The main concept behind this amplifier design is that it is fairly device independent, requires no trimming or fiddling with, and gives great performance for those who don’t care about distortions they can’t hear.
If you want to get your DIY on and build an amplifier that just works, this project is for you.
Some notes on design decisions:
Though not strictly necessary, adding a current source and current mirror to the long tailed pair reduces output offset by a useful amount (it should be under 100mV in this arrangement) and it also greatly reduces the dreaded turn-on thump. That aids the no-fiddle goal of this amp by avoiding an offset trim.
The dominant-pole, or Miller capacitor (between the VAS base and collector) is 220pF. This is higher than you will see for most audiophile designs. A lower value would help reduce distortion at high frequencies, but I don’t really think that is a worthwhile endeavour.
It seems silly to use a bootstrap current source, when you could just use the voltage reference from the long-tailed pair current source for a second current source, but it makes routing the PCB a little simpler, plus the bootstrap has built-in current limiting if you decide that you want to add some VI limiting later. That’s the beauty of DIY.
Zobel network on the output for high-frequency stability. Pretty standard.
No inductor. Who wants to mess around with wrapping coils of wire around resistors?
I used a Darlington arrangement for the output stage because it is dead simple. The extra volt or so of output range gained is simply not worth the confusion of a complimentary pair. I’m not sure why, but I can never put a complimentary pair together correctly. I always end up with a short circuit somewhere. Besides, it gives a little wiggle room for the most important piece of this whole design.
The output biasing. It was developed using my output stage bias tester. A standard red LED has a pretty consistent 1.8V drop at the VAS current of 10mA. I hate messing with bias trimpots, and stupid thermal coupling, and never really being sure that I won’t end up with thermal runaway. With this setup, the driver transistors are biased into class A operation by a solid 600mV, but only 1.3mA with the driver emitter resistors shown. The Darlington setup ensures that the output transistors will always be underbiased by a safe margin, no matter how hot they get.
The result is surprisingly low crossover distortion.
This design is fairly tolerant of transistor substitutions, just make sure you use devices that have sufficient voltage and power ratings. Price was the motivating factor for most of my decisions.
I specified TIP35C/36C for the output devices because they have decent power ratings and only one pair is needed for use with 4Ω loads. Their large size also means they can transfer heat to a heatsink pretty well. You could use a single pair of TIP3055/2955 (or MJE or equivalent) TO-220 transistors for 8Ω loads if you have them lying around, or you could use another pair in parallel for 4Ω. The same thing goes with 2N3055/2955, but that TO-3 package is such a pain to mount that they are not recommended for a no-fuss amplifier such as this.
The long-tailed pair current source transistor runs warm. You might want to swap it out with a BD139 or something similar for a little more dissipation, and/or increase the current source resistor, R1, to 220 or 270Ω, if it makes you more comfortable.
Feel free to use a different supply voltage if you want, but be careful about voltage and power ratings. You can go up to ±40V if you stick to 8Ω loads and increase the value of R1, or you can go down to about ±10V if you want to work really hard to make a big headphone amp.
Standard audio guidelines apply here. Star grounding, .1uF caps on the supply rails. My PCB is laid out almost exactly like the schematic, my preamp (in the same box) is a few inches from the amp board and they are connected with a short shielded cable. Distortion is inaudible, and (when no music is playing) I have to get my ear within 6 inches of a 92dB sensitivity loudspeaker to hear any noise at all.
I will be adding a toner-transfer ready PCB layout soon(ish), maybe with some pictures, and hopefully also with measurement results.
Yes, the name is a ripoff of Doug Self’s “Blameless” Class B Amplifier.