distortion on leading edge of note

[pula58]

I had 1.5K grid stopper on the grids (right at the sockets) of the output tubes. I just tried an experiment where I added an additional 2.2K of resistance.
Unfortunately, no change in behavior.

I tried some other output tubes, no change in behavior.

[Firestorm]

I had 1.5K grid stopper on the grids (right at the sockets) of the output tubes. I just tried an experiment where I added an additional 2.2K of resistance.
Unfortunately, no change in behavior.

I tried some other output tubes, no change in behavior.

Pentodes have a really tiny grid to plate capacitance. You be able to go very large on the grid stopper before you can hear it. At least to check whether it makes a difference in your problem.

[pula58]

I had 1.5K grid stopper on the grids (right at the sockets) of the output tubes. I just tried an experiment where I added an additional 2.2K of resistance.
Unfortunately, no change in behavior.

I tried some other output tubes, no change in behavior.

Pentodes have a really tiny grid to plate capacitance. You be able to go very large on the grid stopper before you can hear it. At least to check whether it makes a difference in your problem.

Hmmm...true..according to old data sheet: C grid-to-plate is 0.6pF. The measured voltage gain of the output tube from grid to plate (measured in my circuit) appears to be roughly 10 or so (i.e., 20dB). So, the Miller effect would make the cap ~10x larger..so 0.6pF becomes 6pF. Plus, the input capacitance (grid to cathode) is about 10pF. So, 16pF total, and 20kHz gives maximum resistance of: 50kohm or so assuming that other parastic grid wiring capacitances are negligible (not sure about this assumption). Maybe I'll try 15k, that's 10x more than the original 1.5K I started-out with.

[Firestorm]

It's why they invented the screen grid. Take advantage of it just to see.

[pula58]

I tired 15K on both grids (in addition to the 1,5K that was already there). No change in behavior.

I may post a photo of the chassis so that someone might spot something silly that I may have done....

[pula58]

chassis guts shot

[jelle]

Move your signal wires away from the tonestack wires. Look at how fender blackface amps were done. Please let us know! Good luck.

Jelle

[pula58]

I cleaned-up the wiring around the tone stack...It seemed to make no difference.

I can get this to happen if I use a signal generator to drive the input of the amp. I increase the output amplitude (at the speaker) using the master volume until I can hear the low-freq sounds I have been chasing. It seems to occur at the point where the grids of the output tubes start to clip.

Attached is a scope image of the output (at the speaker) and one of the PI outputs. By the way, I have 16.5K total grid stop resistance. 1.5K of it is right at the tube socket, and the other 15K is on the main circuit board.

the yellow scope trace is the voltage at the speaker.,

the blue scope trace is one of the PI outputs, on the PI side of the 16.5K grid stop resistance.

BTW: I love my O-scope! USB port. YES!

[surfsup]

Looks like you have a nipple trace in your pi, i'm on my phone what kind of pi is this? Valvewiz had a solution for this.

[pula58]

Looks like you have a nipple trace in your pi, i'm on my phone what kind of pi is this? Valvewiz had a solution for this.

it is a standard long tailed pair. 470 ohms sets the current, then, 22k underneath it to 50 ohm and 820 ohm resistors that set the feedback ratio from the speaker output tap.

1Meg ohm R's from each grid (of the PI) to the junction of the 470ohm and 22k ohm tail R's.

[pula58]

I tried driving the input of the phase inverter directly with a signal generator. A very successful experiment because the low-freq noise still occured. I think this means that I can rule-out any funny business with the preamp, tone stack lead dress (even though I cleaned it up a bit), etc.

So, it's got to be something with the output stage, PI, power supply or bias.

I tied a new choke - made no difference
I tried new 1st and 2nd power supply caps (at "A" and "B" in schematics) - no difference.
I doubled the capacitance at the screens (at "B") - made no difference.
I tried new ouput tubes - no change
I tried a new PI tube - made no difference.
I removed the circuit board, and tried a new, cleaned-up circuit board - made no difference!

I added a new bias and solid state rectifer board - made no difference
I tried three different OT's, it made no difference.

The amp sounds great when not played loudly, but gets this low-freq sound when cranked.


I am pulling my hair out on this one!

[surfsup]

Your PI trace (blue) shows a band of frequencies being emphasized.

Your Speaker trace in yellow shows a band of frequencies being reduced or it could be poor frequency response.

Looks to me that you have a high frequency issue. (imo - perhaps someone more experienced will correct me if I am wrong). I should say I have only been building amps for a year and 9 months-ish. So take my statement with a grain of salt.

[Jana]

My two cents.

It looks to me like your phase inverter is driving the grids of the power tubes into saturation. That "nipple" is a tell-tale sign of that (cold showers will do it too). I suggest reducing the gain of the PI so that the phase inverter will start to clip before its peak voltage swing exceeds the bias voltage. That way, the grids of the power tubes will never be driven positive (when they are driven positive, they start to conduct current which the phase inverter cannot supply--hence the nipple).

You might try increasing the size of the dropping resistor in the B+ string (the one between the choke and the PI node).

[Tillydog]

the yellow scope trace is the voltage at the speaker.,

the blue scope trace is one of the PI outputs, on the PI side of the 16.5K grid stop resistance.

Something feels wrong here - that speaker trace doesn't look right for that PI output. Obviously you're driving the output grids into conduction a little, but not excessively, I think.

Can you get a trace showing both PI outputs simultaneously? Are they reasonably balanced?

Have you biased the output valves to the same level? What is that?

What is your OT rated for? I think you've got >50W output on that trace - perhaps the OT is saturating?

At the moment, I suspect that something in the PI / output stages is pretty unbalanced.

0.02p

Andy

[pula58]

the yellow scope trace is the voltage at the speaker.,

the blue scope trace is one of the PI outputs, on the PI side of the 16.5K grid stop resistance.

Something feels wrong here - that speaker trace doesn't look right for that PI output. Obviously you're driving the output grids into conduction a little, but not excessively, I think.

Can you get a trace showing both PI outputs simultaneously? Are they reasonably balanced?

Have you biased the output valves to the same level? What is that?

What is your OT rated for? I think you've got >50W output on that trace - perhaps the OT is saturating?

At the moment, I suspect that something in the PI / output stages is pretty unbalanced.

0.02p

Andy

Andy,
Both tubes are biased at around 60% (18W each). 466V plate voltage. The tubes are a matched set, and, are biased within 0.1W of eachother (by the way, my initial bias board for this amp had a bias adj as well as bias balance adj. So, I could exactly match their bias, and still had this problem).

The OT is rated at 40W, but, I had this very same problem when I used a 55W outpu ttranny with this amp.

And the problem, to re-state, is that when I play it loud, there is a low frequency non-harmonic artifact that I can hear briefly when I really dig-in to the strings. I can also get it to occur in the lab using a signal generator. AND, when I slowly increase the signal input into the amp (or adjust it by driving the PI input directly (bypassing the preamp) I can start to hear this low freq sound at pretty much the moment when the grids of the output tubes begin to conduct.

The output at the speaker (and I couldn't figure out a way to show this) (i.e., the yellow trace above) fluctuates at its max and min values. That is, the tops and bottoms of the output waveform traces flicker around. The scope trace (above) only captured one moment of that flickering.

I shall post a new trace of both PI outputs. I have 100K and 91K plate R's and a 10K trim pot for balance adjust. I adjust it to reduce the 2nd harmonic seem at the speaker output.