I will add those.
In the Douglas Self Blameless there is 220uF and 100nF.
It is quite different of those exemplars.
With a little more for the P channel.
Thanks.
Lineup, I would suggest you add a Hum Breaking Resistor (HBR) as shown in the attached pic below. This will help to attenuate cross-channel ground loops and give the amp a good measure of protection from this type of noise. A good value is 2.2-4.7 Ohms. The volume control pot lower end must also go to the left-hand side of the HBR.
The RC input filter seems pretty aggressive. R22 at max provides 2500R combined with R21 gives 7200R. Add 100R for a typical active source and you up to 7300R for 265KHz corner.
Maybe drop R21 to 1K? Assuming the same 100R active source, that gives an frequency of 539KHz with the pot and 1.76MHz without. If the builder doesn't use the input pot, they could increase C8 to 220p for a 658Khz corner (assuming a 100R source again)
Dropping R21 from 4.7K to 1K will also lower noise.
Maybe drop R21 to 1K? Assuming the same 100R active source, that gives an frequency of 539KHz with the pot and 1.76MHz without. If the builder doesn't use the input pot, they could increase C8 to 220p for a 658Khz corner (assuming a 100R source again)
Dropping R21 from 4.7K to 1K will also lower noise.
I just use a 4.7 Ohm resistor for HBR in my amp because it's the same resistor I use for the OP transistor base stoppers. The main function of the HBR is to attenuate cross-channel ground loop currents arising inside the amplifier where the grounds of the two channels form a loop that is closed at the source equipment ground. If this loop gets cut by stray mag fields from the power transformer inside the amplifier, this loop current can flow out of the amplifier in the interconnect screen to the source equipment and then back again in the other channel screen. Any volt drop across the interconnect screens is in series with the source signal and degrades SNR. The HBR forms a divider with the screen resistance and gives very effective attenuation of cross-channel loop currents.
The other technique to reduce this effect even further is to connect the signal grounds of the channels together where they come into the amplifier - i.e. put the connectors right next to each other and bond the signal grounds together at that point. This traps cross-channel ground loops inside the amplifier and prevents them from flowing in the interconnect screen.
You will still have cross-channel ground loop currents, but the resistances available for developing noise voltages in series with the signal will be greatly reduced and the HBR's will lower the loop current. Of course, it goes without saying that ground loop areas must be minimized wherever possible - but this is an implementation issue to be addressed during mechanical and wiring layout design.
The other technique to reduce this effect even further is to connect the signal grounds of the channels together where they come into the amplifier - i.e. put the connectors right next to each other and bond the signal grounds together at that point. This traps cross-channel ground loops inside the amplifier and prevents them from flowing in the interconnect screen.
You will still have cross-channel ground loop currents, but the resistances available for developing noise voltages in series with the signal will be greatly reduced and the HBR's will lower the loop current. Of course, it goes without saying that ground loop areas must be minimized wherever possible - but this is an implementation issue to be addressed during mechanical and wiring layout design.
I do not like that inter-ground resistor very much, after I have experienced serious failures of such amplifiers when this resistor was burnt by error ground current. I prefer two separated power supplies for 2 channels and the left and right ground connected on the input connectors. Just at this one and only point.
I believe bypassing the HBR with anti-parallel diodes protects this resistor in a scenario like you described
Lineup, it might be a good idea to sim the amp's power supply rejection for both rails. That will give an idea of hum in relation to distortion products and also what to expect when a loud bass drum plays on one channel and what that produces in the (supposedly silent) other.
Solution that prevents this is described in detail by Rod Elliott and Bonsai in his grounding guide paper.
Along with diode bridge, I use large value resistor (50 -150Ω) because I want to minimize part of signal traveling through power cord ground wires. Power cord ground wires connect through common power bar chassis of audio components and form a parallel ground connection to interconnect cable. So, interconnection ground will be partially shared through power earth wires, close to unshielded mains. Signal distribution is of course determined by Ohm law and impedance at audio frequencies of interconnection cable ground and power earth wires + inserted breaker resistor.
Direct connection of amplifier ground to chassis and power earth is perfect by safety regulations but not best for audio.
Adding to this, would splitting the 6.8k R7 and bypassing the midpoint with a 10u to 47u cap to the negative rail help with PSRR?
Apart from a low power separate supply, I prefer this fix to improve PSR: voltage loss across R40,41 in this case is ~0.5V but PSR improves from ~68 to ~110 dB.
I will see what I can do. Maybe make R21 2.2k.
As for now the lowest frequency is 300kHz. With pot at 50%.
The upper frequency now without input filter is like 900kHz.
The PSRR from the positive is now 58dB with freq 100Hz.
This is really bad.
So, I will have to add supply filters.
This is really bad.
So, I will have to add supply filters.