Peavey Ultra 60 distorted overtones (clips attached) - filter caps?

[Enzo]

So it only happens when the reverb is ON. SO that means the tubes have nothing to do with it, so waste no time on tubes. Besides, the tubes have no way of knowing if the reverb is on or off. You determined that the Q1 functioned, because it turns on and off. What if your car engine made a loud screechy grinding noise whenever it was running? But the car drove down the road OK. That test doesn't tell us the motor is OK. It suggests the motor has an issue. Just as Q1 can function without being OK.

And design flaw??? If that were the case, ALL of these amps would do it, along with all the others using this circuit.

The clamp circuit again could be bad despite the fact it functions. There is more than one failure mode for most parts. But the clamp circuit has no idea if the reverb is on or off. Your issue is noise.

You determined the reverb works OK and adds no crap when ON. You found the reverb control affects the amount of crap. Sounds right. You report the crap only happens when the reverb is turned OFF. OFF is not the same thing as set at zero. Q1 is the element to turn it OFF. My first suspect would be Q1 is noisy.

Q1 is turned off by a gate voltage. That gate has a diode-like relationship to the S-D path. Not at all rare for a JFET to fail in that region.

Without a new JFET, you could test for this: remove the JFET. Does the problem stop? I'd be hard pressed to say how that would not work.

As tedmich said Mouser has 450 of them in stock at a little over $3.
https://www.mouser.com/ProductDetail...YO8iqZag%3D%3D

[ACHIEVEIT]

Thanks so much for the replies guys and thank you Enzo for providing illuminating reasoning (hopefully it'll help me get better)!

Indeed removing Q1 removes the artifact.
At this point, it seems that Q1 is most likely the culprit.
I will get my hands on some replacements and report back.

Thanks again!

[ACHIEVEIT]

So I believe I am onto something. Warning: this is going to be a long post.

So, I substituted Q1 with a new replacement (respecting safe handling practices) and you can imagine my surprise when the artifacts didn't go away.

That made me take a deep look into the circuit and try to understand how it works. I dusted off my understanding of the superposition theorem and calculated the Q1 gate voltage when R49 is grounded and when not. My measurements in the circuit agreed with the theory.
In essence CR7 cathode is either -4.83V or 4.87 V and this makes Q1 either cut off or pass the signal.

So, unable to comprehend the existence of the artifacts when Q1 is supposedly cutting off, I decided to simulate the circuit. Given the opamp gain and assuming it's not designed to clip, I assumed a signal of 30mV peak at the output of the reverb tank.

The problem was then obvious - Q1 was conducting on the negative cycles.

Without overly analyzing it, I went through the options in the LTSpice N-Channel JFET library and found 2SK170-GR to be a good fit.

At that point I thought it would be over (mind the different pinout between J231 and 2SK170-GR) but again to my surprise it wasn't. This time (armed with a newly-acquired oscilloscope), I was able to measure the exact signals in the circuit and found that it's totally possible to get a signal of ~80mV peak at the reverb tank output (high input, bridge humbucker, clean channel volume fairly high).

This was what I measured after Q1 with the original J231. The yellow signal is the guitar signal at the input and the blue signal is Q1 source/Reverb pot.

And this is what I measured with 2SK170-GR

This was a vast improvement already but there were still some faint distorted overtones at strong signals. Again, the simulation agreed with my experience when I increased the reverb pan output signal to 75mV peak

What is more, the simulation showed that even when the reverb would be ON, a portion of the positive signal would be distorted

And then I understood what's going on: for Q1 to be in cut-off, the Q1 gate needs to be lower in voltage compared to Q1 drain by the respective Vgs(off) stated in the datasheet. But in this case, with a 75mV peak signal before the opamp, the output after the opamp is swinging between -6V and 6V approximately. But as stated in the beginning, CR7 cathode is either -4.83V or 4.87 V. So, in the negative peaks, Q1 gate was actually at a higher voltage than Q1 drain and this made Q1 conduct. Respectively, in the positive peaks, Q1 gate was at a lower voltage than Q1 drain and this made Q1 cut off. (This also explained why J231 performed poorly compared to 2SK170-GR. J231 has a Vgs(off) between -1.5V and -5V while 2SK170-GR has a Vgs(off) between -0.2V and -1.5V)

So, the only solution I could think of was to make Q1 gate more negative (when the reverb is OFF) and more positive (when the reverb is ON). Unfortunately, this proved to be quite the challenge. I used Wolfram Alpha to try to find a solution to the two equations I had at hand (that represented Q1 gate voltage as a function of R48, R49, R50, R51 when the reverb is OFF and respectively ON) but that was impossible. I even created a small LabVIEW program to see in real-time how changing the values of R48, R49, R50, R51 would affect CR7 cathode voltage in both conditions (reverb OFF/ON). It quickly became apparent that no matter how much I would tweak the values, the existing configuration would never grant me the desired ~+/-7V at Q1 gate.

And then I had an idea: I noticed how we had +/-24V available in other parts of the circuit. Disconnecting R48/R51 from the +/-17V supply and connecting them to the +/-24V supply while keeping R48, R49, R50, R51 the same, would probably give us the desired +/-7V at Q1 gate. Indeed the simulation agreed and in fact all artifacts/clipping were removed.

So, the absolution came when I implemented this idea on the actual circuit - no sign of distorted overtones anymore even at the strongest of signals!

Luckily, it is easy to get the desired +/-24V from the nearby R96/R97 (don't forget to substitute J231 with 2SK170-GR)

To sum up. I think I stumbled upon one of Peavey's design flaws. My interpretation as to why not many have noticed or reported it is because they probably mistaken it with preamp distortion (remember, the reverb sub-circuit is fed all the time the signal coming from the preamp, so the higher the signal it gets, in other words the higher the volume in the clean channel in this case, the higher the chance of the distorted overtones appearing). However, I am using a Two Notes Torpedo Live and headphones, so it was pretty obvious that something was not right, as I am not misled by sheer volume. Hope this helps anybody else in the future, it was quite an adventure to figure it out

[g1]

To sum up. I think I stumbled upon one of Peavey's design flaws.

That is amazing work, the effort is very much appreciated.
I will point out however that it hinges entirely on the output level of the tank. You were only seeing the beginnings of the glitch at about 20mV RMS. The articles I have read say the return circuit should be designed around 1 to 5 mVRMS, Rod Elliot's site says at max. drive to the input, the output will usually be less than 10mV (for 'B' output impedance tank).
So that is about the only other thing I can think that might cause such an issue, some odd or defective tank. 50mV out seems very hot.

If you are 100% sure there is no other reverb circuit issue that might be giving too much output, this is a good workaround. Either way, it has worked out for you and that is what counts.

[ACHIEVEIT]

That is amazing work, the effort is very much appreciated.
I will point out however that it hinges entirely on the output level of the tank. You were only seeing the beginnings of the glitch at about 20mV RMS. The articles I have read say the return circuit should be designed around 1 to 5 mVRMS, Rod Elliot's site says at max. drive to the input, the output will usually be less than 10mV (for 'B' output impedance tank).
So that is about the only other thing I can think that might cause such an issue, some odd or defective tank. 50mV out seems very hot.

If you are 100% sure there is no other reverb circuit issue that might be giving too much output, this is a good workaround. Either way, it has worked out for you and that is what counts.

Thanks a lot for sharing this g1!
Indeed that also crossed my mind but I dismissed the suspicion around the reverb tank, as the (basic) measurements appeared correct. I am measuring 58 Ohms DCR at the tank input and 203 Ohms DCR at the tank output and for that type, 8EB2C1B, I should anticipate 58 Ohms/200 Ohms according to websites.

So that means that maybe too strong a signal is sent to the reverb tank?
I haven't spent too much time on the left side of the reverb sub-circuit (going to the tank input) the truth is... However, I had measured all resistances, diodes and supply voltages (also no cap is shorted) and they are ok. Also, from a previous test I had done, the 4558 opamp is not to blame, so that leaves only Q2 and Q3. Could they be amplifying the signal more than they should? I was under the impression that this configuration (is it push-pull emitter follower?) was used to increase the output current (in order to drive the tank) and doesn't affect the voltage.

By the way, could you share the articles you are referring to? I'd be very interested to read them!

[g1]

Your DC resistance readings for that tank indicate it is the correct one. They translate to 600 ohm input impedance, 2250 ohm output impedance.
I may have misread the Elliot article thinking that the 10mV tank output was a max. capability. But he did state that if he over-drove the input 10 times more than recommended, it only doubled the output voltage.
Here's the links to the 2 articles:
https://sound-au.com/articles/reverb.htm#s7
https://www.amplifiedparts.com/tech-...d-and-compared

[Helmholtz]

I found that Fender BF and SF amp drive the transducer into heavy core saturation. Maybe something similar is used here.

Also the RMS to peak conversion by square root of 2 only works for steady state sine signals.
With other signals the instantaneous peak voltage can be much higher for the same RMS value.as RMS involves averaging.

[tedmich]

NICE! A quite impressive collection of data and logic (my day job!)

Do you think you could show a concise schematic of the modified circuit? Very similar ones abound in Peaveys, of which I have several.

[ACHIEVEIT]

Thanks a lot for your comments guys and the links g1!

Sure, here are the changes noted in red. For clarity I have isolated the reverb sub-circuit. As mentioned, convenient places to get the +/-24V are R96 and R97 (not visible here, refer to the full schematic, bottom-left section).

By the way I realized I never posted an oscilloscope photo of the output after the reverb tank. So, for completeness, here it is (again, high input, bridge humbucker, clean channel turned quite high, strong strum). The photo was taken when I was still using the stock +/-17V on R48/R51, which is why you see the spikes on the blue trace. Now there are none.

[Helmholtz]

So you're seeing up to 150mVpp. Does your scope calculate RMS?
I doubt a DMM can cope with such high crest factor.

[tedmich]

My Ultra 120 212 is a bit different

with an added Q8 (MPS-6531 from BOM, 6530 in schematic) where your R48/R51 are. It should probably handle the higher voltages fine, but will need testing.

Man those old Toshiba 2SK170's are hella expensive now! Sigh I guess everything is...

[ACHIEVEIT]

My Ultra 120 212 is a bit different

with an added Q8 (MPS-6531 from BOM, 6530 in schematic) where your R48/R51 are. It should probably handle the higher voltages fine, but will need testing.

Man those old Toshiba 2SK170's are hella expensive now! Sigh I guess everything is...

I ran a quick simulation with the sub-circuit you posted (Ultra 120 212) and it's working better. When the reverb is OFF the Q1 gate voltage is so low that there is no chance that Q1 will conduct because of the signal swing in Q1 drain. Respectively, when the reverb is ON, the full signal is passed without any alteration.

My interpretation is that the engineers at Peavey understood their mistake and improved the circuit in that model.
This is a better solution than what I posted - in the one I posted the margin is rather small and if a guitar with stronger output from the pickups in combination with the channel volume totally maxed (resulting in bigger signal at the tank output), there is still a chance that some spikes might occur (although this hasn't been yet the case in my real-life experience). So, if someone's willing to go the extra mile in the modification, this is probably a better solution - if someone wants minimal intervention, the one I posted should be fine too.
Thanks a lot for posting this!!

Ultra 120 212 Reverb OFF:

Ultra 120 212 Reverb ON: