diyAudio (
-   Headphone Systems (
-   -   Single ended class-A headphone amp using two transistors: T2 (

Mark Johnson 27th May 2019 10:49 PM

Single ended class-A headphone amp using two transistors: T2
10 Attachment(s)
Headphone amps are very popular as DIY projects, and quite a large number are available. To name a few: Beta22, Objective-2, the famous CMOY amp, Sjöström QRV01-09, Butte, and many more. In addition to those, the diyAudio store right here on this very site also offers the (vacuum tube) Starving Student II headphone amp, and the (ICs + discretes) WHAMMY headphone amp. So why build yet another headphone amp? How is this one different from all the others?

I think the answer is, T2 has a unique combination of features that are not found together in other headphone amp designs:
  • All-discrete, solid state: no ICs, no vacuum tubes
  • Single ended, Class A circuit operating at 150 mA bias current (exceptionally high for a headphone amp)
  • No mains voltage inside the DIY chassis. Power supply is a commercial, safety rated, DC wall wart followed by a 2 stage supply filter on the PCB
  • 100% thru-hole parts, mounted on a single PCB with plenty of spacing between components. Very easy to stuff and solder, suitable for first time DIYers.
  • Designed to be a comfortable and uncrowded fit within typical headphone amp chassis, including the "Galaxy 1U" chassis from the diyAudio store. Board is 182 x 112 mm
  • Headphone output (6.35mm TRS) on front panel, plus linestage preamp output (RCA jacks) on rear panel
  • Pre-made front and rear panel with silkscreened logos and control labels, for those DIYers who prefer not to drill + label the panels themselves. These are black PCBs, 2.0mm thick, cut to match the 1U Galaxy front & rear panels. PCB silkscreen labels.

T2 is powered by a 24V DC wall wart. With 150 mA of bias in each of 2 channels, that's 7.2 watts of power dissipation, just in the output stages (!). And it is housed in a chassis with no heatsink fins at all (Galaxy)? True. Figure 1 shows an infrared photo of T2 after it has reached final operating temperature. 90 minutes after turning T2's power on, I took this picture with a thermal imaging camera, using its High-Low mode. The camera automatically marks the positions of the highest temperature (37C) and the lowest temperature (27C). To prevent heat transfer between T2 and the tabletop during this experiment, a large wooden board, 1m X 1m, was used as a thermal insulator. The board laid on the tabletop and T2 sat upon the board. The exterior metal was slightly warm to the touch, not hot at all.

Amplification is performed by a two transistor circuit, which is not an especially new idea in itself. Two device amplifiers have been aroung a long time, in hybrid tube+transistor circuits (Fig 2 top) and in all-BJT circuits (Fig 2 bottom (from D.Self)), to name two examples. The T2 headphone amp simply scales up the output stage bias current to 150 milliamps, and uses high wattage power resistors as the "collector load" of the output stage.

Figure 3 shows the business part of the T2 amplifier circuit. The polarities of the two amplifying devices have been reversed; the input transistor is now P-type and the output transistor is N-type. This lets us exploit a serious advantage of Nchannel MOSFETS: they are majority carrier semiconductor devices, whose gain is proportional to electron surface mobility. Pchannels' gain is proportional to the surface mobility of holes. Since electron surface mobility is about 2-3X higher than hole surface mobility, Nchannels have about 2-3X greater gain (per unit area, per unit capacitance) than Pchannels. All we have to do is flip the power supply upside down, and we get 2-3X higher gain. An easy decision! T2 uses a positive-ground, negative-supply arrangement, with the supply rail "NEGPWR" at approximately -22 volts DC. (BTW positive-ground was common in germanium transistor circuits back in the 1950s and 60s.)

The input signal is applied to the base of input transistor Q71 through coupling capacitor C72 and RFI filter R74-C74. Since the input signal is referenced to ground (by definition!), the base of Q71 is referenced to ground. The positive-ground arrangement means the emitter of Q71 is also referenced to ground, so the base-to-emitter voltage across Q71 is quite immune to supply variations, and PSRR of this stage is high.

Q71 is loaded by a constant current source whose manufacturer calls it a "current regulating diode", D75. This drives the gate of output transistor Q72. Q72 operates at 150 mA of bias current, set by potentiometer R73. Its load is a 12 watt, 68 ohm resistor which is synthesized by a series-parallel arrangement of four 68 ohm, 3 watt resistors. Total dissipation in these resistors is < 2 watts, whereas their aggregate rating is 12 watts, so a conservative margin of safety is in effect. Nevertheless the resistors do get fairly warm when powered up.

Negative feedback from the output stage to the input stage is arranged by connecting the emitter of Q71 to the midpoint tap of R75-R78. This sets the amplifier's overall gain to +6.0 dB, which ensures the volume control pot will be cranked up past 25% of rotation, just about always. Typically, channel to channel tracking in pots is worst at low rotation, but improves as you crank the knob up higher and higher.

Diodes D72-D74 elevate the source voltage of MOSFET Q72, which in turn elevates its gate voltage. This biases current source D75 to its sweet spot on the I-V curve for best operation, comfortably above the "knee voltage Vk" discussed on the E-153's datasheet.

An amplifier with a single ended power supply needs an AC coupled input and output; capacitors C72 and C76+77 provide these. C72 and C77 are a high quality film types. Go ahead and calculate the low frequency rolloff corner frequencies introduced by AC coupling; they are far below the audio band.

Some may not agree that the T2 circuit can be called a two transistor amplifier. For one thing, darlington "transistor" Q71 is, in reality, an integrated circuit with two transistors on the same die. And for another, it feels like cheating to call D75 a diode and not a transistor. Even though D75 has only two terminals, even though it cannot provide voltage gain, current gain, or power gain, D75 just seems like it needs to be called a transistor. To those who feel this way, I gently reply: you may be right. However I really like the name T2 and I plan to keep using it. Feel free to call it by another name (T2++ ??) if you prefer.


24V DC from the wall wart is applied to the power supply filter circuits shown in Figure 4. A resettable PolyFuse protects against short circuits, and a bidirectional TVS (Transient Voltage Suppressor) diode protects against overvoltage. High frequency spikes and hash from the SMPS are removed by a ferrite core, Common Mode Choke "L1", working in conjuction with capacitor C1. Bridge rectifier D2 allows the use of either polarity DC plug on the wall wart; the amplifier doesn't know and doesn't care whether it is center-positive or center-negative. The supply is filtered again by L2-C4, to further reduce high frequency noise.

Component L2 is a ferrite core from Laird Performance Materials, specially engineered to attenuate switch mode power supply noise in the frequency range between 100 kHz and 10 MHz. I've snipped out a graph from Laird's datasheet and presented it below as Figure 5. Notice that attenuation increases as the square of the number of turns on the core. Laird anticipated that customers would use big thick wires carrying great big currents, so at most three wire turns through the center hole would fit. But T2 draws less than 400mA, meaning we can use wires even thinner than AWG-28, which allows at least eleven turns through the Laird core. Mentally extrapolate Figure 5 to eleven turns: Cowabunga! That's a lot of attenuation.

Potentiometer R2 and fixed resistor R7 are current limiters which allow builders to continuously vary the brightness of the power-on LED, as high or as low as they wish. Perhaps I should say, as bright or as dim as their spouses wish.


Figure 6 shows the T2 PCB bolted into a 1U Galaxy chassis from the diyAudio store. The front panel, at the top of the photo, holds the power-on LED, the power switch, the volume control potentiometer, and the headphone jack. The back panel, at the bottom of the photo, holds the DC input jack (wall wart connector), the stereo input RCA jacks, and the output RCA jacks. Close scrutiny of the photo also observes green grounding wires which connect to the front panel and to the rear panel. These implement a solid metal Faraday cage surrounding the T2 amp on all six sides, shielding it from external noise and hum. Don't try to build T2 without an all-metal grounded enclosure (Faraday cage)! The risk of audible hum is too great. Not visible in the photo is the electrical connection between the metal body of the ALPS RK27 potentiometer, and the Faraday cage. This incorporates the pot itself into the Faraday cage. Thank the mechanical designers at ALPS for including a metal alignment peg which also serves as a ground connection point.

Figure 7 shows a fully assembled T2 using the blank front and back Aluminum panels that ship from the diyAudio store. I drilled them myself on my $200 tabletop drill press from Amazon. There are no labels but really, why do you NEED labels? It's blazingly obvious that the 1/4" hole on the left is the headphone jack, the knob in the center is the volume control, and the switch + LED on the right are the power on/off control. The rear panel is another story of course; you do need to indicate which RCAs are inputs and which are outputs. And you do need to indicate which are Left and which are Right. Some builders may not care whether the labels on the rear panel are "beautiful" or not, since they are not visible during normal use. Any labels at all, no matter how ugly, would be adequate. Other builders would be offended at the idea of permitting any ugliness at all, anywhere on the build, whether visible or not.

Figures 8 and 9 show a T2 using the low cost front and back panels made by a PCB fabrication house. I designed these as plain ordinary PCBs, including the correct size drill holes for the jack, the pot, the switch, RCAs, etc., and also including the Faraday cage conductors on both copper layers. I selected "matte black" color, and ordered them from a low cost fab house in China. The boards came back drilled and plated beautifully, with acceptable quality silkscreen markings.

The final attachment is a Bill Of Materials for Mouser's BOM tool.


Playing into my Audio Technica M50X over-the-ear headphones, the T2 was nuanced. Female vocals (Joan Diener, Marianne Faithfull, etc) were ethereal and otherworldly. And yet full orchestras sounded detailed and crisp, forceful with no detectable muddiness. "Mars" from Holst's "The Planets" suite was suitably terrifying. And ZZ Top kicked assz like only ZZ Top can do. Put on "Tres Hombres" and toss back a couple shots of El Tesoro Anejo. Bliss!

edit: reviews / listening evaluations from members who built their own T2, are included in posts #64 , #66 , #110 , #141, #151-152, #165, #180 of this thread , and I'll try to update this list every now & then


Two other members of diyAudio, whose mechanical expertise & fabrication skills far exceed my own, are each putting together their own T2 using this PCB and these schematics. I'll call them Beta Testers. This process has turned out to be extremely helpful, as the Beta Testers have already uncovered difficult problems in the build which have been quickly remedied. Nothing which affects the schematic or the PCB (not yet anyway!), but several pitfalls and gotchas in the build notes have been revealed, where doing something a different way turns out to be 20 times easier than the original plan.

These folks are snapping photographs as they build their T2, and I'm confident their pictures will turn out much better than mine. I'll add those photos to the first few posts in this thread, as they become available, and readers can begin to see what Real Craftsmanship looks like.

Meanwhile I've got about ten extra T2 PCBs and ten extra pairs of (matte black) front and rear panels, predrilled and silkscreened, which I am pleased to let other DIYers have at my purchase cost, plus my cost to ship to you. For a set of 3 PCBS (red circuit board, black front panel, black rear panel) that comes ot to USD16.50 for US buyers, USD 26.50 for rest of world. Send PM for PayPal instructions, limit one set of 3 boards per buyer.

I have created a BOM at Mouser, for every T2 part that Mouser sells. You may not WANT to buy all of them at Mouser, but they are all included in the BOM. Example 1: ALPS RK27 Blue Velvet potentiometers are sold at much lower prices elsewhere. Example 2: you may prefer to use a different style of RCA jack than the Gold ones in the BOM. Example 3: You may already have a suitable wall wart and don't want to pay for another one. Etc.

Use the BOM manager to upload this .xls file as a BOM, let Mouser process it, then move it to your shopping cart. Finally start deleting things from your cart which you DON'T want to buy from Mouser.

Although I have tried hard to make the BOM correct and complete, it might still contain mistakes. I hope not but people make mistakes all the time, including me. There may also be parts which T2 needs but which Mouser doesn't sell. I can't think of any but I might have missed a couple. Check against the schematics, spend a little time and make sure you don't have to re-order to fix ordering mistakes.

I also solicit any feedback members may have, about whether a Group Buy might be a good idea if these ten boards sell out. Either (a) PCBs only, (b) PCBs + some of the parts, or (c) PCBs + all of the parts. Thanks for your advice and opinions!

Mark Johnson 27th May 2019 10:50 PM

8 Attachment(s)
Updated (for clarity) Schematics --


I made a few measurements on T2 today, examining output drive capability. I began with this excellent table of headphone data from the website: link

They tabulate the voltage and current (AC RMS) needed to make each pair of hedphones produce 110 dB SPL, which is very loud. I like this approach because it gets rid of intermediate variables like sensitivity, impedance, and so forth. The voltage & current data tells you exactly what you want to know: how to drive each of the 83 pairs of headphones to maximum volume.

I've charted their data on the scatter plot below. Each red diamond is one datapoint from head-fi's table. Overlaid on the plot is measured data from the T2 headphone amplifier, plotted in blue. The blue line shows where T2's output just barely begins clipping: for X output voltage, T2 clips when output current exceeds Y milliamps. 80 of the 83 headphones in's table, fall below the blue line. Which means that T2 will drive all 80 of them to 110dB SPL or louder. The ones T2 cannot push to 110dB are:

HiFiMan HE-4, HiFiMan HE-6, HiFiMan H3-500

whose drive requiremnts border on the insane. T2 is not insane and it cannot drive them to 110dB SPL. Sorry.

I also measured T2's output impedance today. With the input grounded, we inject an AC current into T2's output, and measure the resulting AC voltage. Computing the ratio (Voltage/Current) gives the output impedance. It's Ohm's Law. When the injected current was 13.3 milliamps AC, I measured T2's output voltage to be 19.0 millivolts. Their ratio is T2's output impedance: 1.43 ohms.

Finally, there are a few oscilloscope measurements attached below.
Max amplitude before clipping, with a 22K ohm load, is about 6V RMS. At both 1 kHz and 50 kHz.
Small signal square wave testing at 200 kHz, shows thoroughly damped rising and falling edges with no ringing.
Rise time is 300 nsec, corresponding to a closed loop bandwidth of 1.1 Megahertz (at a gain of +6 dB)


Mark Johnson 27th May 2019 10:50 PM

10 Attachment(s)
Another place holder. More goodies coming.

Mark Johnson 27th May 2019 10:51 PM

8 Attachment(s)
Fourth place holder. This ought to be enough, I hope so.

6L6 27th May 2019 10:55 PM

Absolutely fantastic project!

Here is a populated Mouser cart to build the T2 - Mouser Electronics

I'll have photos and insights to add to the thread soon. :yes:

xrk971 28th May 2019 01:07 AM

Nice project and looks like an ideal build for someone looking for a complete ready to go project with shipping cart and case.

I think the DCA Headphone Amp satisfies almost all your criteria except premade front and rear panels. But I find that $30 CNC all alum HP amp cases on EBay/Aliex work with minimal mods or fuss.

xrk971 Desktop Class A (DCA) Headphone Amp

To be all through hole requires use of TO92 JFET instead of BF862. But those are the only SMT and not hard to solder.

145mA bias is possible with change to source resistance but it runs nominally at 125mA which is plenty.

I have used these cases before. Just need to make a small plate to hold 5.5mm coax DC wallwart jack to go where IEC is.

Detail Feedback Questions about Douk Audio black Aluminium Case Amplifier Enclosure Headphone Amp Chassis on | alibaba group

gannaji 28th May 2019 03:45 AM


Bigun 28th May 2019 04:23 AM

nice one Mark!

EUVL 28th May 2019 06:05 AM

1 Attachment(s)
Can be simpler still with a J175 ?

I prefer NJFET input for lower noise.
These days hardly any difference between NMOS and PMOS.



blackdod 28th May 2019 02:23 PM


All times are GMT. The time now is 05:39 PM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Copyright ©1999-2019 diyAudio