1176 Compressor - HF oscillation in transistor stage - help?

[tired eyes]

Hoping to get some ideas here, this one has me stalled.

I'm dealing with a Universal Audio reissue 1176LN compressor - full schematic here: http://mnats.net/files/MNATS1176REVD.pdf (it's the DIY kit schematic, but virtually identical to this reissue).

There is a slight high frequency noise at the output which varies with the output gain level. I traced it to the 3-transistor gain stage in the attachment, where a high frequency (in the MHz) oscillation of about 15mV appears at the collector of Q2, and continues forward through the signal path. The oscillation is present with or without input signal (it is clearly superimposed on the waveform at lower signal levels). The end result is about 50mV of output noise at full gain.

The base of Q2 has no oscillation. With Q2 removed, Q3 does not oscillate. With Q3 removed Q2 is not biased so obviously does nothing. (this unit has sockets for the transistors so swapping is just easy to do)
These are all supposedly hand picked components, 1% resistors, transistors matched for HFE, etc. so I don't know what to suspect. I've checked a few resistors and caps, swapped some transistors, all good so far. I don't want to mess with the original circuit if at all possible, but this noise has to go.

Any suggestions?

[jrfrond]

Let's see..... R9 is negative feedback. Is it open? Has it changed value? This could destabilize the circuit. C2, the RF snubber cap at the base of Q2... has THAT opened? Have you tried a 27pF cap at the base of Q3 to see if it removes the RF? How about the transistor sockets themselves? Have you cleaned/treated them?

Just for openers.....

[Enzo]

SO this is a kit you built? This is an important question. WHen a piece of gear works for a long time then fails, we can make certain assumptions. Things like all teh parts are correct and installed correctly. WHen we build something and it doesn;t work, it is the same as something that NEVER worked. And we cannot then assume all the parts are corect, or installed correctly, or all soldered. We cannot overlook possible 470 ohm resistor where 470k belongs, and so on.

Don't assume all problems are due to bad parts, most problems are not. Bad or missing connections are a lot more common.

You have some noise signal at the transistors. Is that signal on the power supply rail? Is that 30v really at about 30v? And is it clean?

I see a couple trim pots, do they have any effect?

[Brian Swatton]

I have known 78 series regulators to produce ripple at MHz, app notes usually call for a 0.1uF at the output, as close as possible to the device. I note there only seems to be fairly large electrolytics in the schematic, which probably won't be very effective at RF. I'd be inclined to try adding one and see.

[tired eyes]

JR - thanks for the solid opener. R9 and C2 tested good. I will try the cap on Q3 base when I get home, as well as clean the sockets.

To clarify, this is NOT a DIY build, it is a reissue that I'm fixing for someone else. I don't know the whole history of it, but it did have a leaking filter cap on the positive side bleeding 60hz into it. With that fixed, I'm not getting any visible (or audible) PS ripple.

The solder joints on this unit look like jewellery, nonetheless I reflowed around the circuit in question. + rail is 30.2V and clean. Trim pots have no effect.

Thanks everyone for the tips.

[km6xz]

It is not likely an oscillation, which requires positive feedback to sustain itself. To do that the noise present needs to be amplified a great deal to oscillate, doing so, the amplitude of the oscillation will be a lot higher than 15 mV. You may have a parasitic which would be a different mechanism of generation. I would first suspect probe interaction and pickup of RF. What type of noise are you hearing in the output? Before replacing parts in a valuable piece of gear make sure the observation is what you assume it to be. Although the LN stands for low noise, 1176's were never really low noise. Have you measured the gain and the output to determine what the noise equivalent input is? Have you measured other 1176's?
Do you have a spectrum analyzer that can cover that range to see if there are spurs in the output or whether the noise is regular 1/f noise with a predictable broadband signature?
I would stop right now until you can determine for sure that there is an oscillation that is present inside the unit and without your probes near it. The steel covers have to be on it for it to meet any published spec since they tended to have RF pickup problems without the covers.

[tired eyes]

Update

After experimenting with a 22pF cap to ground, there seem to be two separate oscillations. The oscillation at Q3 is hundreds of MHz (my scope can just barely make it out at 1uS/cm), but at Q14 (the output) it is about 800kHz. The MHz oscillation does not show up at the output.

I tried a 22pf cap between the base of Q3 and ground - that reduced the amplitude of the 100's of MHz oscillation very slightly. The same cap from the collector of Q3 to ground eliminates it altogether. However, this has no effect on the 800kHz oscillation at the output. So I guess I'm now focusing my efforts on Q14, or as just suggested, putting the unit back together.

I did a quick calculation for published S/N ratio of 80dBm at +10 line level, but did not measure noise with input and output signal matched to this level. Going by the specs, noise at the output should be .7mV. I figured 50mV was so out of the ballpark I must have a problem, notwithstanding the gain is cranked and the covers are off. I got the same noise with it in the studio fully assembled, minus the 60Hz which is now gone. Next I will put the covers back and test the output at +10.

The noise sounds like part 1/f white noise, part HF buzz. No spectrum analyzer, just my ears.

If this is simply the nature of the beast, so be it. It is 70's gear after all.

[Steve Conner]

Try adding a resistor in series with Q14 base, about 100 ohms. Also try the same with Q3. Should be easy to rig up as a test, if they're in sockets.

New transistors can have considerably more beta and Ft than the originals of the same type number. They often oscillate in circuits that were stable when built with the originals. The JEDEC spec was a "meet or exceed" thing.

[harrydanielsono]

I would say you probably moved some of the internal wiring, particularly wiring connected to the output, and it came a bit closer to the gain reduction fet. Be very aware of capacitively coupling hf (consider layout and electrical fields, the air, the board, etc) at the gates of fets especially in a circuit such as this one where they are forming antennas to a degree.

Also I don't like the sound of some of the mods suggested here. Would be interesting to know what the outcome of this was.

[J M Fahey]

Donīt stone me for this, but just to check the hypothesis that oscillation=noise, I would temporarily add a small ceramic cap (say, 22pF) base-collector of Q2, check that the oscillation dissappears and what happens with perceived noise.
Yes, it will probably turn the sound into mud or whatever, thatīs not the point, just a test.
Good luck.

[km6xz]

This is an old thread so I don't expect the OP has the unit anymore.
I am not sure there is a problem with oscillations, it changed or appeared based on scope probe position, for example none when monitoring the output, yet 800khz at one spot and hundred(s) Mhz at another spot. Those are not high gain/ bandwidth circuits. I would propose that with careful attention to testing setup, the measured problem goes away.

[J M Fahey]

Agree, on principle.
On max scope gain settings, he might be picking some stray RF field produced by something else close by .... even the cellphone in his shirt pocket or belt!!!

[doctor]

Agree, on principle.
On max scope gain settings, he might be picking some stray RF field produced by something else close by .... even the cellphone in his shirt pocket or belt!!!

BTW, just today I opened 6176 Channel Strip manufactured by Universal Audio.
In accordance with the specification it is composed of the 610B tube preamp and
the 1176LN limiting compressor.
This is true. There are tube preamp and compressor boards inside the device.

But to my surprise built-in compressor does not correspond to any 1176LN
schematics known to me (revisions AũH and K).

Whether someone has correct schematic?
I could do with a 610B schematic as well.