Basic tube amp decoupling question

[soapfoot]

Hello! I've got an old Signal Corps BC-730-B limiter. It sounds great, but if the limiting threshold trimmer (R10)  is advanced too far clockwise, motorboating occurs (all caps are fresh and new; all resistors are new 1% metal films)

It has always been this way, and I suspect it's inherent in the design as the power supply doesn't look to have too much decoupling.

I notice that if I add a modest amount of capacitance to (2µF seems to do the trick) between R7 and ground (at the junction of the plate of V4) the amp stabilizes and the motorboating goes away.

My simple question is--will this modification have any side effects besides stopping the motorboating? Will it impact the sound of the limiter in any way? V4 is essentially the detector circuit of the limiter. Is it fair to assume that adding capacitance there may alter the bandwidth of the detected signal?

I admit to knowing only enough to be dangerous, so I figured better to ask for help before modifying.

Thanks in advance!

 

[ruffrecords]

A 2uF cap from the junction od R7 and V4 plate simple shorts the output of V4 so there is no longer any control voltage produced. I would suggest you check the values of C1 and C6.

Cheers

Ian

 

[soapfoot]

Okay, thank you so much!

Clearly I was just watching for motorboating on the scope and not testing the limiting action with an input signal, which allowed me to miss that.

C1 is the correct specified 1µF.

C6 is specified as "250Mmfd." in the documentation (attached), which is 250pF, correct?

I think this cap is actually missing in my unit (it had been worked on by someone else previously). I'm a little embarrassed that I hadn't noticed. Is it fair to assume that this could be the cause of my issue?

 

[ruffrecords]

Yes, it is 250pF. Try a 220pF or 270pF. Hard to say if it will affect the motor boating but it is worth a try.

Cheers

ian

 

[abbey road d enfer]

I think this cap is actually missing in my unit (it had been worked on by someone else previously). I'm a little embarrassed that I hadn't noticed. Is it fair to assume that this could be the cause of my issue?

I would doubt it very much.
This type of compressors is based on feedback. As such there are stability criteria that have to be met. Any feedback compressor will end up motorboating when loop gain is increased. The trimmer is there to optimize operation by establishing "how much it can go too far" and provide a nice setting where reactivity and transient response are ideally balanced. Since this point varies with the tube's parameters, it must be adjustable. I would think the maintenance manual would explain how to increase the sensitivity up to the point the unit starts to become unstable and back off a little.

 

[soapfoot]

I would doubt it very much.
This type of compressors is based on feedback. As such there are stability criteria that have to be met. Any feedback compressor will end up motorboating when loop gain is increased. The trimmer is there to optimize operation by establishing "how much it can go too far" and provide a nice setting where reactivity and transient response are ideally balanced. Since this point varies with the tube's parameters, it must be adjustable. I would think the maintenance manual would explain how to increase the sensitivity up to the point the unit starts to become unstable and back off a little.

Makes perfect sense, and thanks!

Indeed, the original service manual describes a calibration procedure, but it revolves around target voltages and levels of attenuation. When I tried to go through the steps, it was right at the edge of motorboating (low bass frequencies would kick it into oscillation for the duration of the note).

Backing it off just a touch solved that issue, so that's most likely where I'll leave it.

The only other thing I notice that's a little awry: in the service manual there's a helpful list of voltage specs (attached).

With one exception, all measured voltages are very close--closer than I'd expect in this era of device, actually, considering modern line voltages.

The outliers are Pin #6 (grid 2) of both 6SK7. Bogey voltage there is 9 VDC, and they both measure closer to 12v, even with R8 at its minimum. R6 is a wirewound with a bogey value of 15k and its actual measured value is 14.87k (I doubt that's far off enough to cause an issue here?)

The unit is mostly working and performing as-desired, so maybe I shouldn't rock the boat too much*

*when fed white noise and the output viewed on the spectrum view of FabFilter Pro Q3 (quick and dirty) it does seem to show a little bit of an unusual lift (resonance?) at the very upper end of the audio band. Maybe that's what C6 was there to correct? The manual says C6 is "to flatten frequency response when the limiter is operating."

 

[emrr]

This is same as the more well known Federal.  Look at commentary around it, if you haven't.

I suspect it was not intended to ever pass and compress lows such as we feed things today.  True of many communications limiters.  If you HPF the source, what happens?  If you lower the value of the side chain coupling cap by 1/2 or 1/4, what happens? 

 

[soapfoot]

Yes indeed, the precursor to the AM864u (but full-bandwidth).

You make great points!

It does indeed sound really good, better than any of our Federals (it lives in the rack next to the SA-39b you're currently looking at for us!) so I'm a bit loath to experiment too much with coupling cap values. I'd much rather just back off on the limiting threshold a touch, as we've done for years.

The only reason I'm even in here, really, is that some of the old round-ended "Continental Carbon" carbon comp resistors had gone noisy. I just figured I'd poke around and do an overall health check while I'm in here.

 

[abbey road d enfer]

The outliers are Pin #6 (grid 2) of both 6SK7. Bogey voltage there is 9 VDC, and they both measure closer to 12v, even with R8 at its minimum. R6 is a wirewound with a bogey voltage of 15k and its actual measured value is 14.87k (I doubt that's far off enough to cause an issue here?)

I wouldn't worry about that. Remote cut-off tubes have a wide control range; it's not a sutrprise that this voltage is 30% off. 14.87 vs. 15 is actually better than 1% tolerance.

*when fed white noise and the output viewed on the spectrum view of FabFilter Pro Q3 (quick and dirty) it does seem to show a little bit of an unusual lift (resonance?) at the very upper end of the audio band. Maybe that's what C6 was there to correct? The manual says C6 is "to flatten frequency response when the limiter is operating."

I read that as "compressed program usually seems to lack treble, so this cap tames slightly the side-chain at HF"

 

[soapfoot]

Understood!

I may add the cap; I may not (everyone at our place has loved the character of this thing for years... it's not known around here for being "transparent" by any stretch, so I may prefer to just keep it sounding familiar to us)