Randall RG80 popping noise

[pontiacpete]

Good to hear Im not the only one with upload problems.
I'm trying to identify the corresponding parts that you are talking about with the correct schematic. One obvious difference - R45/46 with the cap in between to ground, this amp doesn't have that, it has one resistor 8.2K with a 47pf cap in parallel. So where would I tie it to ground?
Also, this amp has 2 power transistors 2SA1941 and 2SC5198. The next set of transistors moving back from the output are 2SC4793 and 2SA1837. These are the drivers or the pre drivers? I'm not sure. Then there is a 2SC1815 and 2SC1815 facing the opposite direction from those listed above and they are tied to the center rail that the 8.2k resistor is found.
This looks like a really great troubleshooting method that you are showing me and I greatly appreciate it! But I don't want to screw it up and start burning parts.
Do you think it's better to just wait for the upload issue to be resolved? Or do you think that it's not that risky?

[nevetslab]

In the circuit you had posted, which is what I've been going by, this circuit is Quasi-Complimentary, where the outputs and drivers are NPN's. The new circuit has a complimentary output circuit, where the 2SCxxxx parts are the NPN's with their collectors on the (+) supply, and the 2SAxxxx parts have their collectors on the (-) supply. The pre-drivers you describe 2SC1815, both being NPN's, seems odd for them to use those as pre-drivers. Let's wait until the upload problem has been fixed. Then there's no guess work and trails of smoke to contend with.

That 8.2k with 47pF across it....that's 412kHz, too high for the feedback network to the inverting side of the front end. I'll have to see where that is. See ya on the back side of the uploading.

[g1]

This should be the power amp schematic:

[pontiacpete]

That’s it. Thanks!!

[nevetslab]

I see the attachments are working again. This version of the amp is considerably different. Compound NPN/PNP complimentary output stage. Not so commonly used, but one more approach to making a power amp circuit and not doing the same-old, same-old approach. Opening the loop on this to isolate the front end from the back end is still a similar approach. Though, before doing that, I'd take and lift the current limiter coupling diodes D6 & D7 from the bases of Q12 & Q13. Now, if you have a light bulb current limiter for powering thru for the amp, you might want to use it just in case. This removes the current limiter circuit, while we're still searching for the cause of the popping.

The bias xstr here is Q8, and I would expect it is mechanically coupled to the heat sink. The bias is adjusted with R17, where Q8 is basically a diode, but with some voltage adjustment by R17. If you measure the DC voltage across the emitter resistors R24 & R25, it will be in the very low mVDC range. When the popping occurs, I'd expect there to be a transient voltage across these resistors, though you won't be able to see it without a very fast-reading meter, or a differential scope connection to look across the resistors. Most DMM's have a reading rate of 2-5 times a second. Often enough to tell something just happened, but the transient is gone before any accurate reading is made. Still, I'd watch the current readings thru both R15 & R16, which is the voltage gain stage that feeds the output stage. If you see the reading increase thru the output stage emitter resistors R24 & R25, but not thru R15 & R16, then I'd be surprised. The transient is fed back to Q2 thru the feedback network, then amplified by Q6, and passed on to the voltage gain stage via Q9. That's why I'd want to open the loop and make the front end stable so we can look at it to see if the origin is in the front end before the output stage.

Lifting up diodes D6 & D7, and emitter resistors R24 & R25, and current sense input resistors R20 & R23, then there's no DC path thru the upper and lower output xstrs. By grounding the output buss, that takes feedback resistor R12 to 0V, and should make the front end circuit stable DC-wise. At present, the collectors of Q7 and Q9 should be around +0.6V and -0.6V, maybe less since R17 is adjusting the Vbe drop on Q8. My mind is telling me to remove Q12 & Q13, just to be dead certain the output stage is fully out of circuit and the bases of Q12 & Q13 aren't ripped out.

That then leaves you to watch the collectors of Q7 & Q9 for any transient. If you watch Q7 & Q9 collectors with the circuit fully connected as is, you should see the same transient there as you do on the output, since the output stage is just a compound emitter follower, providing all the current to drive a loudspeaker. With the loop open and set up to make the xstr string of Q7, Q8 & Q9, you should still be seeing the same DC voltage at the two collectors (with the output buss grounded.

Any comments or advise, in case my thinking is out to lunch is appreciated.

[pontiacpete]

Hope I did the correctly. No smoke was a good sign. I put in a 40watt bulb in my light bulb current limiter.
With D6 and D7 lifted and measuring across R16 upon power up there's 2.2v and quickly the first pop occurs the voltage drops to 0v. Then it will pop again it will go backup to about 2 v and then down again to 0. Across R25 emitter resistor it ramps up to about 40mv then the pop and then it goes down to about 10mv and then up to sometimes as much as 60mv to 80mv.
On the other side, it's a bit different,, R15 has the same ~2.2v and when the pop comes it drops to 0v, but across R24 it ramps up to 36-40mv and when the pop it drops to 0mv.
I lifted R20, R23 and removed Q12 and 13 and lifted R24 and 25 and grounded R12 on the opposite side of Q2(is that where meant for me to ground this?) and measured voltage at the collectors of Q7 and Q9, These are both seeing about 34v or 35v(not sure I have polarity correct, one of those might be a negative voltage).
Does any of this say anything? I certainly can't tell.
Again thanks so much for taking the time to look at this with me.

[nevetslab]

Hope I did the correctly.

I lifted R20, R23 and removed Q12 and 13 and lifted R24 and 25 and grounded R12 on the opposite side of Q2 (is that where meant for me to ground this?) and measured voltage at the collectors of Q7 and Q9, These are both seeing about 34v or 35v(not sure I have polarity correct, one of those might be a negative voltage).

Does any of this say anything? I certainly can't tell.
Again thanks so much for taking the time to look at this with me.

The load side of R12 is what needs to be grounded, not the Q2 side. Is that what you meant? So, if that's the case, and the collectors of Q7 and Q9 are latched up to the positive rail, I may have missed something is this set-up. I would expect there is sufficient current thru the 2nd stage Q6 to turn on Q9. There should be equal current thru Q7 & Q9 (looking at the voltage across R15 & R16. If they are equal, then I'd expect the voltage at Q7 & Q9's collectors to be about 0.6VDC above and below 0V, where the output buss would be sitting. If Q2 isn't conducting, then Q3 won't turn on, and Q9 won't turn on, allowing it's collector to go north to near that of Q7's collector.

Backing up to the operational circuit, before surgery:

Looking at the operational circuit with the current limiters taken out of circuit, we can see the transient pop isn't coming from either limiter, so that circuit is not the cause. So, with 2.2V across R15 and R16 (before transients) there's 47mA flowing thru that stage. That seems reasonable On the output stage, if I read you correctly, you're seeing 10mV (when the circuit is stable) across R25, and assume same across R24. The transient shows that to increase for the duration of the pop, then drops back down, maybe to 0V. We're seeing a nominal current of 45mA thru the output stage where it's biased, and we're also seeing the current changing considerably from the transients. I'd expect that. That's also why it's difficult to find who's misbehaving when there is a feedback loop around the amp...you see the results everywhere, and not the cause.

You can remove the light bulb current limiter, since nothing abnormal occurred with the current limiter circuit disconnected. If you're saying one side of the output stage isn't conducting (10mV nominal) thru the emitter resistors, then is the output voltage at the speaker connection is NOT at 0VDC? What was that voltage? Trying to understand for sure that both the upper (Q12/Q14) and lower (Q13/Q15) power xstrs are conducting normally. Nominal condition is 0VDC at the output of the amp (+/- low mVDC).

[nevetslab]

I don't know why I didn't comment on the delay-turn-on circuit before. The constant current source for the front end, which turns on Q1 & Q2 is that circuit attached to the base of Q3. If for some reason Q3 doesn't remain turned on, once the RC network R5/C5, at the base of Q5, you would get the amp turning off. If it's randomly cycling, it could be the cause of the transients, though this type of network is generally silent in turning on, as well as turning off.

The circuit associated with Q16 mutes the input to the amp (from the preamp) when the power supply is turned off. It doesn't show just where diode D3 connects, but probably on the secondary of the power xfmr.

[pontiacpete]

The load side of R12 is what needs to be grounded, not the Q2 side. Is that what you meant?

Yes that's how I grounded R12

Put everything back place except for D6 and D7. The monitoring the output without the light limiter there's -.8vdc and that drops to about .35v and then back up to .8v

I made a youtube clip of the my meters across R24 and R25.

[nevetslab]

It looks like there's not any time when the output sits at 0VDC, and the nominal trickle-current is flowing thru the output stage emitter resistors. With this behavior, I'd expect similar behavior on the current flowing thru the voltage gain stage Q7/Q8/Q9, where it's NOT a steady state 45mA current flowing thru R15/R16.

When you had the output stage disconnected, and just the front end powered up, was there steady current flowing thru all the branches (Q1/Q2 running about the same current), Q3 turned on and the same current flowing thru R13/R14, and the same current flowing thru R15/R16? That's the nominal stability we're looking for, before connecting the output stage and closing the feedback loop. Now, with the output stage disconnected, we could jumper the collectors of Q7 & Q9 together, and tie that to the output buss / feedback resistor R12. That closes the loop, making the circuit a voltage amplifier, with the same gain as the complete amp, just can't drive speaker current. It almost looks like LF oscillation...motorboating, with the upper and lower halves of the output stage conducting, but the current not flowing into the mating half. I'm assuming it's doing this without a load, but I don't know, not having asked.

[pontiacpete]

Disconnected the output stage, lifted R20 and R23, R24 and R25, removed Q12 and Q13 and grounded R12.
Across R13 there's a steady 6.2vdc, across R14 there's a fluctuating voltage between 1.2vdc and .93vdc
Across R15 there's .5vdc repeatedly bouncing up to 1.2vdc.
Across R16 there's .36vdc and the same bouncing to 0v.

When I connect the collector of Q7 to Q9's collector to R12, do I disconnect the ground from R12, first?

[nevetslab]

YES, disconnect R12 from ground first.

As you're seeing the collector current of Q6 fluctuating, see if there is similar behavior thru R11 (collector current of Q2). I'm thinking we should be seeing similar current thru R14, though there would be some base current flowing into Q9. I'm suspicious of Q6. With it fluctuating, that is driving the voltage gain string Q7/Q8/Q9 to do the same. I'd change it out, if you have any. If not, you'd need an 80V rated PNP, moderate gain. The 2SA970 is a 120V part, if memory serves, and a low noise part at that, but for checking behavior, you could take liberties, as long as the pin-out is accounted for (base is on one end, not in the middle).

[pontiacpete]

opa! smoke!
I tied the collectors of Q7 and Q9 to the output buss R12 and R16 burned.

This is with a new Q6 2SC970 in place.

Double checking my work now

edit: r12 had a whisker touching a jumper. now it's not smoking.
Across R11 voltages jumps from 6.3v to 0

[nevetslab]

opa! smoke!
I tied the collectors of Q7 and Q9 to the output buss R12 and R16 burned.

This is with a new Q6 2SC970 in place.

Double checking my work now

edit: r12 had a whisker touching a jumper. now it's not smoking.
Across R11 voltages jumps from 6.3v to 0

Ouch! What was that jumper attached to?

So, you're seeing similar behavior thru R11 as we were seeing thru Q6. And, I"m also guessing the same is happening with Q1 (collector resistor R10). That points to the constant current source circuit. What is happening at the base of Q3? Is it also changing, or steady. It should be about 10V less negative than the negative buss voltage. Ahead of the base is the delay-turn-on circuit. If that circuit is cycling, then it will be cycling the front end and everything else. Could be the cap C5 is bad, causing all this....assuming this is the problem making our 2nd & 3rd stages cycling up and down.

In the scheme of things...this temporary connection of Q7/Q9's collectors tying to the output buss feedback resistor is VERY similar to the complete circuit with the output stage connected. The compound NPN/PNP output xstrs are just emitter followers, with Q7 & Q9 feeding their bases, and there being a bias circuit between Q7/Q9's bases to get the output stage close to turning on. I'm guessing that 'whisker' touching R12 was responsible for the R16 overheating as well. Did Q9 survive?

We could step back to leaving the feedback connection out and leave Q7/Q8/Q9 as it was, and grounding R12 again. The search is for the source of this LF oscillation making everything fluctuating.

[pontiacpete]

The jumper is connected to the collector of the 2sa1941. I hope it's ok. I will pull it and test.
R10 -- it's hard to get a reading across it, voltage bounces all over the place. When measuring to ground it's just the 32 to 35 vdc back and forth.
The base of Q3 to ground there's a steady -37v, not exactly less than the neg buss.

Q9 did not survive. I replaced it.
Still have to check the output transistors.