Peavey Windsor Doug Hammond Mod Phase Inverter

[bobloblaws]

I'm in the process of converting a Peavey Windsor 100W head to a quasi-JCM800 according to the "Doug Hammond" mods. https://www.aronnelson.com/gallery/m...H/amp/windsor/

Preamp is working fine as far as I can tell but the long tail phase inverter is not. I've tried replacing V3 and I've triple checked my wiring but I am obviously missing something. Attaching a couple of pics showing expected voltages in and around the PI circuit and my actual voltages. Anyone have any thoughts on this?

- Bob

[Enzo]

My first thought: what is wrong with it? What does it not do or what unwanted thing does it do? Without a problem it is hard to advise action.

[bobloblaws]

My first thought: what is wrong with it? What does it not do or what unwanted thing does it do? Without a problem it is hard to advise action.

There is no signal beyond the coupling cap that precedes the PI. On the Marshall schematic it is C9 (22nF) and on the Windsor it is C13 (100nF) (not shown on the pic I uploaded, it's connected to "sig".) The other side of the cap is connected to the master volume wiper and I have signal there but not on the PI side.

[Enzo]

OK, you HAVE a Windsor, you are trying to alter the circuit to resemble a MArshall model. So C13. C13 is on the preamp board. You have ssignal on the preamp side of C13. Where exactly does the signal disappear. I get that it is not at pin 2 of the tube. But is it even at the power amp end of C13 itself? From there it crosses the preamp board to pin 2 of the ribbon connector. got it there? Now pin 2 of the other end of the ribbon, on the power tube board. And between that ribbon and pin 2 of the tube, there is a wire jumper in the middle of the trace. Does it make it past that?

All that to detect a break in the path. But the path could be shorted to ground too. You have moved parts around and I guess cut traces? A mistake might have happened: a blob of solder, a part removed and not replaced. Or a wrong part installed. Or two parts swapped places. WHo knows.

Oh, and I just re-read. You have signal at the master voume, and not past C13. You don't actually say you have signal AT C13. From the master volume, the signal path goes through a couple of relay contacts and/or cutouts on J4 FX return.

One quick test, do you still get signal coming out the FX send jack?

And the PI itself, I don't know what your green voltages are on th Windsor drawing, but a long tail pair with zero volts on the cathodes tells me the tube is not conducting. Are the heaters even glowing?

[bobloblaws]

Oh, and I just re-read. You have signal at the master voume, and not past C13. You don't actually say you have signal AT C13. From the master volume, the signal path goes through a couple of relay contacts and/or cutouts on J4 FX

return.

Yes, signal on the master volume side of C13 but not on the other side. And I definitely confirmed continuity from C13 to pin2 of V3. The FX loop/relay stuff muddies the waters. The guy I bought it from told me it was stock. When I turned over the preamp board I found a bunch of jumpers and cut traces which as far as I could tell changed the loop from parallel to serial. I thought he was full of crap but he swears he bought it new and knew nothing about it. Then I found a pic on a forum showing virtually the same mod and the poster believed it was done at the factory. So who knows. Anyway, I changed it further so that neither the FX loop or the boost circuit are in play at all.

The last time I checked whether V3 heaters were working they were. But I couldn't be positive that was the case when I jotted down those voltages so just now I went to confirm. This time I removed the tubes (V3 to V7) and pulled out the power amp board and reinstalled the tubes so I could troubleshoot from both sides of the board. Anyway, now I'm actually getting positive voltage readings in the PI circuit (albeit quite a bit lower than what I've seen documented for JCM800's.) Maybe the tube was seated improperly, I don't know. I'm still seeing 0V at pin 2 and one Marshall diagram I saw suggested 28V was normal at that point. I can audio trace the signal to the power tube grids but I'm getting nothing at the speaker output and nothing past C13 responds to the master volume control. Here's the new voltage readings.

[Enzo]

So right at C13, signal on one end and not the other? Check that the no signal end is not shorted to ground. And if not, then I sure would try a new cap in that spot.

That FX loop circuit is weird. If we connect something to the FX loop, with the loop turned on, it works. To turn off the loop, they added relay SK102 to short across it. Unfortunately, that sets up a feedback loop sending the return signal right back through the send. So, many complaints, and I don't doubt the factory modified a bunch of them.


Remember, you don't HAVE a MArshall. I don't see any reason the voltages in this Peavey would match a JCM800 model. I would look for reasonable voltages, not exact duplicates. Right off the bat I'd be surprised of the B+ voltage was the same in both amps.

If both halves of the tube are heating, and you still have zero volts on the grid, that lends credence to my shorted to ground notion.

There IS NO NORMAL grid DC voltage when measured to ground. When someone says 28v, that means in HIS amp with HIS meter. Look at your MArshall drawing. It shows 45v on the cathodes - I'll believe that. It shows 28v on the grid. That grid is 17v more negative than the cathode, Imagine how poorly that stage would perform if it was really biased that far down. The problem is your meter. It has an internal impedance. The grid voltage comes from the bottom end of R16. I'd expect maybe a volt drops across that resistor. So 44v for discussion. Theree is no grid current, so that 1meg resistor does not drop any voltage while in use. The real grid voltage is then 44v. But when you go to measure, your meter impedance and R17 form a voltage divider. like this: 44v, through R17, and through your meter. SO it reads artificially low, REAL low. But meters are not all the same, so there is no way I can predict that your readings will match his.

You want to knw the real grid voltage? Measure DC voltage across cathode to grid directly. SUbtract that from the cathode to ground voltage and you will have the grid to ground real voltage.

HAving explaained all that, if I am, servicing one, I will check to ground, and if I see something in the range of half the cathode voltage, I'll consider it good, I don't need to calculate the "real" voltage.

And when I blame your meter, there is nothing wrong with it, my nice FLuke acts the same way. Rather it is a limitation of your gear and procedures you must be aware of.

[bobloblaws]

So right at C13, signal on one end and not the other? Check that the no signal end is not shorted to ground. And if not, then I sure would try a new cap in that spot.

Sorry, my last post was not logically self contained. At first I responded based on where I was at in the previous post and your inquiry, then I went and did additional testing. So " signal on the master volume side of C13 but not on the other side" did not reflect current status.

[Enzo]

OK, then what IS current status? I can only respond to what you present.

[bobloblaws]

"I can audio trace the signal to the power tube grids but I'm getting nothing at the speaker output and nothing past C13 responds to the master volume control."

Sorry for the confusion.

[Enzo]

The master has no effect on what you see/hear at the power tube grids? That sounds like crosstalk. WHat kind of level is the signal at those grids? Plug a signal into the FX return, what comes out?

[bobloblaws]

So right at C13, signal on one end and not the other? Check that the no signal end is not shorted to ground. And if not, then I sure would try a new cap in that spot.

That FX loop circuit is weird. If we connect something to the FX loop, with the loop turned on, it works. To turn off the loop, they added relay SK102 to short across it. Unfortunately, that sets up a feedback loop sending the return signal right back through the send. So, many complaints, and I don't doubt the factory modified a bunch of them.


Remember, you don't HAVE a MArshall. I don't see any reason the voltages in this Peavey would match a JCM800 model. I would look for reasonable voltages, not exact duplicates. Right off the bat I'd be surprised of the B+ voltage was the same in both amps.

If both halves of the tube are heating, and you still have zero volts on the grid, that lends credence to my shorted to ground notion.

There IS NO NORMAL grid DC voltage when measured to ground. When someone says 28v, that means in HIS amp with HIS meter. Look at your MArshall drawing. It shows 45v on the cathodes - I'll believe that. It shows 28v on the grid. That grid is 17v more negative than the cathode, Imagine how poorly that stage would perform if it was really biased that far down. The problem is your meter. It has an internal impedance. The grid voltage comes from the bottom end of R16. I'd expect maybe a volt drops across that resistor. So 44v for discussion. Theree is no grid current, so that 1meg resistor does not drop any voltage while in use. The real grid voltage is then 44v. But when you go to measure, your meter impedance and R17 form a voltage divider. like this: 44v, through R17, and through your meter. SO it reads artificially low, REAL low. But meters are not all the same, so there is no way I can predict that your readings will match his.

You want to knw the real grid voltage? Measure DC voltage across cathode to grid directly. SUbtract that from the cathode to ground voltage and you will have the grid to ground real voltage.

HAving explaained all that, if I am, servicing one, I will check to ground, and if I see something in the range of half the cathode voltage, I'll consider it good, I don't need to calculate the "real" voltage.

And when I blame your meter, there is nothing wrong with it, my nice FLuke acts the same way. Rather it is a limitation of your gear and procedures you must be aware of.

Yes, I wasn't expecting exact duplicates, but I would have expected in the ball park. For example, my cathode voltage is 19V as opposed to 45, yet my plate voltages are quite a bit higher than the nominal JCM800 counterparts. So am I wrong to think that my recorded voltages indicate a problem? They seem lower than reasonable to me.

Thanks for the tip about the meter impedance setting up a voltage divider. That didn't occur to me. As it happens, I went ahead and measured direct from grid to cathode and the grid is 18V more negative than the cathode.

[bobloblaws]

The master has no effect on what you see/hear at the power tube grids? That sounds like crosstalk. WHat kind of level is the signal at those grids?

I'm not sure how to quantify the level. I'm using a simple audio trace and my scope can't seem to detect a signal, at least not in the way that I'm accustomed to. Maybe I need to flip the probe to X 10.

Plug a signal into the FX return, what comes out?

As I said, the FX loop has been taken out of the equation.

[Enzo]

Can you inject a signal at the point the loop used to be? I am looking to bypass everything C13 and before to just test the power amp.

One way to characterize level is to listen at one point and listen at the next without touching the controls of the tracer. If it gets louder, we have gain in the stage. If for example we have a good strong signal at the master or at C13, but at the power tube grids it is much quieter, that tells us the signal is not being amplified.

Just thinking aloud: the cathode voltage of the PI comes from the current through the tail, If it is half what you expect, that could mean one side is not conducting. ANy chance pins 3 and 8 are not well connected on that PI?


Higher than normal plate voltages and low cathode voltages mean the tube is not conducting as much.


What are your B+ voltages here? And what is B+ in your Marshall model?

Got a scope? Good. Can you apply a test signal to the regular input and see a good signal at the master or at C13. Don't turn up the scope vertical from then on. The signal will not shrink in the PI or power stage, so if you were having to turn it up to see, then we have missing signal level.

[bobloblaws]

Can you inject a signal at the point the loop used to be? I am looking to bypass everything C13 and before to just test the power amp.

One way to characterize level is to listen at one point and listen at the next without touching the controls of the tracer. If it gets louder, we have gain in the stage. If for example we have a good strong signal at the master or at C13, but at the power tube grids it is much quieter, that tells us the signal is not being amplified.

Just thinking aloud: the cathode voltage of the PI comes from the current through the tail, If it is half what you expect, that could mean one side is not conducting. ANy chance pins 3 and 8 are not well connected on that PI?


Higher than normal plate voltages and low cathode voltages mean the tube is not conducting as much.


What are your B+ voltages here? And what is B+ in your Marshall model?

Got a scope? Good. Can you apply a test signal to the regular input and see a good signal at the master or at C13. Don't turn up the scope vertical from then on. The signal will not shrink in the PI or power stage, so if you were having to turn it up to see, then we have missing signal level.

As far the scope goes I had better luck with my old analog scope than I did with my digital one, but the bottom line is, there is in fact a net signal loss from V3 pin 2 to the power tube grids that I can detect with the audio probe, instead of the expected amplification. I believe the connection between the cathodes is solid but I had not re-measured V3 pins 1 and 6 once the tube started conducting where apparently it hadn't been previously, so the updated diagram I uploaded actually need more updating. Providing that now. I don't know why the voltage on pin 6 is so much lower than pin 1. R66 measures 100K in circuit.

B+ for the PI node (they call it B++) is 270. I'm also uploading one of the Marshall diagrams I used for reference showing 390V on that node.



JCM800_2204.pdf

[Enzo]

If the PV B+ is 120v lower than the MArshall, that right there might exoplain why the voltages are lower than expected.

I don't know why the voltage on pin 6 is so much lower than pin 1.

Your B++ is only 270v to start with and you measure 268 at pin 1. Pin 6 is not "so much lower", pin 1 is so much higher. There is no voltage drop across R45, which means that triode is not conducting. No conducting, means high plate voltage and low cathode voltage and no amplification.