So Good... I built it twice!

It was in the comments, I think

 

I think you may well be right. OM2.7 pair #2 is now sounding very good!!

I did experiment with moving the speaker returns to the PCB and back to the power star. In the past I've preferred to use the power star. It seems to make minimal difference on this amp, maybe because the connections are very short in this particlar setup.

 

The decent (1000uF) bypass capacitance on the pcb directly links the fat 0v connection track where the AC speaker current returns to the output stage collectors, keeping Mr Kirchoff happy and with a nice small loop-area.

 

So, I managed to get the paralleled output transistor mods done this weekend. I used matched pairs to get best current sharing, although i found all the devices I had were very similar anyway so it probably wasn't needed. I was only impressed by how closely the NPN and PNP output devices HFEs are matched, all in the mid 80's, these are very good complementary pairs.

Also I've been following the "Wolverine" build thread on DIY Audio (I actually have a set of bare boards I may build at some point) and they have a very similar schematic to these OM2.7s, almost identical input stage and also the Locanthi T output stage. In their BoM they are fitting a 1uF speedup cap for the second driver stage but for the first driver stage the cap is omitted, despite showing 0.1uF on the schematic. Interestingly they are running a higher bias current for the first driver stage, so I've followed suit, dropping R29 to 560R.
Another point of note is that the input cap C2 is quite small on the OM2.7 schematic at 1uF, and I used 4.7uF to make sure there is no potential low end roll off.

I'm not sure if it was my imagination but I do think the sound of my boards had filled out somewhat over the last week or so, though I did also move my speakers back a couple of inches to get a little extra bass reinforcement. They are still perhaps a touch mid forward but generally sounding more balanced, if anything the extra output transistors seemed to have helped giving a bit more bottom end substance.

The extra output transistors have improved the performance quite a lot across the spectrum, better space and dynamics, even more detail but also a fuller and more effortless sound. These are sounding really quite good now.

 

A Wolverine is also in my future I feel. Its specs are mind bogglingly good, even better than the Benchmark AHB. If your reaction to that is so what, then you’ve probably never heard such an amp eg a Neurochrome amp. FWIW I also love class A and the whole dominant 2nd harmonic thing, but EXTREME transparency and neutrality has its own charms as well. It’s a classic dichotomy

 

Found it #6: https://www.diyaudio.com/community/...amplifier-only-music-2-7.374372/#post-6716816

 

I have translated the OM2.7 web page from Russian to English (thanks Google!) including the BoM.

If anyone wants a copy, let me know.

 

As an experiment this evening I tried moving the gnd return point from the OM2.7's from the "output" end of the Hackercaps to the "input"/rectifier end. To be honest I wasn't expecting much, if any, difference but to my surprise I was wrong...

The sound changed quite significantly and was noticeably less bright and forward (actually verging on dull sounding) but the timing and coherence was all over the place and very muddled sounding. Not obviously "broken" sounding but definitely worse overall.
I'm not sure I can explain why this would be as the hackercap is star grounded so the "output" ground point isn't sharing any common impedance with the smoothing caps as far as I can see...

..anyway this has got me thinking about grounding and whether there might be some improvement in the upper mid forwardness I'm hearing by running separate grounds for the front end of the OM2.7 back to the power board along ( and also moving the speaker return to there) and then leaving the existing ground connection to the OM2.7 to just carry the returns for the decoupling caps and zobel network.

I must admit this is where my understanding runs out a bit - the OM2.7 has a 1ohm resistor R3 that provides a local ground connection for the front end and feedback to the main ground point on the board. This would appear to be common practice as the Wolverine amp does this also, but using 2.2R. With all that said the Wolverine amp also runs a 0.1uF bypass cap across that resistor as well as the feedback cap... another possible point of experimentation!

 

Thanks for the heads up, have ordered 10 OM2.7's and also the 3 to try. Not that I need more amp PCB's to build, test and use!