New Princeton Reverb build PI issue

[Apexelectric]

Been struggling with an issue on a new build. It’s a Princeton Reverb Blackface type with a tremor-nator circuit for the tremolo.

Amp functions and sounds fine during the first couple minutes of playing but eventually looses volume and starts to distort at the same time. I’ve isolated the issue to the Cathodyne phase inverter. The tube seems to have a loss in current flow as the voltage increases at the plate resistor and the voltage drop decreases across it significantly. Starts out at around 80v drop then down to 3V. Output tubes stay biased properly and are drawing 24ma each.

I’ve checked tubes, coupling caps, solder connections, plate/cathode resistor values, tremolo circuit has been isolated, etc.

Any thoughts on what might be causing the change in PI behavior? Thanks in advance.

[sluckey]

Got a schematic? Especially interested in how you have the Trem-O-Nator connected.

[Apexelectric]

It’s built as per the Hoffman layout and the trem taps after the coupling cap on V3b and before the PI grid. I used a .68 250V cap between the tap point and the intensity pot. Seems to work well.

Haven’t bothered drafting a new schematic since I used other known schematics and layouts. I used revision three for the trem circuit. 

[sluckey]

Pictures, schematic, and layout that shows what YOU did to Hoffman's circuit would be mighty helpful. My crystal ball says there appears to be an error in the V4 circuit, but it doesn't reveal the details. Only you can do that.

[Apexelectric]

Gut shots. Wires have been removed from intensity pot and one end of tremolo tap cap has been removed from the turret.

[sluckey]

I see a 470K resistor on V4 pin 7 (directly on the socket). What's that about? Where does it connect?

[glass54]

Don't mean to be offensive, but can you clarify the wires on pot as highlighted at bottom? Or is this just temporary, hard to tell from photo.

[Apexelectric]

470K resistor is just a grid stopper to tame the distortion character. Issue occurred before this addition.  PI plate voltage was also taken from the unused C node of the power supply instead or the D node, as per the schematic. Also not part of the issue as it was a problem before moving power supply nodes.

Wires were removed from the intensity pot to eliminate the tremolo as a cause of the problem.

Other mods to the stock circuit are adjustable NFB, either stock value, double the value or no NFB at all. Testing done with stock value. Cathode bypass cap/resistor switch for V1A, .68/2700k or 22uf/1500k. Swamp switch for tremolo and voltage switch for B+ for the two different PT HV taps. Tested on the higher voltage that yields about 420V B+

One other interesting characteristic to this problem is that the phase inverter seems to stay stable until the volume is turned up past four or five then the voltage drop across the plate resistor changes and doesn’t recover until the amp is turned off for quite some time.

[Apexelectric]

Hopefully this schematic helps.

I’ve made changes to the Hoffman version to reflect what I’ve done on mine.

[PRR]

> looses volume and starts to distort at the same time. I’ve isolated the issue to the Cathodyne phase inverter. The tube seems to have a loss in current flow

That's a strange way to inject vibrato. I would not expect any gain-change from wobbling a cathodyne. It would change gain in the output stage, except in *opposite* way in the two tubes, and the grid-coupling would not pass low-rate trem.

That aside: I bet it is not wired as the plan shows. Or a part is bad. Disconnect the 0.68u cap, it may be leaking.

I don't think the 470k at cathodyne grid is reasonable. Cathodyne rarely distorts before the stage before or after it. And if there is a bogus solder joint near here, that would explain why the bias goes wacko in a few minutes.

[Apexelectric]

Vibrato has been disconnected including the .68 cap. 470k resistor was added after the problem started but will remove and eliminate as a contributing factor.

I did chopstick the whole amp to look for a bad connection with no luck.

[66Strat]

My bet is on a bad cathode resistor on the cathodyne triode.

[2deaf]

That's a strange way to inject vibrato. I would not expect any gain-change from wobbling a cathodyne. It would change gain in the output stage, except in *opposite* way in the two tubes, and the grid-coupling would not pass low-rate trem.

That aside: I bet it is not wired as the plan shows. Or a part is bad. Disconnect the 0.68u cap, it may be leaking.

I don't think the 470k at cathodyne grid is reasonable. Cathodyne rarely distorts before the stage before or after it. And if there is a bogus solder joint near here, that would explain why the bias goes wacko in a few minutes.

He's not wobbling a cathodyne.  He is changing the load on V3B from 50K to something way less thereby radically changing the gain of V3B. 

If he has a bogus solder joint so that the grid becomes disconnected, then the current through the tube would increase from lack of bias instead of decreasing as he has observed.

[Apexelectric]

My bet is on a bad cathode resistor on the cathodyne triode.

Both the plate and cathode resistors measure ok but I was thinking there might be a breakdown on one under load?

[2deaf]

The drop in current described above is what happens when you ground the grid of the cathodyne.  It doesn't quite shut the tube down, but it's close.  I'm sure there are other ways to achieve the same current loss, but I'd be looking for any way that grid could become grounded first. 

[Apexelectric]

The drop in current described above is what happens when you ground the grid of the cathodyne.  It doesn't quite shut the tube down, but it's close.  I'm sure there are other ways to achieve the same current loss, but I'd be looking for any way that grid could become grounded first. 

Makes sense but why the gradual decline in signal and the extended time to recover? Sounds more like a heat related cause. Would a bad resistor act like that?

[sluckey]

Could you temporarily replace those giant blue caps with a couple Xicons so we can examine all the stuff underneath them? Also give us the grid, cathode, and plate voltages.

[66Strat]

My bet is on a bad cathode resistor on the cathodyne triode.

Both the plate and cathode resistors measure ok but I was thinking there might be a breakdown on one under load?

My thought was that the cathode resistor may be failing (open) when heated during operation. This would cut-off or severely limit current flow causing plate voltage to rise.

[Apexelectric]

Could you temporarily replace those giant blue caps with a couple Xicons so we can examine all the stuff underneath them? Also give us the grid, cathode, and plate voltages.

Rather not have to pull the Sozos. I’ve gone through all the voltages of the preamp and output and they are pretty much dead on with the schematic when the amp is acting normally. The only voltage changes seem to be the ones in the PI. With the PI being fed from a different power supply node it’s a bit higher than the schematic normally(240v) when it’s behaving.  I’ll swap out the plate and cathode resistors in the PI later this afternoon when I get home and see if I have an issue there. I was thinking of using a 1W 82K instead, not sure I have 56k 1W.  If that doesn’t change anything, I’ll post the voltages on the other triodes.

[2deaf]

My thought was that the cathode resistor may be failing (open) when heated during operation. This would cut-off or severely limit current flow causing plate voltage to rise.

Changing the bias resistor from 1K to 100K (or so) would cause the symptoms described.  Maybe measure the voltage drop from the socket lug to the negative end of the 1K?  The socket itself may be creating resistance between the lug and the tube pin, so the test isn't 100% conclusive.

[Apexelectric]

My thought was that the cathode resistor may be failing (open) when heated during operation. This would cut-off or severely limit current flow causing plate voltage to rise.

Changing the bias resistor from 1K to 100K (or so) would cause the symptoms described.  Maybe measure the voltage drop from the socket lug to the negative end of the 1K?  The socket itself may be creating resistance between the lug and the tube pin, so the test isn't 100% conclusive.

Was thinking of upgrading the two 56K resistors to 1W, primarily due to the voltage present. I’ll change the 1K bias resistor as well. I certainly wouldn’t change it to 100K but I think I know what you’re getting at.

[66Strat]

My thought was that the cathode resistor may be failing (open) when heated during operation. This would cut-off or severely limit current flow causing plate voltage to rise.

Changing the bias resistor from 1K to 100K (or so) would cause the symptoms described.  Maybe measure the voltage drop from the socket lug to the negative end of the 1K?  The socket itself may be creating resistance between the lug and the tube pin, so the test isn't 100% conclusive.

Yarp, that could do it too. I think it is most likely related to heat induced drift in resistor value. Or, maybe the socket pins are expanding in some weird manner when heated. It wouldn't hurt to clean and re-tension the socket.

[Apexelectric]

My thought was that the cathode resistor may be failing (open) when heated during operation. This would cut-off or severely limit current flow causing plate voltage to rise.

Changing the bias resistor from 1K to 100K (or so) would cause the symptoms described.  Maybe measure the voltage drop from the socket lug to the negative end of the 1K?  The socket itself may be creating resistance between the lug and the tube pin, so the test isn't 100% conclusive.

Yarp, that could do it too. I think it is most likely related to heat induced drift in resistor value. Or, maybe the socket pins are expanding in some weird manner when heated. It wouldn't hurt to clean and re-tension the socket.

Will certainly check but it’s new and I’ve wiggled tubes to check on connections during the brown out period so probably not that. Hopefully a resistor failure. Fingers crossed!

[2deaf]

Was thinking of upgrading the two 56K resistors to 1W, primarily due to the voltage present. I’ll change the 1K bias resistor as well. I certainly wouldn’t change it to 100K but I think I know what you’re getting at.

I would do some non-invasive testing before replacing parts.  Whereas getting 3mA from the lug to the negative end of the 1K is inconclusive, getting something like 3V when the cathodyne is malfunctioning is very conclusive.

[2deaf]

I think it is most likely related to heat induced drift in resistor value.

If the resistance of the 56K resistor connected to ground rose to some ridiculous value (mega Ohms?), that would also cause the symptoms.  Measuring the voltage drop across that 56K resistor while the cathodyne is malfunctioning should determine if the resistor is doing that or not.   

[Apexelectric]

Was thinking of upgrading the two 56K resistors to 1W, primarily due to the voltage present. I’ll change the 1K bias resistor as well. I certainly wouldn’t change it to 100K but I think I know what you’re getting at.

I would do some non-invasive testing before replacing parts.  Whereas getting 3mA from the lug to the negative end of the 1K is inconclusive, getting something like 3V when the cathodyne is malfunctioning is very conclusive.

I have definitely observed firsthand that the voltage drop across to 56K resistor will go from 80 V to 3 V during this process.

[2deaf]

It is not necessary to completely ground the cathodyne grid in order to nearly stop the current through the tube.  A 100K resistor from grid to ground is almost as effective.  Even a 1M resistor will seriously reduce the current, which brings me to my point.  If your volt meter has a 1M input impedance, it will give you a very erroneous reading for the grid voltage.  Even a 10M input impedance will alter the grid voltage significantly.

[Apexelectric]

It is not necessary to completely ground the cathodyne grid in order to nearly stop the current through the tube.  A 100K resistor from grid to ground is almost as effective.  Even a 1M resistor will seriously reduce the current, which brings me to my point.  If your volt meter has a 1M input impedance, it will give you a very erroneous reading for the grid voltage.  Even a 10M input impedance will alter the grid voltage significantly.

I might not be on the same page here. How would reading the voltage drop across the PI plate resistor to check current flow relate to the partial grounding of the PI grid?

[2deaf]

I might not be on the same page here. How would reading the voltage drop across the PI plate resistor to check current flow relate to the partial grounding of the PI grid?

We are definitely not on the same frequency.  I was just saying that you cannot directly measure the grid voltage and gave one reason why.  So when you go to measure the various voltages, don't bother trying to measure the voltage on the cathodyne grid.

[Apexelectric]

Gotcha now! Thanks for clarifying. Totally understand how that would not be helpful.

[sluckey]

It's 4 resistors and 3 caps. You can replace all of it in 15 minutes. If it turns out to be a wiring error or bad solder joint you will probably find that in the process.

When you get it fixed I suggest you don't use that .68 cap. Connect the INT pot to the left side of C15 (as shown on the schematic).

[2deaf]

Gotcha now! Thanks for clarifying. Totally understand how that would not be helpful.

It would be better to measure the voltage across the 1M resistor instead of trying to measure the grid voltage because then the meter won't affect the performance of a properly operating cathodyne.  The voltage drop should be zero or very nearly zero.  Voltage across the 1M resistor would indicate a path to ground at the grid end of the 1M resistor, which would cause the diminished anode current.

[Apexelectric]

Gotcha now! Thanks for clarifying. Totally understand how that would not be helpful.

It would be better to measure the voltage across the 1M resistor instead of trying to measure the grid voltage because then the meter won't affect the performance of a properly operating cathodyne.  The voltage drop should be zero or very nearly zero.  Voltage across the 1M resistor would indicate a path to ground at the grid end of the 1M resistor, which would cause the diminished anode current.



Looks like it was the 1M resistor. Replaced it and it seems stable now.

It's 4 resistors and 3 caps. You can replace all of it in 15 minutes. If it turns out to be a wiring error or bad solder joint you will probably find that in the process.

When you get it fixed I suggest you don't use that .68 cap. Connect the INT pot to the left side of C15 (as shown on the schematic).

So tap at the plate resistor on V3a before the coupling cap not after the coupling cap? I had some issues with a decrease in signal strength when I tried other tap points in this amp before settling on the one I ended up with.

[sluckey]

So tap at the plate resistor on V3a before the coupling cap not after the coupling cap?

NO! Sorry, my bad. It seems that C15 is shown twice on the schematic, once on the preamp page, then again on the power amp page. I was looking at the power amp because that's where the PI is shown. I would connect the INT pot to V3 pin 7 via a resistor. I'd start with 470K and work my way down until satisfied. Are you using a 50K pot? May need to try a 100K or 250K pot.

Using a cap as you did will work but it probably introduces a varying tone effect as well as varying volume. Kinda like a weak automatic wah wah pedal. If that's OK  with you then connect it as you originally had it.

[Willabe]

Using a cap as you did will work but it probably introduces a varying tone effect as well as varying volume. Kinda like a weak automatic wah wah pedal.

That might sound good. 

[2deaf]

How about if he really does have two C15's and they are connected in series and the tremolo pot is connected to their junction with no .68uF cap?

[sluckey]

How about if he really does have two C15's and they are connected in series and the tremolo pot is connected to their junction with no .68uF cap?

That would work if there really were two C15s. But that's not the case. There is only one on the layout and only one on his actual board. Easy to make it so though.

[PRR]

> better to measure the voltage across the 1M resistor instead of trying to measure the grid voltage because then the meter won't affect the performance of a properly operating cathodyne.

True. But if you realize the beast is mostly a Cathode Follower, then if the cathode is right the grid must be right.

[Apexelectric]

How about if he really does have two C15's and they are connected in series and the tremolo pot is connected to their junction with no .68uF cap?

That’s what I wanted to try next. I figured I would double the value and use two, then tap the junction of the two caps. I wanted to minimize the signal loss as much as possible while still preserving the intensity of the effect. Seems like all the tap points I tried created an unacceptable amount of loss and the current configuration was the best but still lossy. Although I didn’t use a resistor after the tap. Would that have lp? It was better with a 250KA pot but I don’t have a RA type on hand.

[2deaf]

That’s what I wanted to try next. I figured I would double the value and use two, then tap the junction of the two caps.

You don't need to double the value.  The input impedance of that Cathodyne is something like 16M, so the cut-off frequency is ridiculously low with either a .022uF or .011uF cap. 

[Apexelectric]

In an effort to keep it the same cutoff frequency I figured two .047 in series would match the current value but create the tap point I need and isolate the PI grid from the tremolo circuit.

If I recall correctly, I had an issue with the signal before I used a cap to isolate it from the grid of the PI. The larger value I used, the better the effect and the stronger the overall output of the amp was. But that was having the cap in series with the intensity pot

[2deaf]

True. But if you realize the beast is mostly a Cathode Follower, then if the cathode is right the grid must be right.

True.  And, believe it or not, I do realize the similarity between an AC-Coupled Cathodyne, a Self-Biased Cathode Follower, and an Un-Bypassed Gain Stage.  But if the cathode is not right and I suspect the grid has found an extraneous path to ground, then it would create a voltage drop across the 1M grid-leak resistor that I could detect without the meter itself creating an additional malfunction.

 

[Apexelectric]

True. But if you realize the beast is mostly a Cathode Follower, then if the cathode is right the grid must be right.

True.  And, believe it or not, I do realize the similarity between an AC-Coupled Cathodyne, a Self-Biased Cathode Follower, and an Un-Bypassed Gain Stage.  But if the cathode is not right and I suspect the grid has found an extraneous path to ground, then it would create a voltage drop across the 1M grid-leak resistor that I could detect without the meter itself creating an additional malfunction.

 

Problem turned out to be a misbehaving 1M grid leak. The earlier post was hard to discern the way it got sandwiched in the quoted reply. Thanks for pointing me in the right direction!

[Apexelectric]

Regarding the tremolo, creating a tap point in between two caps doesn’t do anything more to help the signal loss problem but it does sound better, more clarity, so it’s a step in the right direction there. Adding resistance after the tap does not help either. So I think the best compromise at this point is going to be the 250kra pot with the coupling caps in series. Guess I’ll be ordering a few pots.  Will mess with it some more to see if I can find a better combination. 

[sluckey]

I would connect the INT pot to V3 pin 7 via a resistor. I'd start with 470K and work my way down until satisfied. Are you using a 50K pot? May need to try a 100K or 250K pot.

Have you tried this?

BTW you don't have to use a RA pot.

[Apexelectric]

I would connect the INT pot to V3 pin 7 via a resistor. I'd start with 470K and work my way down until satisfied. Are you using a 50K pot? May need to try a 100K or 250K pot.

Have you tried this?

BTW you don't have to use a RA pot.

Will give that a shot