Bias circuits: Is there a link???

Are bias circuits between left & right channels linked in any way?

I have been working on a G-6700 for quite a while, and have observed cases when I was adjusting one channel down and the other would rise. Admittedly this happened in situations when the initial bias in each channel was higher than desired.

I have been measuring bias as the voltage drop across the two 0.47Ω(5) emitter resistors in each channel and the desired bias current is 10 mA, so the ultimate desired voltage is 9.7mv. I am still working on a DBT with a 60W bulb and I realize that the bias increases as the size of the DBT bulb is increased, or I take it off the DBT. So, please take these numbers with a grain of salt.

Starting with the 60W DBT I measures B+ at 34V and Right bias at 36.8. Then I changed the DBT to 75W and B+ went up to 37V and Right bias rose to 53mv, so I swapped back to the 60W DBT. I try to always measure the rail voltages because P/S voltages will change as the line (mains) voltage changes.

For this case I started with rail voltages of ±34.5V, Left bias= 28.9mv and Right bias = 42.4mv.

Adjusted Right bias (VR04) down to 10.0mv and Left bias rose to 43mv.
Adjusted Left bias (VR03) down to 25.4mv and Right bias rose to 12.4mv.
Found a balance at 13.4mv each, but it was not a null.
Adjusted Left bias down to 10.0mv and Right bias rose to 14.2mv.
Adjusted Right bias down to 10.0mv and Left bias rose to 10.2mv.
Adjusted Right bias down to a null at 9.2mv.
Adjusted Left bias down to 4.0mv and Right null rose to 9.6mv.
I did not want to adjust the left any lower.

Now I am hoping that some one can explain how the bias in one channel can affect the bias in the other.

With the DBT in place, you are placing the PSU in the area where it is most sensitive to changes of current drawn, due to the series resistive effect of the bulb. The small increments of current caused by adjusting or changing the BIAS will cause comparatively large changes in rail voltage - these changes in themselves also affecting bias current.

The PSU is effectively throttled back, unable to supply the high current required when you are 'cranking it' at high volume. But also 'so' throttled back that even mA increases in bias pull the voltage down. Remember, there is one helluva non-linear resistor in series with the transformer(s) - so all the voltages will be affected.

In short, other than a quick test there is almost no point in adjusting bias with any accuracy when the DBT is in place. Check it does actually adjust and the range for each power amp is similar, then remove the DBT and adjust the BIAS for real. Because that's what BIAS adjusted with a DBT in place is... not for real.

Oh, and ALWAYS remember to back the BIAS trimmers down to lowest BIAS before removing the DBT for a full power test.

Hyperion said:

With the DBT in place, you are placing the PSU in the area where it is most sensitive to changes of current drawn, due to the series resistive effect of the bulb. The small increments of current caused by adjusting or changing the BIAS will cause comparatively large changes in rail voltage - these changes in themselves also affecting bias current.

The PSU is effectively throttled back, unable to supply the high current required when you are 'cranking it' at high volume. But also 'so' throttled back that even mA increases in bias pull the voltage down. Remember, there is one helluva non-linear resistor in series with the transformer(s) - so all the voltages will be affected.

In short, other than a quick test there is almost no point in adjusting bias with any accuracy when the DBT is in place. Check it does actually adjust and the range for each power amp is similar, then remove the DBT and adjust the BIAS for real. Because that's what BIAS adjusted with a DBT in place is... not for real.

Oh, and ALWAYS remember to back the BIAS trimmers down to lowest BIAS before removing the DBT for a full power test.

Click to expand...

Yes, indeed this is it. Thats pretty much why I don't use a DBT, I use a variac, and a killawatt meter. Adjust things at full line voltage.

Some of the gear I work on wont work on a DBT also.

The other thing that could be happening is that the two SVT diodes are on the same heatsink, and small changes in temp will affect the other channel as well as the diodes are in there for thermal tracking.....

Thanks! I am glad to know that it wasn't magic.

I am only doing this now to find a stable baseline and hope that I don't blow my OP TRs, which is why I am using the DBT. Next phase is to restart my efforts to see if I can identify any oscillations. I'll run through a sequence of input frequencies on a 60W bulb, then 75W, then 135W, then without the DBT.

If I can get that far and still be in one piece I'll set up the DCO & Bias and see if it will play OK.

kevzep said:

The other thing that could be happening is that the two SVT diodes are on the same heatsink, and small changes in temp will affect the other channel as well as the diodes are in there for thermal tracking.....

Click to expand...

The left & right outputs and the SVT diodes are currently mounted to separate heat sinks. I was monitoring the bias with two DMMs and observed that adjusting one would cause an immediate inverse change in the other. So I guess that this falls in line with John's explanation.

ghazzer said:

Thanks! I am glad to know that it wasn't magic.

I am only doing this now to find a stable baseline and hope that I don't blow my OP TRs, which is why I am using the DBT. Next phase is to restart my efforts to see if I can identify any oscillations. I'll run through a sequence of input frequencies on a 60W bulb, then 75W, then 135W, then without the DBT.

If I can get that far and still be in one piece I'll set up the DCO & Bias and see if it will play OK.



The left & right outputs and the SVT diodes are currently mounted to separate heat sinks. I was monitoring the bias with two DMMs and observed that adjusting one would cause an immediate inverse change in the other. So I guess that this falls in line with John's explanation.

Click to expand...

Oh right, I couldn't remember to be honest, but certainly, yes put the scope on it to look for your oscillations...

with a shared power supply and throttled down supply current/ you will see things like you asked about . all you have to do is monitor the supply voltages to see why .
even if its separate supplies and dbt is on main supply it will also happen .

Thanks guys. Now I can move on.

I think I will go back to my original thread about this G-6700, as that could have been the underlying problem from the beginning. If the search for oscillation(s) is fruitful I may finally be able to get it playing for more than just a few minutes.

Besides, I cannot remember reading (here on AK) about anyone searching for oscillations, so the processes may be useful for others.

Well stated John, only caveat I will toss out there is that this particular unit tends to have a sudden unexplained bias rise under mains power after anywheres from 30 seconds to 30 minutes and once this happens it is too quick to save the outputs. I might suggest moving up to a 100w, then a 150 or 200w bulb in the absence of a variac and ammeter.
A 200w bulb would allow 2amps roughly of current capacity which should slow any sudden high current draw yet allow for a better feel for the behavior of the bias circuit. Of course Steve you are welcome to drive by for a visit and use my variac
-Lee

Tracking this further & re-reading my notes, I see that the cross channel tracking decreases significantly as the bias gets lower and the nulls in the Left channel are always lower than the Right channel. Isolating the STV-3H diodes and measuring the resistance I get ∞/25 MΩ in the Left side and ∞/37 MΩ in the Right ch. Is this why Right ch bias is always higher than the Left?

Or maybe there is a problem with the Left ch STV-3H, resulting in a runaway bias as things eventually heat up?

Anyone have any answers, or even a good theory?

I have another pair of STV-3H diodes that I got with an 'extra' pair of f-3058 bds that I bought on flea-bay. One has a broken lead that is VERY short. I dug into the potting material to get enough lead to hold with a grabber. It measures ∞/26MΩ. The other one measures ∞/39 MΩ.

None of these diodes show a reading on the "diode" function of either of my DMMs. Three 1N4148s in series measures ∞/16 MΩ.

Do you think it would be worthwhile to pull the 3058 bds and replace one of them so that I would have a pair with closer readings?

I would test your bias diodes with a power source and a limiting resistor, whose value would deliver the same current through the diode as when it is in circuit. Then measure the voltage across the diode. (Example calculation = resistor value say 1KΩ = 6.5mA, using 9V battery and assuming 2.5V dropped across the diode(s) ) - Look for a matching voltage across 2 diode(s) - one for each channel.

(Then you'll discover it was a low gain driver transistor causing all the problems anyway )

Looking at actual measured data in these bias circuits, I see that the voltage drop across R37 or R38 (10Ω) is typically +0.726V to +0.632V, or 0.094V. (0.094/10) = 0.0094, or ~ 10mA.

Voltage readings either side of the STV-3H diodes is typically 0.632V and -1.123V, or Vd = 1.764V

I have four sets of data readings for each channel (DBT=60W) and all have been in the same ballpark.

Does this sound like what you were describing?

Thanks - - -

Pretty much, I was suggesting testing them out of circuit with a battery and a resistor - for speed, also I would be looking for two diodes whose voltages as close together as possible.

If I have taken these measurements correctly, then the voltage readings across the two diodes currently installed are very close indeed.

1.764V
1.744V
1.755V
1.724V

You could try fitting the diode with the highest voltage drop in the channel with the lowest bias, and vice-versa - see if that balances things up a bit?

I would say that these reading are very close. The slight variance in the voltage drops reported in the last post were most likely due to changes in the mains over the times that the measurements were taken. B+ with the DBT varied from 34V to 38V. I live in a rural area so I always try to remember to measure B+ when I take voltage readings so I'll know how relevant the readings are.

I think that I may pull the OP TR bd in the left channel and replace the ∞/25 MΩ diode with one that measures ∞/39 MΩ. The diode currently installed in the Right ch measures ∞/37 MΩ.

AAARRRRGGGHHHH ! ! !

Just found another loose, intermittent connection. It is the jumper going to the base of the NPN OP TR in the left channel. Most of the measurements I have made over the last couple of weeks are suspect, and will have to be repeated.

The good news is that this is the channel with the STV-3H that only reads 26 MΩ, so I can replace that while the F-3058 bd is out.

Time for an update?
-Lee

Tonight I just finished re-drawing the track layout for the F-3058 bd in TurboCAD. I made all the pads larger for better adhesion to the bd. I now have F4 copper clad bds, so the copper is thicker. Next step is to etch a new set of bds.

I'm having difficulty with consistently reading the resistance (with DMM) of the STV-3H diodes I have. Looks like I need to follow the example of EchoWars, MarkTheFixer, & Hyperion and breadboard a rig to measure the voltage drop across the diodes with a fixed current.

This week I finished etching a new set of boards. Now I must drill the holes and replace the first set with Rev 2.


With guidance from MTF I set up a breadboard and tested a set of STV-3H diodes using a 9vdc battery and 1075Ω of resistance. The drop across the diodes was 1.71V and 1.69V, so I guess these are good to go.

I also have one with a broken lead and I will follow Mark's example to dig it out, solder a thin wire lead to it, then close it up with a JB Weld casing.