Squealing/positive feedback - explanation of what is happening

[sleipnir]

If you read the thread he was having some real problems with the amp assembled before.

That is exactly right - the whole thing didn't start with me testing the amp like this on the bench from the start - when I originally finished the build and played the amp, I eventually realized I would get some squealing which depended on proximity to the amp and the amount of gain. I never noticed this with my Super Reverb, which is why I started wondering if my new amp had a design/layout issues. Having done a lot of searching and reading online (including R.G.'s site), none of those fixes seemed to work for me. Perhaps this is just a case of some amps (like evidently several Marshall amps as mentioned by folks in this thread) being unstable when run with controls high or maxed. However, given that I pretty much gutted the amp in the process, I am hoping to eliminate with as much certainty the possibility of there being something wrong with my layout/wiring.

I am hoping R.G. will weigh in on the results of the test with the 100k resistor on the cable before I put the amp back together. I am not sure what to make of it, but he might have some ideas?

Chuck - here is a pic of the current OT wiring with the temporary single tap Super Reverb OT. I moved the secondary wires (black and green) to a chassis hole farther away from the preamp. Do you see any issues with this as well as the NFB and presence wiring at this point?


I can definitely mount the OT to the chassis and put the chassis back in the cabinet along with the shielded back panel, but unfortunately, I have no reason to believe the squealing would be gone as nothing seems to have changed since I started trying out various fixes. It might still be a good to do as it's been a month since I last played it that way :-)

[R.G.]

I gave this a shot. I tried wrapping a 100k resistor across the tip and sleeve on one cable and also tried soldering a 100k resistor on another just to ensure a good connection. I also tried a cable with no resistor.

Nice video. This tells you a couple of things. One of those is that you're having a capacitive feedback problem - not acoustic, as in my earlier wild-eyed theory - and another is that it's a grounding issue. (I think. that's the theory of the moment right now )
A resistor-terminated guitar cord is a distributed inductor-capacitor with a resistor at the end. For most purposes the inductance is too small to matter. I was hoping to pick out an RF resonance with the resistor out at the end. But your setup showed that the frequency is too low for an RF resonance, and that the frequency depends on how close the input cable is to the wiring. The frequency changes as you move the wires, and since there's not much in the way of current flow possible, that mostly eliminates inductive coupling and leaves capacitive coupling.

The resistor being at the input-jack end of the cable might change things a bit. I'll have to think about that some.

My take on the video is that the cable capacitance is letting signal picked up on the shield of the cable couple into the input wire. The only way that can happen is if the cable "ground" isn't grounded, or not solidly grounded. It would be very interesting to see what happens if you repeat the same test as the "100K" video, but with a wire wrapped around the sleeve and tip of the free end of the cable, and wave that shorted end around the OT as in your video. That would separate out capacitive coupling to the shield versus coupling to the input wire.

Another thing I'd do is to use a clip lead between the tip and sleeve of the input jack at the input *with the cord plugged in* so it shorts the cable signal at the input jack. Repeating the video test then tells you if the shield itself is carrying in the feedback signal, or if it's coming in to the signal wire in the cord.

I figured I would also give it a shot with a metal cover/shield over the chassis while also moving the OT and its wires to the other side of the chassis away from the power tubes.

That's instructive. The scope trace shows a nice sine wave when it squeals, instead of a clipped on. Lower loop gain.

In addition to a separate safety ground, I have a power ground, preamp ground and the speaker jack ground. In the photo below, you will see six wires connecting at the power ground. Two are from the output tubes, one from the PT center tap, one from the filament supply artificial CT, one from the bias circuit, and one from the filter caps - specifically the first two sections which connect to either side of the choke:

Well, that's the source of the buzz you're hearing. The PT center tap wire needs to go to the first filter caps, and NOT to the common ground. But that doesn't have anything to do with your squeal.

The pots and input jacks are isolated from the chassis with the exception of one input jack which serves as the single grounding point for the preamp section. There is a wire that runs along the back of the pots which is connected to the one grounded input jack. The various grounds from the preamp circuit are grounded here, as are the remaining two filter cap sections. I currently have the presence pot grounded here as well, although I have tried grounding it to the power ground and speaker jack ground with no improvement. I did try grounding each pot to the chassis but that did not make a difference. I tried grounding ALL the filter caps at the power ground, but that did not change anything. I also tried grounding the speaker jack to both the preamp ground and power ground, but it did not change anything

Hmm. This is what I'd call an ad hoc grounding scheme. Most amps are ad hoc. Using the chassis for a signal ground is something that amp builders mostly get away with, not good practice, much like running yellow lights in your car. Usually it works, but sometimes it gets you into trouble. In an ideal world, you could make one and only one connection to the chassis, and in an almost-ideal world, two connections. The ideal world one has one connection to the chassis, either the power supply ground star point or the input jack. That means that it's impossible for current to flow across the chassis, unless it's intercepted RF or capacitively-coupled current, and that's what shields are for. If circuits are grounded to the chassis at multiple points, it's possible for currents to flow in the chassis, and also possible for circuits that use the chassis as a 0-volt reference to see the tiny voltage drops across the chassis resistance as a signal voltage to be amplified.

This relates to your case in that your input jack is connected to the chassis, and so is the ground return point, but at different places. There are also grounds from the controls to the pot-case buss, so those signals get fed through the chassis back to the ground return point as well. That includes a presence control, which is attached to the outputs stage, a potent source of high voltage and high current signals.

You mention a speaker ground, but I didn't see it in the write up or photo. It is really good practice to NOT wire the speaker jack to the chassis, but to insulate it and run a wire from the speaker jack sleeve to the OT "common" winding point. This keeps the speaker current out of the chassis. From there, a zero-volt reference wire would run either to the star ground (in a theoretically ideal world) or to the PI stage local ground, as you're also running the output signal from the speaker jack back to the PI for feedback and "presence". The speaker winding is actually a floating differential signal, and grounding it to the 0-V reference point has to happen somewhere, but it's better for stability to keep it isolated to the circuit it feeds - the PI. The PI is then kept referred to the bigger amplifier ground.

So I don't see any definitive "ahah!" points yet, but the symptoms, pictures and video point to issues where the grounding is allowing capacitive pickup to get back into the input.

[sleipnir]

I was hoping to pick out an RF resonance with the resistor out at the end. But your setup showed that the frequency is too low for an RF resonance, and that the frequency depends on how close the input cable is to the wiring.

R.G. - thanks for this detailed response! I don't know if matters much, but when I was preparing to do the video, I believe I saw an oscillation way above what I could hear in the 40-50kHz range which might be what you were expecting? At that time, I did not have a wire clipped from the chassis to the OT frame, but when I added it that wire, this high frequency signal went away. I can try to recreate this if you think it is significant.

The resistor being at the input-jack end of the cable might change things a bit. I'll have to think about that some.

I can easily redo the test with the resistor at the other end if you think it makes a difference

It would be very interesting to see what happens if you repeat the same test as the "100K" video, but with a wire wrapped around the sleeve and tip of the free end of the cable, and wave that shorted end around the OT as in your video. That would separate out capacitive coupling to the shield versus coupling to the input wire.

Just to clarify - are you saying leave the 100k resistor still on the cable at the input end while shorting the other end of the cable? This would then effectively make the shield and the core wire one and the same for the whole length of the cable (except at the actual end where the 100k resistor is), right?

Another thing I'd do is to use a clip lead between the tip and sleeve of the input jack at the input *with the cord plugged in* so it shorts the cable signal at the input jack. Repeating the video test then tells you if the shield itself is carrying in the feedback signal, or if it's coming in to the signal wire in the cord.

And to clarify this - no 100k resistor at all on this cable, but just short the end at the input jack, right?

Well, that's the source of the buzz you're hearing. The PT center tap wire needs to go to the first filter caps, and NOT to the common ground. But that doesn't have anything to do with your squeal.

I am definitely interested in any improvements I can make to the grounding while I have the amp the way it is. I had read somewhere that it is good practice to separate power and preamp grounds, which is what I had attempted to do, but I am more than willing to try a single ground setup if that is ultimately the ideal scenario. If I want to strive for just one ground, wouldn't the PT CT and first filter caps also be grounded at this one point? Since my pots are already isolated, could I simply isolate the one remaining chassis grounded input jack and run a wire from it to the power ground, or do I need to break the wire on the pots into smaller sections with each section having a wire back to the main ground? I guess at least the presence ground should be separated from the other pots and have a wire direct to the main ground? I can also isolate the speaker jack and run a wire from its sleeve to the power ground. In the above scenario, only the safety ground would be separate from this one star ground point. Please let me know if you think this would work as the ideal setup, or if I am missing something.

Thanks!

[R.G.]

when I was preparing to do the video, I believe I saw an oscillation way above what I could hear in the 40-50kHz range which might be what you were expecting? At that time, I did not have a wire clipped from the chassis to the OT frame, but when I added it that wire, this high frequency signal went away. I can try to recreate this if you think it is significant.

It may be significant or not. That translates in my mind to "the internal capacitances of the OT will couple high frequency signals to each other in a way that can make the amp oscillate if the frame isn't grounded." But the OT will never not be tied to the frame in actual use, so I would tend to note that and only come back to it if you get to a point where there's an ultrasonic oscillation.

A lot of my debugging thinking is devoted to divide-and-conquer: do tests to separate things, then squash things that are found. This will uncover other issues until you finally fine the last one that you have the energy to fix.

So I'd say ignore it for now.

I can easily redo the test with the resistor at the other end

I'd put this in the "worry about it later" category. There are other issues.

Just to clarify - are you saying leave the 100k resistor still on the cable at the input end while shorting the other end of the cable? ...

And to clarify this - no 100k resistor at all on this cable, but just short the end at the input jack, right?

The 100K was put there just to fake the DC path into/out of a typical guitar, and to ensure that there was no issue with grid leakage on the input of the amp. I actually meant to put the 100K out where it would look like a guitar. But the test you did with it on the input jack end was OK.

What I would do if I was at your work bench would be to short the guitar end of a cable and put that into the input jack, then repeat the test you did on the video, "sniffing" for oscillation. Then I would leave that short and also short the input jack with a cliplead and try that. Shorting the guitar end leaves the whole cable as a distributed RF resonant circuit that looks like a short circuit at audio frequencies. Shorting at the clip lead tends to remove the cord's antenna effects up through frequencies where the cliplead itself looks like a loop antenna. Both would be interesting in telling you whether the feedback is coming in through the cord when it's plugged in or through the grounding of the input jack to chassis.

I am definitely interested in any improvements I can make to the grounding while I have the amp the way it is. I had read somewhere that it is good practice to separate power and preamp grounds, which is what I had attempted to do, but I am more than willing to try a single ground setup if that is ultimately the ideal scenario. If I want to strive for just one ground, wouldn't the PT CT and first filter caps also be grounded at this one point? Since my pots are already isolated, could I simply isolate the one remaining chassis grounded input jack and run a wire from it to the power ground, or do I need to break the wire on the pots into smaller sections with each section having a wire back to the main ground? I guess at least the presence ground should be separated from the other pots and have a wire direct to the main ground? I can also isolate the speaker jack and run a wire from its sleeve to the power ground. In the above scenario, only the safety ground would be separate from this one star ground point. Please let me know if you think this would work as the ideal setup, or if I am missing something.

Grounding is HARD to conceptualize. "Ground" isn't one thing - it's a bunch of things hiding under one word.

There is only one real, true ground. That's a thick copper rod driven deeply into wet soil to form a solid electrical connection to the whole planet; and that is why we call it "ground" (or "earth" if you're from a UK heritage) in the first place. The entire planet as a whole is the only universally-agreed-upon source of truly zero volts. But we all know that electronics work if they're run from batteries that are not connected to a grounding rod, so it became common to refer to the reference potential in any circuit as "ground" out of deference to the One True Ground.

Ground is the place in a circuit where we hook the "zero" lead of the oscilloscope or voltmeter. It's the "0.0000000V" point that we measure everything else against. But it's more than that. There are several kinds of ground:
1. True Earth Ground: that copper grounding rod
2. AC safety ground: connected to a copper conductor into the dirty by your building's AC power network, but "dirtied up" by any current flowing through it back to the real earth
3. Power Line Neutral: connected to AC safety ground back in your breaker box and out on the power pole, from whence it's connected to earth ground.
4. Power ground in your circuit: the power supply makes some voltages, and one of these voltage points is designated as the point we will consider as 0V.
5. Reference ground: this is the point where you hook the voltmeter (metaphorically), and where your circuits refer their inputs
6. Power return ground: I call this "sewer ground", because all of the used electricity flows back on the power return ground lines to the power supply to be recycled.
7. Shield ground: generally a metallic covering around the actual circuit to intercept and drain away any electromagnetic waves which come through. Done properly, this forms a "Faraday Cage" which prevents RF internally.

As you can imagine, things get dicey when we call several things "ground" and use their conductors interchangeably for crossed purposes. The problems come up because there are no superconductors that we can use in our amps yet. For normal conductors, there is always some resistance (and some inductance, when you get into RF). Any time a current flows through a real wire, a voltage drop is generated across the wire by Ohm's law and by its extension to AC circuits for inductance and AC signals.

Ignoring inductance, the voltage is V = I * R. R may be small, but it's always there. So
>> THE ONLY WAY TO HAVE ZERO VOLTS ACROSS A CONDUCTOR IS TO MAKE THE CURRENT THROUGH IT ZERO.<<

If you want the millivolts of signal from your guitar to be coupled to the input grid of your input tube with no interference from other places, you must take the "signal ground" from the guitar right to the tube grid leak resistor ground side, because the tube thinks the bottom of the grid leak is ground, no matter what else you try to convince it. We used to have a saying at work that you can't b*llsh*t electrons. The tube will amplify whatever comes in as a voltage difference from the grid to the bottom of its grid leak, usually the resistor at the bottom of the cathode.

If the incoming guitar signal ground shield (notice that the shield is cross-used for both RF protection and carrying return current in that cord) is connected to an input jack that is 'grounded' to the chassis, and only the signal wire goes to the first tube grid, and the cathode bottom resistor and grid leak go off to someplace else on the chassis, then any current flowing through the chassis prys the tube's reference ground and power return grounds apart, and this signal, however small, is injected into the tube grid with the guitar signal.

A really ugly situation is set up if the input jack is "grounded" to the chassis and the speaker return is also "grounded" to the chassis, and that large return current happens to flow across the chassis in a way that the voltage it generates in the chassis metal generates a voltage on the input jack "ground". Now you have direct feedback from speaker output to the amplifier input. With enough amplifier gain, this *will* oscillate. [notice that this may not be your specific problem, but it is a valid example of the issue]

There are several ways to fix this. One is to ignore it and build the amp, and hope that the current doesn't happen to flow through the chassis metal where it will cause the problem. With inputs on one end and outputs on the other, most of the output current flow is probably at the output end, little at the input end, so it doesn't oscillate. Ducked a bullet. Another is to isolate one or the other or both of the input jacks and output jacks from the chassis where they enter it, and take an explicit ground reference (for inputs) or ground return (for speakers) wire to where the corresponding current wants to get, so the signal current doesn't flow through the chassis, which is after all there only for mechanical strength and for RF shielding. Isolating the speaker jack return "ground" from chassis and taking it all the way back to the OT, where that current wants to go anyway" removes large currents from the chassis and cuts the possible oscillating potential a lot.

Isolating the input jack from chassis and taking a signal ground wire to the first tube's grid circuit keeps interfering signals from chassis off the signal input - and lets you pick up radio. The better way to do this is to isolate the input jack from chassis and install a small ceramic cap from speaker ground to chassis right next to the input jack. The cap now siphons off the RF and lets the audio go to the tube.

You'll see a common theme here: where does the current flow? We're educated to think about voltages, not current paths. This is HARD until you practice it a lot.

I said all that to get to this. The currents from the rectifiers into/out of the filter caps come in short, sharp pulses. These travel through wires, and those wires have resistance. If any part of the wires get between your circuit's reference ground and the filter cap negative side, those pulses appear as ground noise in your circuit. You hear them as a hum-toned buzz. The way to do this right is to take the PT center tap (or negative-side rectifiers if you use a bridge) to the negative terminal of the first filter cap, and nowhere else, period. The negative side of the first filter cap then can be connected to the circuit as a "zero volt" point, and it's not being jerked around by rectifier pulses, near as the circuit can tell.

From there, how does the circuit "ground" get connected up. Easy- where does the current flow?

A lot of it, probably 80+%, comes back from the output tubes. So it would be clever to use a wire from the output tube cathode circuits directly to the first filter cap's negative terminal, so that current, which pulses in synch with the signal, is not ghosted into the rest of the circuit by moving the circuit's "ground" up and down. Notice that this also takes care of the screen currents.

Next? What's the next biggest current? Probably the PI. The PI has modest currents, but has to be hooked to the OT's signal circuit to get feedback, so the OT's secondary-side "common" terminal needs connected to the PI's ground, and the OT's secondary signal output needs connected back through the presence control setup to the PI's cathode circuit. It would make sense to take this current off the chassis, and run it on individual wires back the first filter cap negative terminal, to keep it from interfering with the preamp signals.

You see where this is going, right? In order of highest to lowest current, return wires are moved to the first filter cap negative. We're slowly constructing a star ground.

Star grounds can be proven to work without further knowledge of the circuit and physical setup. There are an infinite number of non-star-grounds that work, too. But all of them depend on special cases of the mechanics, or where the ground currents are not too bad, or where ground currents cancel - that can happen, too.

The first filter cap negative terminal is critical for the rectifiers and output stage return. The speaker jack return is important for keeping audio signal off the chassis. What about those other filter caps?

If you think about where the current goes, two things come up:
1. The DC all goes back to the first filter cap negative terminal.
2. The AC signal and any interference follows the lowest AC impedance path.

The AC path is what causes oscillation. You can't, by definition, have DC oscillation, after all. So you have some options. You can use the caps as local bypasses to divert all the AC currents through small local loops, or you can take all the caps in the DC dropping chain back to the first filter cap negative terminal. But the caps further down the power distribution chain need to be big enough and of high enough quality to keep AC (even millivolts!) off the return path if you distribute them out to the circuits. Even then, the DC return current will raise the DC level of "ground" by the return current of the circuit times the DC resistance of the wires.

And that's why I sometimes say "I'll have to think about that." I have to go think about what currents flow through which wires.

If it were me at your workbench, I'd just go do the following:
1. Pick a place for safety ground and move safety ground to the chassis. Use this bolt to the chassis for nothing else, for safety reasons. Do not use a transformer or other mounting bolt for this, for safety reasons.
2. Move the PT CT to the first filter cap negative terminal.
3. Isolate the speaker jack from chassis, and return its sleeve to the OT common terminal.
4. Run a wire from OT common terminal to the PI circuit's local ground. This last could go to the whole-amp star ground, but the wiring gets complicated, and grouning it here will do just as well.
Then test. Still oscillate? Well, this stuff certainly didn't hurt the amp, and probably removed one or two things that would have been uncovered after the squeal was killed.

[Mike Sulzer]

What I would do if I was at your work bench would be to short the guitar end of a cable and put that into the input jack, then repeat the test you did on the video, "sniffing" for oscillation. .

In effect, that was done way back in the first post:

The squeal will go away if I sufficiently lower the volume or tone on the guitar,....

The transformer sitting on the bench makes a big oscillating electric field (especially if the case is not attached to the chassis). If the end of the cable is not shorted (and the presence of a guitar affects the details of this) then it can pickup this field, developing a voltage across the cable. Move close enough and there will be a range of frequencies where there is enough gain to get an oscillation. We expect large phase shifts at the higher frequencies, and so it will oscillate at some frequency. There may well be grounding issues, but this simple mechanism would do it even if there are not.

[R.G.]

In effect, that was done way back in the first post:

In an ad hoc way. We're looking for definition.

The transformer sitting on the bench makes a big oscillating electric field (especially if the case is not attached to the chassis). If the end of the cable is not shorted (and the presence of a guitar affects the details of this) then it can pickup this field, developing a voltage across the cable. Move close enough and there will be a range of frequencies where there is enough gain to get an oscillation. We expect large phase shifts at the higher frequencies, and so it will oscillate at some frequency. There may well be grounding issues, but this simple mechanism would do it even if there are not.

As I said, anything will oscillate if the gain is high enough. He's looking for a few more DB so it doesn't squeal by moving near the amp while playing the guitar if I read it correctly.

[sleipnir]

What I would do if I was at your work bench would be to short the guitar end of a cable and put that into the input jack, then repeat the test you did on the video, "sniffing" for oscillation. Then I would leave that short and also short the input jack with a cliplead and try that. Shorting the guitar end leaves the whole cable as a distributed RF resonant circuit that looks like a short circuit at audio frequencies. Shorting at the clip lead tends to remove the cord's antenna effects up through frequencies where the cliplead itself looks like a loop antenna. Both would be interesting in telling you whether the feedback is coming in through the cord when it's plugged in or through the grounding of the input jack to chassis.

I finally got some time today to try this out and made a video:
http://vid270.photobucket.com/albums...queal%2010.mp4

I did not notice any squealing while shorting the cable at the guitar end nor when shorted at both the input and guitar end. I'd be curious if you can draw any conclusions from this and also whether the other observations are of any value. I definitely do not understand why the squealing is MORE sensitive with the frame of the OT grounded to the chassis...

If it were me at your workbench, I'd just go do the following:
1. Pick a place for safety ground and move safety ground to the chassis. Use this bolt to the chassis for nothing else, for safety reasons. Do not use a transformer or other mounting bolt for this, for safety reasons.
2. Move the PT CT to the first filter cap negative terminal.
3. Isolate the speaker jack from chassis, and return its sleeve to the OT common terminal.
4. Run a wire from OT common terminal to the PI circuit's local ground. This last could go to the whole-amp star ground, but the wiring gets complicated, and grouning it here will do just as well.
Then test. Still oscillate? Well, this stuff certainly didn't hurt the amp, and probably removed one or two things that would have been uncovered after the squeal was killed.

[/QUOTE]
I have read this informative post several times and I think I am starting to grasp the grounding concept somewhat. As I am using a tweed Bassman chassis, the filter caps are located in the "doghouse" on the back, so grounding the first filter cap and PT CT involved a long cable run one way or another. If I run a wire from the first filter cap negative lead into the chassis and connect it to the PT CT, is there truly a benefit to doing this and then running a wire from this connection to the star grounding point versus just having the two wires (i.e. PT CT and first filter cap ground wire) meet at the star ground?

[R.G.]

If I run a wire from the first filter cap negative lead into the chassis and connect it to the PT CT, is there truly a benefit to doing this and then running a wire from this connection to the star grounding point versus just having the two wires (i.e. PT CT and first filter cap ground wire) meet at the star ground?

Yes there truly is.

The PT CT is sucking current out of the negative side of the first filter cap. That's where the rest of the amp draws its DC from, and only the PT CT removes the "used electricity" from the negative side of the first filter cap. The current pulses happen twice as often as the AC line frequency, either 100Hz (in 50Hz countries) or 120Hz (in 60Hz countries). The pulses can be several amps tall, because they only have a millisecond or less to suck out all the electrons that piled up there since the last rectifier pulse. The pulses can only come out of the first filter cap negative, so no matter what else is hooked up there, the pulses come out of the first filter cap negative, flow through any wires needed to get to the PT CT, and go back to the transformer. They cause a voltage to appear across the wires they travel through on their way.

A long wire from the PT CT to the first filter cap negative adds a little resistance to the CT wire, but because this wire doesn't connect to the circuit ground except at the first filter cap negative, it cannot make the circuit "ground" voltage be any different. It just lowers the power supply voltage a tiny fraction, and the filter caps keep this out of B+.

If you think about it, this dodge moves the rectifier-pulses jitter of a few millivolts from ground, where it will cause trouble, to B+ where it can be filtered out.

[R.G.]

I finally got some time today to try this out and made a video:
I did not notice any squealing while shorting the cable at the guitar end nor when shorted at both the input and guitar end. I'd be curious if you can draw any conclusions from this and also whether the other observations are of any value. I definitely do not understand why the squealing is MORE sensitive with the frame of the OT grounded to the chassis...

The video did help. I think what's going on is that you have two or three different oscillation modes that become prominent at different times.

The OT primary wires are carrying a bit less than twice the B+ voltage in peak to peak signal. That's a lot of signal voltage, so it can transfer a fraction of a volt to an input cable, even with only tiny capacitances.

Shorting the input cable at one end and then the other and having the squeal disappear shows that the signal is getting in through the signal wire, not the shield as an antenna. That's really good to prove, as if the shield was conducting the stuff in, the short would not have helped. It also shows that it's capacitive, not inductive pickup, because inductive pickup would not have been helped and might have become worse with the short on the free end of the cable.

One characteristic of grounding problems is that they're very sensitive to what is "grounded" to which exact point on the chassis. The OT frame being grounded to the chassis through a clip lead to the left end of the chassis is different from it being bolted in place inside the chassis.

As to why the transformer frame being grounded or not makes a difference at all, the primary of the OT has hundreds of voltage of signal on the internal wires, and each internal wire has some tiny capacitance to the transformer frame. This stew of capacitances wants to deposit some fraction of the (big) signal voltages on the frame. Hooking up a clip lead lets that current deposit on the chassis where the clip lead goes. Enough gain, and it sings.

[sleipnir]

R.G. Thanks for your thoughts on this. I think my next step then will be to mount the OT to the chassis and implement the grounding improvements as best I can. It will be interesting to see if the change in grounding will help with the squeal, but I realize there is a good chance the squeal will remain, so I will only remain cautiously optimistic :-)

I'll report back once I have had a chance to work on the amp some more this week!