• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

Sony TA-F3A Bias adjustment

Interesting.. Might it have been a poor welding/false contact after all?

You mentioned the fuse-resistors did not blow, and Q707 always blows first. That would suggest a false contact some ware between R715* and Q707. Or across R717*. Narrows it down quite a bit. I'd analyze the hell out of that title part of the PCB.

*not 100% sure I read the numbers correctly
 
Build a DBT and try the 0.22R without speakers first.

I wouldn't power the unit without a DBT after all this problems.
 
OK, I did put the amp on DBT and got the resistors to 10R (did not have the courage to go to 0.22R yet) and powered the amp. No fireworks! I ran it for a bit and got enough courage to remove the DBT, adjusted bias to 6mV and added signal to the amp aux.
Now, bias behaves, I think, strange:
- the bias on the channel I had no issues with is increasing slowly to 24mV and is stable there
- the bias on the channel with issues is going up and down, from 5-6mV to 1V, almost keeping the pace with the music!
The Output transistors seem fine, not getting too hot after running the amp for 10 min at 1/3 of the volume, maybe close to 30C (86F).
I checked the PCB again to make sure it is OK, but I cannot see any defects. I measured the traces with my ohmmeter and everything checks.
Any idea what can cause the bias to bounce around like this?
 
I normally remove the final output power pair and bring up the system using a DBT. The amp circuits before the power transistors should behave about the same with and without the power transistors, **most of the cases**. Check all the voltage against the schematic. In your case, I will make sure it's about .6V across both R712 and R715. This .6V is about the proper level of bias for the output transistor.

You also mentioned that both Q704 and Q705 are quite warm. This might be a sign of over-biasing, leading to thermal run-away of the power transistor. What's the position of the bias pot? You might want to "zero" it all out first to cut of the bias initially. Your new power transistor might have thermal tracking problem with this circuit.

You also mentioned heat shrinking some diodes. Is it the bias tracking one (D751 or 701)? Are they mounted on the heatsink? Heat shrinking them is not a good idea as they are normally used to sense temperature around them.

Last but not least, DBT is a must for this situation.

Hope this help.
 
I would also check to make sure the protection circuit is in fact working properly just in case to protect the beloved speakers....
 
Yes, the position on the bias pot is close to zero, and I was able to start from zero with the adjustment. Everything looks fine until I apply signal to the channel, that is when the bias voltage starts dancing.
I have replaced both bias tracking diodes and both are the same, 2 x 1N4148 in shrink tubes, pressed down on the heatsink by thermal conductive tape. The other channel works fine with the same type of diode, so I don't think is the diode at fault here. Also, the bias dancing starts right away, not after the temperature of the Output transistors rises.
 
When you said "Everything looks fine until I apply signal to the channel, that is when the bias voltage starts dancing." Did you mean you measure the quiescent DC bias voltage while applying input to the amp? If so, this is normal as the instant DC voltage component changes with the AC input. The quiescent DC bias voltage should be measured with no input signal.
 
That is what I thought too! What through me off is the fact that this is not happening at all on the other channel! On the other channel it is just raising slowly to 22-24mV.
 
How about try to measure the DC output voltage again on both channels without input to see if they behave the same.

It's just me, I would not put heat shrink around the bias diode as the thermal tracking depends on the thermal resistance between the diodes and the heatsink too. Was there thermal paste to thermally connect the diodes to the heatsink in the original design?
 
Already did that, and with no input they behave the same!
The original diode was 1 piece only, and no thermal paste, or if it was there before it was gone by the time I got the amp. I can remove the shrink and use it only for the terminals. The thermal conductive tape is thick and soft so I think it does a decent job because it is pushing the diodes on the heatsink by a metal clamp that holds the output transistors also.
 
I don't think a piece of tape or a layer o heat shrink tube will make the bias so unstable.

Anyway, check it without signal. "bias voltage" is an idle state adjustment, you are setting a small current flowing at the output transistors. The current raises when you apply a signal.

Can you detail where are you measuring that voltage, and what values are you setting?

I've seen bias "dancing" due to some transistor in the bias network. Or, a cold solder in that board.

Also had a very hard to find problem in a Sansui amp, high bias, caused by leaky "black flag" capacitors. So don't discard a leaky capacitor, even if it's film or ceramic. If in doubt, replace any of them you suspect, or swap it to the other channel.

But I'd do all that after you confirmed the bias is not stable without signal. Let the amp warm up for 15' monitoring the bias, let another 15', check again.
 
I have measured the bias voltage across the 0.22R resistors:
- channel with no issues, still has all original transistors and resistors, new caps and bias diode only, I measured across R766 & R767 - I adjusted the bias for 0.06V with no signal applied
- channel with issues, only Q701, Q702, C709, and RV701 are original, everything else is new, I measured across R716 & R717 - I adjusted the bias also to 0.06V with no signal applied
I left the voltmeters attached to R717 and R767 to can compare, I applied signal and the voltmeter attached to the good channel, R767, is not showing any dynamic ("dancing") change, it just increases to 0.22-0.24V and holds it there stable.
The other voltmeter attached to R717 is the one showing the variations in voltage. I thought that is normal until I compared it with the readings from the good channel!
I resoldered everything on that portion of the board, but the board is old, so I tested all the traces and all checks fine. I will replace C709, it is a Polypropylene cap.
 
Are the two meters of the same make and model? How about swapping the meters to rule out measurement issues. See if the symptom follows the meter or the resistor.
 
No, meters are of different brands, Fluke and Mastercraft. I also did the swap to make sure I am not hallucinating, but the readings are not following the meters.
 
Aux is connected to my iPhone running a Flac player. No, I did not check the output with a scope yet. Should I look for something specific on the scope? Also, I am still running the channel that was defective on 10R for R716 & R717.
The preamp section still has all original components, nothing was replaced there.
 
The 10 ohm resistor as compared to 0.22ohm will make quite a difference. Assuming the output power level is about the same on both channels, the output current will be about the same too. The voltage developed across the 10 ohm will be 45 times higher than the .22 ohm. So your observation sounds normal to me.

I would suggested to, with no input and zero down the volume pot, compare both output using the "DC" input with the highest sensitivity on scope to see if the bad channel output level drifts more than the good channel. This will give you some insight if there are other steady state problem.
 
0.06V (60mV) across one 0.22 ohm resistor is 270mA.

0.06V across two 0.22 ohm resistors is 136 mA.

Both values a bit high for an average amp. I don't know your unit but I'd set it 10 times lower than that. Try 20mA both channels and let it idle for 20 or 30 minutes and readjust.

Then if you have an oscilloscope check a sine wave and square waves at the output. No load by now. Check for anomalies. If the signals perform OK, try wih a load.

You could monitor the voltage at the 0.22 resistors while playing a sine wave. But measuring while playing music doesn't make sense. No objective measurement that you can get from that.

Try to handle the Bias current as "mA". Thinking the bias as Voltage can be confusing. You are setting a current flow. Voltage will be different depending of the Resistors values.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom