A Fender-Tone Tube Headphone Amplifier.

by Alex Cavalli

cavalli0a

This project grew from several conversations that I had with John Broskie of the TubeCAD Journal. I was looking for a good headphone amp and John and I discussed some of the advantages and disadvantages of several of his published circuits. I must say that for someone who creates so much good material, he is also very generous with his time. One particular discussion led me to a White cathode follower design that seemed elegantly simple, that could easily drive 300ohm headphones, and might do respectable duty driving 32ohm phones. The original circuit appeared in the April-May 2001 issue of TubeCAD Journal.

The more I looked at the circuit, the more I liked it. Naturally, because it was so simple, I decided to make it more complex. An issue for me at the start of the project was that I didn’t have any test and measurement equipment (except my ears), so I needed to find a design that I was pretty sure would work before I committed to building it. I spent many hours simulating various topologies with OrCAD PSpice using triode models from Norman Koren’s Vacuum Tube Audio Page that I modified for better accuracy. I also created a new 5687 model and added it to the library. The more designs I simulated the more I liked the simple White cathode output stage.

While Chu Moy and I were discussing the design for this article, he asked me how closely related Broskie’s amp was to the Morgan Jones amp featured in the HeadWize projects. Thus began a long conversation between us that resulted in the update to the MJ article with the new designs, the OrCAD PSpice circuit simulations and the tube libraries. For Bruce Bender’s 6N1P OTL amplifier, I suggested new values for some of the amplifier resistors and a higher voltage power supply, which Bruce generously agreed to try and was able to get much better results.

The Circuit

The Broskie Headphone Amplifier

cavalli1
Figure 1

Figure 1 is Broskie’s original design for 300-ohm headphones. The plate resistor in this circuit is calculated so that the drive signals to the upper and lower tubes are balanced (see The White Cathode Follower for a discussion about optimizing this configuration). Balancing the drive signals is the biggest issue with this kind of push-pull design. Given the component values, the output impedance of this circuit will be about 53 ohms. It can easily push 20mA peak-to-peak into a 300-ohm load with 0.5V input.

The TubeCAD Journal article at the link above contains a description of this circuit. To summarize here, the output is an optimally designed White cathode follower. A White cathode follower is a push-pull output topology where the signal at the plate of the upper follower is fed back into the grid of the lower triode causing it to follow out of phase with the upper triode (hence the push-pull). To ensure that the tubes both see the same amplitude drive signal the plate resistor must be the reciprocal of the triode’s Gm. The feedback lowers the output impedance below that achievable with a simple cathode follower.

The input stage is a grounded cathode with an active load. Its voltage gain is mu/2, which for the 12AU7 is about 8.5. Broskie selected 5687s for the output tubes because, for miniature triodes, they have very high maximum ratings and good linearity. Maximum current is 30mA and maximum plate dissipation is 3.75W per section. The high maximum current allows the tubes to be biased at 20mA in class A mode. In the push-pull circuit, each triode can swing 10mA up or down with low distortion, giving a total current swing of 20mA. With a Zo of about 53 ohms, the output section has an open loop voltage gain of about 0.92. Furthermore, voltage gain and output impedance stay constant with dropping load impedance.

By comparison, a straight cathode follower using a single 5687 with the same plate voltage, same bias voltage, and open loop voltage gain of 0.76, has an output impedance of 94 ohms. With a 32ohm load the voltage gain drops to .2 and output impedance drops to 24 ohms. Even with a pair 5687s in parallel, the gain only increases to 0.32. Clearly the White cathode follower is a superior power buffer. But even so, Zo of 53 ohms is still not low enough for 32-ohm headphones. The total forward open loop voltage gain of the amplifier is about 7.8 (.92 x 8.5). There are other features of this design relating to power supply noise that are described in the TubeCAD article.

The New Design

As I was pondering this design, several things happened more or less at the same time:

  • Although I wanted a respectable tube headphone amp, I also wanted to experiment with some other ideas.
  • I realized that I frequently listened to music with some bass and treble boost. I knew that I would not be happy with a new amp if it didn’t have at least some bass/treble boost.
  • I realized that the output was out of phase with the input (this is not really as big a deal as some think, but there was an opportunity to correct it anyway).
  • I had some extra 5687s, so that if I paralleled output tubes I could lower the impedance by at least half.
  • John Broskie published another article in TubeCAD Journal discussing a variety of possible sonic controls that could be incorporated into a tube amplifier design (The Missing Sonic Controls).

This all led me to augment John’s design by doing the following:

  • Adding and additional set of output tubes.
  • Adding a Fender tone stack.
  • Adding an input stage to align the phase and compensate for the loss incurred by the tone stack.
  • Adding stereo blend and spatial controls that were now possible between the input stages of each channel.

cavalli2
Figure 2

Figure 2 is the schematic of the final design including the tone stack but leaving out the volume, balance, and sonic controls. The right half of the circuit is John Broskie’s original White cathode amplifier with a pair of 5687s as output tubes. This lowers the output impedance to about 27 ohms (gain remains the same) and allows me to set the quiescent current at 17mA per section to preserve tube life while still achieving high current swings with lower distortion. I’ve increased the size of the output capacitor to get a better lower frequency response at lower load impedances and added good quality audio caps in parallel with the electrolytics. I also added a fuse after the output capacitors to protect the headphones just in case the electrolytic fails someday.

The 10M resistors (R5 and R13) on the grids of the 12AU7s are not strictly necessary. I included them, however, because I was going to build the amp in stages. I knew that at some point I would want to power it up to measure bias voltages before the volume and tone controls were wired. I needed these resistors to ground the grids and set the proper bias points. I could have just as easily used temporary clip leads to ground the grids when measuring voltages, but adding resistors was not that much extra trouble.

The Fender tone stack is a pure boost tone stack. Unlike the equally well-known Baxandall tone stack, the Fender stack cannot provide bass or treble cut. But, since I never listen with bass or treble cut, the Fender stack offered fewer components and complete capacitor coupling thereby avoiding the use of an additional coupling capacitor from the plate of V1. Like all tone stacks, the Fender tone stack has an insertion loss, in this case approximately 18dB. The “boost” actually comes relative to this loss. In the Fender stack, the mid control is only really active if there is bass and/or treble boost. When the bass/treble controls are off the mid control becomes another volume control.

The parts of the stack form several high-pass filters. The upper part is a high-pass filter through the 250pF capacitor into the treble control whose wiper feeds the next stage. The high-pass bass filter (through the 0.47uF capacitor) sits under the treble control. The high-pass mid filter sits under the bass control. Any signal at the top of the bass or mid controls is fed directly to the treble control’s wiper. If the bass control is all the way on, then the low frequencies are passed directly to the wiper. If the bass control is all the way off (zero resistance), then the low frequencies are shorted through the mid control to ground.

The mid control has a higher pass frequency than the bass control. It bleeds off some of the signal that otherwise would pass through the bass section to ground. Because the turnover point is higher than the bass turnover point, and because, when there is treble boost, high frequencies go through the treble section, the mid control has the effect of creating a mid-range notch that in this design is at about 400Hz-1KHz depending on bass/treble settings.

When bass and treble are off, the output sides of the lower two capacitors and the treble control wiper are shorted together and the response is basically flat. The mid control acts simply as a resistance load which, when set to zero, shorts the entire signal to ground. The 27K resistor prevents V1 from seeing a pure AC short to ground. An extremely useful tool for calculating the characteristics of various tone stacks is the Tone Stack Calculator from Duncan’s Amp Pages. I used this tool when designing this circuit.

Because the Fender stack introduces an 18dB loss, it was necessary to recover some of this loss with the additional input stage. This stage has a voltage gain of about 5.5 (with a bypass capacitor the gain would have been a little over 10). It does add some distortion to the original design (which has very low distortion), but I was willing to accept this to experiment with these additional controls. When bass/treble are off and mid is halfway the total open loop voltage gain is about 7. This is plenty of gain for CD players.

I had to compromise the design here because the spatial control (described below) requires unbypassed cathode resistors. To get enough gain from the first stage I used a fairly high plate resistor. Together these features give the input stage an output impedance of about 8K. The input impedance of the tone stack is about 27K at its worst (everything turned off). A higher ratio of impedances would be better, but I decided not to mess with the design any further than this.

cavalli3.gif
Figure 3

Figure 3 shows the input section with the balance, volume, and the spatial and blend sonic controls. The balance pot has an “M-N” taper, where at center position, there is zero resistance in each channel The output is shorted for half of the pot rotation and is linear taper for the other half, and each section operating in reverse compared to the other channel. Turning the control to the left attenuates the right channel without affecting the left channel and vice-versa. The blend control performs a standard stereo-mono blend between left and right channels by more or less shorting the plates of V1A and V1B together.

The spatial control is quite interesting. It introduces an in-phase signal from one cathode to the other. When an in-phase signal is applied to the cathode of a grounded cathode stage, this is negative feedback. Here’s an example that helps to explain its operation. In the “soundstage,” a source to the left of center will have a larger presence in the left signal than in the right, but will have a presence in both channels (under normal conditions). Feeding an in-phase left signal into cathode of the right channel will cancel some of source signal that is already in the right channel. And vice-versa. This will tend to make the channels sound more separated or “farther apart” because they no longer contain as many common sounds. [Editor: this type of spatial effect is also called ambience enhancement.] I am able to hear this effect, although it can be very subtle.

Power Supply

cavalli4b
Figure 4

The power supply, Figure 4, is a simple solid-state supply, although it is slightly overbuilt. During simulation I discovered that if the plate voltage for the input stage were drawn from the B+ (high voltage tap) of the output stage the circuit would oscillate, even with standard isolation techniques (R-C filtering). The prototype amp hummed slightly. I put the 5H choke in to track down the hum problem. In the original power supply, I tapped the first stage B+ from the very first filter section. When I installed the choke I had to take all four filter sections from the output of the choke. Both channels have the first two filter stages in common, but different final filter sections.

The heater supply uses a 12V, 5A low-dropout regulator (Linear Technology LT1084-CT-12). One problem with the amplifier design is the range of heater-to-cathode voltages present. If the heater supply were grounded the upper tubes would have heater-to-cathode voltages greater than 125VDC, exceeding 100V specification for the 5687. I solved this problem by borrowing a technique from Bruce Rozenblit (see the OTL design in his book, Audio Reality) where the heater circuit is allowed to DC float, but is AC grounded through a capacitor at the mid-point of V1’s heater. In my case, I attached the capacitor to the negative rail of the heater supply.

There are two power switches; one to turn on the heater supply and the other to turn on the B+ supply. The B+ switch is wired to the heater switch as a fail-safe. Studying the circuit I can’t see any places where adverse voltages would be applied if the B+ came on before the tubes were conducting. But, I turn the heaters on first anyway.

CONSTRUCTION

Component Sources:

Allied Electronics – chassis
Antique Electronics – knobs, audio caps, RCA jacks
Audio Electronic Supply – Noble balance control
Avel-Lindberg – toroidal power transformers
Digikey – wire, regulators
Mouser – resistors, capacitors
Radio Shack – switches, fuses, and assorted parts
TubeBuilder – terminal boards
TubeWorld – matched 12AU7s and 5687s

Stage 1

The front of the amplifier is shown at the beginning of this article. The input and output jacks are on the far left, the power switches on the right. The inputs include right and left RCA jacks and a single stereo mini-jack wired together appropriately. The outputs are both a 1/4″ phone jack and another stereo mini-jack also wired together appropriately. The controls are not labeled yet, but they are from left to right: balance, volume, bass, treble, mid, blend, spatial.

cavalli5

First, I constructed both power supplies and the Tubebuilder boards (ordered from TubeBuilder.com). Laying out a TubeBuilder terminal board is much like laying out a PC board, except the runs are made with point-to-point wiring and the components are connected directly to the tube sockets or the binding posts. Each board has tube sockets, binding posts, and copper ground plane with standoff mounting spacers underneath. I really like the Tubebuilder terminal boards because they let me do a lot of the wiring outside of the confines of the chassis. It’s much easier to wire the boards first, install them, and complete the power and signal wiring.

Then I drilled, punched, and painted the chassis. Unfortunately, I experimented with the paint on this project. I used appliance-grade black paint whose durability is not what I had hoped for. If I were starting again, I would use standard spray paints and good quality primers that can be found at most hardware stores. Since the chassis is metal, I would bake the paint on carefully. The aluminum chassis and cover plate are made by Hammond. The chassis dimensions are 17″ x 3″ x 10″. Once I painted the chassis black, I decided to paint the screws blue. Since the transformers and caps were blue, it was an easy choice for the screw color.

In the first prototype, there was a small fan on the top of the chassis to provide ventilation. I’ve removed the fan completely because it was too noisy. Now that the regulator is heatsunk to the chassis, the ventilation is not really necessary. Since I had a big hole in the top of the chassis, I made a cover plate with holes in it to let some of the internal heat out anyway. It doesn’t really have much effect, but I had to cover the hole somehow.

The sockets on the Tubebuilder boards are connected to the copper ground plane, which I grounded to the circuit ground. Thus, it is good practice that the tube sockets not touch the chassis to avoid ground currents. Normally the Tubebuilder boards are mounted so far beneath the surface (see below), with only the tubes sticking out, that this is not a problem. But I wanted to be able to see all of the tubes, so I shortened the spacers making it possible for the sockets to electrically contact the chassis. To fix this I drilled an extra large hole for the input tube and cut rectangular openings for the power sections using a saber saw. I filed the hole and cutouts to clean them up. Then I drilled all of the rest of the mounting holes and painted.

The toroidal power transformers were from Avel-Lindberg. (A-L were extremely helpful in taking their time to help me with a very small order of just two transformers.) They are mounted on the top of the chassis, because I like the way they look and because the heater transformer gets very hot. Tube circuits are already hot. No sense subjecting the internal enclosed components to more heat than necessary. Wires for the transformers are fed through rubber grommets.

The heater rectifier and regulator are heat-sunk to the inside of the chassis. Between the two of them they dissipate a lot of power and after a while the entire chassis is pretty warm. It might be possible to use a 3A regulator in place of a 5A regulator, because the heater requirements are 2.7A. However, the “cold heater” current of the tubes may cause the initial surge current to exceed 3A. Current limiting regulators will shutdown and may prevent any heater current from flowing to warm up the tubes so that the current demand can then drop below 3A.

However, in either case, the regulator must be a low-dropout type regulator. Standard regulators must have an input voltage at least 2.5V above the output voltage. But, unless you’ve got a huge 12V transformer, the voltage input to the regulator is likely to drop to or below 14.5V. At this point a standard regulator will stop regulating. On the other hand, a low-dropout regulators need only about 1V difference input to output. In my case, under load, the input to the regulator is about 14.4V, on the edge for a normal regulator, but well within specs for a low-dropout type.

The tone stack capacitors are inserted into the perf board that is suspended in mid air near the controls on the front. The input stage Tubebuilder board is underneath the perf board. The output stage high voltage supply is on the right and the input stage and heater supplies are on the perf board on the left. The non-electrolytic output capacitors are in the center with the fuses.

cavalli7a

An aluminum shield installed between the high voltage supply/transformer and the rest of the circuit reduces radiation problems. After mounting the power supply and Tubebuilder boards to the chassis, I wired up the as much as can be wired without connecting the jacks and controls (this includes wiring the power switches). The power cord ground and the circuit chassis ground are connected to a single chassis ground point.

I tested the 12V supply first without tubes, including testing to make sure that it was floating. I installed all of the tubes and tested the heater supply again and powered up the high voltage and waited for smoke. When none appeared, I started measuring the bias voltages according to figures 2 and 4. In my case, the actual B+ voltages were little higher than designed, but well within tolerance. When these checked out, I pulled the tubes and started Stage 2.

Stage 2

At this point I mounted the jacks and controls, positioning their various connections for ease of wiring. All of the jacks’ grounds are insulated from the chassis using plastic and fiber washers, again to avoid ground loops. The Noble balance pot has a small circuit board that is mounted to its pins. Wiring is then done to the circuit board.

cavalli8a

Good design made the control wiring very simple. Once I got this far, I built the tone control board. This board contains the three tone capacitors and the 27K resistor for both channels. The board is mounted in mid-air above the input stage (or below depending on your orientation). Because it is so light, the board is held in place by the wires connecting it to the rest of the circuit. I used shielded wire for the signal wires to an from the tone stack and from the volume control to the input stage.

The last step was to mount and wire the output electrolytics. To do this, I had first drilled holes in chassis about 1/4″ smaller diameter than the capacitor itself. Then, as part of early construction, I mounted a perf board flush underneath the chassis covering these holes. I used the screws that fastened the perf board to also attach the fuse holders. The caps are snap-mount types with fairly short, stiff leads. I pushed the leads through the perf board from the top and bent them over, which essentially fastened the caps down to the chassis. Then I wired the parallel audio caps and connected these pairs to the output of each channel, through the fuses to the output jacks.

cavalli9a

Anything resembling a wire bundle is held together with plastic cable ties. Once the controls and tone board were wired, I attached the knobs. The knobs are available from Antique Electronics (see below). They are 1″ black anodized aluminum. I also bought the Solen caps and the gold-plated RCA jacks from these folks.

Except for the balance control, the pots are all inexpensive Alpha potentiometers that I bought from Mouser. The dual balance pot is Noble part number 220Y100K(W)X2-9920, purchased from Audio Electronic Supply. ALPS also makes this kind of pot, but I couldn’t find the ALPS pots on the web, so I bought the Noble. If you had to substitute, you could just use an ordinary dual 100K linear pot wired as a balance control. This will cut the signal strength in half immediately at the input, but there is plenty of gain to compensate for it. I just don’t like to do this because gain always adds distortion and noise.

The values for the “mid-point” resistance of the blend and spatial controls, calculated using the formulas in the TubeCAD Journal, are around 10K. After some experimentation I finaly chose a 50K linear pot for the blend control and a 25K linear pot for the spatial control. Both pots have rear-mounted switches so that the controls can be completely removed from the circuit.

cavalli9b

The small capacitors in the amp (figure 2) are metal film, except for C1. C1 is a ceramic capacitor, because I was not able to find a 250pF film capacitor. I have heard the various discussions about capacitors for audio equipment. But, I am not enough of an expert on this to know one way or the other. Plus, I don’t have any equipment and I am not able to try experimenting with various caps. I did later see some 250pF silver mica caps at Antique Electronic Supply and would have used them just to be safe. In truth, however, I don’t know how much difference it would make. The amp sounds OK to me and the highs (where these caps matter) don’t sound harsh or brittle or whatever.

Here are the component layout and wiring diagrams for the TubeBuilder boards:

cavalli10

cavalli11

cavalli12

PERFORMANCE

As I mentioned above, I don’t have test and measurement equipment so my specs on this circuit are derived from simulations (which are generally better than reality). The performance of the amp, as determined from simulation, is shown in figures 5 through 8.

Note: the gain of the amplifier, described above, is not really that important. After all if the 0.5 volt output from a CD player line-out could drive 32-ohm headphones with enough current, it would be delivering 8mW (pretty loud). What any power amp really does is convert the voltage at the input into current into the load. As everyone knows power amps are just buffers that translate a high-impedance voltage source into a low impedance current source. In figures 5 to 8, therefore, I’ll be looking at current into the load.

cavalli13
Figure 5

Figure 5 shows the current supplied to a 300ohm resistive load. The input is 0.5V. The bass and treble controls are set at midpoint (in terms of resistance, not rotation). The midrange control is all the way on. The power output into the load (using P = (I^2 * R)/ 2) is about 15mW. For most headphones this should be more than adequate power.

cavalli13a
Figure 6

Figure 6 shows a Fourier analysis of this same trace. I’m using the Fourier analysis to estimate the distortion. For those who don’t know what a Fourier analysis is, a Fourier transform translates a time-varying signal into the frequency domain showing all of the frequency components that comprise the signal with their respective amplitudes. In this case, the input signal is a pure sine wave at 1KHz. If any other frequency components show up in the current at the load, then the amp is adding harmonic distortion (harmonic because the frequencies occur at harmonics of the base signal).

Before reviewing the Fourier analysis, it is important to point out that this is simulation. Real circuits exhibit many characteristics that are not accounted for in some of the models, including the fact that simulated components are perfect unless specifically created to be otherwise. In this case, the resistor and capacitor models are perfect; the distortion is coming primarily from the non-linearities in the tube characteristics. We should use these estimates with caution, although they should give a basic picture of what the amp is doing.

The large peak at 1KHz is, obviously, the primary signal (it goes up off the screen). There is no significant signal at the fourth harmonic or above. The total current flowing at 2KHz and 3KHz (the small peaks) is about 3ua. An estimate of the THD is given by dividing this number by the total current flowing at 1KHz, about 10mA. Thus, 3uA/10mA = 0.0003 = 0.03%. This is a very low distortion figure. The real number will be higher than this. The only way to get this number is to measure it. However, the simulation indicates that the amp should have very good distortion figures. Even if the real distortion is ten times higher, it will be around 0.3%. Still a good number.

cavalli14
Figure 7

Figure 7 shows the current supplied to a 32-ohm resistive load, 0.2V input at 1KHz. The bass and treble controls are set at midpoint. The midrange control is all the way on. At 22mA into 32 ohms, the amp is producing 8mW. This is still a lot of power for most headphones.

cavalli14a
Figure 8

Figure 8 is the Fourier analysis for this output trace. Adding up second and third harmonics we have about 85ua. Performing the same calculation as before, the THD is approximately 0.38%. The actual distortion may be higher than this. As you can see the amp does better with a 300-ohm load than a 32, but it still performs well at 32 ohms.

At 15mW and 8mW power delivery into the loads of Figures 5 to 8, the amp does not exhibit any clipping behavior. This is because the quiescent current in the output section is about 36mA, making it possible to swing ±32ma in class A operation with minimal distortion and without clipping. Beyond this, the grids in the real circuit will start to be driven positive. The simulation models do not show the effect of this very accurately.

Frequency Response

cavalli15
Figure 9

Figure 7 shows the amp driving a 32-ohm load at 40mA peak-to-peak. At 20mA peak-to-peak, THD is about 0.15%. The THD in both cases seems fairly constant from 20Hz-20KHz. With the tone controls set at no boost, response is better than 3db flat from 10Hz to 300KHz, shown in Figure 9 (PSpice AC analysis result).

cavalli16
Figure 10

Figure 10 is the frequency response with bass and treble at full boost and the mid control at half. The volume control for this simulation was set lower to keep figures 9 and 10 near the same scale. The center of the notch is about 400Hz.

RESULTS

cavalli17a.gif

My current headphones are Sony MDR-V600 Studio Headphones (45 ohms) that can be found at most consumer electronics outlets. My music source is a Sony portable CD player line output. There is much to be desired in replacing this combination, but even with these components I can say that the sound from the amplifier is very clean and crisp. It is also very quiet. The tone controls give me enough flexibility to satisfy my listening habits. The stereo blend control works exactly as expected. At first, I couldn’t really hear the effect of the spatial control. But after listening carefully, I can hear the sound stage get wider and narrower as the control is brought through its “mid-point” value. The amount of effect, of course, depends on how the music was recorded to begin with.

The performance numbers from simulation are borne out to the extent that with 45-ohm headphones I can barely crack the volume control to achieve a comfortable listening level, even with the tone stack set flat. With the volume near halfway, the sound level is unbearable for me. This leads me to believe that this amp could adequately drive Grado 32-ohm headphones. I have tested the cheap 32-ohm headphones that come with portable CD players. The amp easily drives these.

If I were to build another amp like this one, I would add a crossfeed section based on the designs described in HeadWize, and I might experiment with some other tone controls and equalizer designs. Actually, I am very happy with the Fender tone stack. I have found a combination of bass/treble boost that I like (as I thought I would) with my Sony headphones: in terms of rotation – Bass 1/2, Treble 3/4, Mid 1/2. This is a very pleasing sound for me and I find myself leaving this combination for almost all recordings. The effect of the spatial control is to separate the soundstage. It is not that substantial. Furthermore, I generally don’t have problems with headphone ambience. I guess I’m not a true headphone connoisseur!

I would probably use the Fender stack again. I would eliminate the ambience control and design a crossfeed filter. I might also remove the regulator on the heaters and just put a giant electrolytic (10,000 – 22,000uF). This would be easier, although it would require a voltage dropping resistor that would still dissipate a lot of power.

The next step for me is to get some better headphones and music source! In fact, if someone wants to contribute an article on building a tube CD player, “I’m all ears!” In the meantime, this project has exceeded all of my expectations. My thanks again to John Broskie for his great work in the TubeCAD Journal and for taking time to give me his insightful comments.

Appendix: Simulating the Amplifier in OrCAD PSpice

Alex Cavalli has provided the project files for simulating this amplifier using OrCAD Lite circuit simulation software. OrCAD Lite is free and the CD can be ordered from Cadence Systems. At the time of this writing, OrCAD Lite 9.2 is the latest version. OrCAD Lite 9.1 can be downloaded from the Cadence website (a very large download at over 20M) and should work as well. There are 4 programs in OrCAD 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 Cavalli Headphone Amplifier

Download OrCAD Triode Simulation Libraries

After downloading cavalli_sim.zip and orcad_triodes.zip, create a project directory and unzip the contents of the mj_sim.zip archive into that directory. Then extract the contents of the orcad_triodes.zip archive into the \OrcadLite\Capture\Library\PSpice directory. The files triode.olb and triode.lib libraries contain simulation models for several popular types of triode vacuum tubes including the ones used in this amplifier. They are based on tube SPICE models found at Norman Koren’s Vacuum Tube Audio Page and Duncan’s Amp Pages. Put the files triode.olb and triode.lib into the \OrcadLite\Capture\Library\PSpice directory. Note: heater connections are not required for any of the triode models.

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.

cavalli_orcad

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 cavalli.opj.
  2. In the Project Manager window, expand the “PSPICE Resources|Simulation Profiles” folder. Right click on “Schematic1-Response” 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). 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) on the toolbar and place a marker at the junction of R6 and the grid of U7. Place another marker above RL at the amp’s output.
    orcad4
  9. The tone controls are set full on, so the frequency response is not flat. To get a flat response set R10A to 250K, R10B to 1, R11 to 1, and R12 to 5K. The treble pot has to be represented by two resistors because PSpice doens’t have a native variable resistor model.
  10. To run the frequency response 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.
  11. 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). To see the bias currents displayed on the schematic, press the Enable Bias Current Display toolbar button (orcad6).

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=0.25, 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-Transient” 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 the junction of R6 and the grid of U7. Place another marker above RL 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). 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) 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. 2002 Alex Cavalli.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.