Menahem Yachad wrote:

... all original BC5xx, matched where necessary... 

Did you also match the pair in the differential amplifier? And what type did you use here?

Martin

Just read it again and realized you have the same offset in both channels.
50mV is not the end of the world, but please tell me that you did not fit low-ESR caps in the output stages and as power supply filter caps?
And have you checked for oscillation with a good scope?
What happens if you ground the inputs to the output stages (TP107/207)? Still hum?

Martin

I've done further checking.

The only TP which is not according to the SM, is TP111, 211 - the Base of the Feedback TR113/213 in the Diff Amplifier = 64mV. That's out of spec - 0+/- 50mV.

So, I'm focused on the Diff Amp right now.

In the Diff Amp, I used matched BC550C, which I think may be causing oscillation (The SM's specs for these stated BC237C and BC547C, but the other specs were BC182B). I'll check that tomorrow.

I can't think of a better scope than my Tek 2465B!

I'm planning on replacing the Diff Amp with BC546B, with a lower HFe of 300.

I'll also change C138/238, according to the SM - even though it's new, perhaps the output DC messed it.

I don't use super-low ESR caps in amplifiers anymore. The regular long-life 105°C with a slightly-higher ESR is my standard choice.

If that doesn't work, I'll ground TP107/207, and see what happens.

 

BC546B would be a good choice for the differential amplifier (or BC549B if you can get them - they are basically the same, only with better noise specs).
It's a good handrule to never use C-rated transistors in B&Os output stages. Things get too eager.

Normally you don't have to match the two transistors scientifically. In most cases it'll be fine if you can just fit two transistors from
the same production batch. They are often placed physically close together to experience the same operating conditions, temperature etc. for the same reason.

Martin 

OK, so here's what I did - for the time being I'm working on the LEFT channel only (in order to determine results, before later duplicating any changes on the RIGHT channel as well)

I removed the Diff Amp 2x BC550C HFe 467/465 

I installed 2x BC546B HFe 300/300 - same batch, same DoM.

I removed the C138 47uF, and replaced it with a new one.

TP111 went UP from ~50mV, to 94mV.

I was really disappointed!

 

Then I removed C132 1uF, which connects the PreAmp to the Output Amp.

No change at TP111.

 

Then I connected the junction of R153/C232 (C232 is still uninstalled) to PCB ground.

TP111 dropped to 33mV.

 

I'm sure that you're about to teach me something significant here!

If the offset got worse, try swapping the two transistors and see, if it gets better.

Martin

Did you replace the MPSH54 transistors Menahem? I seem to remember I had trouble with those until I could get hold of NOS ones. Offset was too high (but no hum).

The Beomaster 4400 produces a switch-on thump but nothing when switched off, and it’s perfectly silent.

The one I use at home (what a lovely sound!) has been in service for 8 years and that’s several hours of daily use.

Why did you put Wimas everywhere?

chartz wrote:

Did you replace the MPSH54 transistors Menahem? I seem to remember I had trouble with those until I could get hold of NOS ones. Offset was too high (but no hum).

Why did you put Wimas everywhere?

Yes, I replaced them with brand-new OnSemi MPSA55 from Mouser - I had no option - all the TR's were toast.

I also have 2SA970's in stock - they are listed as a substitute - I'll try with those

The Wima's are not "everywhere" - I use them ONLY in the coupling positions in the signal path, and only 4.7uF and smaller. Much improved signal quality, over an electrolytic.

 

Dillen wrote:

If the offset got worse, try swapping the two transistors and see, if it gets better.

Thanks Martin - I'll try that tomorrow.

But in addition, can you explain the significance of the massive drop in the Offset voltage at the Feedback side (94mV to 33mV), and also Why is there still 33mV, and where should I look to start eliminating that?