Harman Kardon PM640 Amplifier Repair

[sean0118]

I'm halfway through repairing this, I'm a bit stuck so thought I would make a post while I do some more research. I'm not sure if there is a fault or not. 

Originally I was just going to do a recap, upgrade the trimpots and replace R421, R422, R423 and R424 because they were cracking from too much heat. But after testing I found one of the channels was much quieter than the other and breaking up. I'm not sure which channel it was, I don't really want to plug it into my speakers again in case it blows them up. I thought it might just be the volume / balance control trimpots that need cleaning but after checking voltages against the schematic I'm not so sure.

1 - R421, R422, R423 and R424 were damaged from too much heat (outer layer was cracking off some of them). I replaced them with new 0.25W resistors, but I found R42 and R422 dissipate 0.22W which seems a bit close. I think part of the problem is the A and B rails are unregulated, they are meant to be +/-35V but mine are closer to +/-37V. I measured the line voltage as 245V AC and the voltage selection switch is set to 240V AC. It seems like it's a design issue or the original resistors were poor quality? Maybe I should replace them with 0.5W versions?

2 -  I've trimmed the bis voltage to about 33mV as outlined in the service manual (across R447/R445 and R446/R448). But the voltage is not evenly distributed across the current sense resistors, giving me about -7mV DC on the output to the speakers. I've attached a schematic with the voltages annotated, it's the emitter voltages of Q427, Q429, Q428 and Q430 I'm most concerned about. But the base voltages for Q401, Q403, Q402 and Q404 are far from the nominal values listed, and the current sink on their emitters is off too.

Does it look like there's a fault? Or is it just that the A and B rails are above nominal? I expected to find differences in voltages between the L/R channels, but they seem fairly similar.


edit: I was going to attach the service manual but the file size is too big, it's available here:
https://www.hifiengine.com/manual_library/harman-kardon/pm640.shtml

[bob91343]

7 mV on the speakers is trivial.  Close enough to zero, although one would think it should be possible to move it through zero.  It will drift as the amplifier warms up.

[sean0118]

Thanks, I guess it is probably mostly working then. I was thinking the current through the push / pull output transistors wasn't balanced, but looking at it again they are both in the 60-70mA range.

What makes the output voltage 7mV rather than 0mv? Is it the ratio of the 0.22 ohm resistors (R445/R447 etc)? They are integrated into a dual resistor package, but I pulled one and both internal resistors seems fairly well matched (~0.226ohm).

[magic]

I haven't read your post nor looked at the schematic, but input offset is typically due to matching of the input pair transistors (including their temperature), matching of their bias currents and matching of source resistances driving their bases (unless they are JFET). The last factor may be the biggest in BJT input amplifiers and it may vary with signal source, if the input isn't AC coupled.

Any offset in the output stage is suppressed by full open loop gain, by factor like 10000x or more.

[sean0118]

ah, so the base voltages on the input stages need to be matched for the output to be 0mV?


Bonus question, they used film for every 0.1uF capacitor except for C420 (which is electrolytic). Is there a reason why they did this? I was thinking it might be to dampen oscillations, but it has so much resistance in series and parallel I don't see how it could oscillate?

[magic]

That's the goal. In practice, it may take a few mV difference to zero the output - same thing as input offset voltage in IC opamps.

The reason why both bases are below ground is because of base current flowing through that 56kΩ resistor at IN+ and whatever impedance at IN-, which should also be close to 56kΩ by design.

As for the input stage current source, it's the current which matters. This is measured by dividing the voltage across R413 or R415 and comparing with the reference value.

[sean0118]

Thanks, that makes sense, it never occurred to me how much audio amps have in common with OpAmps.