[till]

I just built a CHA47, just a basic circuit with crossfeed, no volume control yet. Power is supplied by one 9 V battery, the rails are split by a TLE2426 and a BUF634. The power supply splits the voltage quite evenly, like -4.13 V vs. 4.17 V.
There is no sound on one channel and without a source I get a DC offset of about 650 mV, in contrast to 30 mV on the channel that is working. I know that this is a bit too much, too (probably caused by the slightly inequal power rails). Well, I just measured again with OPA2132s instead of OPA2604s, and I get 3 mV on the good channel and a whopping 3300 mV on the other!
I checked all resistors with a meter, heated all solder joints and checked all connections... no luck finding the mistake so far. Ideas, anyone? What can cause this large offset?

 

[Pars]

Well... I would first check to make sure you have the appropriate voltages on the power pins (V+ and V-) on both opamps. Since you a) get no sound from one channel, b) get excessive offset from that channel (the other channel at 30mV is pretty marginal, I personally would not settle for that one either, but that might just be me), and c) have swapped opamps with no change (at least for the better), you can pretty much rule the opamps being bad as the problem.

Assuming you socketed the opamps, I would pull both, and using your trusty DMM, ohm out pins to ground, and pin to pin, checking between the two opamps (I am assuming you used the CHA47 design where each channel uses both halves of the same opamp) and the schematic to correlate what you measure to what you think you should measure. For instance, pin 2 on most dual opamps is -in... this usually has an input resistor of something like 100K to ground. Measuring from pin 2 to ground without the opamp in, you would expect to see 100K. And so on. Since you have no sound in one channel, even after going over the solder joints, there is some sort of wiring problem.

And I don't think your offset problems are caused by the very slight rail difference you are seeing. CMoys built with the voltage divider resistors can have 2V or more difference on V+ and V- when plugged in, yet still have acceptable offset.

hth, and good luck,

Chris

 

[till]

Thank you, Chris. Did that... the voltages on pins 4 and 8 are OK. I pulled the opamps, and the resistances between all pins and to ground do match on both channels.
To avoid the large DC offset on the good channel, I now only do tests with OPA2132s (with them, I get 3 mV). On the bad channel, there is now an offset of 100 mV. Strange. But I'll continue testing despite this thing makes me feel stupid

I reckon that there must be a bad connection somewhere because the DC offset on the bad channel decreased after I heated the connections again. OTOH, a bad connection would result in a larger-than-expected resistance, right? And I don't see that anywhere. I have a strong feeling that this may be a dumb question, but could the offset be caused by a damaged input cap? I cannot rule out that I applied too much heat. If all connections prove to be OK, maybe switching the caps will give the answer.

 

[till]

The problem must have been a cold joint that resisted all of my re-heatings. After I switched the input capacitors to no avail, I just took the bad channel apart and rebuilt it - working now.