by Kevin Gilmore
I bought the Omega II headphones without the amplifier ($1995 + shipping from EIFL Corporation in Japan). I would love to listen to the SRM-007t or SRM-717 amplifier, but really do not want to fork over $4000 to do so. I have been working on this solid state Stax headphone driver for a long time. It satisfies all of the design requirements. Of course it sounds absolutely amazing which is clearly the goal here. There are no capacitors in the signal path. Its fully DC coupled. No expensive parts, and can be built by just about anyone.
The amplifier operates primarily in the current domain. The first stage is a voltage controlled current sink. The second stage is a current-controlled voltage source. The fourth stage is a constant current sink. The main advantage of current domain amplifiers is speed. Standard voltage gain amplifiers with lots of gain are affected by the Miller Effect which prohibits extended frequency response.
This solid state amp is so much better than my tube amp that I no longer listen to it. I’m not a solid state snob; it’s just plain better. The people who have listened to this amplifier (some of whom were giants in the industry in their day) love it, much more than my tube amp. I love it too. I can’t stop listening to it. The tube amp has moved into a secondary position in my listening rack.
How It Works
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 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.
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 approximate voltage gain of this stage is 5. But it 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. The current source feeds a common base amplifier. The common base amplifier feeds a modified Vbe multiplier. I believe a famous designer is now calling this circuit a current tunnel. Its 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 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 transistor (which is wired as a vbe multiplier). This generates the 13 volts (referenced to – rail) necessary to properly bias the 3rd stage. The bottom transistor acts like a zener diode in series with a resistor, except a lot less noisy.
The third stage is another differential amplifier feeding another common base amplifier. The simple differential amplifier has a voltage gain of about 100. The common base amplifiers are used to reduce the miller effect on the differential pair. Since the miller effect depends on both gain and output voltage swing, reducing the output voltage swing of the bottom differential transistors significantly improves the speed of this circuit.
The fourth stage is an emitter follower driven by a constant current source (gain = 0.99). This output stage dissipates 12 watts total (3 watts per transistor x 4 transistors). The main design goal was low output impedance. For example, my electrostatic tube amp has a 50K load resistor and thus has a 50k output impedance. This amp has a 25 ohm output impedance (actually a little less with feedback) The result is a much more extended high end. The slew rate of the solid state amp is more than 5 times that of the tube amp.
For the output stage, each 2SC3675 sources or sinks 9 mA at a quiescent output voltage of zero volts referenced to ground. For the driver stage, each 2SC3675 sinks 1.1 mA, resulting in 1 VDC at the collector (referenced to ground). The bases of the 2SC2705s sit at about 16 volts (referenced to – rail). The overall open loop gain of the amplifier is about 2000, but feedback reduces it to 1000. Even without any feedback of any kind the total harmonic distortion of the amp is still under .02%.
My first prototype, the unit in the pictures, uses an unregulated power supply. Given the stiffness of the capacitors, and the fact that the amplifier is pure class A, there is absolutely no fluctuation in voltage when signal is supplied. Of course, a regulated supply is always better. A regulated design is shown above. The 2SC3675 and 2SA1968 are mounted on heatsinks (the small tab ones are fine). The transformer is a Thordarson 24R22U (Allied # 704-0952). Adjust the pot to get 580VDC for the bias voltage.
The ±15 volt supply is an encapsulated fully regulated power supply brick from Sola Linear (Allied part number 921-9215), which retails for $117. I used a 60 mA version, but thats overkill, because the total current drain is about 12 mA for both channels. Lots of companies make these. It’s the black brick in the picture. It is NOT a switching supply. I do not use switchers in audio stuff if I can possibly help it.
This project involves working with high voltages, so be extremely careful! Keep one hand behind your back at all times. 600VDC across both arms might possibly stop your heart.
All resistors are 0.5W. Most do not need to be. The 300K resistors in the top of the 3rd stage need to be 0.5W. The 150K resistor in the current drive in the last stage needs to be 0.5W. I am trying to find 2SA1968 transistors, which are 900 volt PNP types. If they are fast enough, then the two 300k resistors can be replaced with current sources instead, making the amp 100% current source driven.
The LEDs in the amplifier circuit are voltage references (1.7 volts types in the prototype) which track changes transistor voltage with temperature (low voltage zener diodes have tracking problems). They also serve to show that the unit is running properly. If the LEDs are not lit, something is wrong. You could always replace each LED with 3 1N914 diodes in series, but the LEDs look so pretty (reminds me of the glow of a vacuum tube).
I am using standard regular brightness red LEDs. The blue and green ones run at different voltages (blue = 2.6 volts, green = 2.1 volts). Using LEDs with voltage drops greater than 1.7V can affect biasing. Higher LED voltage drops in the first and second stages will tend to cancel each other out, and the numbers will be the same. That is, a higher voltage diode will increase the current sources from 2mA to maybe 3 mA (each), but at the same time, the current sink in the first stage will go from 2mA to 3 mA (total), so the net result is zero.
However in the final stage, a higher voltage diode will increase the standing power. As long as the heatsinking is good, an increase from 12 watts per channel to 15 or so is just fine. The transistors are actually good for 10 watts each, so it is possible to increase the bias to 40 watts per channel.
All 4 output transistors are mounted on one aluminum angle that bolts through the front panel to the heatsink. The mounting heatsink is 4″ x 5″ x 1/8″ aluminum plate, punched and then bent along the short axis. There are 4 holes that hold the transistors to the angle, and 5 holes that bolt the angle to the heatsink. The blue-finned heatsinks I found on some old power supplies. I used them because they were big enough and pretty at the same time. The 2 2SC3675 drivers have small standup heatsinks.
The two pots balance the output voltages to 0V referenced to ground. Begin the adjustment by putting a voltmeter between + output and – output and setting the first pot for zero volts. Then put a voltmeter between the + output and ground, and set the second pot for 0V. After the amplifier warms up for 30 minutes, adjust the pots again. I adjusted my unit once, and keep checking it every so often. The output voltages on my unit are less than ±200mV. Compared to the 580 volt bias, that is close enough to 0V. And that is over a 1-month period.
Assemble the output stage with care. The full output voltage swing exists between the bases and the collectors of the bottom output transistors. Poor soldering techniques combined with excess flux can cause an arc which may damage the transistors. It happened to me once.
The Stax jack is Allied part number 719-4043. For all headphones except the Stax Omegas, the plug fits in all the way. On the Omegas, the plug is a little fatter and does not fit in all the way, because the plastic center of the jack is about 0.25″ below the base of the metal rim. So I put the jack in a lathe, and took 0.25″ off the metal rim so that it is flush with the plastic insert. This modification does not affect the fit of other Stax headphone plugs. For details on how to wire the jack, see All-Triode Direct-Drive Tube Amps for Electrostatic and Electret Headphones.
The 2SC3675 is made by Sanyo. The 2SA1968 and 2SA1156 are from NEC. The rest of the transistors are from Toshiba. Here are the current prices:
In the USA, all of the Japanese semiconductors are available from B&D; Enterprises. B&D; takes credit cards. The entire semiconductor cost not including the power supply is about $50 USD. The parts are also available from MCM Electronics, Farnell and Newark Electronics. Since they are all the same company, these parts can be purchased just about anywhere in the world.
There are no recommended substitutes. No American manufacturer makes 900V PNP or NPN transistors with a low Cob anymore. Neither does Phillips of the Netherlands. The only manufacturers of these transistors are Sanyo and Toshiba, and only because they are heavily used in dynamic focus applications for large CRT monitors.
The enclosure is a Mod.U.Line by Precision Fabrication Technologies Inc. (part number 03-1209-BW) and is available from Newark Electronics, probably Allied too. It measures 3″ x 12″ x 9″.
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 +300V or -300V, then something is seriously wrong and needs to be fixed. An oscilloscope really helps.
The amp can output 800Vp-p or 1200Vp-p with headroom. At 800Vp-p, THD is less than .008% from 20Hz to 20kHz. The actual frequency response is 0 to 45khz (-3db at 45kHz) 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.
Last weekend, I took home a standard dummy head, and measured the SPL in Omega 2 headphones driven by this amplifier. With a drive signal of 800 volts peak to peak per side, the resulting spl is 106db. THAT’S LOUD! The amp can put out 1200 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.
[Editor: Contact the author to discuss the possibility of obtaining pre-etched PC boards for this amplifier.]
2/21/01: Corrected mislabeled transistor part number: 2SC1815 (was 2SA1815).
9/5/01: Corrected mislabeled transistor part number: 2SC3675 (was 2SC367).
2/12/2002: Richard Albers built the following version of the CDEA amp with some interesting modifications of the original circuit. He writes:
I have changed the 2SK389 FET for a MAT02 Dual Transistor in the first Stage of the CDEH-Amp. There were no problems, and it all worked fine from the start. It sounds much cleaner then with the Dual-Fets now, and I guess they add less harmonics to the music.
In the third Stage of the CDEH-Amp, I have changed the Voltage-Divider 350K/20K, which sets the Bases of the SC3675 at ca. 20V. For the 20K Resistor i have put in a 20V, 1.3W zener. For proper working, I set the current through the zener at 7mA. Two 25K ohm, 5W Mills non-inductive wirewounds replace the 350K, dissipating ca. 2.2W of heat. To reduce the zener noise, I have put a 4.7uF tantalum together with a 47uF electrolytic capacitor in parallel with the zener diode. Noise is no problem.
The Cabinet is a very simple construction, with the advantage of ease changing components or parts. There is only a wooden base with two side-panels. The front is the large heatsink together with an aluminium-angle. A suitable top-cover is under construction. The whole construction could be made way smaller, all parts on one pcb, with a smaller toroid-transformer, and all built in a industrial case, but for my own usage, it’s ok.
The two smaller transformers under the wooden cover are the 10H-chokes for the high voltage power supply. The little transformer on the bottom generates the bias-voltage. The oversized big-one is a special-made 250W transformer, from Experience-Electronics in germany. The electrolytics are from EPCOS (Siemens).
This is a further way to tune-up this fantastic machine. Together with the MAT02 dual-bipolar input device and using only the best parts you can get, such as non-inductive Caddocks, very low ESR electrolytic caps in the high voltage section, and so on, there is no better electrostatic headphone amp in the world. It sounds just fantastic!