note), but ROE, SPRAGUE, MALLORY are good substitutes.

In the last years, two trends have emerged about the capacitors to use in a high-end amplifier: MKP polypropylene (output stage and interstage), electrolytics (in the power supply) and paper oil (interstage). The MKPs in many tests rank at the top for low distortion, low internal resistance and maximum speed under dynamic conditions. Electrolytics have higher distortion and lower speed, but very good electrolytics are now available, such as Black Gate and ELNA Cerafine, both designed for audio.

I have not tested these new capacitors, because I am searching for a good source with a good price before I buy. At this time, I am using the ROE electrolytics and the SOLEN MKP. In listening tests, the MKP has an inclination to smooth out contrasts. So make your choice! In this design, the output capacitor of the power supply and the output capacitor of amplifier can be replaced with 220uF or higher value MKP types or other audio types. This list of capacitor sources is very useful for finding the right shop:

THL Audio
Angela Instruments
Audio Note
Sonic Frontiers
Solen Inc.

ciuff1_4.gif
Figure 4

ciuff1_5.gif
Figure 5

ciuff1_5a.gif
Figure 6

About the power supply, the filaments can be powered by AC or DC except that the 3A5 (version 2 of the amplifier) requires a DC filament supply. Each supply has a timed relay that mutes the audio output until the tubes stabilize. I have designed 3 different power supplies – regulated, passive and cheap passive:

  • The supply in figure 4 is regulated with a Darlington hybrid (an IRF power MOSFET and a Motorola BUX48 power transistor) to implement a virtual battery (the circuit is based on a Technics design). The MOSFET/NPN pair can conduct up to 400V at 15A and will easily handle the surge currents of large power supply capacitors. The BUX48 should be heatsinked to dissipate about 6W.
  • The supply in figure 5 is all-passive with inductors (higher cost).
  • The supply in figure 6 is a “cheap” version of the all-passive and uses the slow turn-on filament supply in figure 7 to also energize the relay.

The audio output relay prevents turn-on thumps and allows the tubes to stabilize by muting the amplifier output for about 150 seconds or 2.5 minutes (RC = 330K * 470uF) after startup. The relay is a 12VDC DPDT (7-10A max. load) with a coil resistance of about 150 ohms. See the amplifier schematics (figures 3a and 3b) for details about wiring the relay connections.

ciuff1_6.gif
Figure 7

This amplifier runs quietly without a DC filament power supply, but for a more stable filament voltage, I suggest the slow turn-on supply (figure 7) to replace the filament supplies in figures 4 and 5 (the cheap passive supply already has the slow-on circuit). Also, this circuit is excellent for the 3A5 of version 2 of the amplifier, which requires a DC filament supply. The slow-on supply will also increase the life of tubes! The MJ15004 should be heatsinked to dissipate about 12W.

Full-sized PC board layouts are available for the slow-on power supply and the audio output mute circuits as well as the copper boards for mounting the Allen Bradley resistors. See below for the pc board stuffing guides and suggested chassis layout.

ciuff1_7a.gif

ciuff1_7b.gif

PERFORMANCE

Some achievements of my amplifier (regular version):

  • Gain = 21.5db
  • Rout = 33 ohm
  • THD = 0.88% with Vout = 9.5v on 300 ohm => 200 mW
      • 0.40% with Vout = 5.9v on 300 ohm => 80 mW “MAX POWER FOR Sennheiser 580”
      • 0.17% with Vout = 2.9v on 300 ohm => 15 mw
      • 0.07% with Vout = 1.2v on 300 ohm => 2.5mw
  • Freq = 0.7Hz to 1GHz

ciuff1_8
Figure 9

The graphs in figure 9 are the THD (distortion) decays for the standard amplifier at 9.5V, 2.9V and 1.2V: PERFECT! They show a linear harmonic decay (and the lower the power, the lower the distortion).

[Editor: For the latest version of Andrea Ciuffoli’s design, see A No-Compromise OTL Tube Headphone Amplifier.]

Addendum

9/17/98: Updated figure 4 regulated supply: replaced single MOSFET with MOSFET/BJT Darlington pair to better handle surge currents from large output capacitors. Updated figures 4 and 5 power supplies: grounded filament supply and removed connection point “E” from amplifier schematic (figure 3). Also added a “cheap” version of the passive supply (figure 6).

9/22/98: Added version 2 of amplifier schematic (figure 3b) and revised power supply schematics (figures 4, 5, 6 and 7).

9/24/98: Added Rin to both amplifier schematics (figures 3a and 3b) and removed 6VDC output from slow-start power supply schematic (figure 7). Also corrected value of R7 and fixed spelling of DCC90 name for V1 tube in version 2 of amplifier.

9/29/98: Updated version 2 amp schematic (figure 3b) – added R8, R9. Updated slow-start filament supply (figure 8). Added layout diagram for mounting 7x22K resistors (figure 3c) and suggested chassis layout diagram (figure 7). Also added paragraph on capacitor selection.

9/30/98: Added PC board layouts and chassis layout.

10/15/98: Added 6SN7 version of the headphone amplifier. Also added notes re heatsinking BUX48 and MJ15004.

c. 1998, Andrea Ciuffoli.
For commercial use of the circuits in this article, please contact Andrea Ciuffoli.
From Andrea Ciuffoli’s Home Page. Republished with permission.

 

Output Transformerless Tube Headphone Amplifier.

by Kurt Strain

portman1.jpg

The tube headphone amplifier in figure 1 is a high current mu follower, with a 12AU7 cathode follower driven by a 6CG7. Both sections of each tube are paralleled to minimize noise. It’s a zero global feedback, OTL design of pretty high quality. I use it and like it for Grado, Sennheiser, and Sony headphones. The maximum gain is about 25 dB with the 6CG7, so it pays to watch the attenuator level. R5 and R6 total 20K ohms as a stepped attenuator. A log pot of 20K to 100K ohms would be a good substitute. This design can be used as a line preamp as well.

strain1.gif

The power output depends a lot on headphone impedance. The circuit has significant output impedance (something like 500 ohms) which some headphones don’t mind, some do. It will put out around 20 mA peak current. The output impedance is pretty much constant and independent of load, but any tube changes will greatly affect it.

portman3.jpg

Layout is crucial to avoid motorboat oscillation due to parallel operation. Short wires across tube pins help this, and have them cross each other perpendicularly.  Once the oscillation is gone, it stays gone. If this problem persists, separate cathode resistor R7 into 2 cathode resistors, 100 ohms each to ground, and add 100 ohm plate resistors to V1 on each side, to B+. This will isolate the mismatch between units better and prevent a bistable bias condition.

strain2

Both channels run off a well-filtered 6VDC, 2.5A filament supply and a 250VDC supply with separate filter capacitors for each channel to minimize hum and crosstalk (figure 2). The supplies don’t need to be regulated. The 12AU7 will take the raised cathode potential so you don’t need to raise the voltage of the filament supply above ground level. The Hammond 269EX power transformer works very well in this application and can be obtained from Angela Instruments.

Use shielded twin lead to wire the filaments. Run one lead to each tube separately from the transformer to keep current on the line at a minimum. Keep the filament wire away from the signal and DC power supply wires. Also, keep the power supply away from the amplifier circuit, especially V2.

portman2a.jpg

The way I built it was to put it in an enclosed box that’s about 7″ wide by 4.5″ tall by 12″ deep with vented covers. The inside has a deck with cut-outs for tubes and wires and caps. It is light and compact (weighs about 10 lbs). [Editor: The amplifier shown above was built by Jason Portman, and uses an 8″ x 11″ aluminum chassis from Hammond, which he polished before constructing the amplifier.]

This headphone amp sounds very good with the Sony MDR series. I have found that even though the Grado’s are a low impedance load, they don’t really require a low impedance amp to sound good if the amp works out synergistically.

For Sennheiser 580 headphones, the amp will not put out as much bass definition and detail as other circuits could. The Sennheisers tend to sound soft, so a little tighter amp seems to help. The Grados tend to sound sharper, and so the highish output impedance OTL seems to soften it to where I liked it.

Addendum

12/8/98: Revised power supply (figure 2) for improved ripple rejection and channel isolation.

5/3/99: V2 changed from 6922 to 6CG7/6FQ7 for improved stability. The 6CG7/6FQ7 is pin-for-pin compatible with the 6922 and no other circuit changes are required. The author advises that this substitution be made for existing builds because the 6CG7/6FQ7 tube is better able to handle the power dissipation of the circuit and will last longer.

Also upgraded filament power supply (figure 2) from AC to DC for reduced hum and greater stability.

5/21/99: Added pictures of OTL amplifier built by Jason Portman.

5/25/99: Ole W. Saastad built this version of the OTL amplifier:

saastad.jpg

It runs off a 330V supply, which necessitated changing R2 to 5.6K ohms, 2W. The 6VDC filament supply is regulated with 7806 regulators. All instructions, including a chassis drilling guide and parts layout, can be found at http://www.uio.no/~olews/otl/otl.html.

7/16/99: Revised volume control in figure 1.

1/14/00: Jason Portman writes: The OTL is now fully tweaked and working beautifully. In fact after the conversion to DC heaters it was MUCH MORE STABLE! I have been experimenting with 3 different front end tubes. The 6CG7, 6FQ7 and my favorite, the Svetlana 6N1P. The GE 6CG7’s with “white” logo seen second best in terms of level and sound quality. The 6N1P’s are really much better though; they are much less muddy than the 6CG7’s, have better low end, better imaging and much more transparent. Also an added benefit of their 300ma heaters is they glow quit nicely!

7/15/00: Eduard Orvisky built this headphone amplifier using ideas from Kurt Strain’s and Eric Barbour’s designs. It uses two 6SN7s in the mu follower (compared to Eric Barbour’s circuit, R4 is reduced to 2K, 2W) and solid state and rectified heaters (he also tried a 5Y3 tube rectified supply, but preferred the solid state version for its faster sound). The chassis for this prototype was made from a sheet of plexiglass. It has no bottom, and permits quick substitution of parts.

orvisky1.gif

9/1/00: Eduard Orvisky moved his amp to a new Hammond steel chassis (available at Parts Express) and gave it a stunning paint job. He writes: “I sprayed it matte black in the beginning, but did not like it. Did not look bad, but it was too obvious. Then I tryed silver metallic. That was worse, not as real polished metal. Then I painted acrylic sapphire blue on it. With brush, hence the strokes.”

orvisky3
orvisky4.jpg

c. 1998, 1999 Kurt Strain.
Author’s website: Tube Odd-e-o-file Page.

Brute Force In A Line Stage.

by Eric Barbour

barbour1.jpg

The need for a high-gain stereo preamp has diminished, because the CD has replaced the LP record in most audiophile homes. Most CD players produce enough voltage to drive nearly all amplifiers directly to full power, so you require only a volume control, or a switch if there is more than one source signal, as from an FM tuner. Also, you need a line stage preamp if there isn’t enough output voltage from one of the sources. The CD may have rid us of the preamplifier as a necessity, but not all line-level sources are compatible, and some gain may be required.

Another, often forgotten, purpose of the preamp is to drive your headphones. Although most of today’s “audiophile” preamps lack a headphone jack, the solid-state ones could adequately drive phones. Tube preamps are another matter. A semiconductor driver usually feeds the low-impedance load, as with the Melos SHA-1. Its tubes drive a MOSFET output. Sometimes, especially late at night, I listen through headphones with a pair of AKG K240s (600 ohms), which is a relatively easy load to drive. But I wanted to design an all-tube line stage and driver that could power them, without resorting to output transformers

Design Criteria

My design philosophy was to achieve “purity.” My experience with the common weaknesses of existing preamps helped me determine the following basic rules for designing a “pure” line stage:

1. All tube, no semiconductors if possible;
2. No electrolytic capacitors, if possible;
3. Best possible insulation materials;
4. Best possible tube types without resorting to rarities;
5. Regulated power supply;
6. Low output impedance without iron to drive 600 load;
7. Sufficient voltage gain of 10-20;
8. No feedback, stable, minimalistic.

 

Experimentation

I have many opportunities to experiment with other equipment. I have tried innumerable preamps based on 12AX7 and 6DJ8 types, including Dynaco PAS-3s with every modification imaginable; McIntosh C11, C20, and C22; Scott 130; various Audio Research units; old NYAL Minuets; Lafayette KT-600 (probably the best of the vintage units); and some home-brews.

Not one of them had the real magic. In my opinion, most 12AX7s have a “lumpy” sound and do not maintain similar frequency response between channels, because negative feedback tends to “amplify” response aberrations, especially when tubes age and drift. The 6DJ8s have a wiry, irritating sound to me. I believe that some listeners confuse this kind of distortion with extreme sonic detail.

Although probably the best of the manufactured preamps, the old MFA Luminescence (currently out of production) isn’t perfect. It is too complex, the phono stage is a bit of a monstrosity, and the line stage uses unnecessary negative feedback. It uses all octal-based tubes, which tend to produce better sound than similar miniature types. Maybe octals have a less microphonic character, but of a more pleasing type.

Two of my acquaintances, who each have a Luminescence, both complain about the unit running too hot to touch. And the line stages drift, so they often must look for closely matched 6DN7s. Also, the 7591 in the power supply and all the 5691s in the phono stage add to the expense of keeping this unit running. (As a substitute for the 5691s, Luminescence owners could try 6SL7s, which work just as well and at a fraction of the price.)

I believe octal tubes are the best choice. It is difficult to beat a good 6SN7 for audible excellence at a low price. You can use other tubes, but for both present and future availability, the SN is tops. It’s like an old friend who doesn’t let you down. Even the Russian versions sound better to me than any 12AU7.

Circuitry

A cathode follower based on the 6BL7 or 6BX7 is ideal for output. I used a mu-follower circuit with no feedback, a 12SN7 as the gain stage, and a 6BL7 as the cathode follower/current source. (I have many old GEs I wanted to use, and converting to 6SN7 means simply changing the hookup to the 24V filament transformer). You might argue that the current source should be a pentode of highest possible gain, but pentodes can be noisy, and one of my design goals was minimalism.

barbour2.gif

The line stage (Fig. 1) is a straightforward mu follower, with medium-mu triode 12SN7GT below and low-mu triode 6BL7GT above. I paralleled both sides of each tube for maximum lifetime and for minimum output impedance. I chose resistors from my junk box, so there is nothing sacred about the values shown.

The 10F capacitor on the output was calculated to give a -3dB low-frequency cutoff of 26.5Hz when working into a 600 load. You can build all the circuitry point-to-point on ceramic terminal strips of the type from old “junker” Tektronix tube oscilloscopes (mine were surplus NOS) or from Triode Electronics. All the sockets are premium ceramic octals with silver contacts. You probably don’t need DC for the filaments, because the filament wiring is run inside grounded copper shield braid right up to the sockets.

This circuit should work fine with a 6J5, 12SX7, 6CG7, or 12AU7 as the lower tube, and a 6BX7, 12B4, or triode-connected 6BQ5 as the upper tube. Any of the miniature TV dual triodes such as 6DE7 or 6EW7 can serve as both top and bottom triodes since they contain a medium-mu and a low-mu section. (This wouldn’t be my choice, since the octals seem to sound better, but you may choose the tubes you like.)

I rescued a four-section Alps volume control from a scrapped Kenwood amp as a potentiometer. It is very similar to the Old Colony unit. The center shaft is volume, and the outer ring is a balance trimmer. I provided two stereo inputs, selected with a DPDT switch. All the RCA jacks are on top of the chassis, very close to the audio circuits to allow for minimum path lengths and easier cable hook-up (I hate reaching around behind components to mess with interconnects). Heat from the tubes is not the problem you might think; I laid a cable on top of a 6BL7 for eight hours (they run hot), and the insulation didn’t soften.

Power Supply

barbour3.gif

The power supply (Fig. 2) is regulated, has slow warmup inherent in its operation, and is simply designed. I chose the 6AS7G for looks; it’s much bigger and lights up more impressively than a 6080. The only problem was with the voltage standard. As I’ve stated before in GA, gas tubes are capable of impressive noisemaking.

Also for appearances, I installed two OD3s, but was never able to clean them up enough for optimal results. There was always at least 10mV of hash on the 6AS7 output, even when I selected the quietest among 40-plus OD3s and added all kinds of LC iltering. They also have a charming tendency to oscillate at audio frequencies as they age. So I left them hooked up to the raw power supply (they look impressive with their purple glow) and used a string of 51V zeners on the 6AS7 grids. So much for “purity.”

The 5Y3 rectifier is an old classic still in production, but a 5V4 will also work here without changing any wiring. The power transformer is a “radio replacement” type surplus unit I bought many years ago. It must have been more than 30 years old, as the original $5.88 retail sticker was still on it. Its regulation is not very good, without an electronic regulator and LC filter. I used a 10F Sprague 735P as a supply filter simply because I had some surplus units, which fit into the narrow chassis. There is only one electrolytic in the whole preamp; C2 had to be changed from a film type to minimize hum.

You do not need a vast filter capacitor with this circuit; the AC ripple on the 6AS7 plate is less than 10mV. The choke is an old 20H Triad unit, probably military surplus. If you can’t find one, try two conventional 10H units in series. R1 is simply a small 10 resistor serving as a safety fuse.

This supply is good, with no-load to 50mA regulation of about 5V DC. With both stages putting out 91V P-P sine wave into a 100k load, I put an oscilloscope on the regulated 310V supply line and observed less than 20mV P-P of audio signal.

Rave Review

As it turned out, with my AKG headphones plugged into the output jack, the -3dB cutoff was below 20Hz. Plate current of this stage is about 25mA with 310V plate power. Gain is about 17.4 and seems very stable. After a few months of regular use, it hasn’t decreased substantially. With the AKG headphones as a test load, frequency response was from below 20Hz up to 86.4kHz.

There is plenty of headroom: I measured 91.2V peak-to-peak before the onset of clipping (with a 100k high-Z load on the output). This is an excellent circuit to use in a single-ended 300B amp as the driver stage. IM distortion was 0.03% with 1V P-P into the 600 load. I measured an output impedance of about 300 with the 600 phones plugged in, which could be decreased by bypassing the cathode resistors (R6) with a large capacitor of 1,000F or more at 25V. This wasn’t necessary because by the time the phones cause audible load distortion, they are unbearably loud.

Purity Attained

The resulting sound was excellent. Used as a line stage, this design is virtually immune to the bizarre “tweak” cables that plague audio today. I tried a few borrowed types, from preamp output to power amp input, and there was no noticeable difference in sound quality. As a headphone driver, it produces excellent space and detail without the harsh transistor quality of other headphone drivers, and is genuinely euphonic and accurate at the same time.

TABLE 1

PARTS LIST

For each line stage channel:

R1,R5 100 1/4W carbon
R2,R6 100 1W metal film
R3 1M 1/4W metal film
R4 10k 25W wirewound
R7 10k 1/2W metal film
VR1,VR2 Part of Alps quad volume/balance control
C1 2uF 400V polypropylene
C2 10uF 400V polypropylene
V1 12SN7GT (GE) or 6SN7GT (see text)
V2 6BL7GT (GE)

For power supply:

R1 10 1/2W metal film
R2,R3 20k 10W wirewound
R4 10k 1/4W metal film
R5 18k 1/2W metal film
R6 56k 2W metal film
R7 27k 2W metal film
C1 10uF 400V polyprop
C2 220uF 400V electrolytic
C3 0.1uF 400V Mylar
C4 2uF 400V polyprop
T1 power transformer, 250-0-250V 50mA, Thordarson 24R09 or similar
T2 filament transformer, 24V CT 2A, Mouser 41LK020 or similar
L1 choke, 20H 50mA, two Triad C-3X in series (see text)

Other: (8) ceramic octal sockets, (5) 7-position ceramic terminal strips, (6) chassis-mount RCA sockets, aluminum chassis 15″ 4″ 3″, AC power cord with strain relief, knobs for pot, DPDT miniature switch, Teflon-insulated wire, 1/4″ copper tubular braid, stereo 1/4″ panel-mount phone socket, assorted hardware.

Addendum

10/7/98: Changes to power supply schematic (figure 2): R1 restored. Also an EL34 with a 100 ohm resistor connecting the plate to the screen can substitute for the 6AS7.

Options for the amplifier: A 6AG7 can work in place of the 6BL7 and 1/2 of a 6SL7 for the lower tube. Set R4 = 68K and R5 = 820 ohms as a shunt regulator for the 6AG7 screen.

7/15/00: Eduard Orvisky built this headphone amplifier using ideas from Kurt Strain’s and Eric Barbour’s designs. It uses two 6SN7s in the mu follower (compared to Eric Barbour’s circuit, R4 is reduced to 2K, 2W), and solid state and rectified heaters (he also tried a 5Y3 tube rectified supply, but preferred the solid state version for its faster sound). The chassis for this prototype was made from a sheet of plexiglass. It has no bottom and permits quick substitution of parts.

orvisky1.jpg

9/1/00: Eduard Orvisky moved his amp to a new Hammond steel chassis (available at Parts Express) and gave it a stunning paint job. He writes: “I sprayed it matte black in the beginning, but did not like it. Did not look bad, but it was too obvious. Then I tried silver metallic. That was worse, not as real polished metal. Then I painted acrylic sapphire blue on it. With brush, hence the strokes.”

orvisky3.jpg
orvisky4.jpg

Eric Barbour is an applications engineer with Svetlana Electron Devices and holds a BSEE degree from Northern Arizona University. He has been a staff editor with VACUUM TUBE VALLEY magazine since its founding in 1995, and has been a regular contributor to GLASS AUDIO magazine since 1991.

c. 1995, Audio Amateur Publications, Inc., P.O. Box 576, Peterborough, NH 03458, USA. All rights reserved.
From Glass Audio, Issue 4/1995, pp. 1, 6, 8, 10. (Republished with permission.)

Blue Hawaii Hybrid Electrostatic Amplifier for Stax Omega II Headphones.

by Kevin Gilmore
(Project Editor: Chris Young)

gilmore4_0.jpg

The Blue Hawaii amp is my latest design in my search for the perfect amp to pair with my Stax Omega II headphones. The genesis for this hybrid electrostatic headphone amplifier occurred when I was in Hawaii on vacation, at a fancy hotel on Maui. Sitting at the bar on the beach, drinking “Blue Hawaiis,” I drew the schematic for the amp on a placemat. The design is my conception of the mysterious and rare Stax T2 amp, which I have never been able to find at anything resembling a rational price.

I searched out any information I could find on the T2 in an attempt to create my own version. I was able to determine that it used EL34s as output tubes in a grounded grid configuration, which is the lowest distortion tube output circuit known. It also used 6DJ8s as input tubes with some solid state in the second and third stages. My design uses the first and second stages from my solid state electrostatic amplifier coupled with a third FET stage and then the final grounded grid stage.

My design ended up with a fairly large amplifier pulling significant amounts of power which results in a very smooth and extended frequency range from DC to over 200khz (-3dB at 400khz). Of all my electrostatic amps, this one has the largest output voltage swing. This is not an amplifier for the timid, nor is it a good idea to build this as your first project, though some, however, have actually done so.

The Circuit

gilmore4_1a.png
gilmore4_1b.png
Figure 1
(Click here to see single-image schematic of amplifier.)

Figure 1 is the amplifier schematic. The entire amplifier has a differential topology from input to output to get a balanced input and for lower noise, less ground loop problems. The first stage is a differential amplifier with feedback directly from the output stage. It works equally well with both balanced and unbalanced audio input sources. The step attenuators from Goldpoint make good volume controls for this stage. The JFET device (Q1) is a dual JFET all on one wafer. It is known for extremely low noise and excellent matching, and is used in a number of expensive designs, such as the Nelson Pass amplifiers. Q17 is a current source that sinks 3mA.

Because the amp is totally DC coupled from input to output, drift in the input stage is a bad idea. Since the first two stages run in current mode, the JFET input is more linear than a pair of bipolar transistors. Dual transistors all on one wafer suitable for audio use are hard to find these days. The FETs steer current away from the current sources Q2 and Q3. Together Q2 and Q3 each supply 2mA or a total of 4mA. The Q17 current source takes away 3mA leaving 0.5mA in each of Q4 and Q5, but some of the sink current is coming from the output feedback, so each FET is actually using somewhere between 0.5mA and 1mA.

The approximate voltage gain of this stage is 5; this stage really runs in current mode. The unit was designed to work equally well in both balanced and unbalanced mode. For single-ended signals, ground either the + or – input and apply signal to the other. The much higher impedance of the JFET works better when one side is grounded for unbalanced inputs.

The second stage starts with a constant current source (Q2 and Q3). The current source feeds a common base amplifier (Q4 and Q5). The common base amplifier feeds a modified Vbe multiplier. I believe a famous designer is now calling this circuit a current tunnel. It’s the most linear way of translating the voltage down to the bottom rail. The voltage gain of this section is about 4. The basic idea of the first two stages is to supply the third stage with a very fast low impedance drive signal that is referenced to the bottom rail.

The currents flowing into the common base amplifier (Q4 and Q5) are the difference between what Q2 and Q3 are supplying and what the FET is taking away. The rest of the current goes down the tunnel to the vbe multipliers (Q6 and Q7) that convert the current back to voltage. The current sources in the second stage supply 2 mA each. With no signal, the FETs take 1 mA, leaving 1 mA going through the common base amplifier into the bottom transistors, which are wired as Vbe multipliers (like a zener diode in series with a resistor, except a lot less noisy). This generates the 13 volts (referenced to – rail) necessary to properly bias the third stage.

The third stage is another differential amplifier (Q13 and Q14) being driven via another constant current source (Q10 and Q16). The voltage gain is about 200. Q11 is the power supply for this stage and makes a 100 volt power supply with -400V as the reference. The power supply voltage for this stage is kept down to 100 volts to reduce the Miller effect and keep the frequency response up. The higher output impedance of this stage is lowered by the use of 2SJ79 transistors, which are used as zero voltage gain emitter followers. The use of FETs in this stage coupled with the current source further reduces the distortion and provides for a solid low impedance drive signal for the output stage.

The 4th and final stage is a tube in grounded grid configuration (V1/Q8 and V2/Q15), similar to the common base amplifier in the 3rd section of my solid-state current-domain electrostatic amp. Q9 and Q12 are high compliance current sources and supply 25mA of bias current. Think of them as linear pull-up resistors for current (in fact, one builder has replaced the current sources with large resistors). The use of a current source here instead of load resistors acts to further linearize the output stage and reduce output distortion. V1 and V2 are the equivalent of common base amplifiers and do the entire rail-to-rail output voltage swings.

With feedback, the overall voltage gain of the amp is exactly 1000. The frequency response is kept high due to the low impedance cathode drive. The EL34s are biased at 10 watts and have an 800V voltage swing (by comparison, the output tubes of my original DC-coupled electrostatic amp are biased at 2 watts with a 600V swing), resulting in a frequency response well in excess of 100kHz into a 150pF load. (+0/-0.1dB).

gilmore4_2.png
Figure 2

A regulated power supply design is shown above. The ±15V supply is made with the standard 7815/7915 regulators. The high voltage supply is a pair of 400 VDC supplies, glued together at the output (P-channel MOSFETs are a lot more money than the equivalent N-channel MOSFET). In each section, beginning with a 460V raw supply, a PNP transistor (2SA1968) is used as a current source to feed the 400V zener reference. Then a N-channel FET is used as a high impedance, input voltage follower and outputs 400VDC. By the way, the same exact supply with a 350V zener reference string instead and a slightly smaller transformer (without filament windings) is what I use now for the solid state current domain headphone amp.

The bias supply is a voltage doubler with an adjustable reference. It has a range of about 350VDC to 650VDC. For low bias Stax headphones, put a 10M resistor to ground at the end of the 4.7M. to make the output voltage .66 times the voltage before the 4.7M, which puts it in the range for low bias.

Construction

gilmore4_3.jpg
(Click here to download pc board patterns in pdf format. 1)
(Click here to download pc board patterns in pdf format. 2)
(Click here to download pc board patterns in pdf format. 3)

Caution: This project involves working with high voltages, so be extremely careful! Keep one hand behind your back at all times. 800VDC across both arms might possibly stop your heart.

This amp was assembled on three printed circuit boards (two for each channel of the amp and one for the power supply) and housed in separate enclosures. A complete set of pc board patterns (pdf format) can be found above. They could be sent to just about any circuit board manufacturer to have boards made. The top of the board is almost all groundplane. All the parts, including the tubes, are mounted on these boards – the tubes are installed in pc-mounted ceramic tube sockets from Parts Express. The tubes must be exposed through the chassis. They dissipate 20W each (actually 10 to 12 watts of plate dissipation plus another 6.3V * 1.6A = 10 Watts of filament power).

gilmore4_5.jpg

It was so much easier to do a pc board for this amp, but if I were to make a prototype, I would again use the 0.1mm perf board; the layout look much like the circuit board. (Note: For a layout in a single chassis, see the interior view of Headamp.com’s Blue Hawaii amp below.) 99% of the wiring would be on the bottom, and it would be, therefore, rather flat. Mounting the tubes would be trouble though, and would cause mechanical problems. The tubes are fairly heavy and get stinking hot. Each chassis measures 12″ x 10″ x 3.5″. (Note: Headamp.com is selling the Blue Hawaii design in a single chassis measuring about 16.5″ W x 16.0″ D x 3.5″ H and may sell the Blue Hawaii pc boards. Please contact Headamp for more information.)

I have Mullard EL34 tubes, but keep them put away due to what I could sell them for if I wanted. I actually used the National Union tubes from Richardson Electronics which cost $11.50 US each.

gilmore4_6.jpg

All of the parts except the 2SA1968 have lots and lots of sources such as Digi-Key, MCM Electronics and Mouser Electornics. Only B&D; Enterprises has the 2SA1968 in the United States. In Japan and Canada, they can be ordered from Sanyo direct – the minimum order is 100 at a time, but then they are $1.25 each or so.

Q9 and Q12 are each made of six 2SA1968 transistors in parallel with one 2SA1968 as the driver. Matching the transistors is not required – unless one of the 2SA1968s is way off compared to the rest in which case it might get way too hot.

All resistors are 0.25W except where labeled. It is important to have all the pnp current source transistors correctly mounted to a large heatsink with silicon impregnated washers. If any one pnp transistor gets too hot it can short out the whole current source.

gilmore4_7.jpg

Standard tab heat sinks will do for the 2SK216 and 2SJ79 transistors, but the 2SA1968 and 2SC3675 transistors must be mounted a big heatsink (one for each channel), capable of dissipating 20 Watts of heat. I obviously fabricated them, but otherwise they can be obtained from Conrad Heatsinks cut to length. The IRFBC30 MOSFETs in the power supply must be heatsinked too: Mouser part number 532-529902b25.

gilmore4_8.jpg

The Stax SRC-5 headphone jack came from AudioCubes.com. Since the price has gone up to $19 each (I paid $10), it may be more cost effective to use the Allied jacks (see the current domain amp project article). Allied has a $25 minimum order, the cost of three pieces. Then they must be filed down on a lathe. Actually I am buying the male connectors from Allied, because no one else sells them. The male connectors are much easier to convert to standard Stax plugs. The power supply-to-amplifier connectors are the Amphenol military 12-pin connectors. The 4 connectors (two male and two female) were $130.

gilmore4_4.jpg

The custom Victoria Magnetics power transformer has these specs: 2 x 330VAC/150mA, 36VCT/100mA, 2 x 6.3V/5A (filament supplies). Everything with Victoria Magnetics is custom. I paid $110 for the transformer with shipping. They know about the Blue Hawaii design and will supply the correct transformer on request. For safety, I recommend a 2A/110Vac fuse located in the input line to the power supply.

Setup and Results

Test voltages (with the amp at idle) are shown in red on the schematic and are with respect to ground. To set up the amp, adjust the two pots in each channel of the amp. P1 controls the differential output voltage. Put a voltmeter between the 2 stators of one channel of the headphone and set this pot for zero. P2 controls the voltage with respect to ground. Put a voltmeter between any stator and ground and set for zero. Then repeat both adjustments a few times. The plates of both tubes should measure 0 volts with respect to ground when the pots correctly adjusted. Once the pots are adjusted, that’s it – there’s no change from headphone to headphone.

Setting the bias voltage depends on the headphones. For Stax headphones that can accept a high bias voltage, adjust the pot for 560V. I do not think that the Omega II headphones can be damaged by this amp unless the bias is set way too high. If the bias is set right, the outputs are close to 0V at idle, and all the LEDs are lit, then the amp pretty much has to be working correctly. Now if one or more of the outputs is stuck at +400V or -400V, then something is seriously wrong and needs to be fixed. An oscilloscope really helps.

gilmore4_9.jpg

Adjust the pot to 580 volts for Sennheiser HE-90 and HE-70 headphones or leave it at 560V. For Koss headphones, adjust the bias for 600V. To use the Sennheiser HE60 headphones with this amp, I made the adapter shown above. Those are RS-232 female connector pins that fit the HE60 pins perfectly. By hand I cut a circuit board with lands exactly 3.5mm apart put the pins on the HE60 connector, lay them down on the circuit board and solder. Then attach wires and a standard stax plug.

The amp can output 1500 V p-p measured stator to stator. At 800Vp-p, THD is less than 0.004% from 20Hz to 20kHz. The actual frequency response is 0 to 100kHz (-3dB at 150kHz) into an Omega II load. Compared to the sound of my previous tube amplifier, the bass is no longer tubby; it’s very sharp and tight. The high end is no longer rolled off, so female voices sound much more real. If the bias supply is reduced to 280V, the amplifier will drive all electrostatic headphones. I tried it last night on a pair of SRX’s. I never ever heard them sound so good.

Previously with a standard dummy head, I measured the SPL in Omega 2 headphones driven by this amplifier. With a drive signal of 800Vp-p per side, the resulting spl is 106dB. THAT’S LOUD! The amp can put out 1500 volts peak-to-peak, and thats louder! I just ordered a pair of Stax SR-001 MkIIs, which can reach up to 120dB. My ears distort before the amplifer/headphones do. It is quite loud at clipping, but the clipping is a hard clip with no oscillation or ringing. To use the amplifier with electret headphones, delete the bias voltage. And probably keep the output swing under 200V. Electrets phones when driven with this amplifier can probably get very very loud.

Several of my previous electrostatic designs are available in the Headwize Projects section. Comparing the Blue Hawaii to my all solid-state current domain amplifier, they really are more the same than they are different, but in general, the differences are the differences between tubes and solid state, such as a much smoother high frequency response, which in the case of the Blue Hawaii goes well beyond 500kHz. Additionally, the four times power consumption of the Blue Hawaii means a much stiffer and tighter bass response. Even though both are flat to zero and test similar, the BH bass is much more apparent and tighter.

c. 2004 Kevin Gilmore.

All-Triode Direct-Drive Tube Amps for Electrostatic and Electret Headphones.

by Kevin Gilmore

gilmore1_1.jpg

Electrostatic headphones, like electrostatic speakers, work on a push-pull arrangement. When one of the stators is going up in voltage, the other stator must decrease in voltage so that the diaphragm can move due to the static charge. These two direct-drive electrostatic tube amplifier designs will drive any existing electrostatic headphones without step-up transformers. The first is AC-coupled. The second is fully DC-coupled from input to output. These amps can also drive electret headphones – which are just permanently biased electrostatics – as long as the voltage swings are not too high.

I have built a total of 4 of the first amp, and one of the second. These are not beginner projects by any means. Since I have access to a chemistry electronics shop, I used lots of expensive and wonderful parts. The coupling caps I used are .22uF mica caps (probably 30+ years old) from military surplus (which today go for about $25.00 USD each). Due to the Apex op-amp DC regulators, the second design is a whole bunch more money. Both amps sound better than the Stax SRM-T1S unit, and virtually the same as the Stax Omega (if you could find one). And it definitely sounds better than the Sennheiser Orpheus.

BACKGROUND

Why I decided to design and build these headphone amplifiers. There are lots of reasons. First let me say that there are some things transistors do better than tubes, like low output impedance, which works better with low efficiency speakers like the Wilsons.

Then there are things that tubes do better than transistors. Like high voltage. Electrostatic headphones are very high-voltage devices. With a maximum voltage swing of 550 volts peak-to-peak, transistors or MOSFETs just are not good enough to do the job. The idea here was to design and build something good enough to make it into Stereophile’s Class A category.

Here is what is available in electrostatic headphone/amplifier combo’s:

Stax Omega and tube amp (approx. $10,000 a pair).
With the re-organization of Stax, this system is no longer available. The amplifier in this system is an all tube DC-coupled unit. When it was available, it was absolutely the best thing in the world!

Stax Nova Signature and either SRM-T1S or SRM-T1W amp ($1500/$1700).
The SRM series amplifier has tube output, but MOSFET input and MOSFET gain stages. The sound is nowhere as good as the Omega, but strictly due to the amplifier used.

Sennheiser HV60/HEV70 combo ($1500).
The solid state drive amplifier is horrible. Headphones themselves are excellent.

Sennheiser Orpheus ($12,000).
A wonderful system, although the inboard D/A converter is awful. Otherwise excellent.

Koss 950 (about $500).
The headphones are fairly decent, but the original transformer drive box is horrible. And the latest solid state drive box is also awful.

There are lots of good old headphones out there: old Stax units – both low bias and high bias (about 7 different models), the original Koss headphones, and the Jecklin Floats. The tube amps described in this project will drive any of them (but I recommend using only the AC-coupled amp with the low-bias Stax phones – see below).

THE DESIGNS

gilmore2.gif
Figure 1

Both amps are OTL and will output 500 volts peak-to-peak (measured from front to back) for an input of 1 volt peak-to-peak. The AC-coupled amplifier (figure 1) uses a series of single-ended pure class A gain stages (the second tube is a phase splitter). The first two tubes make up the first gain stage with a gain of about 20 and the second two tubes make up a gain of about 30.

gilmore3.gif
Figure 2

The DC-coupled amp (figure 2) uses two differential gain stages (the first stage is also a phase splitter). The output stage is single-ended pure class A. The first stage gain is about 7, the second stage gain is about 5 and the output stage has a gain of about 25. Decreasing the overall gain is easy – add an input pot. Increasing the gain is unnecessary as virtually any preamp can put out 1 volt, and that’s quite loud (about 100dB).

The reason for a fully direct-coupled amp is that interstage coupling caps, regardless of quality (even the silver mica units I use), are not perfect. The transformers used in some electrostatic speakers and early Stax drive boxes are also far from perfect. So in this amp, from input to output, only tubes touch the audio signal. It’s absolutely the best thing I have ever done.

I believe in feedback. Not a lot of feedback, but enough to fully stabilize the circuit. In both designs, the first two tube sections generate a stable pair of signals perfectly matched, but 180 degres out of phase. Then each of these signals goes to a voltage booster once again with local feedback to drive the diaphragm. I have listened to these units for literally thousands of hours, trying various kinds of tubes, ranging from NOS Mullard, RCA, GE and Raytheon to current-made junk. The feedback makes all these tubes sound virtually alike. Not at all like some tube preamps that sound dramatically different depending on the tube you use.

The DC-coupled amp is also rather expensive. The 6 Apex op-amps alone are $240 in parts. It goes without saying that the build quality will ultimately effect the sound. I built several of these units on a pure copper chassis, with ceramic silver tie points, just like Tektronix used to do in their tube oscilloscopes (the prototype shown in the pictures used an aluminum chassis and standard tie points). It’s insanely time consuming, but the result is worth it.

CONSTRUCTION

This project involves working with lethal high voltages, so be extremely careful! Keep one hand behind your back at all times. 600VDC across both arms might possibly stop your heart. The author accepts no responsibility for any harm caused by the construction of this project!

An electrostatic element has one diaphragm in between two fixed stator elements. Each amp has front and rear stator outputs. Call the front and back whatever you want, so long as the left and right ear pieces are wired the same, so that they are in phase.

The PA42 is a 350-volt opamp from Apex Microtechnology Corp. (I use a lot of Apex products in research applications). It’s a high voltage monolithic MOSFET unit. Since the opamps are strictly used for DC bias control, they do not influence the “TUBE” sound. They keep the output voltage at +300 volts. This is so that the output voltage can swing to zero and then to +600 volts. Note: the Burr-Brown 3583 high voltage opamp is not pin-for-pin substitute for the Apex; however it should work.

The Headphone Cable Connectors

All of the low-bias Stax units use 6-pin plugs (Amphenol microphone plug). All high-bias units are the same plug with 1 pin missing, and both diaphragms tied to the one wire. That way a high-bias headphone can plug into a low-bias driver, but the low-bias phones cannot plug into a high bias driver.

The Stax plug wiring scheme is as follows:

left front: pin 2
left rear: pin 5
bias: pin 1
right front: pin 3
right rear: pin 4
bias: pin 6

The high bias headphones do not have a pin 6. Instead, the bias for both elements is tied to pin 1.

The jack I used is an Amphenol 78-S6S. I am not sure where you get them. I have a bag of them that’s probably 30 years old. The were used as speaker connectors when speakers did not have permanent magnets.

It looks like this from the pin end:

  o
o o o
 o o

except that it’s not that regular.

  3
2 6 4
 1 5

Above it is shown wired.

The Sennheiser HV60 uses a 6-pin inline plug. Here’s the Sennheiser plug wiring scheme:

pin 1 (the corner notch): left front
pin 2: left diaphragm (bias)
pin 3: left back
pin 4: right back
pin 5: right diaphragm (bias)
pin 6: right front

The Chassis

I built the prototype shown in the pictures on an aluminum chassis (later units on a solid copper chassis that I bent up from .075″ thick copper myself). The first amp is built on a 13″ x 15″ x 2″ chassis. The second amp is built on a 15″ x 17″ x 2″ chassis. I spot-welded the corners together and cut out holes for tubes and other parts before painting the chassis to avoid scratching it.

gilmore1_3.jpg

The outside of the chassis is painted black. I put it in an oven for two days to get the paint real hard (set the oven to 200 degrees Fahrenheit for at least 4 hours bakeout). The chassis should be degreased with alcohol before painting it. I used a special primer (do not remember which) and then sprayed on Krylon flat black. If an aluminum chassis used, have one copper sheet on the inside to make sure that everything has a really good ground.

The 1 inch thick EBONY wood side panels make it look cool. I only used ceramic tube sockets (once again military surplus). For added effect, I built these units in a style similar to that of 40+ years ago. Everything is on ceramic tie points. No circuit boards of any kind. Building things to silly standards is part of the fun.

The Power Supply

Since the majority of people will be scrounging parts, and some still like to build power supplies with tube rectifiers, I did not make elaborate power supply schematics. For me, solid state power is just fine. The diodes and capacitors I used were the best that I could lay my hands on. For the DC-coupled amp, standard tube rectifiers cannot generate 860V without breaking down. Solid state here is definitely the way to go.

The 12AX7 is either a 6-volt filament or a 12-volt filament, depending on how you wire it. In the schematic, I show it wired for 6 volts. The 6S4 is strictly a 6-volt filament tube. I use AC for the filaments. You could certainly run DC for the filaments, if you wish. It might result in less hum, which is probably already unmeasurable if not inaudible.

The transformers: I have a lot of scrounged parts and actually used two transformers, one for the high voltage, and another dual unit for the filaments. I just looked through a bunch of catalogs for transformers available. It’s no longer a tube world. Allied (Hamilton Avnet) still sells the right transformers. Model 6K7VG is a 75-watt unit with 750 VCT and 6.3V. Model 6K94HF is a 25-watt unit with 12.6 VCT. The first unit is certainly overkill power-wise. The outputs run class A and are set at 4 watts, meaning 16 watts on the 600-volt line. The 300-volt line draws less than 2 watts.

gilmore4.gif
Figure 3

The AC-coupled amp supply: Any high voltage transistor can be used for the 600 volt series regulator. I used a Motorola MJ16018 (they are still available). MJ12005’s work better. All zener diodes are 1 watt units. All electrolytics are the best I can find (typically Black-Gate).

gilmore5.gif
Figure 4

The DC-coupled amp supply: In the DC-coupled amp, the stators are at 300 volts nominal for a signal swing of 0 to +600 volts. The bias voltage is 860V, but it is still effectively 560 volts between the diaphragm and the stators, because EVERYTHING is lifted 300 volts. (In the AC-coupled amp, the stators are at 0 volts nominal with a signal of -300 to +300 volts, and the bias is at 560 volts.) The diaphragm is a capacitor – all it cares about is the differential voltage, which is still the same. However, I do not think that it would be a good idea to use the DC-coupled amp with the old low-bias Stax phones. Since the stators sit at +300 volts, bad things might happen with the low-bias phones.

The 100V reference regulator can be built in two ways. The simplest is the same sort of arrangement as the 600-volt regulator: a 100-volt, 1 watt zener diode, a resistor and a pass transistor. A 100K, 1/2 watt resistor will be fine. The reference draws virtually no current.

A better arrangement, which as yet I have no schematic for, is another Apex amp rigged as a x10 multiplier, and a precision 10 volt reference like the Burr-Brown REF10. This is what I use. The output of REF10 goes to + input of opamp. The output of opamp goes to – input with 99K resistor. Then connect the – input of opamp goes to ground through a 1K resistor. You have to rig yet another power supply to come up with +15 volts to light up the REF10.

SETTING UP THE AMPLIFIERS

gilmore6.gif
Figure 5

Biasing the headphone diaphragms: The diaphragm bias network for the AC-coupled amp is shown in figure 5. The AC-coupled amps will work with any electrostatic headphone, if the bias voltage is correctly set. Older Stax units are low-voltage bias (~330V). Most of the newer Stax are high-bias (560V). I do not know what the Koss units use (check the schematic or owner’s manual). The Sennheiser HV70 uses 580V. Anything about right, but not over the recommended voltage is fine. The DC-coupled amp should drive only high-bias electrostatic headphones (do NOT use the DC-coupled amp with low-bias Stax phones).

Balancing the gain for the AC-coupled amp: The gain adjustment control is Rp in the schematic. Put a 1kHz squarewave on the input. Then adjust the pot for equal amplitudes at the outputs. It is best done with an oscilloscope. If you use a 1kHz sine wave, most AC digital voltmeters will also work.

THE FINAL RESULTS

gilmore1_2.jpg

How do these amps sound? Exactly what you would expect from an all-triode tube amp. The high end is liquid and sweet. No harshness of any kind, especially when compared with the Sennheiser unit. Clipping (you have to be listening quite loud) is virtually undetectable, whereas the Sennheiser unit overloads (lights the red light) and sounds crackly. I can literally listen for 4 hours at a time – something I could not do without extreme fatigue with any of the other amps.

The Stax SRM-T1S is actually a really good unit, but it uses 6FQ7’s which really cannot withstand the 600 volts. So as it gets louder and louder, the sound becomes more restricted, and because of the lower bias power, as the sound becomes louder the amount of high-frequency energy slowly disappears. I have heard of others complaining about this also. If you are in search of absolute excellence, the all-triode amps described here are the best electrostatic headphone amplifiers currently available, anywhere in the world.

Addendum

9/3/1998: The DC-coupled amp schematic (figure 2) was corrected as follows: R14, R22 are 5W resistors; R15, R23 are 1W resistors.

12/22/1999: The author writes: “I am using omega 2 headphones on my amplifier. The sound is so amazing especially with my new SACD-1.”

8/31/2001: Added high resolution pictures of the amplifier, including a new shot of the inside of the chassis.

2/4/2003: Corrected value of C9 in figure 1.

c. 1996, 1998, Kevin Gilmore.
The author’s website: The Homepage of Kevin Gilmore.

 

A Simple Tube/Opamp Hybrid Amplifier.

by Alex Cavalli, Mark Lovell and Bill Pasculle

cavalli2_ampshot.jpg

Introduction

After seeing many of the excellent and eye-catching tube-solid state amplifiers of others, we’d like to present a slightly different topology of hybrid amplifier design, using the same two basic components of a tube and an opamp. This amp is simple and has less than $50.00USD worth of parts and no lethally high voltages. It makes an ideal tube headphone amp for those who like the sound of tubes but worry about the safety of high voltage equipment.

Bill was looking for an amp that sounded good and also was a good companion for his Rio Carbon portable player. He also suspected that others might like something similar as well. With high bit-rate MP3 or WMA files, the Carbon can produce excellent sound, but as with many other portable players, benefits by the addition of a decent headphone amp. He also wanted to use the amp in his cubicle at work – requiring good sound without taking up too much space.

The design of the amp started when Bill suggested that he wanted to build one of the YAHA (Yet Another Hybrid Amplifier) amps , but using a different design. Alex, being mostly interested in tube amps, was not initially interested but as Bill and Mark began to suggest requirements for the amp and some options for the input tube, it was obvious that there was room to improve the typical hybrid topology. After more thought, Mark provided a list of some requirements for the design that he felt were necessary to meet in order to improve on the hybrid topology.

Alex went to work taking the design discussions and turning them into a draft design. Our first name for the amp (as a joke) among the team was Stoopid Opamp Headphone Amp (SOHA). The name, as so often happens with skunk-works project names, eventually stuck and finally we are just calling this amplifier the Stoopid or the SOHA. Another name for the amp, with the same acronym, might be the Simple Opamp Hybrid Amplifier.

The original prototypes assured us that it is possible to make a surprisingly good performing amp utilizing a tube at relatively low voltage and while still keeping the build cheap, easy, and reasonably electric shock-free. After altering some of the power supply and circuit values and testing the prototype, we ended up with something that was stable and fun to listen to for extended periods. It applies some compression but that’s part of its charm, and it would be a rather sterile sounding amp without the 12AU7/ECC82 altering the sound the way it does with its 40V plate voltage.

Amplifier Circuit

Most of the hybrid amps that have appeared in HeadWize threads and elsewhere (such as the Millet hybrid) have used the same B+ for both the tube and the opamp. Some of these amps are designed to be portable enough to run from a battery, but most are really constrained to a low voltage DC supply of some kind plugged into the line and so are not truly portable.

In addition, it is generally true that tubes that are not designed for low voltage use will not perform well at 12-24V (which is why the Millet uses special low voltage tubes), so we decided to try to provide the tube with higher B+ to get better performance, while still keeping the voltages fairly low. This meant that the amp could be small and portable although requiring AC power. Like the other hybrid amps, the SOHA is designed to give the sound of tubes while avoiding the high voltage risk that some builders don’t like. Still, providing a higher B+ permits us to get excellent sound from a more commonly available tube like the 12AU7/ECC82, which is in good supply from NOS and current production sources and which gives a wide variety of choices for tube rolling. Having this wide selection also makes it easier for the amp to be constructed in any part of the world.

The way in which most other hybrids use a common B+ for both tube and opamp has two detrimental effects on a hybrid amp:

  1. It puts the opamp in a single ended configuration where it needs an output cap to block half the B+
  2. It forces the B+ on the tube to be low so as not to exceed the maximum opamp rail voltages.

The first design decision was to decouple the B+ for the tube and opamp. Doing this makes it possible to use a standard bipolar supply for the opamp, eliminating the large output cap (as in the Chu Moy’s pocket amp for example) and requiring only a small coupling cap between stages (see power supply discussion below).

The tube is loaded with a constant current source (CCS) for two reasons:

  1. The resulting non-linear performance associated with low voltage operation is partially offset by the high dynamic impedance of the CCS
  2. A CCS has a much better PSRR than a simple resistor making it possible to have more ripple in the B+ and, therefore, simplifying the B+ PS.

The original amp is designed to run with FET input opamps. See note 1 (the BJT opamp section) for a version using BJT input opamps.

After extensive prototyping by Mark and Bill, and after posting the first design to the HeadWize forums and receiving feedback from several builders, we modified the original design. The most notable change was to the heater circuit. Originally the heater voltage was supplied directly from the AC secondary with voltage dropping resistors. This approach was implemented initially to maintain simplicity. However, because there is so much variation in transformer regulation, line voltage, and heater characteristics, the simple resistors were replaced with a regulator circuit. A side-benefit is the elimination of power wasted as heat since the dropping resistors reached 105°C – 115°C under normal operating conditions.

The Amplifier Circuit

The basic amplifier circuit uses a very small number of components, as shown:

cavalli2_fig1.png

R1 100k Log Pot C1 1000u 10V Electrolytic
R2 300R 1/8W C2 100n 100V
R3 560R 1/8W D1, D2 1N4148 or similar
R4 2k Trimpot U1 OPA2134 or similar
R5 1M 1/8W V1 12AU7/ECC82 or equivalent
R6 150R 1/8W

Figure 1 – Basic SOHA Amplifier

The topology of the amp is a simple grounded cathode gain stage coupled through a capacitor to an opamp wired in unity gain mode. The standard SOHA uses LND150 depletion mode MOSFETs for the CCS for reasons discussed below. The amplifier circuit is, thus, very simple. Trim pots are provided as part of the cathode bias resistors to adjust for variations in tubes to set the plate voltages to ~40V. Each CCS is set to regulate at approximately 1mA.

An advantage of this design over many of the other hybrid designs is that there is no large coupling electrolytic at the output. The required inter-stage coupling capacitor is small making it possible to use good quality film/audio capacitors here at nominal additional cost.

With a 12AU7/ECC82 the input stage has a gain of about 12. This is sufficient for almost any source driving almost any headphones which is why the opamp is simply operating as a unity gain current buffer. However, with such high gain it is possible to exceed the input voltage tolerances for the opamp with just 1Vp at the input. The data sheet for the OPA2134 (and many similar opamps) indicates that the maximum input voltage is (V-) – 0.7V to (V+) + 0.7V. This means that the input swing must not exceed the supply voltage by more than one diode drop. The diodes ensure that this does not happen.

If the diodes conduct, the excess current passes into or out of the bipolar power supply. What happens thereafter depends on the ability of the tube to source/sink current and the ability of the PS to sink/source it. In this case, the tube will source/sink in the range of hundreds of micro amps which will find their way to the output caps of the bipolar supply which are in turn supplying current to the opamp V+ and V-. The output of the regulators will fluctuate some, but at this point the amp would not be operating properly anyway.

The standard CCS for the basic SOHA uses a single LND150 MOSFET in this configuration:

cavalli2_fig2.png

R7 1k 1/8W R8 360R 1/8W

Figure 2 – Standard LND150 MOSFET CCS

For a discussion of why this was selected as the standard CCS, see note 2 (the CCS comparison section) at the end of the article. One limitation on the standard SOHA CCS is the uneven availability of the LND150 MOSFETs globally. To ensure that this amp can be built almost anywhere, we have created two variations that use other devices for the CCSs. The first uses J113 JFETs and the second uses 1N5297 current regulator (CR) diodes. Here are the schematics for both variations:

cavalli2_fig3.png

R8

1k5 1/8W

Figure 3 – Alternate CCSs using JFETs and CRDs

Care should be exercised when building the SOHA with the J113 JFETs. Their maximum Vdss is 35V. Under normal operating conditions they will see only about 15-20V, but if the plate voltage on the tube is too low it is possible to exceed this maximum and destroy them. To protect the JFETS, the minimum plate voltage should never be set below 20V (see below the warning about adjusting the trimpots). The 1N5297 CRD has a 100V maximum and should withstand all of the normal voltages in this amp. The J505, noted in parenthesis, will also work but has only a 50V maximum. This makes the J505 a little more robust in this circuit than the J113, but less desirable than the 1N5297.

Power Supply Circuit

The key to this amp is the power supply. Initially the amp used an easy-to-acquire 30VCT/200mA transformer. As noted above, during the development and testing process, including builds by several HeadWizers, we changed the heater supply from AC to regulated DC. To accommodate the 150mA DC drawn by the heater it is necessary to increase the current spec on the secondary to 400mA. This will also give some headroom for the amp itself. Eventually, we chose the Amveco TE70053 toroid to replace the original split bobbin transformer. Another benefit to using the toroid is less EM radiation in the box and, since the SOHA also designed to be small, this reduces or eliminates problems with PS buzz. Other transformer possibilities are in the Power Supply section below. You can use a higher current rating transformer without difficulty, but if you increase the voltage be careful about not exceeding the maximum input voltage for the regulators. The bipolar opamp supply is a conventional regulated supply using 78L12/79L12 inexpensive regulators. They have a maximum input voltage of 40V.

The trick to the power supply is the use of a 1.5x full-wave voltage multiplier to generate the B+ for the tube. To make the voltage multiplier, the entire secondary of the transformer is rectified through a pair of coupling capacitors and bootstrapped on top of the V+ of the bipolar supply. With a typical transformer with 25% regulation and with no load on the B+ for the tube, this generates over 80V (this is marginally dangerous and will give you a pretty good sting so be careful).

The power supply, including the heater circuit is shown below:

cavalli2_fig4.png

R9 2k2 1/8W BR1, BR2 100V 1A Bridge Rectifier
R10 1k3 1/8W VR1 12V Fixed Regulator 78L12
R11 11k 1/8W VR2 -12V Fixed Regulator 79L12
C3-C6 100u 100V VR3 Adj. Negative Regulator LM337
C7, C8, C11 47u 16V D3, D4 1N4002
C9, C10, C12 470u 35V T1 30VCT 400mA Transformer

Figure 4 – the SOHA Power Supply

When the B+ is loaded with the tube, with each triode drawing ~1mA, the voltage is pulled down to between +55-65V. This means that there is plenty of headroom in the B+ to run the tube at+40V while still leaving space for driving 7-10V into the opamp. And this seems to give very good performance. As noted above, using a CCS for the plate load relieves ripple requirements on the B+ so a much simplified and less expensive filter section becomes possible. For the components as drawn the B+ ripple is about 1mV. The capacitor values are kept low and, hence, the capacitors are small and inexpensive. Again, for CCS PSRR comparisons see below.

The heater supply uses a full-wave rectifier into a negative 12.6V regulated supply. Pay careful attention to the orientation of the rectifying diodes. The heater supply is attached to the negative half of the bipolar supply. This was done because the heater current will pull down the input to the filter section of whichever half of the bipolar supply to which it is attached. Since we are using the positive supply to bootstrap the B+ for the tube we don’t want the heater supply to pull this voltage down. Therefore, it is derived from the negative supply because if the negative input to filter drops by a volt or two the regulator will not be affected. Pay careful attention to the orientation of the rectifier diodes and capacitors in the heater circuit since it is a negative supply. A power indicator LED can be attached to the heater supply taking care to note the polarity. The negative regulator should be heatsunk to dissipate about 2W.

Construction

The full schematic for both channels and the PS is shown below with the complete parts list less some miscellaneous components such as enclosure, power switch, etc.

cavalli2_fig5.png
Click here to see full-size schematic.

R1 100k Log Pot C3-C6 100u 100V
R2, R12 300R 1/8W C7, C8, C11 47u 16V
R3, R13 560R 1/8W C9, C10, C12 470u 35V
R4, R14 2k Trimpot D1, D2, D5, D6 1N4148 or similar
R5, R15 1M 1/8W D3, D4 1N4002
R6, R16 150R 1/8W U1, U2 OPA2134 or similar, dual or single
R9 2k2 1/8W V1 12AU7/ECC82 or equivalent
R10 1k3 1/8W BR1, BR2 100V 1A Bridge Rectifier
R11 11k 1/8W VR1 12V Fixed Regulator 78L12
R7, R17 1k 1/8W VR2 -12V Fixed Regulator 79L12
R8, R18 360R 1/8W VR3 Adj. Negative Regulator LM337
C1, C13 1000u 10V T1 30VCT 400mA Transformer
C2, C14 100n 100V

Figure 5 – Full SOHA amplifier, both channels and PS

For the J113 version eliminate R7, R17 and change R8, R18 to 1k5 1/8W. For the 1N5297 version simply replace the entire CCS with the single diode.

The amp has been built several ways by different HeadWizers [click here to see forum member Neurotica’s (Jim Eshleman) SOHA build narrative]. Mark and Bill initially built the prototypes using point to point wiring on perfboards and several others did so as well. Bill eventually also made a homemade PCB while Alex designed a PCB using the commercial package ExpressPCB (see below). Most builds to date have been like the prototypes with the PSU and amplifier circuits on the same board. Pictured below is a pictorial drawing showing how the SOHA can be wired point-to-point on a 4 x 6-inch perfboard.

cavalli2_fig6.png
Figure 6a – Bill’s SOHA constructed by point to point wiring on a perf board
Click here to see full-size layout.

Part of our purpose with the design and component specs is to keep everything as small and cheap as possible. The parts list shows the parts from the usual American suppliers. Mark was able to source similar parts from Farnell and RS in the UK.

Part # Mouser Catalog Number Description Qty. Price Total
R9 270-2.2K-RC Xicon 2.2k 1/8W

10

0.11

1.10

R3, R13 270-560 Xicon 560R 1/8W

10

0.11

1.10

R11 270-11K Xicon 11k 1/8W

10

0.11

1.10

R10 270-1.3K-RC Xicon 1.3k 1/8W

10

0.11

1.10

R5, R15 270-1.0M-RC (regular CCS) Xicon 1.0Meg 1/8W

10

0.11

1.10

270-100K-RC (mu follower) Xicon 100k 1/8W

10

0.11

1.10

R2, R12 270-300 Xicon 300R/1/8W

10

0.11

1.10

R6, R16 270-150-RC Xicon 150R 1/8W

10

0.11

1.10

R4, R14 652-3306K-1-202 Bourns 6mm 2K pot

2

0.56

1.12

C7, C8, C11 140-HTRL16V47 Xixcon 47uF/16V

3

0.07

0.21

C3, C4, C5, C6 140-HTRL100V100 Xicon 100uF/100V

4

0.42

1.68

C9, C10, C12 140-HTRL35V470 Xicon 470uF/35V

3

0.3

0.90

C1, C13 140-HTRL16V1000-TB Xicon 1000uF/16V

2

0.25

0.50

C2, C14 1429-1104 Xicon 0.1uF (100nF)

2

0.44

0.88

CCS Options
512-J113 J113 JFET

4

0.24

0.96

R8 270-1.5K Xicon 1.5k 1/8W

10

0.11

1.10

OR
689-LND150N3-G LND150 MOSFET

2

0.55

1.10

R8, R18 270-360 Xixon 360R 1/8W

10

0.11

1.10

R7, R17 270-1K-RC Xixon 1k 1/8W

10

0.11

1.10

  OR
610-1N5297 1N5297 CC Diode

2

4.29

8.58

BR1, BR2 821-DB102G 1A 100V Bridge

2

0.33

0.66

VR1 512-LM78L12ACZ LM78L12 Regulator

1

0.27

0.27

VR2 512-MC79L12ACP LM79L12 Regulator

1

0.34

0.34

VR3 512-LM337T LM337 Regulator

1

0.5

0.50

567-273-AB Wakefield Heatsink

1

0.38

0.38

Power LED 351-3310 Xicon Blue Led 3mm

1

1.5

1.50

271-560-RC 560R 1/4W LED Resistor

10

0.09

0.90

D1, D2, D5, D6 78-1N4148 1N4148 Diodes

10

0.03

0.30

D3, D4 512-1N4002 1N4002 Diodes

5

0.10

0.50

R1 313-1240-100K Taiwan Alpha 12mm pot, 100k

1

2.84

2.84

575-393308 IC Socket

1

0.36

0.36

J3 161-3502 3.5mm Headphone Jack

1

0.92

0.92

J1 161-1052 RCA Jack Black

1

0.82

0.82

J2 161-1053 RCA Jack Red

1

0.82

0.82

Total

38.04

OR
DigiKey
U1, U2 OPA2134PA-ND OPA2134PA

1

2.63

2.63

T1 TE70053-ND Amveco 30V CT 500mA

1

12

16.22

Total

18.87

Optional Sources
Newark
U1, U2 75C4624 OPA2134-PA

1

2.37

2.37

18C6948 J113 JFET

2

0.2

0.40

VR1 34C1091 LM78L12ACZ Positive regulator

1

0.28

0.28

VR3 34C1076 LM337T Regulator

1

0.67

0.67

R4, R14 46F1092 Bourns 6mm 2K pot

2

0.27

0.54

Antique Electronic Supply
V1 T-12AU7-JJ JJ 12AU7 Tube

1

8.95

8.95

V1 P-ST9-511 Tube Socket

1

1.95

1.95

Miscellaneous
Knob
Power Switch
Wire
Fuse holder and Fuse (0.25A)

The Amveco toroidal transformer (30VCT/500mA) is available from Digikey. Remember with 15-0-15 VAC (nominal) secondaries and the poor regulation of these inexpensive transformers, you will see over 21V with no load at the inputs to the bipolar power supplies and ~85V with no load for the B+. Because of the poor regulation make sure to use capacitors with voltage ratings that meet these off-load conditions. Note that the PS parts table shows 100V capacitors for the B+ section. If you use a higher voltage transformer make sure that the input to the regulators does not exceed their maximums (typically about 37V).

Some other possible split bobbin transformers are: Dagnall D3019 (0-240 pri), D3023 (0-115,0-115 pri). Both are 12VA. Other possible toroids are: MULTICOMP MCTA015/15 (0-115,0-115 pri), MULTICOMP MCFE015/15 (0-230V pri), or MULTICOMP MDCG015/15 (0-230V pri).

This design is optimized for 12AU7/ECC82 and its exact equivalents (5963, 6189, and 6680) rather than a close equivalent (or other types of tubes such as 6922).

All three flavors of CCS provide a degree of PSRR and some immunity from power fluctuations. They differ in availability worldwide and in maximum voltage ratings. The best overall CCS uses the LND150 MOSFET which is not available everywhere. The J113 FET is widely available but its maximum DC voltage is only 35 volts. Normally the FET wouldn’t see more than 15-20V unless the plate voltage gets too low. The 1N5297 CRD has a maximum voltage of 100 VDC but is not as widely available and is also expensive. Nevertheless, working amps have been built using all three types of CCS. Trim pots located at the cathodes are used to adjust the plate voltage. In order to prevent burning out the CCS FETs the cathode trim pots should always be turned to their highest resistance when swapping in a new tube.

OPA2134 and its relatives are fairly common opamps for audio. This was a good place to start. The authors would like to know how other FET input opamps perform and welcome feedback from builders. The OPA551, for example, is a FET input opamp that drops right into the Stoopid. However, it only comes in single packages so you will have to account for this with the build.

FET input opamps are preferred because there is a risk that BJT input opamps may tend to excessively load the tube and defeat the effect of the CCS. To use BJT input opamps see the section below for modifications to do this. The authors welcome feedback on the performance of the SOHA with BJT input opamps.

A BUF634 could easily be put into the unity gain feedback loop of the opamp to give super high output. One change that might be necessary if really trying to pull 200mA is to increase the size of the input capacitors in the bipolar PS to more like 2200uF. Even larger values may be required to get full bass.

Mark added 150 Ohm resistors (R6) at the outputs as this enables the amp to drive low and high impedance headphones without experiencing a large change in volume. These can be left out of the circuit at the builder’s discretion, however, their use is recommended. Likewise the pairs of 1N4148 diodes connected to the non-inverting inputs of the opamps are optional, but serve to protect the opamp inputs from overload and their use is recommended.

Here are a few details to pay attention to during construction and double check before applying power to your SOHA:

  • Wiring the Triad transformer is not intuitive; the pins are not numbered consecutively. Study the datasheet carefully.
  • The 78L12 and 79L12 do not share the same pinout.
  • The capacitors in the heater supply (as well as those in the negative half of the bipolar supply) have their positive leads grounded.
  • Use of a star-ground is highly recommended.
  • Use of shielded cable from the input jacks to the pot, from the pot to the tube grids, and from the opamp to the output jack is highly recommended. Attaching the safety ground to the star ground is optional. Most builds have worked fine with the star ground floating but an occasional unit has been quieter with the safety ground connected to the star ground.
  • Grounding the pot body is usually required to eliminate static/hum.

Wire dress is important in this amp to avoid hum. Keep all signal wires away from the transformer; keep the filament wires as far away from the audio circuit as possible.

PC Boards

We’ve created Express PCB boards for the SOHA. These are related to the full schematics with part numbers shown.

For the J113 version eliminate R7, R17 and change R8, R18 to 1k5 1/8W. For the 1N5297 version simply replace the entire CCS with the single diode.

ExpressPCB and PDF files for both the amp and power supply are included below. The tube socket on the amp board is in the center of the board. Note that the tube socket mounts on the foil side of the board. With this configuration you can easily mark a hole in the center of the standoffs and punch it out to pass the tube through so that the tube can stick up through the chassis while the components are sticking downward.

The copper layer in these PDF and ExpressPCB files can be used for home etched boards.

SOHA Amplifier Board (PDF)
SOHA Power Supply Board (PDF)
SOHA Amp and PS boards (ExpressPCB)

Techniques for making home PCBs were suggested by HeadWizer Bill Blair. Here are some links that Bill used to make his own SOHA boards using the single layer PDFs:

EasyPCB Fabrication
HomeBrew Printed Circuit Boards

The boards can be jumpered to use all three versions of the CCSs and to operate as standard plate drive or as source follower drive. This is the stuffing guide for these possible configurations.

cavalli2_fig12
Click here to see full-size stuffing guide.

Figure 6b – Bill’s stuffing guide for the SOHA amplifier PCB

Setup

Wire everything up but don’t put the tube/opamp in yet. Measure the voltages at the B+, V+, V-, and heater. They should be >80V, +12V, -12V, and -12.6V respectively. If they are not then there is a problem that must be fixed before inserting either the tube or the opamp.

If voltages are good and nothing has fried, power down and then insert the tube and opamp. Before powering up again, dial your trim pots so that they are in the maximum resistance position. This will put the maximum bias on the tube. Measure the voltage at the plates (pins 1 & 6) and adjust the associated trim pot until the voltage comes down to 40V for each plate. After these adjustments, measure the B+ again. It should be between 55-65V. Occasionally you may find a NOS tube does not work well in this circuit. You may need to replace the tube to get good results. If so, the tube is probably outside of its published operating characteristics. Each triode of 12AU7/ECC82 draws only 1mA from the B+ supply and at these low currents there can be a wide variation in operating characteristics, particularly among tubes that may be marginally within spec.

Results

OK, how does it sound? Well, in short, stoopidly good. When first powered up, the prototype plate voltage was only 17V and the amp sounded decidedly solid state. Very “steely” and just tonally “off”. As the plate voltage was raised the sound became more lush and tube-like. At 40V the amp began to perform extremely well. The SOHA easily rivals the Cavalli-Jones/Morgan Jones which costs over three times more to build! It’s got decent amounts of bass, classic sweet tube midrange and plenty of top end extension. Also the soundstage is extremely wide and respectably deep. This thing is just plain stoopid fun to listen to!

The amp drives headphones of any impedance between 16 Ohms and 300 Ohms without problems, which covers most that are currently available.

The compression applied by running the 12AU7 with 40V at the plate allows an unexpectedly refined sound with no sharp edges, yet without being mellow. It will reproduce transients when required and has a respectable dynamic range. The overall result is something than can be listened to for extended periods with no “listening fatigue” and providing a pleasingly wide and reasonably deep sound stage.

As is the case with tube amps, a warm up time is required and in this respect the authors agree 20 minutes is required for it to sound its absolute best, but of course it’s up and running after 30 seconds.

Mark has compared this amp to three other headphone amplifier designs available at HeadWize having built them: namely the CJ, the CL MkII, and the BCJ MkI, (the BCJ MkII was unavailable). All of the alternative designs used for comparison tests are more expensive to build, all require potentially lethal voltages, and all are optimized to ensure the tubes are working at their optimum.

Clearly, the SOHA would be the worst of the bunch? Not so. It proved itself to equal the CJ and gets closer than expected to the CL MkII. That’s pretty impressive stuff for a tube amp deliberately designed to be cheap and not use lethal voltages.

Tube rolling in this amp is also a lot of fun. Mark and Bill, who listen mostly through Sennheiser HD-600’s, found that grey-plate 5963’s from GE and RCA and Brimar sounded better than other tubes. Some other Headwizers with low-Z cans seemed to prefer black-plate versions of these tubes. Among the new production tubes, the Electro-Harmonix 12AU7 seemed to approach (but not exceed) the performance of the NOS tubes while the JJ 12AU7 was a somewhat distant second. Differences between tubes seemed to be in the clarity of the top end and the amount and quality of the bass.

Here are some photos of Bill’s SOHA in its final home.

cavalli2_fig7.jpg
Figure 7 – Bill’s SOHA Top Side and Figure 34 – Bill’s SOHA The Guts

Note 1: BJT-Input Opamps

As noted above the SOHA was designed to use FET input opamps. However, to permit opamp rolling, we’ve created two minor variations that permit the use of BJT input opamps.

Bipolar-input opamps like the TSH22IN, NE5532, or NE5534 can substitute for the 2134. But bipolar opamps will have lower input impedance than the FET input opamps. This will increase the loading on the tube and increase the distortion.

One way around the increased loading is to configure the CCS as an active load source follower. This variation requires only one change in wiring at the CCS and will only work for the LND150 and the J113 versions. An active load source follower is a variant of a well-known tube topology where the CCS that is acting as a plate load is also utilized as the output device in a follower configuration. With this topology the output impedance of the gain stage drops considerably and its ability to supply current increases in proportion. With both FET topologies we can wire the FETs as source followers to make a hybrid follower configuration for the first stage.

If you’re using the LND150 CCS you can convert the CCS into a source follower by simply changing the point where the coupling capacitor is connected. The FET then becomes a source follower with low output impedance and the ability to drive higher currents into the load.

cavalli2_fig8.png
Figure 8 – Changing the LND150 CCS for BJT-input Opamps

If you’re using the J113 CCS you can covert it to a source follower using the same technique:

cavalli2_fig9.png
Figure 9 – Changing the J113 CCS for BJT-input Opamps

Although the 1N5297 CRD is actually a JFET wired as a CCS we cannot access the source of the device so the CRD cannot be used when driving BJT opamps.

For example, a full amplifier schematic for the standard LND150 CCS with BJT opamp is:

cavalli2_fig10.png
Figure 10 – Driving BJT input opamps

If your amplifier exhibits high DC offset with BJT opamps, you can decrease the value of R5 from 1M to 100k or even 50k without overloading the gain stage. Note that decreasing the value of R5 while leaving C2 at 100nF also reduces the low frequency response of the amplifier. To correct for this, increase the value of C2 so that the product of R5 x C2 is the same as 1M x 100nF. For example, if you decrease R5 to 100k, then to maintain the same low frequency response, increase C2 to 1uF.

For BJT input opamps, the full schematic is this:

cavalli2_fig11.png
Click here to see full-size schematic.

R1 100k Log Pot C3-C6 100u 100V
R2, R12 300R 1/8W C7, C8, C11 47u 16V
R3, R13 560R 1/8W C9, C10, C12 470u 35V
R4, R14 2k Trimpot D1, D2, D5, D6 1N4148 or similar
R5, R15 100k 1/8W D3, D4 1N4002
R6, R16 150R 1/8W U1, U2 BJT opamp, dual or single
R9 2k2 1/8W V1 12AU7/ECC82 or equivalent
R10 1k3 1/8W BR1, BR2 100V 1A Bridge Rectifier
R11 11k 1/8W VR1 12V Fixed Regulator 78L12
R7, R17 1k 1/8W VR2 -12V Fixed Regulator 79L12
R8, R18 360R 1/8W VR3 Adj. Negative Regulator LM337
C1, C13 1000u 10V T1 30VCT 400mA Transformer
C2, C14 1u 100V

Figure 11 – Full SOHA with BJT opamp, both channels and PS

Note the part changes shown in red. These are the only changes necessary to use BJT opamps in the SOHA. The layout of amp does not change.

Note 2: CCS Comparisons

The choices for standard CCS and acceptable variations are derived from three criteria:

  1. maximum breakdown voltage of the CCS
  2. current regulating ability
  3. PSRR

These comparisons were done using PSpice simulations. These simulations are not likely to give absolute accuracy, but they are good at providing a relative comparison among the various CCSs.

Simulations were done for the following CCS types:

  • Single LND150
  • Single J113
  • Single PN2907A
  • Single 1N5297
  • Single PN2907A with CRD bias string
  • Cascoded J113
  • Cascoded PN2907A
  • Cascoded PN2907A with CRD bias string

This table shows the current variation, PSRR, and breakdown voltage (BV) for these various configurations:

Current Variation

Ripple at Plates

BV

300mVp 1kHz Input

1mVp 120Hz Ripple from PS

Delta I (uV)

Delta V (mV)

PSRR (db)

Topology

Cascoded JFETs (J113)

0.01

0.002

-54

35

Single MOSFET (LND150)

0.9

0.0025

-52

500

Cascoded BJTs with CRD

0.26

0.0095

-40

60

Single BJT w/ CRD

1.2

0.0135

-37

60

CRD (1N5297)

6.6

0.017

-35

100

Single JFET (J113)

13.3

0.034

-29

35

Cascoded BJTS (PN2907A)

0.34

0.058

-25

60

Single BJT (PN2907A)

1.2

0.06

-24

60

The cascoded JFETs have the best current regulation, followed by the MOSFET. The cascoded BJTs with CRD and without CRD have the next best current regulation. This might make these the next best choices. But, we must look at the PSRR and BV tables too.

The cascoded JFETs also have the best PSRR but they have a low BV. The LND150 has nearly the same PSRR (indistinguishable as a simulation result) but a very high BV. The LND150’s current variation comes in fourth behind the cascoded BJTs. However, the PSRR for the cascoded BJTs is 12db and 27db less than the LND150. Furthermore the BV for the BJTs is on the margin of where the voltages in the amp may be, and the BJT CCSs require many more parts than the either the JFETs or the MOSFET.

Taking all of these results together, the LND150 rises to the top for the standard SOHA because of its good current regulation, excellent PSRR, very high BV, and low parts count. The cascoded JFETs come in second because of their excellent regulation, PSRR, and low parts count. The CRD comes in third because of its good regulation, high BV and extreme simplicity (only one part).

Appendix: Simulating the Amplifier in OrCAD PSpice

Alex Cavalli has provided the project files for simulating this amplifier using OrCAD Lite circuit simulation software. The simulations will run in OrCAD Lite 9.1 or 9.2 only (later versions of OrCAD Lite and OrCAD Demo are more restrictive and will not run the simulations). The installation files for OrCAD Lite 9.1 or 9.2 can be downloaded from various educational sites on the internet. Search for them using the keywords OrCAD or Pspice and 9.1 or 9.2. OrCAD 9.1 is the smaller download (27MB). If you have trouble finding these files, email a HeadWize administrator for help.

There are 4 programs in OrCAD Lite suite: Capture, Capture CIS, PSpice and Layout. The minimum installation to run the amplifier simulations is Capture (the schematic drawing program) and PSpice (the circuit simulation program).

Download Simulation Files for SOHA Headphone Amplifier

After downloading cavalli2_soha_sim.zip, create a project directory and unzip the contents of the cavalli2_soha_sim.zip archive into that directory. Move the .lib and .olb files into the \OrcadLite\Capture\Library\PSpice directory. These are the component libraries containing the SPICE models for the vacuum tubes, MOSFETs and opamps used in the SOHA. (Note: heater connections are not required for any of the triode models.) In OrCAD’s Capture program, open the stoopid.opj project file.

The two basic types of simulation included are frequency response (AC sweep) and time domain. The time domain analysis shows the shape of the output waveform and can be used to determine the amplifier’s harmonic distortion. They both run from the same schematic, but the input sources are different. For the frequency response simulation, the audio input is a VAC (AC voltage source). The time domain simulation requires a VSIN (sine wave generator) input. Before running a simulation, make sure that the correct AC source is connected to the amp’s input on the schematic.

cavalli2_sim.png

The following instructions for using the simulation files are not a complete tutorial for OrCAD. The OrCAD HELP files and online manuals include tutorials for those who want to learn more about OrCAD.

Frequency Response (AC Sweep) Analysis

  1. Run OrCAD Capture and open the project file stoopid.opj, if not already open.
  2. In the Project Manager window, expand the “PSPICE Resources|Simulation Profiles” folder. Right click on “Schematic1-ac” and select “Make Active.”
  3. In the Project Manager window, expand the “Design Resources|.\cavalli.dsn|SCHEMATIC1” folder and double click on “PAGE1”.
  4. On the schematic, make sure that the input of the amp is connected to the V4 AC voltage source. If it is connected to V3, drag the connection to V4.
  5. To add the triode library to the Capture: click the Place Part toolbar button (orcad1.gif). The Place Part dialog appears. Click the Add Library button. Navigate to the triode.olb file and click Open. Make sure that the analog.olb and source.olb libraries are also listed in the dialog. Click the Cancel button to close the Place Part dialog.
  6. From the menu, select PSpice|Edit Simulation Profile. The Simulation Settings dialog appears. The settings should be as follows:
      • Analysis Type: AC Sweep/Noise
      AC Sweep Type: Logarithmic (Decade), Start Freq = 10, End Freq = 300K, Points/Decade = 100
  7. To add the triode library to PSpice: Click the “Libraries” tab. Click the Browse button and navigate to the the triode.lib file. Click the Add To Design button. If the nom.lib file is not already listed in the dialog list, add it now. Then close the Simulation Settings dialog.
  8. To display the input and output frequency responses on a single graph, voltage probes must be placed on the input and output points of the schematic. Click the Voltage/Level Marker (orcad2.gif) on the toolbar and place a marker at grid of U6. Place another marker above R9 at the amp’s output.
  9. To run the frequency response simulation, click the Run PSpice button on the toolbar (orcad3.gif). When the simulation finishes, the PSpice graphing window appears. The input and output curves should be in different colors with a key at the bottom of the graph.
  10. The PSpice simulation has computed the bias voltages and currents in the circuit. To see the bias voltages displayed on the schematic, press the Enable Bias Voltage Display toolbar button (orcad5.gif). To see the bias currents displayed on the schematic, press the Enable Bias Current Display toolbar button (orcad6.gif).

Time Domain (Transient) Analysis

  1. On the Capture schematic, make sure that the input of the amp is connected to the V4 sinewave source (VAMPL=0.4, Freq. = 1K, VOFF = 0). If it is connected to V3, drag the connection to V4.
  2. In the Project Manager window, expand the “PSPICE Resources|Simulation Profiles” folder. Right click on “Schematic1-signal” and select “Make Active”
  3. From the menu, select PSpice|Edit Simulation Profile. The Simulation Settings dialog appears. The settings should be as follows:
      • Analysis Type: Time Domain(Transient)
      Transient Options: Run to time = 80ms, Start saving data after = 40ms, Max. step size = 0.001ms
  4. To display the input and output waveforms on a single graph, voltage probes must be placed on the input and output points of the schematic. Click the Voltage/Level Marker (orcad2) on the toolbar and place a marker at grid of U6. Place another marker above R9 at the amp’s output.
  5. To run the time domain simulation, click the Run PSpice button on the toolbar (orcad3). When the simulation finishes, the PSpice graphing window appears. The input and output curves should be in different colors with a key at the bottom of the graph.
  6. To determine the harmonic distortion at 1KHz (the sine wave frequency), harmonics in the output waveform must be separated out through a Fourier Transform. In the PSpice window, press the FFT toolbar button (orcad7.gif). The PSpice graph changes to show the harmonics for the input and output waveforms. The input and output curves should be in different colors with a key at the bottom of the graph.
  7. The fundamental frequency at 1KHz will have the largest spike. The other harmonics are too small to be seen at the default magnification. In the PSpice window, press the Zoom Area toolbar button (orcad8.gif) and drag a small rectangle in the lower left corner of the FFT graph. The graph now displays a magnified view of the selected area. Continue zooming in until the harmonic spikes at 2KHz, 3KHz, etc. are visible.
  8. Harmonic spikes should exist for the output waveform only. The input is an ideal sine wave generator and has no distortion. To calculate total harmonic distortion, add up the spike values (voltages) at frequencies above 1KHz and divide by the voltage at 1KHz (the fundamental).

Note: simulations only approximate the performance of a circuit. The actual performance may vary considerably from the simulation as determined by a number of factors, including the accuracy of the component models, and layout and construction techniques.

c. 2006 Alex CavalliMark Lovell and Bill Pasculle (remove _nospam_).

Build These Noise-Canceling Headphones (plus: Binaural Mike Headset, Audio Probe and Parabolic Mike).

by Jules Ryckebusch

rychke1.jpg

In today’s hectic and noisy world, we are all searching for a little peace and quiet. Well, you might not be able to slip off to a tranquil forest for an hour or two, but you can block out background noise with the Noise-Canceling Headphones. The theory behind this project is that by picking up ambient sound with a microphone and reproducing it out of phase, we can actively cancel or “null” out background noise. In fact, several commercially available devices perform the same function. However, by building your own headset, you can add features not otherwise available and have fun while doing it!

Along with noise-features, the Active Noise-Canceling Headphones let you mix in an auxiliary line-level signal from a CD or tape player. That allows you to minimize background noise while quietly listening to music. The project also has a phase switch that will let you keep the microphone signals in phase, thus amplifying background sound. In addition, the design of the Noise-Canceling Headphones lends itself to several other interesting functions, which we will look at later.

How It Works.

The electronics consist of three op-amp circuits; each built around one half of an NE5532 dual op-amp. Each circuit uses that op-amp in a different configuration. The first circuit is a non-inverting pre-amp, the second is a unity-gain phase-inverter, and the third is an inverting headphone amplifier. Since the Noise-Canceling Headphones is a stereo device, the circuit is actually two identical circuits side-by-side. Only one channel will be described; the second channel works in exactly the same way.

rychke2.gif

Fig. 1. The Noise-Canceling Headphones is a simple phase-inverting amplifier. All inverted sounds played back through the headphones cancel out the original sounds, leaving nothing but silence. The amount of canceling can be adjusted for different situations. A CD player or cassette tape can be listened to if you want to “fill the quiet.”

The schematic diagram in Fig. 1 shows the design of the electronics portion of the project. A headset-mounted microphone is connected to J1, a 1/8-inch stereo jack. Electret-condenser microphones need a 2- to 10-volt bias voltage for their internal FET pre-amps. That is supplied by R2. A voltage-dividing network, which also decouples the bias volt-age from the power supply, is provided by Rl and Cl. That is necessary due to the high gain of the entire signal chain.

The signals from the microphone then go to ICl-a. an NE5532 set up as a standard non-inverting pre-amp. The gain is set to one plus the ratio of R8/R6 in the feedback path. The total gain for that stage is about 31 dB. Resistor R4 provides a ground reference for the pre-amp. A pair of high-pass filters is formed by C2/R4 and C4/R6. Those filters block any DC that tries to slip through the pre-amp.

From the output of the pre-amp, the microphone signal is sent down two different paths. It feeds both one pole of Sl-a and the phase-inverter. The phase inverter is nothing more than a second NE5532 configured as a unity-gain inverting op-amp (IC2-a). The output of IC2-a is connected to the other pole of Sl-a. That way, Sl-a can select either the inverted or the non-inverted signal. The selected signal on Sl-a’s common pole goes to potentiometer R14-a. That potentiometer sets the level of the microphone signal feeding the headphone amplifier.

The headphone amplifier is built around IC3-a, a third NE5532 wired as an inverting op-amp stage. The gain here is set by the ratio of R19/R15. That type of op-amp configuration can be easily modified to add a summing feature by the inclusion of R17. The second input comes from an auxiliary line-level input that is attenuated by potentiometer R23-a.

There is a reason why 10K-ohms was chosen for the value of R15 and R17. Besides keeping the values of R19 manageable, the 10K-ohms resistors interact with the 100K-ohm linear potentiometers. The potentiometers then behave in a logarithmic fashion.

This is how that feature works: One end of the potentiometer is tied to ground because we are using it as a voltage divider. Because the sum-ming junction of an op-amp is at a virtual ground, the 10K ohm resistor is also essentially tied to ground. That affects the response of the potentiometer, As the potentiometer is rotated, there is a more pronounced increase in the output as the end of the potentiometer’s travel is reached. That causes a smooth increase in perceived loudness of the signal. Potentiometers with an audio taper are, of course, available, but a linear-taper unit is easier to obtain and costs less.

The output of the headphone amplifiers is coupled to output jack J3 through R21. That resistor provides overload protection to IC3-a in case the output is shorted. If you have never used an op-amp for driving headphones before, you are in for a nice surprise. The NE5532 will supply a 10-volt rms signal into a 600-ohm load with very little distortion. That works out to 166 mW of power. Most personal stereos only supply 20 to 30 mW of power to headphones.

A final note on using operational amplifiers as headphone amps: Most generic ones will not supply enough current to function properly. Some substitutes for the NE5532 that are known to work include the 0P275 from Analog Devices, the OPA2604 from Burr Brown, and the LM833 from National Semiconductor. Those components are available from several sources, including Digi-Key, Allied Electronics, and Jameco.

Construction.

There are two parts to this project: building the electronics and modifying a pair of headphones. The circuit is relatively simple and can easily be assembled on a perfboard. One style of perfboard that simplifies construction is one having a pre-etched copper pattern on its solder side that connects groups of holes together. One example of that type of perfboard is Radio Shack #276-150.

The etched pattern on that board has a pair of buses that run the length of the board. Those buses are very convenient for power distribution. If you use that type of board for the Noise-Canceling Headphones, it is best to start by spacing out the three ICs on the board so that they straddle the buses. Then attach the power supply leads from each chip to the buses. It is then a simple matter of point-to-point wiring the rest of the circuit.

Check your work often while building the circuit. A common mistake many hobbyists make is not checking their work thoroughly enough. Often a few components are accidentally wired in backwards. The usual result is that the circuit will probably not work, the ICs could be damaged, and the electrolytic capacitors might explode!

rychke3.jpg

Fig. 2. The circuit board and batteries fit neatly into a simple project box. Keeping the wiring neat and following the layout shown here makes assembling the unit and changing the batteries easier.

When wiring the jacks, it is a good idea to follow the audio industry standards as to which jack connection is for which stereo channel. Normal standards for stereo connections are to connect the right channel to the ring and the left channel to the tip. The board and batteries are mounted in a suitable enclosure. A suggested layout for the components and control panel is shown in Fig. 2.

When selecting a case for the project, be sure that it is large enough to hold the circuit board and the two 9-volt batteries comfortably. After the front panel is laid out and drilled, check to make sure all the controls and jacks will fit. One method for labeling the front panel is to spray the entire panel with a flat color such as white or yellow. After applying transfer let-ters, seal the panel. Use several thin coats of a clear coating such as Crystal Clear by Krylon. The results are worth the effort. While the front panel is drying, we can start on the headphones.

rychke4.jpg

Fig. 3. The microphones are mounted on the earpieces of the headphones with a dab of silicone sealant. Tie both wires together in order to make the headphones more comfortable to wear (A). Follow the diagram in (B) when wiring the microphones. The left mic is connected to the plug’s tip and the right is connected to the ring. The ground connection on an electret microphone cartridge is easily identified by the solder connection between the terminal and the mic’s case (C).

Headphones with Ears.

The headphones are a standard pair of aftermarket Walkman-type units. They sell for about $20 at most record or electronics stores. The headphones are modified by mounting two small electret-condenser microphones on the head-phones, one on each earpiece. That modification is shown in Fig. 3.

The key to making the headphones wearable is to use thin wires running to the microphones. An excellent source of thin audio cable is to buy another set of cheap headphones – the cheapest you can find. Cutting the wire off them will yield a shielded stereo cable that is thin and flexible. As an added bonus, the wire will have a l/8-inch stereo plug molded on to it already!

The best way to strip that type of wire is to roll a razor blade very carefully over the insulation without cutting the wire underneath. Once the insulation is cut, carefully pull it away from the wire. That method works especially well on Teflon-insulated wire. After you have prepped the wire, mark the wire that is connected to the ring and the one that goes to the tip of the jack. An ohmmeter makes that task easy.

Carefully solder the wires to the microphone elements. The easiest way to do that is to pre-tin the wires and melt the little dab of solder on the microphone element with the tinned wire beneath the soldering iron tip. Look carefully at the microphone. As shown in Fig. 3C, the terminal that is connected to the case of the microphone is the ground connection. The other terminal is the actual microphone output. Holding the microphone element in an alligator-clip holder will make the job much easier. After soldering on the microphone elements, it is good idea to test them prior to gluing them to the headphones. The wiring should follow Fig. 3B.

Mount the microphones on the headphones as shown in Fig. 3A. One way to attach the microphones to the headphones is to use a dab of silicone sealant. Using a toothpick or other suitable substitute, mold the silicone around the edges of the microphone element to smooth everything off. Be careful not to get any on the black felt surface – that is where the sound enters. Obviously, the left and right microphones should be attached to the left and right sides of the headphones, respectively. Trying to cancel out a sound on the right with a sound from the left will not work.

After the glue is dry, gather and bundle the wires together with several nylon tie wraps along the length of the wires. With the headphones complete, it is time to experiment with the Noise-Canceling Headphones.

Creating A Quiet Zone.

For testing purposes, you should be in a quiet room with just a little background sound, such as a heater or air-conditioner fan. Plug in the microphone jack and the headphone jack, and put on the headphones. Turn both controls all the way down and turn the power switch on. Slowly turn up the microphone level. You should either hear the background sound increase or start to fade. If it increases, change the position of the phase switch. At some point, you should reach a “null” point where the background sound is at a minimum. If you adjust beyond the null, background sound will become louder as the out-of-phase signal exceeds the ambient sound level. Try talking aloud. If it sounds like you have a massive head cold and can barely hear yourself, the circuit is functioning properly.

Note that it is impossible to eliminate all incoming sound. Many things affect the ability to cancel out noise. The loudness of the incoming sound, the specific frequencies involved, and the position of the sound source all play a part in how well the headphones do their job. Feel free to experiment.

If everything is working fine, try connecting a CD player to J2. You will need a 1/8-inch -to- 1/8-inch patch cord similar to the ones used to connect portable CD players into a car stereo. After connecting the CD player, slowly turn up R23. It should sound clear with no distortion. Experiment with combining low levels on the CD player and canceling out room noise. The Noise-Canceling Headphones is the perfect device for environments that have a loud ambient sound level, such as rooms with loud ventilation systems.

Beyond Peace and Quiet.

This project lends itself to many other uses. Several interesting applications will suggest themselves that do not require any additional hardware. For example, by switching the microphones to “in-phase,” the unit can be used to assist hearing or improve hearing. Areas that can benefit include outdoor activities such as hunting or just observing nature.

Another unusual application for the Noise-Canceling Headphones is in binaural recording. Since we already have two microphones mounted in essentially the same place human ears are, all we have to do is send the headphone output to a tape recorder input. Binaural recordings put the listener directly in the sound field. The two microphones capture the exact phase and timing relationships of sound as we hear it. Those are the clues our ears use to determine the location of a sound.

Try this little experiment: record a person talking to you while you are wearing the headphones and have them walk around you in a circle. Then listen to the recording on the headphones. You will hear the person walk around you! The microphone elements used in this project feature full 20-Hz to 20-kHz frequency response. They provide a signal with surprisingly high fidelity.

rychke5.jpg

Fig. 4. Mounting both microphones angled apart at the end of a long stick makes an audio probe. It is very useful when you need to listen at a location that you can’ t reach.

Other interesting tools can be created by building different types of housings for the microphones. If two microphones are mounted on the end of a length of 1/2-inch dowel, an audio probe is the result (Fig. 4). It is wired similarly to Fig. 3B. That device lets you listen to things up close that you wouldn’t normally hear. It can be used to “sniff” out problems in mechanical equipment or to record things like hamsters chewing on cardboard. With the microphones mounted at an angle between 90 degrees and 120 degrees, you will have a stereo image of the sound source too!

rychke6.jpg
Fig. 5. A parabolic dish or lamp reflector makes a usable “Big Ear” microphone. The microphone is mounted at the focal point facing in towards the dish. Either one or two microphones can be mounted in the dish. If you build two, you can pick up stereo sounds.

An extension of the shotgun-style microphone is a “Big Ear.” The general arrangement is shown in Fig. 5. The main component is a small parabolic dish. Place one or both microphones at the focal point of the dish and experiment away. Sources for parabolic dishes can be as close as a local hardware store. A simple reflector for a light bulb can be found at a very reasonable price. Another source of true parabolic dishes is Edmund Scientific, 101 East Glouchester Pike, Barrington, NJ 08007. For some advanced experimenting, use two dishes (one for the left microphone, one for the right) and experiment with stereo reception of a distant sound.

Parts List

Resistors (1/4W, 1% metal film):

R1 – 4.7Kohm
R2, R3 – 2.2Kohm
R4, R5 – 1M
R6, R7 – 1Kohm
R8, R9 – 33Kohm
R10-13, R15-18 – 10Kohm
R14, R23 – 100Kohm pot, dual-gang, linear taper
R19, R20 – 100Kohm
R21, R22 – 47 ohm

Additional Parts and Materials:

IC1-3 – NE5532 dual audio op-amp
C1 – 33uF, 25WVDC, electrolytic capacitor
C2, C3 – 0.01uF Mylar capacitor
C4, C5 – 10uF, 25 WVDC, electrolytic capacitor
J1-3 – Audio jacks, 1/8-inch, stereo
S1, S2 – Dpdt toggle switch
B1, B2 – Battery, 9 volt
Microphones (Digik-Key P9967-ND or similar), headphones, PC board, case, wire, hardware, etc.

Addendum

10/5/1998: The following sentence: “The signals from the microphone then go to ICl-a. an NE5532 set up as a standard non-inverting pre-amp” originally identified an NE5522. Corrected.

7/8/2000: Marc Goodman built a noise-cancelling headphone amplifier for his Sennheiser HD580 headphones based on the circuit in this article. He writes: Just thought I’d report some results on building a noise cancelling headphone amp for my Sennheiser HD580’s. I started with the Ryckebusch project from this site, but I ended up making a fair number of changes. My goals were, in order of importance, 1). get the best audio quality possible driving my HD580’s from my Pioneer portable DVD player (PDV-LC10), 2). cancel enough noise, on demand, to make airplane listening/movie watching possible, 3). maximize listening times between battery recharging, 4). make the resulting project with enclosure small enough to fit into a camcorder bag along with the DVD player, headphones, and a selection of DVDs (i.e., not too big but no need to make it as small as, say, a CMOY amp).

1. Optimizing Audio
The Ryckebush NC project uses a single NE5532 inverting op amp stage for line-in amplification. I didn’t like the fact that the audio signal gets inverted, so I first tried changing the circuit to use a non-inverting amplifier. It sounded OK to me, but when I went to a unity-gain inverting op-amp as a first stage with an inverting amplifier as a second stage, it sounded cleaner, so that’s what I went with. I also used LM6172’s for both stages. As other people have noted, the LM6172 requires a fair amount of voltage. More about this in section three.

2. Noise Cancellation
Ryckebush feeds the output of electret condenser microphones through a non-inverting preamp and allows switching of this output through a unity-gain inverting op-amp to select between cancelling the noise or amplifying the noise. The inverted or non-inverted signal is then summed with the line-in signal through the final inverting amplifier. I was more concerned with conserving power than I was with being able to amplify external sounds, and feedback was a real problem with the Senns in any event, so I opted to drop the inverter from the circuit and to separately switch the power on/off for the entire mic/preamp stage. Since the preamp is non-inverting and the final audio stage is an inverting amplifier, the output of the preamp can be fed directly (via a potentiometer) to the input of the final stage. I would call the result “noise reducing” rather than “noise cancelling,” but the effect is quite noticable especially for lower frequency sounds. There is also a faint but noticable hiss when the noise cancelling circuit is enabled, so I wouldn’t want to use this stage unless the environment was pretty noisy to start with.

I wasn’t thrilled with the notion of gluing microphones to the outside of my Senns, both because it would look ugly and also because I carry them around with me everywhere and they’d be bound to get snagged/knocked off. Fortunately, the Senns have removable covers over the speakers with plenty of space between the speaker and the cover. In order to both reduce higher frequency noise and to provide a secure mounting for the microphones, I cut and shaped pieces of low-density foam (squishy) to fit the inside of the cover and give the speaker a little room. Though this closed the speakers off a little, the effect was not significant enough to be noticable (to me, anyway). It also teneded to acoustically isolate the microphones from the speaker elements which reduced the likelihood of feedback. I found the best results to be when the mics were placed to one side of the foam cover, directly in front of my ear canals.

3. Batteries
My first implementation of the circuit used dual 9V batteries as in the original Ryckebush circuit. However, I was completely bummed by how short a charge lasted. So, I tried using 6 C rechargeable batteries instead with a ground tap in the middle. These batteries held around 1200mAh as opposed to around 140mAh for the 9V’s, so they lasted a long time. But, I was driving the circuit with only around +/-4V instead of +/-9V, and while this turned out to be enough for the LM6172’s to power up, there was a huge amount of clipping at higher volumes with my HD580’s.

My solution was to build a separate DC voltage doubler using the ICL7662. The 7662 is a higher-voltage version of the 7660 that avoids latch-up on startup. It operates at 10KHz, which is smack in the middle of the audible range so ripple is a big issue. I found that by using 22uF low-ESR tantalum caps for the reservoir capacitor and by carefully routing all wires away from this capacitor to minimize inductive coupling, the ripple was beneath the threshold of my hearing. Note that getting the wires positioned just right was extremely important, as was putting the voltage doubler on a separate circuit board. Also, I had to shorten the wire lengths to all of my input/output jacks. It was kind of a pain, but not as bad as listening to a barely audible high-pitched whine ;).

4. Enclosure
It’s about the size of a paperback novel with roughly half the space taken up by the 6 C batteries in three rows of 2 lying in a Z and the other half taken up by the PCBs and controls (two DPDT switches, two pots, three 1/8th” jacks and a power on/off LED).

Conclusion
All in all, kind of fun and only around three times as expensive as if I had bought something from a store ;).

7/15/2000: Laura Balzano, Anna Huang, Eric Rombokas and Yasushi Yamazaki at Rice University built noise-cancelling headphones based on the Ryckebusch design for their project: “Sound Cancellation by Signal Inversion.”
As you know, we were using low-quality headphones, and when we mounted the mics on the outside of the headphones, we got little beneficial effect. We found that we could get better results using nice headphones, but they still were not very effective. We moved mics inside headphones to decrease sound delay effects. This greatly improved performance, but introduced feedback until we rigidly pointed the mics away from the headphones’ speakers and covered the mics with headphone foam (from the pair of headphones that we had to sacrifice)…. We found that due to sound propogation delay, non-inverted output (in our case) produced better effect than inverted output. We also observed that headphones were ineffectual for high frequencies, so to reduce noise produced by the circuit, the input was put through a simple RC lowpass filter with Wo=1K.

Results: The apparatus heavily attenuates sounds of very low frequencies, and somewhat attenuates all frequencies <= 1Khz. The headphones also get rid of the part of your voice that “echoes” in your inner ear– the extra echo that you typically hear when your ears are covered and you speak.

2/10/2002: Value of C3 in figure 1 corrected to 0.01uF. Also posted larger version of images and cleaned up figure 1.

c. 1997, Gernsback Publications (Popular Electronics and Electronics Now).
From Electronics Now, September 1997. (Republished with permission.)

Poor Man’s Surround Headphones.

by Steve Connors

Disclaimer: Try this project with a pair of inexpensive or disposable headphones first. Otherwise, purchase one or two additional sets of ear cushions for experimenting, so that the headphones can be restored to original condition if desired. Neither the author nor HeadWize is responsible for any damage to headphones or earbuds that results from building this project.

I came up with an idea to make surround sound headphones from a pair of old but nice stereo headphones to use with a SB Live soundcard. I have a surround sound system with 5 speakers, but there are a huge number of gamers that just can’t blast the neighbors and family late at night. Quad headphones ? … a blast from the past but they seem to have disappeared. My main problem was finding a good reasonable set for my SB Live.

connors1.gif
Figure 1

These surround headphones use three sets of earbuds to supplement the stereo headphones for front/back sound – 2 rear channel earbuds and 1 front channel earbud per side. My main headphone is a Sony Digital Reference MDR-CD60. I got them at WalMart a couple years ago for 10 bucks. For the back speaker effect, I’m using Sony Walkman earphones MDR – E821Lp with Mega Bass. I use Koss earbuds for the front speaker effect. Someone said you couldn’t get front/back effect. Well, I beg to differ. As long as you can tell where sound is coming from, the brain will figure it out, and the surround immersion is complete.

[Editor: Adding an acoustic simulator.to the front channels may enhance the realism of the soundfield.]

Having two earbuds for the rear channel does help compared to just one to help resolve the characteristics of behind vs front. In my setup, the back is a tad louder and bass is pumped up. To power the rear channels, I have an old Labtec PC stereo speaker set that has a headphone jack (I’m sure there are endless ways to get the signals to the back earbuds). The front channel headphone/earbud combo is just plugged into the headphone jack of the SB Live.

Here are step-by-step instructions for making the surround headphones:

1. Make buttonholes in each ear cushion – 1 front, 2 back. Earbuds along the back of the ear ended up being the best location for me.

2. Use two sony earbuds with mega bass as buttons in back; these need the bass turned up just a tad and the volume cranked up.

Note: they are to be used in a way they really weren’t made to be, i.e., kept loud as possible just to point before they distort. Yes it will take two earbud, but hey! I had a couple of them sitting around.

3. I’m using Koss earbuds in the front ear cushion buttonhole.

Note: all earbuds face toward ear. [Editor: for ideas on how to position the earbuds, see the section on multichannel headphones in Technologies for Surround Sound Presentation in Headphones.]

4. Two stereo splitters hooked to 20ft extension chord. I bought my stuff at Radio Shack. Get the gold-plated versions.

5. The back earbuds go into one splitter/20ft extension.

6. The front earbud and main headphone speaker go into the other 20ft splitter extension.

7. Make it neat and put connection mess in a little sack. Tie top of sack. This will protect connections, allow easy access and keep them from being pulled apart.

8. Fix up the wires now … group neatly. Earbuds at top taped neatly to the Y of headphones. Headphone and front earbud all neat and secure.

9. The rear channel earbud extension goes to the headphone jack of the Labtec speakers.

Note: I just happend to have a Labtec CS900 in the closet with stereo headphone jack and bass, treble, and volume controls. I’m thinking up ways now to get a graphic equalizer into the fray. Why ?… because it’s there. The nice thing is to have some control of bass and treble, not to mention volume. Getting the rear channel earbuds to work means they have to be pushed. They have to reproduce back whispers, low rumblings, etc. and still sound clean.

10. The headphone/front earbud extension goes to the front headphone jack of your SB Live.

11. Set up your SB Live up to headphones … read the docs … they won’t be much help with the surround headphone setup, but you basically have a quadrophonic surround with 4 speakers.

12. Make all neat. Secure and protect lines to achieve max range of wires. I ended up with good 15 feet.

The headphones look fairly good now that I have wires organized down the cord. I’ll end here but would like to say that the front-back and right-left directions can be tweaked with the SB Live mixer and SB Live value experience positioning toys. You will have to go to Audio HQ – mixer – view -> check the box to show balance sliders on mixer.

They work pretty good, but as you can imagine, getting the depth front and back is the promise land of directional sound. For me what this does is creating an obvious difference when sound is coming from the back. I can really pinpoint where someone shoots at me now.

One drawback thus far. I’m all headphones. I’m going to use a toggle switch from Radio Shack to send SB Live signals to the surround sound receiver. That way I can just change my SB Live settings from the 4-speaker quadrophonic surround to 5 speaker surround… flip a switch and presto, I’m using speakers.

c. 1998 Steve Connors.