Troubleshooting intermodulation distortion

[oxbow_lake]

Hello,

Sorry to ask you all to look at this again, but I built and have been playing through this amplifier which has a top-boost-style preamp and PI feeding a homebrew output stage based on a 6BN11 dual pentode compactron. selected for matched pentodes.

voxbow 1.0.png

I have noticed what I think is intermodulation distortion at any gain/volume level. It happens only when two or more notes are played at once. There is an audio clip attached which shows the ghosting I'm talking about. This is with the volume turned pretty low, and the master loud enough to hear the amp but not to audibly clip the output tube. As I mentioned, it doesn't really matter how low I have the gain set, I hear this distortion at all times.

IMD sample.mp3

Where I need help is figuring out how to troubleshoot it. Switching out tubes makes no difference. Sending the FX loop to another amplifier results in no distortion, so the preamp does not seem to be the cause of IMD. When I plug the guitar directly into the FX return the distortion returns, so clearly it's either happening in the phase inverter or the output stage. I have a scope, but I'm not sure what I'm looking for in this case since it's not just straightforward clipping that I can test with a single test tone.

Or maybe someone sees an obvious flaw with the design that would cause this? I'm probably going to get a lot of "that 6BN11 is just a poor choice," which is maybe fair, but not particularly educational. The pentodes are intended as IF amplifiers, which seem like they should be perfectly acceptable for AF.

Thanks for any help!

[maxkracht]

Have you tried power amp cathode bypass caps?

Edit: I'm an idiot, this would probably do the opposite. Just had an eye out for schematic oddities.

[xtian]

All amps exhibit IM distortion. I can't hear much of it in your audio example.

[LOUDthud]

I would try a balance control on the PI plate resistors, or an active current source in place of the 47K.

[oxbow_lake]

Thanks everyone. For the balance control, does that look something like a 10k pot in series with one of the PI plate resistors? Even if it doesn't fix the (perceived!) problem that's something I'd like to try just to see for myself how it affects the sound.

[cdemike]

Thanks everyone. For the balance control, does that look something like a 10k pot in series with one of the PI plate resistors? Even if it doesn't fix the (perceived!) problem that's something I'd like to try just to see for myself how it affects the sound.

Check out the phase inverter section here. This is one of a few ways of implementing a balance control. I've also seen more radical versions that use 250KL feeding the wiper and the plates getting fed directly from either end of the pot, i.e., with no plate resistors other than the track of the balance pot; you can decide the range of adjustability.

[LOUDthud]

I'd use something more like this:

PI_Bal.GIF

[oxbow_lake]

Tried balancing the PI, surprisingly didn't make any difference that I could hear, at least with gain low enough to avoid clipping.

I also tried plugging into an EL84 single-ended amp I have, and got the same ghosting/distortion. Plugged into a super clean solid state chip amp there is no distortion. So maybe this is just what should be expected using tubes.

Thanks again for your suggestions.

[oxbow_lake]

Hello again,

A slight change of topic here. I reworked the circuit to use 2 6K6GTs instead of the 6BN11. Current schematic with measured DC voltages is attached. It sounds good, but seems like the output power is low. Per the 6K6 datasheet, a push-pull pair should be getting me around 10 watts.

With an 800Hz sinusoid test signal, master volume set to maximum, and the volume control set to set to just-before-clipping, a 100mV p-p input gets to 285Vp-p on each 6K6 plate. Each 6K6 is idling at 25mA, so like 86% plate dissipation which is hot but the tubes aren't red plating or anything.

The output transformer is an old but supposedly NOS Stancor A-2312 which has a primary impedance of 14k, and the box it comes in says "6-12 ohm speaker." Per the Stancor catalog this is a 10W OT.

With the above settings and input signal I get 3.2Vrms over an 8 ohm speaker, which is only 1.3W? It is indeed several decibels quieter than my SE EL84 amp. What am I missing here?

Thanks!

[cdemike]

I'm not certain you have the right primary impedance -- can you confirm the model number? According to this old Stancor catalogue (https://www.earlytelevision.org/pdf/stancor_61.pdf), this OT should be showing your 6K6GTs approximately 28k ohms (page 5, section F). Dropping the impedance of the speaker to 4 ohms should get you a serviceable OT ratio. I'd recommend against using it as-is currently given the impedance mismatch.

Also, biasing that hot will reduce headroom, so I think second order of business is to Increase your cathode resistor's value if they're still biased that hot.

[oxbow_lake]

Here is a picture of the box; the stamp on the OT itself also says a-2312. I agree, the catalog page you link says 14k to 4R, so it's weird that the box says 6-15?

Anyway I reduced the bias to 18mA which is about 60%, and tried sending the signal to a 4 ohm speaker. Now I max out at around 3.5W (4W with the amp cranked and clipped), which is better but still a little lame?

Hammond suggested that their 1750AB OT would be appropriate for PP 6k6s - primary impedance is 9k, output is 8R. I'm reluctant to shell out $75 to try it and see though!

[cdemike]

Yeah I’d feel pretty confident that’s the right part number. Did your power supply change at all from the schematic posted?

I might be doing too deep a read, but it sounds like you might have a scope on hand. If there’s any doubt about the OT specs, you could also use that to measure the winding ratio of the OT by measuring the difference in amplitude between input and output signals without any guesswork.

[oxbow_lake]

I used the heater supply to test the OT. With 3.3V dropping over the secondary (in parallel with that 220R resistor) the primaries showed 177V, so turns ratio of 53.
If the catalog primary of 14k is right, then 14k/53^2=5 ohm secondary?

So ok, the 4 ohm speaker is a better fit, and should be showing the tubes 11.2k? That seems reasonable. Still not sure where the power is going then when I use the 4 ohm speaker?

All the DCRs seemed reasonable too, seems like the OT itself is okay.

The power supply voltages have not changed significantly. Everything is a couple of volts higher than before changing the power tube bias but also the wall voltage is a little higher today.

Thanks again for your help.

[oxbow_lake]

Ok bear with me and my stupidity here folks. I had those 470k resistors before the 6k6 grid stoppers left over from the 6BN11 to keep them from clipping instantly. Removing them gets me to 6.5W, and still the limitation is the headroom in the preamp.

Double checking the loadline for 6K6s it seems like even with a 14k primary the maximum output should hit 9.5 watts, so I think the power amp is working fine here. Thanks again for your eyes.