DC Offset with Cap Coupled Amp

[dbxdx5]

I'm working on an H.H. Scott 348 that initially had a problem with very unstable bias in one channel. (See this thread.) The bias issue turned out to be the due to someone before me inserting a cap backwards in the tone circuit. The bias in both channels now adjusts properly and is relatively stable.

Next I replaced the two 2,000uF output caps, as well as the 1000uF cap that handles the 70V B3+ for the amp boards. I hadn't checked the DC offset prior to doing this, so I have no point of comparison for what I found next. With an 8 ohm dummy load hooked up, I'm seeing DC spikes at the speaker terminals of more than 300mV in one channel and 200mv in the other. My understanding is that, with the circuit loaded, I shouldn't be seeing any DC through these output caps, let alone spikes that high. Thoughts?

 

[rayma]

By spikes, do you mean while adjusting the balance pot, or otherwise?

 

[Osvaldo de Banfield]

With some kind of amps you will allways have a steady dc offset because a part of driver's current flows though the voice coil. See R18 from the feedback network. But not spikes.

What kind of such spikes do you refer to? (Frequency, amplitude. Duration, etc.).

 

[dbxdx5]

By spikes, do you mean while adjusting the balance pot, or otherwise?

I have two bench DMMs hooked up to the dummy load to observe the DC. One of the meters has a peak setting, and with that enabled I saw over 300mV on one channel. When I switched it to the other channel, the peak DC offset was over 200mV. I didn't touch any of the controls, volume at minimum.

With some kind of amps you will allways have a steady dc offset because a part of driver's current flows though the voice coil. See R18 from the feedback network. But not spikes.

What kind of such spikes do you refer to? (Frequency, amplitude. Duration, etc.).

The DC offset is fluctuating, which isn't unusual. What's odd is that I'm getting spikes/peaks of high DC several times a minute. I had test speakers hooked up briefly, and there was audible intermittent noise, which I assume is what I'm seeing on my meters. I'll hook up my scope to the dummy load and see if I can take a few photos of what it shows.

 

[Osvaldo de Banfield]

Clearly the amp is oscillating at a low frequency rate, usually called motorboating. Check all bypass caps in the power buses. Eventually replace them one by one testing between.

 

[dbxdx5]

Thanks, though I could use some help understanding which caps would be the bypass caps in the power supply.

 

[Osvaldo de Banfield]

C404 seems to be small for such an ampifier. In any case try to disconnect the amp from the pre at C3 to see which of them are causing the trouble.

 

[JMFahey]

With an 8 ohm dummy load hooked up, I'm seeing DC spikes at the speaker terminals of more than 300mV in one channel and 200mv in the other. My understanding is that, with the circuit loaded, I shouldn't be seeing any DC through these output caps, let alone spikes that high. Thoughts?

Please SHOW them

Hook scope there and post screen image.

How else can we see waveform, amplitude, width, frequency, to answer you?

 

[dbxdx5]

C404 seems to be small for such an ampifier. In any case try to disconnect the amp from the pre at C3 to see which of them are causing the trouble.

Ah well, I probably should have upped the capacitance of C404 when I replaced it, but I kept it at 1,000uF.

Looks like C1 is the input from the pre, so I'll try disconnecting there. Thanks Osvaldo.

Please SHOW them

Hook scope there and post screen image.

How else can we see waveform, amplitude, width, frequency, to answer you?

Thanks so much for weighing in Fahey.

The peaks that I've measured occur so quickly that I can't snap a photo of my old analog scope screen fast enough to capture them.

 

[indianajo]

These sudden DC shifts can be caused by bad solder joints, bad resistors capacitors transistors or diodes, etc. Heard as pops.
If you have a scope, then you can trace by looking where the pops or DC shifts first occur. You'll need bigger gain towards the front than near the back.
Then at the stage where the pops start, look for bad connections around there. Push on solder joints with a stick.
If it is in a part ( and not unusual for parts in 1965) a heat gun or circuit cool spray may excite them to misbehave differently.
Certainly look at the power supply outputs with the scope. If the power supply voltages are not steady DC, then nothing else can amplify silently.
HHScott was reputed to use AB military grade carbon comp resistors, which I have found to be pretty reliable. Hammond organs are full of them. The one defect they have is that high value ones (over 100k) can experience moisture leak in and decrease the value.
The usual suspect on 50 year old amps is the electrolytic capacitor. Rubber sealed ones are particularly destined to a bad end. There were some sardine tin models that Hammond used that last nearly forever, but most other brands the seal that keeps the water in is some sort of elastomer. 1965 elastomers were nothing to respect greatly. Electrolytic caps are the ones with a plus & minus, a straight and curved plate. The actual part will have a plus on one end. If removing mark the board with plus so as to not put the replacement in backwards. Parts are available now with service lives up to 10000 hours, which should be fairly permanent. These high rated electrolytic capacitors have quite respectable elastomers as sealant.
But 1965 diodes & transistors were subject to dust & dirt floating around the fab areas, and much more likely to be popping than modern parts. Welds inside diodes & transistors were made by humans in 1965 too, with variable results.
Don't neglect the potentiometer, which can connect or lose connection wiper to track randomly in old age. This includes the volume pot, subject to high wear.
Thanks for posting schematics. I've never seen a used survivor HH Scott model <$150. I've never owned one.
Happy hunting.

 

[dbxdx5]

Truly appreciate the suggestions and intel indianajo. In addition to what Osvaldo suggested, this gives me some useful direction for areas to check.

 

[rayma]

Try some freeze spray, on one part at a time. The input stage is more likely to have the problem.

 

[JMFahey]

The peaks that I've measured occur so quickly that I can't snap a photo of my old analog scope screen fast enough to capture them.

Ok, I had understood that they happened continuously, like tick - tick - tick ..... , every few seconds .... "motorboating".

You may try this:
* first look at screen to check you can see it.
* put camera pointing at screen,
* turn room lights off, so practically only light source is screen trace
* no flash of course
that will make all cameras, some phones might do the same, choose a "long" exposure, say a couple seconds, instead of , say, 1/30th of a second.
It may catch the peak.
I also was a Photographer long time ago and s similar trick was used to catch lightning and fireworks, go figure.

If not, but you can see it, draw it on a piece of paper.

* scope supply rail, at or near Q110 collector, is it there?

* what if you short R13 (3k3) with, say, a 10uF to 100uF cap? Does it disappear? (on screen or audibly)

* what if you short R4? (same thing).

we are trying to find where does it get into te amp.

PS: fwiw I also have an old analog scope

 

[dbxdx5]

Thanks Fahey.
I haven't tried everything that's been suggested yet, but here's what else I've found since my last post.

• Video of what my scope is showing at the speaker output of one channel. This is with an 8 ohm dummy load. 0.1V/div 20m/div. Scott Amp - YouTube
• All of the power supply voltages measure within 1V DC of what's in the schematic, with the except of the B3+ rail. This is 75V vs. 70V. All look stable and free from any oddness on my scope.
• I disconnected the lead going to C1 on the amp board in order to separate the preamp board. (Someone check me on this being the correct place, but I believe it is, since the preamp seems to go out to the volume/loudness and then to C1 negative side.) No noticeable change to the waveform with the lead disconnected.
• I tried shorting R13 and R4 with a 100uF cap. This would be jumper from non-ground side of each resistor to positive side of cap and negative side of cap to chassis ground, correct? If so, the waveform became significantly worse. If this isn't right, let me know how I should be shorting them.
• The voltages at the collectors of Q110 and its counterpart Q10 are ~75V. This sometimes moves up slightly, on the order of .5V, I assume due to fluctuations in the line voltage.

 

[indianajo]

There are 4 inputs from the Riaa & tone control sections to Q1 & Q2. Bass treble, loudness, and balance. Any of those could be source of the jumping. I'd take a scope and look at each, for amplitude of any jumping. Don't know what you are doing with capacitor to R13 & R4. R13 establishes operating point of Q3, I wouldn't monkey with it. Except to push on connections with a stick and see if big popping occurs. A trash speaker on output will alert you to what is going on, if your eyes blink or the pop happens between scope sweeps.
Slow variation of +70 rail at collector of Q110 shouldn't cause popping. It should sag in voltage as the amp puts out more wattage, that is normal. But such sags are below 10 hz usually and inaudible.
Again, locating first stage where the voltage changes suddenly with no input is kind of where you are, since your power supply is okay. Shorting the input is a problem on this amp since it has 4 of them.

 

[dbxdx5]

Ok, I checked the four inputs. Those at C3 and R15 are both showing fluctuating millivolts on my scope and meter. I then checked around the caps for the bass pot and my meter displayed the same thing: voltages spiking to over 200mV.

Probably not coincidentally this is the same problem area that led me to that reversed C3 cap that was causing the bias in one channel to be all over the place. I had checked the voltages at the bass pot then, and found a few volts in spots (see attachment). When I replaced that mis-installed C3 cap with a new one, the bias calmed down and those grossly high voltages were gone, however, I didn't pay much attention beyond that. Something else I had noticed was that pressing on the knob of the pot had initially caused those voltages to spike. I retightened the panel nut for the pot and that seemed to help. At this point I'm thinking either one or more of these .068 caps is bad or there's a mechanical issue with the pot itself.

 

[indianajo]

Film caps are usually pretty reliable, but maybe not in 1963-68. I've had good luck with Mallory and Goodall film caps in hammond 1964-68 organs, and sprague in 1966 wurlitzer organs.
Pots, especially those on the front panel, are high wear items. Hard to buy these days though, clarostat & CTS and all mil-spec 1/4" shaft pots, the factories have been closed & demolished.
Then again, don't forget to push connections in that area with a stick while listening. These were hand soldered.

 

[JMFahey]

Good.
I trust capacitors, specially film types, but pots and any mechanical connections, not that much.

Indianajo: I suggested shorting (AC grounding) those resistors not as a "solution" or "messing" but to check whether they killed the mystery spikes.
It´s a "diagnostic" action

That problem worsened makes me strongly suspect grounding (actually loss/worsening of it) or loss of supply decoupling.

Old high ESR rail decoupling caps come to mind.

 

[dbxdx5]

Yep, old film caps. No markings as far as brand. It would be nice if one or more were the problem, because as noted finding a replacement pot is going to be tough. But I'm not optimistic it's the caps.

 

[Osvaldo de Banfield]

I continue suspicing of motorboating because of a bad cap in the supply or decouplig filter.
Film caps work preferably with higher frequencies.