Sony TA-F6B DC Bias problem

I'm working on a Sony TA-F6B, previously the power supply gone bad. After fixing it, everything looks OK and sounds nice and clean on both channels.
Tried to set the DC bias and here comes the problem: measuring on the test points, I got only 0.5 mV and 0.2 mV instead of 22mV no matter how I adjust the pots.
All the elkos are replaced on the amp board, beside that everything is original.
If I measure DC voltage on the speaker ouputs, I can measure 10-200 mVs and it is changing when I set the DC bias pots. Right channel's DC voltage is stable, on the other its kinda drifting down in .1 mV steps.
The CX171s has the +-8.8V voltage, but on the output pin 8,5 there is only around 19V instead of 21V.(if i remember correctly)
Measured both channels 2SK58 diff.amp, left channel's S1-S2 voltage is 0.9V, right channel's S1-S2 voltage ~1.5V.

I'll measure later all the resistors/diodes, but maybe someone has any good tip what would be the problem.
Joined 2007
Paid Member
If both channels are behaving the same I would suspect either a problem in the way you are measuring or perhaps a problem with the main rails from pulse locked supply being low.

Are you measuring directly across the 0.22 ohm resistors?
Yes, I measured on the testpoints which are just 2 pins from the 0.22 ohm resistors, also I tried to measure directly on the resistor, but it's the same voltage.
All 0.22 ohm resistors are good.

About the main rails: good point, I measured +- 48.1 but I thought it shouldn't be a big problem. I set the plps voltage to ~98V on bench and after soldered back main rail wires and screwed together the power supply box, I realized that there is 1V drop on load on each rail.
This 1V drop on the rails could cause this?
I inspected closely the board, and found a few 4.7k resistors with cracked shells. The interesting part is that these resistors are on the same place on both channel. These are 0.25W resistors, I'll replace them tomorrow with 0.4W(i don't have in 0.25W right now).

Screenshot 2023-05-20 at 22.16.56.png
Joined 2007
Paid Member
Those resistors feed what will be a shunt regulator (like a Zener) within the chip. If the rails (the 8.8 volts) are correct then the the resistors are probably OK. It's not uncommon to see the glaze cracked on hot running parts like these.

Wattage needs to be:

49-8.8 gives 40.2 volts dropped across the resistor. Power dissipated is (40.2*40.2)/4700 which is 0.34 watts. A 1 watt metal film would be a suitable choice for replacement. I can't find the originals in the parts list but typically they would be metal oxide which are good for high temperature running.

If the originals read OK when isolated they will be fine.
Joined 2007
Paid Member
1/4 watt is under rated but tht uncommon to see in commercial stuff.

I really don;t know whatto suggest tbh. Given it works and sounds good and given both channels are behaving the same is a bit odd.

Do the output transistors get warm at all? They shouldn't if no bias current is flowing buts lets check it that way as a rough test. Try it with the preset at each end.
I don't feel much warm when setting to each end. In the meantime, I measured a few points and changed those 4.7k resistors, because the coating cracked off as I touched them. Something not ok or I'm mistaking something, because I remember(maybe I was tired) the CX171 rails was the same, but no. The right channel 8.8V rails are ok, also the +-49V mains are bang on now.
The other amateur question: the DC bias should be set without dummy load on the outputs, right?
Because I hooked up an 8 ohm dummy load to the output to set the DC Balance to 0V. When setting it I accidentally turned the Bias pot(other meter was on the bias testpoints) and suddenly it changed from 1.6mV to 16mV and just for curiosity I set it to 22mV. After that I tried to lower the DC balance voltage, but cannot go under 700mV.
But if I try to set the DC balance to 0V (within just 10-30 mV) I got the lowest reading on the output when the bias pot is set to measure 5-6mV on the tespoints(with load).
Tbh I'm clueless given that everything sounds ok, but I'm afraid to leave it like this. If I have time, I'm going to measure every components on both channels(already desoldered and cleaned the bias/balance pots, they are perfect).

Screenshot 2023-05-21 at 15.25.48.png
Joined 2007
Paid Member
Yes, you set the bias current with no load attached. This is because any DC offset will cause a current to flow and that will show in one or other of the 0.22 ohms (depending on the offset polarity).

If you can not set the DC offset to zero then that is the thing to look at first.

The other thing I wondered was if there was still an issue with the PSU and perhaps there was a lot of noise or high frequency ripple present and that was doing strange things.
“Yes, you set the bias current with no load attached. “
Then I did it right at first. I can set the DC offset to almost zero(10-30mV the lowest point). If it’s set to that value, then without load the bias is not changing from 0.6-1mV, no matter where I set the pots.

The PSU is hell of an interesting part, every cap replaced, a few burnt resistors and if I remember right I replaced all of these transistors at the regulator part:
Q601-Q613 -> 2SC2023R replaced with a pair of 2SC2023 with the lowest same hFE (45)
Q602-Q604 -> 2SC1810 replaced with BUX85G
Q603 -> 2SA911 replaced with ZTX758P

The rest of the transistors and parts are ok, desoldered all and measured one by one.

Anyway, it’s worth a shot to measure it with scope, I’ll get one.
Joined 2007
Paid Member
It sounds like you've done a good job with the PSU. They are very fussy on the transistor types used.

Have any of the transistors in the output stage been replaced? although it would have to have been for both channels as both are not adjusting correctly.

Not sure if this would prove anything or not but I wonder if measuring the voltage between pins 5 and 8 first with the preset at one end and then the other would prove anything. I see pin 9 is shown as connected on only one channel. Odd or could be an error.

If you try that be very careful, one slip and it all goes bang.
The amp board apart from the caps looks like stock original, the output transistors too, so I think it wasn’t replaced.

I’ll measure tomorrow between 5&8, today I only measured between gnd and 5,8. Both side is around +-20-21v, not really changing for setting the preset(just a little 100mV fluctuation, which could me measuring error)

Good point, on the right channel pin 9 and 5 connected, on the left 9 is alone. As I see on the CX171 block schematic, pin 9 is unused. As I see on the amp board I suspect there wasn’t enough space to layout the trace from pin 5 to q353, so they use it creatively :)
Mooly, I made a small mistake :) I just realized, that I soldered a wrong resistor to the LED replacement for the VU meter light. So I think the soft start relay never switched...
Today I made a rough calculation to the replacement lights: 1W 100 ohm resistors parallel with the main light rails -> in each meter 2 LED in parallel+ series 0.6W 100 Ohm resistor. The LEDs got ~19-20mA.
From another topic I remembered thath the relay coil resistance is 1622 ohm(24V), and the original lamps was 6.3 70mA so I made the calculation based on that.
The problem is, that the 150 ohm 7W resistor starting to get pretty hot after a few minutes, and the speaker protection relay is taking much more time to click(around 10 sec). Before that it was like 4-5 sec.

I don't think this has to be do something with the bias problem, because I measured it while adjusting the preset and the problem still persist.
Did I made a silly error during the calculation?
Joined 2007
Paid Member
The 150 ohm will get hot. It has to given the voltages involved.

Some variables are whether the relay is intended to run at 24 volts or whether it is slightly under run.

Another variable is the volt drop of your LED replacements. Working to the values you have (1622 ohm relay resistance) it works out like this.

The bulb must have had a hot resistance of about 46 ohms. The 150 ohm will dissipate around 3 watts which will make it hot.

As long as you have approx 20v or more across the relay then it should pull in normally.

Screenshot 2023-05-26 114402.png
Thanks for confirming, then it should be ok. Just the resistor seemed to be really hot(you know, when starting to smell:) ) and I scared that I messed up something, because just 2.8-3W heat should be dissipating and it felt like more.

Anyway, I'm still chasing the bias problem, just started to measure every component. I had a thought that maybe both CX171 gone bad, but if that would be the case, I don't think it should sound/work at all nicely.


  • Screenshot 2023-05-26 at 12.55.17.png
    Screenshot 2023-05-26 at 12.55.17.png
    228.2 KB · Views: 40
Joined 2007
Paid Member
If the voltage across the 150 ohm is as both our diagrams show (about 21 volts) then all is well. 3 watts is more than you think and will heat the resistor significantly.

I think you said both channels have the bias issue and that makes a duff CX171 very unlikely. Does pin 2 of the chip vary correctly as you turn the pot. It should go from zero to -8v or so.
I'll measure the voltage on 150 to make it sure.

Couldn't measure voltage on pin 2 , because I started to replace a few propylen/mylar caps which looks like original. But.. I had an idea and measured resistance between pin 2 and -8v rail. It stays on 3.5 kOhm, no matter how I set the preset(just on one end showed 0).
So I thought what if something shorting out the bias pot? Desoldered C303 33n mylar cap and voila, the pot resistance changes between 0 and somewhere 10k. Ok, what about the other channel? Desoldered the same cap and boom, the preset is changing here too, from 0 to 10kOhm.
Soldered 2 new 33nF MKT cap as replacement. I hope it solves finally the issue and before powering up I'm started to replace the remaining mylar caps.
What are the odds that both cap gone bad and that cause the problem?