manfy wrote:

Is that channel imbalance audible in the speakers?
How far do you have to push the balance slider to get a balanced signal on both channels (according to the multimeter)?

The imbalance is not huge but definitely audible (stereo image shifted to the left). The balance slider needs to be set to +0.5 to compensate.

manfy wrote:

How did you check that? Oxidation is not really easy to see with the naked eye. Did you try any sort of cleaning?

I measured resistance with a multimeter in circuit for both channels. Resistance ranges were lineair (no peaks or dips) when moving the sliders and I noticed no significant difference between channels. 

Yeah, 0.5 on the balance slider sounds a bit too high.

Check that the caps in the tone control filters are balanced. They have a surprisingly high impact on the voltage level. Preferrably check and match them out of circuit. C112 and C114 should be matched to each other and to C212, C214. Same procedure with C115, C116 and C215, C216.

Post the actual values of those caps and I'll run them through a Spice simulator.

Thanks! Here are the values:

C112 9.94 nF

C114 10.17 nF 

C212 10.01 nF

C214 10.27 nF

C115 68.73 nF

C116 67.98 nF

C215 70.65 nF

C216 68.58 nF 

Thanks!
So, I connected a 1kHz sine signal to the input, at 0dBV = 1.0Vrms, 300ohms and measured rms values at following points:
1) Signal-In = output signal generator
2) Filter-Out = cathode of C117
3) Signal-Out = cathode of C121

all voltages in mVrms in order 1/2/3
- ideal cap values as per schematics: 994m/4.89m/952m
- actal values L channel: 994m/4.87m/947m
- actual values R channel: 994m/4.88m/948m

For the fun of it, I changed C112 to 5nF and C114 to 20nF:
- 5n/20n highpass: 994m/3.67m/664mV

As you can see this change lowered to output by more than 25%, which constitutes the proof of concept that these filter caps have a significant impact on the signal level. The change of caps will of course change the cut-off frequencies, hence the overall characteristics of the tone control, which may or may not be desirable.

So, I'd say we forget about fine-tuning the caps for now and look at the output transistors TR101/201.
Since the output signal is fed back to the filter network, this transistor will have an impact on the level of the output signal, thus different hfe in left and right channel will result in imbalance.
Solder back your tone filter caps and check the hfe of both transistors. Since I don't trust most of those "component guessers" and their guessed values, just write those values down for reference and swap TR101 with 201 and viceversa. Now make the very same measurement as you did in post #?? (above from June 16th) and compare with the results from that day:

-) if the R channel shows higher output levels now, one of your transistors is the culprit
-) if the problem remains unchanged, the imbalance stems from some passive components in the tone control circuit. In this case I'd start measuring the DC working points around the filter and TR101/201 amplifier stage.

[edit] PS: Just out of curiosity, what multimeter or cap-meter did you use to measure the caps? They seem oddly consistent. If you used a professional cap-meter, the measuring frequency and ESR values would be interesting and could be enlightening!

Are C117/C217 new?
If not, I suggest you replace them. Ideally also C111/C211/C121/C221.
If they are tantals originally, I suggest fitting tantals again.

Martin

👍 (this thumbsup is in reference to Martin's post!)

Particularly C111/C211 seem critical!
I just simulated different values and ESR from 0 to 3500ohms for all those caps. C117 and C121 showed little change in the signal level (I suppose because of the amplifier feedback to the filter network, which "balances itself" and adjusts the gain level accordingly).
But minor changes in capacitance or ESR of C111 showed a direct impact on the output signal level!

PS: Just a disclaimer: My spice simulations are just simulations and as such they are based on an "ideal world". Real world conditions may be quite different because of parasitic C/R/L values present in every circuit and component and transient effects in a real world environment!

This is awesome, thanks so much!

manfy wrote:

So, I'd say we forget about fine-tuning the caps for now and look at the output transistors TR101/201.

I already tried replacing TR101/201 with matched transistors but that did not make a difference.  

manfy wrote:

-) if the problem remains unchanged, the imbalance stems from some passive components in the tone control circuit. In this case I'd start measuring the DC working points around the filter and TR101/201 amplifier stage.

Yes, I'll do some DC measurements tomorrow and report back.

manfy wrote:

PS: Just out of curiosity, what multimeter or cap-meter did you use to measure the caps? They seem oddly consistent. If you used a professional cap-meter, the measuring frequency and ESR values would be interesting and could be enlightening!

I used a XJW01 LCR meter at 1KHz (not professional by any means but accuracy should be decent). ESR measures around 50 ohm for C112/114/212/214 and 8ohm for C115/116/215/216.  

Dillen wrote:

Are C117/C217 new?
If not, I suggest you replace them. Ideally also C111/C211/C121/C221.

Thanks Martin. Yes, C110/210, C111/211, C117/217, C120/220, C121/221, C122/222 are all new.  

Dillen wrote:

Are C117/C217 new?
If not, I suggest you replace them. Ideally also C111/C211/C121/C221.

Thanks Martin. Yes, C110/210, C111/211, C117/217, C120/220, C121/221, C122/222 are all new.  

krais wrote:

Thanks Martin. Yes, C110/210, C111/211, C117/217, C120/220, C121/221, C122/222 are all new.  

Yes, but since you have a real LCR meter, you really should check C111 and C211 carefully. E-caps usually have a +/-10% or +/-20% tolerance, and that includes brandnew ones. At least part of your channel imbalance could originate from these 2 caps.

krais wrote:

I used a XJW01 LCR meter at 1KHz (not professional by any means but accuracy should be decent).

Not too shabby! That seems to be a real LCR meter that measures with a sine wave signal and not one of those $10.- testers that uses tricks & obscure firmware to guess the component and value.
Looking at the specs that go around on the net, this tool might actually use the dedicated LCR chipset ES51919/51920 by Cyrustek. If so, the basic capabilities of the meter is good! I have an Appa703 and a DerEE DE5000. Both use that chip and both devices have a good reputation.

The only real problem with those low-cost no-name devices from China is that you don't know the quality standards the maker applied. You rarely get a calibration certificate. It's a bit of hit and miss. If you're lucky, you might get a meter with same quality as a $400 branded meter.