OT, Bias and Redplating (Asymmetrical Output from Cathodyne Phase Inverter)

[dankinzelman]

EDIT - I've edited the thread title to reflect emerging ideas about where the problems are. The inital question proved to have more to do with a bad oscilloscope probe than the actual amp problems.

Hi everyone,

I'm working on a Davoli C3 guitar amp. It runs EL34 output tubes and a mix of ECC82 and ECC83 tubes in the preamp, phase inverter and driver stages. It had a shorted output tube when I got it, so I swapped in a new pair and biased them to 17W, then deoxed all tube sockets. Now it turns on and plays sort of ok at low volumes, but turning things up immediately gets me a nasty, metallic distortion (seems to be uniform on all inputs).

I have sent a sine wave through it, and everything looks ok up until the phase inverter, where things get weird. Basically, the plate and cathode of the phase inverter outputs are not even close to symmetrical - the plate has a significantly stronger signal. I don't understand how this can happen.

I've tried swapping the plate and cathode resistors for that stage, and the two interstage caps following them, but it doesn't seem to have helped. I've also tried swapping in a known good tube in the phase inverter stage, and still no joy.

Any suggestions, or something obvious I'm missing?

[astronomicum]

What is the cathodyne cathode dc voltage and the plate dc voltage of the preceding stage?

[shooter]

metallic distortion

disconnect (cr), NFB
post scope shots from the 2.2k/100k junction of the PA tubes grids

[Merlin]

The symptoms seem to imply the cathode output is being unusually loaded. Could the following 1M grid leak be bad?

You're using a small test signal right? Because if the signal is large enough to overdrive the following driver amplifier it would also have a similar effect in loading the cathode.

BTW what reverb unit is that? It doesn't show a reverb transformer, or is that just a schematic shorthand?

[dankinzelman]

Astronomicum: here's the lower chassis with PI, driver stage and output tubes. My voltages are in orange, but I left voltages already on the schematic for comparison.

[astronomicum]

I do not see any issues with the voltages you posted.

[pdf64]

Scoping inside a feedback loop can show whacky stuff; what's displayed is a combination of the real signal and the error correction signal.
I suggest to open the loop and recheck.

[AlNewman]

Your voltages look good in your power section and the gain stages after your PI, then you're about 40v over on everything compared to your schematic. 
Something doesn't seem to be drawing current.  You might have to make a complete voltage chart and compare.

[ghostinthescope]

Might be worth verifying that the plate and cathode resistor values haven’t drifted. 

[tubeswell]

Check the 100nF coupling caps for leakage. One of them may be leaking DC (affecting output current and voltage from the cathodyne)

[dankinzelman]

Hey everyone,

just to let you know I'm hung up at the moment because I discovered one of my oscilloscope probes is bad (only 1.5k DC resistance from tip to ground when AC coupled). It's ancient, so I guess no surprise, but I'm hesitant to trust the readings and certainly don't want to connect it to anything at high voltage as is. So I'm waiting for new probes to arrive, then I'll do some more poking and share what I find.

[dankinzelman]

Ok I got my new probes and did some scoping today. I'm more confused than ever.

I fed 200mV 1khz sine wave to the input of the tremolo channel. All subsequent level variations were undertaken using the amp volume knob. Feedback (C.R. in attached schematic) from output transformer to cathode of first tube was disconnected. Amp was powered through a dim bulb current limiter fed from an isolation transformer, so voltages might be slightly lower than spec.

Output tubes are matched, nearly new EL34 from Electro-harmonix, recently biased to ~16.5W per tube. Output transformer DC resistance measures 33.3ohms and 42.8ohms from center tap to each respective plate.

I've followed the signal through the amp, and although the phase inverter at extreme levels does show gain spiking, it remains clean well into driver clipping. First photo shows EL 34 grids at max clean level, with scope probes on output tube pins. This is just before the driver starts to clip, and is about 40V peak to peak.  Is this a reasonable voltage swing to drive the EL 34s? I've inverted one channel to show waveform symmetry.

In 2nd photo top trace is the same, but I've moved the other probe to the dummy load. You can see how weird the lower trace looks, and this is only about 5V pk-pk across 8 ohms.

3rd photo is the same, but I've brought the amp volume pot down to try to get an output that still looks (kind of) like a sine wave. It only managed about 2.5V pk-pk to the dummy load.

4th photo is the same, but I've raised the volume until the driver stage itself starts clipping. We get a really messed up waveform with ~6.2V pk-pk across the dummy load.

My conclusions:

1. Things look ok up to the EL34 grid inputs (right?).

2. Dummy load shows crazy distortion, except at extremely low levels.

3. Amp sounds like garbage.

I'm wondering now if there's something wrong with the output transformer, or how I've wired it (I'm following factory labels on the back of the amp), or something weird with the output tubes. Any ideas of how to proceed? I also noticed my dummy load doesn't get warm at all, but the output transformer started to heat up a bit over the course of testing (it was still quite comfortable to touch it.

Also, the output to the dummy load appears phase shifted. Is this a normal effect of a transformer?

[shooter]

is about 40V peak to peak.  Is this a reasonable voltage swing to drive the EL 34s?

that seems about right, my drive signal for KT88's is typically 40-50vac

the output to the dummy load appears phase shifted. Is this a normal effect of a transformer?

inductors, (transformers) will have a shift, think ELI the ICE man.  voltage will lead current in an inductive circuit.

the fact that you only have about 4.5W (6^2 /8) at the load tells me something is off in your PA or setup.

are you using 10X probes that you didn't account for in the math?
did you swap out the PA tubes just to eliminate?
what is your G2 (sreengrid) VDC values?

[dankinzelman]

Hi Shooter,

thanks for your response. What's PA?

I counted the 10x probes in my math, but I will double check later, and check voltages of screen and control grid. I was quite amazed myself at the poor output. Does the distortion look strange to you or is it fairly standard looking?

[Willabe]

Shooter;

.... Amp was powered through a dim bulb current limiter fed from an isolation transformer, so voltages might be slightly lower than spec.

[dankinzelman]

Shooter;

.... Amp was powered through a dim bulb current limiter fed from an isolation transformer, so voltages might be slightly lower than spec.

I am not very experienced, but based on reading and watching videos, my understanding is that an isolation transformer is an important safety precaution when probing anything at high voltages. Is it enough to remove the dim bulb current limiter and leave the amp on the isolation transformer?

Also, I don't know what you mean by PA ("something is off in your PA or setup.").

[Willabe]

Just the light bulb limiter. It's used to test for shorts/over current. Once you know you have no shorts, you take the amp off the light bulb limiter. It will give false/low readings with it hooked up.

Shooter means power amp = PA.

[dankinzelman]

Here's voltages for the EL34s (without current limiter):

Pin 4 - G2 364V
Pin 3 - Plate 731V
Pin 5 - G1 -34V
Pin 2 - heater - 7.1VAC

[Willabe]

No, it also means using the light bulb limiter can throw off any scope readings.

[Merlin]

This does seem to point to a defect in the OT. If you swap over the EL34s, does the output signal still look the same or does the 'raggedness' on the signal peak also swap?

[dankinzelman]

I've tried swapping output tubes, but the funny jaggedness in the output waveform remains.

Also I noticed the output tubes red plating when fed the 1khz signal at high volume. I promptly turned the signal down, but on another thread, it was suggested to me that the driver p-p swing (around 45V before clipping) should not be greater than the static negative bias (-33VDC). Could this be part of the problem?

Is there a way to definitively test the funcionality of the OT?

[Merlin]

I've tried swapping output tubes, but the funny jaggedness in the output waveform remains.

You mean the raggedness does not swap to the other side of the waveform?

it was suggested to me that the driver p-p swing (around 45V before clipping) should not be greater than the static negative bias (-33VDC). Could this be part of the problem?

No, that's just the required size of the grid signal before you hit clipping (in a normal working amp).

Is there a way to definitively test the funcionality of the OT?

I guess you could disconnect it from the circuit and hook the primary up to the wall voltage, then see if the secondary waveform is a good representation of the primary. Or you can feed a smaller AC voltage into the secondary and look at the waveform on the primary. Or you can try the neon lamp test.

[dankinzelman]

No, the raggedness seems to stay on the same side. Here's photos when I really push it (can't do this for long because it red plates really quick). Top trace is output of driver coupling cap, 20V/div (before grid stopper). Bottom trace is dummy load at 10V/div.

At some point along the volume pot, I also got this weird hf distortion on each cycle - is this oscillation?

[Merlin]

I think this is looking more and more like a bad transformer. All kinds of craziness can happen if one half of the winding has a short.

[pdf64]

I find 1R cathode current sensing resistors are all but essential to identify an OT that doesn't have a 'hard' short.
Open the NFB loop and compare the waveforms at each cathode with the output.
If the current waveform is good but the output isn't, then the OT is probably causing some nonsense.

[dankinzelman]

You mean I solder in a 1ohm resistor between cathode and ground for each power tube, then probe across the resistor?

[pdf64]

Yes.

[dankinzelman]

What power rating do I need?

[Merlin]

Surely you know how to calculate power? P = I^2 * R
1/4W will do, 1/2W is better

[dankinzelman]

I don't have any 1ohm resistors. Will 100ohms be ok?

[Willabe]

I don't have any 1ohm resistors. Will 100ohms be ok?

No.

Just buy some, our host Doug sells them in his on line store, 1 ohm 1/2 w metal film; 

https://hoffmanamps.com/MyStore/catalog/parts3.htm

You can find them other places too.

[dankinzelman]

As it turns out, the amp circuit is different than shown on the schematic. Namely, there are 50 ohm 4W resistors between EL34 cathodes and ground. They measured correct. Updated schematic is attached.

I measured across these and got the following images (10V/division).

I'll make another post with images from the 'add' setting on the oscilloscope.

[dankinzelman]

Here are scope shots using the 'add' function with each probe measuring across the 50ohm resister on the cathode. Still 10V/division.

Can anyone suggest a reason for the periodic noise in the 2nd shot, and the early clipping?

EDIT The periodic noise stays on the same side even if I swap the tubes. I notice the noise appears at the signal peak, then disappears as I continue to raise the volume. Could this noise be the onset of insulation breakdown on one side of the OT primary which, once exceeded, leaves a near perfect short?

[pdf64]

I was hoping to see the output OT secondary wave on one trace, one cathode current then the other on the other trace?

[shooter]

also with NFB (c.r.) disconnected

[dankinzelman]

NFB was disconnected for all scope images taken today.

Top trace is cathode current, bottom is dummy load. Still 10V/division.

As a reminder, the cathode current measurements are across a 50ohm (not 1 ohm) resistor, as the amp already had this resistor installed.

[Merlin]

Can anyone suggest a reason for the periodic noise in the 2nd shot, and the early clipping?

Looks like oscillation during a fraction of the audio cycle, aka squegging. This could indeed be caused by a bad transformer having wacky leakage inductance due to a shorted turn.

[pdf64]

With the NFB disconnected, what do the waveforms in the ECC82 gain stage and cathodyne look like?

[dankinzelman]

I have a 2-channel scope. Please specify what you want to see on each trace.

[shooter]

I'm interested in the PI output, where from the earlier schematic, the Grid leak, Grid stop and coupling caps all meet.  one probe at each junction.

[dankinzelman]

Sorry if I'm a bit confused. You want the probes at the spots marked in green or marked in purple on this schematic?

[shooter]

pinkish arrows, would expect to see a signal 20-40VAC-ish

[dankinzelman]

I just blew up my DMM due to the high B+ on this amp. Hopefully no damage to the amp itself (it blew a fuse). Any recommendations for a reasonably priced DMM that can handle 750+ VDC?

[Willabe]

You might have just blown the meters fuse.

[dankinzelman]

I don't think so. I opened it up, it looks pretty fried to me.

Wasn't expecting this, the meter is rated 1000VDC Cat III, which is much worse than what I've been feeding it (right?).

[HotBluePlates]

I just blew up my DMM due to the high B+ on this amp. Hopefully no damage to the amp itself (it blew a fuse). Any recommendations for a reasonably priced DMM that can handle 750+ VDC?

... the meter is rated 1000VDC Cat III, which is much worse than what I've been feeding it (right?).

That depends on what/where you were probing when it blew up.  If it was an output tube plate while a signal was fed into the amp, there could easily be >1000v on the output tube plate when swinging towards low/zero plate current.

[dankinzelman]

I was trying to check the bias, probing the center tap of the OT, so there was no signal going into the amp. It read ~724 just before exploding.

I have a feeling the DMM was not of the best quality. There were no third-party (UL, etc) logos on it. Just glad I didn't get hurt.

I have ordered a Brymen meter BM252s, 1kV CAT II, should be ok, right? Consensus according to my research is they make quality products...

What is a safe way to measure high voltages like this? Do I need to make a voltage divider with a couple really high value resistors? Or can I trust the meter?