Presonance issue

[Tonegeek]

I replaced my existing presence circuit with Martin's presonance control in my BM 50 watter.
I ran a test signal in and when I turned it on I got this reaction:
I could hear the presence kicking in when I turned the pot up
when I turned it down toward the resonance side, the amp started a low humming along with the test signal.
Also the presence pot starting smoking!
When the humming starts the AC voltage at the junction of the FB resistor and pot input jumps from .02 VAC to 30 VAC. I assume I am getting an oscillation.

Here are my values based on originally having a 4.7k tail, 100k FB and .1 pres. cap

RFB = 100k (off 8ohm tap)
R tail = 4.7k
C1, C2 = .05
Pres pot = 25k (should be 10k -20k or so, but this is what was in there) I even jumped it with a 24k resistor to no effect.

I also chopsticked the feedback wire, moved it around, and changed the fb resistor to 50k and 27k to no effect.

I will mention that my ipedance switch is set to 4 ohms although the GNFB is coming off the 8 ohm tap. Other than that, this is a standard Marshall type pi

What can I do about this, if anything?

[martin manning]

I'm not sure what to think, Tonegeek. Colossal had an oscillation issue with the presonance, but it cleared up when he tidied up his wiring as I recall. One thought I had at the time was that there is a difference in the phase shift through this network vs. the basic presence as the control is swept through its range, and it could be that it gets too close to positive FB at some point. Reeltarded had good results with it on a plexi-type amp, although he did report some behavior that sounded like an ultrasonic oscillation at one particular setting of the pot. I think it would be helpful to look at this with an oscilloscope to see what's going on.

[Cliff Schecht]

Martin that's a legitimate concern. Putting any sort of filtering element into an amplifiers negative feedback loop can cause unforeseen (and unforecalculated) issues. Maybe some sort of phase boosting network could be added to the NFB resistor off of the OT secondary to prevent oscillation issues without causing audible side-effects.

[Jana]

I second Martin's concerns.

I have experimented a lot with presence/resonance controls and one thing I have found is that there needs to be limits placed in the circuit. Normally, with a fixed NFB resistor and nothing else, the presence cap/pot works fine and doesn't give any problems. But, if you put a cap in the NFB path and don't have any upper limit (a series resistor with the cap) then when the presence gets turned up full there is a point where the high frequencies can really spike and take off. It's hard to explain, much easier to see on a plot. One solution is to place a 820 to 1K resistor in series with the presence cap.

[Tonegeek]

Its odd since NFB is supposed to cancel. In my case when the pot is bringing the cap closer to the OT, for a bit it is canceling some highs but then it takes off. I am thinking there is something to the positive feedback idea since the cap is bound to shift the phase somewhat. The limiting resistor makes sense. You could put it in series with the cap or drop it off the OT line and attach the input to the pot and cap on the other side.

[Firestorm]

How does the amp behave open loop? I can't see how decoupling the loop at low frequencies would be worse than decoupling it completely unless the feedback was supressing an oscillation.

[Cliff Schecht]

I've read in old tube literature about the "risks" involved with adding frequency dependent negative feedback and oscillation is a possibility for sure. Without investigating both the phase and gain margin of a given system with the pot rotated to both extremes it's hard to say exactly what's happening. This would be hard to simulate too because of the extremely frequency dependent nature of the output section of a tube amp (output transformer and speakers mostly). Unless you have good models or have otherwise calculated/extracted the parasitics this might be a tough cookie to crack for general purpose cases.

Good thing there's a bunch of EE's on this forum .

[Reeltarded]

On the low end of travel I get a drop out that sends my output tubes into electron storms.

[martin manning]

I think this might be the smoking gun. Blue is the baseline Express and the green is the presonance circuit. Solid lines are voltage and dashed are phase. Note the hump that gets to about +60 degrees at 100Hz. That's at the full CCW setting. It can be reduced some by putting a resistor between the CCW end of the pot and the FB resistor. A 10k, same value as the pot, limits the shift to 35 degrees or so. That might be enough to tame it, but maybe at some cost to the overall effect...

[Colossal]

Colossal had an oscillation issue with the presonance, but it cleared up when he tidied up his wiring as I recall.

Keep in mind too that my amp is a point-to-point build, a test amp (so parts have been soldered in and out multiple times), and violator of many 'rules'. My issue turned out to be a reversed wire on my PPIMV (the grid connections). Switched two wires (not the output transformer primaries), and the amp was perfect. The oversight was mine, not the fault of the presonance circuit. Once I sorted out the PPIMV, I have had absolutely no issues with the circuit and it is stable at both extremes of the knob.

[Firestorm]

I think this might be the smoking gun. Blue is the baseline Express and the green is the presonance circuit. Solid lines are voltage and dashed are phase. Note the hump that gets to about +60 degrees at 100Hz. That's at the full CCW setting.

Wouldn't the shift have to get past 90 degrees to become a problem? Aren't the "resonance" traces supposed to look that way: approaching a 90 degree shift near the resonant frequency of the speaker?

[martin manning]

There will be some internal phase shift from the PI input to the speaker output. My thought is the sum of that and the presonance circuit shift might get to be too much. I don't recall which way that internal shift goes, though. I latched on to Tonegek's report of a low frequency apearing and it possibly being close to the 100Hz peak. The plot only represents the few components in the FB NW, not the whole amp.

[greekie]

FWIW I solved problems like this by putting two or more caps in series to get the right phase of the NFB. You have to get larger values of course.

[Reeltarded]

Sorry to run away, but..

I am using 100k tapped off the 16 right now to hump up the action of the presonance. I stay away from the ugly area on the pot.

I have a pile of .1 and .68, but I have been using .047s. If I want to increase the reso action into a higher range, what effect would either of these have? I am just trying to get a little hump above the drop out. The pres is too dull down there to use that part anyhow, which is good!

Oh, and UNLIKE Dave, my amp is perfect, and there is nothing I can do to make it more awesome, so it's just chance, and left to the amp gods.

=P

Martin, I reinstalled the normal presence circuit, and ya know what? It had a little extra scratchy right in that same range as the presonance gets funky. Hmm

[martin manning]

If I want to increase the reso action into a higher range, what effect would either of these have? I am just trying to get a little hump above the drop out. The pres is too dull down there to use that part anyhow, which is good!

If you want to push the range up in frequency, decrease the series cap (the one that goes to the wiper), and increase the shunt cap going to ground. For example if you are 0.05 and 0.05, try 0.033 and 0.68 for series and shunt respectively. You might also try just the series cap first and see how that sounds. Reducing the total capacitance will raise the range of the presence end too.

I reinstalled the normal presence circuit, and ya know what? It had a little extra scratchy right in that same range as the presonance gets funky. Hmm

Hmmm is right. Now, what's up with Tonegeek's amp, then?