Shorted Reverb Tank Causes Distortion

[Phostenix]

I thought this odd, so I'm looking to see if it's typical. Long story short, If I short the reverb send (or plug in a tank that's shorted), I will get an output from the amp that is nasty distortion - even with the Reverb control turned all the way down. It works fine with a good tank or no tank at all. This is a Crate GT3500H and the relevant part of the schematic is attached. "F" on the schematic is the output of the preamps - after the level controls.

It seems odd to me that shorting the output of IC5:B to ground would pull down the non-inverting input with it. If that's normal op amp behavior, it's a good thing to keep in mind for future troubleshooting.

[Enzo]

I don;t know if I am convinced of your analysis yet, but maybe. Not about it happening, but about the why. If you want to be scientific, lift C55 and see if the effect goes away. SHorting the output does put 22 ohms to ground on the IC output, and that could do things to the power rails. Scope them while it happens. ALso within the chip, since the recovery stage is the other half of it, there could also be effects we wouldn't predict.

[Phostenix]

Since the output of the other half of the IC is shorted to ground when the Reverb control is all the way down, I figured that eliminated its impact. As for lifting C55, I will be pulling the preamp board down the road to make some other changes, but it's a lot of knobs to remove to do tonight.

So, in your experience, shorting the output doesn't usually pull down the + input?

[Phostenix]

I'll check the voltage on the rails (while shorting the Send output) tomorrow.

[Enzo]

Look again at the other side of that IC, the output of it flows from pin through a cap to the top of the 10k pot. That IC sees the 10k no matter what. Turning the pot down grounds 100k R91 only, not the IC output.

I wasn;t thinking that driving the shorted load would pull down the power rail per se, but it could cause current spikes that could get on the rail and thus be sent into the signal path elsewhere. That is just a working hypothesis for me though.

COnsidering the input pins of the IC are at virtual ground, it it possible for it to react through C55 and affect the signal path there. hence the test.

AS to my experience, I have never checked out this exact scenario. I run a commercial repair bench. if I get this situation I see that a pan is shorted and causing problems, I then repair or replace the pan, and then I am done. IN fact, I will spend more time trying to figure out just WHY it acts this way because of helping you with it, I wouldn;t bother in my own shop. it is a perfectly legitimate question, and if it helps you learn, then i am all for it though.

[Phostenix]

I realized after I posted about the Reverb control shorting IC5:A that I should have been clearer. I meant, wrt the signal path, turning down that pot effectively shorts the output of the reverb return circuit to ground. I would think that anything odd from that half of the IC would have no effect on the amp.

I only thought it interesting because I would expect a shorted reverb tank to simply give you no reverb, not the dramatic effect it had on the "whole" amp. If someone described this issue to me (before yesterday), I never would have started with the reverb tank to troubleshoot.

My interest in this op amp behavior is that, if this is typical, I need to keep it in mind on future projects that a short on the output of an op amp could be causing distortion "upstream" of it. The 22 ohm resistor could well be the thing that makes this a unique situation. Any idea why it's there?

I won't get back to this amp until this afternoon for more test.

[Enzo]

Which is why we have to find out if that is the route your problem takes.

[Phostenix]

I had a few minutes for what my wife has now dubbed, "The Amazing Fiddle Box" ("you never run out of things to test on that amp").

OK, one of the reverb tanks I have has a shorted input xfmr, which is where this started. Last night, for test purposes, I was shorting the Send RCA plug against the case of the tank to confirm that shorting the reverb Send caused the distortion (the output of the tank still connected to the Return cable to the amp).

Today, I put a signal generator into the amp & an 8 ohm dummy load on the output & put the scope on the output. I shorted the Send cable to itself and got no distortion. Turns out, the distortion only happens when I short the Send cable to the Return cable's ground (the tank when it's connected)! Shorting to the amp's chassis wil not produce distortion. Shorting to the ground of the input jack will produce a slight distortion of the sine wave.

I have to come back to this later, but now it looks like I definitely have to pull the preamp board & do some more investigating.

[Enzo]

SO the output of that IC is driving ground directly then, without even that 22 ohm, resistor.

[Phostenix]

Here's what I found.

The distortion is worse at low levels of output than high. It affects all 3 channels. The inverting input of IC4:B (pin 6) develops a square wave of about 10mV p-p on it when this is happening.

I disconnected the "F" end of C55 & the problem went away.

Here are some screen shots. All are taken with the scope probe clipped onto the Line Out jack (equivalent to J2 - pin 1 on the previous schematic). All except #4 are at 10mV/Div on the scope. #4 is 50mV/Div.

The 1st is a clean signal (1v p-p 1kHz sine wave into the clean channel of the amp but reduced by the Volume control).
The 2nd is a shot of the distortion with the shorted reverb tank connected.
The 3rd is the same setup with the Volume turned down a little to show the distortion mire clearly.
The 4th is the same setup with the Volume control turned up, cleaning up the distortion.
The 5th is what the distortion looked like if I took the reverb Send cable & shorted its tip to the input jack. The distortion is less than when connected to the tank. If I had the ground of the scope connected to the input jack's ground, shorting the Send cable to the input ground would NOT produce the distortion.

[Phostenix]

Looks like #5 didn't make it, so I'll add it here.

Now, if I clipped onto the "F" side of R77, the distortion didn't show. Clipped onto the IC4:B side of R77 got me a small distorted waveform that turned into a square wave very quickly as I raised the Volume control. I've included 2 pics of that so you can see the waveform before & after it turned square.

[Phostenix]

And here's part of the preamp schematic that shows the input jack's sleeve switch that ties to point "D" on the previous schematic (the mute for IC4:B).

[Phostenix]

This gets stranger the more I look at it.

Here are 4 more pics - all with the shorted reverb tank connected. All 10mv/Div range.

The 4 pics are:
1. The +16V rail going to the op amps of the preamp.
2. The -16V rail of the preamp.
3. J16 - the Reverb Return ground.
4. J9 The Reverb Send ground.

These signals will all go away if I turn off the signal generator. They will all go away if I disconnect the shorted reverb tank.

BTW, the amp, scope & signal generator are all connected to the same power strip.

[Phostenix]

Now, here's a dual trace of the Reverb Send & Reverb Send ground - J6 & J9 on the schematic with a good reverb tank connected. Scope set to .1v/Div on both channels. The audio signal is clearly on the Send ground at a lower level. R22 allows this to happen. So, if the tank's input xfmr is shorted, that signal can develop over any small resistance to ground that exists on the PCB ground traces. I think that's what's going on here.

[Phostenix]

Here's what the Reverb Send signals look like with a good tank & one with a shorted input.

Scope ground is clipped to the chassis ground of the amp. Scope channel 1 is clipped to the Reverb Send - signal (J9 on the schematic).
Scope channel 2 is clipped to the Reverb Send + signal (J6 on the schematic).
Chanel 1 is set to 20mV/Div & Channel 2 is set to .5V/Div

Picture 1 is a good reverb tank. Picture 2 is with the shorted one.