I checked H727 and it was good, I compared it to H728 and the readings were different. This varistor was bad when I first started and I replaced it with one I built it from information from a tec that use to be at work.
The circuit diagram certainly seems to suggest that three diodes would be required for two of them and two diodes for the other two. I see no reason why that choice of diode would not be fine. My understanding is that these are there to provide a voltage reference that varies with temperature like n x a PN junction and is often used for temperature compensation or as part of a constant current source.
H727 was built from 2 1N4148 diodes epoxied to the original varistor.
Shouldn't H727 be three diodes? And what varistor was the original component mounted to? OK, I see, they call this diode thing a varistor. I presume they're mounted to a heatsink with the output transistors?
I removed H727 and installed one from the left channel board that was very close to H728. Adjusted bias and all was fine. No smoke no oscilliation.
OK, so it was fixed by grabbing the one from the other board. I would try a replacement with three 1N4148s to see if that also works (because I assume you need to replace the one on the other board...)
Lots of
suggestions to replace them with 1N914's or 1N4148's (and these are pretty similar diodes). SO you're clearly on the right track there.
Good to see that you got the bias right
Changed selector knob to FM and feed signal through amp with dummy load. At about 1/2 volume the collector resistors, the 10ohm ones, smoked a little but unit did not start to oscillate.
Yep, smoke is expected. You might want to check the resistance of those resistors and just make sure they're not way different from 10R
I also found out that a ECG605A is suppose to replace the STV-3H.
I don't like them. Firstly because they don't appear to be easily mounted to a heatsink (and I think the originals may be). Secondly, the Vf seems too low. Sounds like 2 PN junctions, not 3.
When you have time, Would you mind looking at the specs and see if you think the 3 1N4148 or the ECG605A diodes will work?
I'd tend to go the three individual diodes.
The right board sounds like it is going much better now. If everything is stable, you might like to:
1) test the DC offset of the output at levels up to 1/2 volume
2) remove the 10R resistors and repeat (If the first test yields acceptable results)
3) continue up to full volume (briefly)
4) try connecting a speaker
Each step involves additional risk, try to be sure before you take each step that everything is stable and within spec. When you remove the 10R resistors, recheck the bias settings. I don't believe they should need tweaking, but best to be sure.
And as you say, then on to the other amplifier.
The advantage of having one channel that is known good is that you can then compare voltages between them. That should make fault finding a case of "spot the difference" (fingers crossed)