BC546/556 are not perfect. Agree. I switched to 2N3904/06 and they look slightly better in the switching region. But now I have to live with 40V devices in 45V powered circuit. Generally it's fine for the 1/2 supply location, but I'd prefer to be safer about that.I used BC546B and BC556B as splitters. The typical characteristics of these are not identical the former is more linear in its' 50uA working range outside of the turn on area. The BC556B plot has a degree of slope
One of the main points of the circuit according to Linsley-Hood was to avoid temperature issues with the output stage. The problem of setting bias currents for the output stage is then removed to the splitters. I used BC546B and BC556B in simulations. These give acceptable but not stellar results at 1kHz with low output quiescent current. Unfortunately things fall apart at 20kHz.
My 4 channels are ready and stable. I think I'll leave them as is for now and in the next few days will post my updated schematic. My goal was just to try the amp, but also to modify a little for 4-ohm load capability. Looks OK to me. The 2A set by Blomley is not enough to drive 4-ohm speakers when close to full voltage swing.
I think there are 2 places where thermals are involved. One is D4, D5 vs splitters, and the second is the D1, D2 vs. output stage. I find the splitters to be stable. Once set will stay pretty much the same. Output stage is not so stable and the idle current may rise some 40% at high power for long time. It definitely does some thermal tracking, but it's not enough. I don't think this is a problem though. For people wanting perfectly stable output idle current, the Onsemi devices with built-in diodes are probably the answer.
Simulation worked fine for me too. With exception of predicting all the parasitics that I have introduced with wires and PCB design. I did the Cf selection experimentally. I played with the splitter diodes as well and must say that they are not super critical. Actually the splitter will work OK even with one 1N4148, but my thinking is that jumping less gap during switching is better, so I added SD101A (replacing BAT46).
The 20kHz performance is not bad in the real amp. But if you try 100kHz, the trimmers would be to a much higher current than 38mA. I find that some 80mA make the 100kHz performance better. Interesting that setting the trimers depends also on the power level. Adjusting at, say, 15W requires more DC current to remove the crossover distortion than adjusting at 1W output power. Must be something related to dynamic parasitic effects... Interesting schematic, no doubt. I like dissecting its performance.
There are two main issues:
- the inputs cannot be left floating
- the start-up mess.
For future builders I attach 2 snapshots showing the crossover distortion at 100kHz/3Vrms/4ohm load. One is with no DC current and the second is with about 80mA.
I'm curious to read more about the re-designs that were proposed in this thread! No matter what, 4-ohm load is kind of a stretch for this circuit. I'm afraid it will become different amplifier if re-designed for that.