Strange oscillations in speaker... Tremolo at fault or bad tube?

[loudthud]

The bias wiggle tremolo counts on the balance in the output stage to cancel the LFO signal in the output transformer. Because the balance is imperfect, some LFO goes to the speaker. If the speakers can't handle it, get different speakers.

[SoulFetish]

The Concertina phase inverter causes the LFO signal to be imbalanced at the power tube grids because the PI's output impedance is different between plate and cathode. Easy to see with an oscilloscope. Smaller coupling caps to the power tubes might make a small improvement. Try 0.022uF. An oversize OT exacerbates the problem. There will always be some imbalance. Live with it.

Ah, yes this makes sense

[Britt Rossman]

Quick Update...

With an output tube bias modulating tremolo circuit like the one you are building it is necessary to set the bias such that the largest tremolo modulation amplitude (depth) does not force the output tubes into cutoff or saturation. Yes, it is a compromise, but you cannot simultaneously have a hot bias and real deep tremolo with the 6G2 circuit. I suggest that you try a cooler bias setting like ~25ma per tube and determine if that reduces the problem. Give it a try and let us know the results.
Tom

Well Tom, I haven't adjusted the bias yet because I think the problem lies with the trem swinging too much voltage. I would rather have a shallower tremolo. I had the Depth control at ~2/3rds up and stopped being able to read voltage swing from the -44v bias point down to -57. Higher than that, my meter couldn't keep up, so I figure the swing is somewhere around +/- 20V or more....

Then again folks, maybe I missed the point of everyone sharing their 2c about bias (all because I said something about maximizing power, I believe). Regardless of power amp design goals, the issue I am trying to fix is the bias wiggler screwing with my output, so I feel like the bias arguments (other than Tom Phillips) may be OT...

The bias wiggle tremolo counts on the balance in the output stage to cancel the LFO signal in the output transformer. Because the balance is imperfect, some LFO goes to the speaker. If the speakers can't handle it, get different speakers.

My OP stage is pretty balanced, I think. I don't have an O-scope to check this or see the trem signal... I feel like if I can simply limit the LFO depth, I may be able to solve this issue. Though admittedly, I may also have to change the bias point to have lower current draw... The tremolo is stronger for this design than any other I've heard - this can't be right...

Specs on the power section for those interested:

L-6L6
35.8mA
437.4V(a)
15.7W
-44.5V dc bias

R-6L6
37.2mA
435.6V(a)
16.2W
-44.5V dc bias

LFO:
Freq range >> 3.2-8.9Hz
AC volts out = 14

[loudthud]

A Voltage divider for the LFO signal is formed between the 220K grid leak resistor for each output tube and the output impedance of the phase inverter. If you can measure the LFO signal at the junction of the two 220Ks, then measure it at the output tube grids. You should see an imbalance of 20% or so.

[Jazz P Bass]

Well Tom, I haven't adjusted the bias yet because I think the problem lies with the trem swinging too much voltage. I would rather have a shallower tremolo.

Then again folks, maybe I missed the point of everyone sharing their 2c about bias (all because I said something about maximizing power, I believe). Regardless of power amp design goals, the issue I am trying to fix is the bias wiggler screwing with my output, so I feel like the bias arguments (other than Tom Phillips) may be OT...




Specs on the power section for those interested:

L-6L6
35.8mA
437.4V(a)
15.7W
-44.5V dc bias

R-6L6
37.2mA
435.6V(a)
16.2W
-44.5V dc bias

LFO:
Freq range >> 3.2-8.9Hz
AC volts out = 14

Lower the bias.

[Tom Phillips]

Well Tom, I haven't adjusted the bias yet because I think the problem lies with the trem swinging too much voltage. I would rather have a shallower tremolo... I feel like if I can simply limit the LFO depth, I may be able to solve this issue...

I agree. You are describing the trade off that I mentioned. Your build is not an exact clone of the original circuit so it's likely that some adjustments are needed. If the tremolo depth is sufficient for your needs without turning the depth control all the way up then you may just need to limit the LFO drive to the bias. In addition, it is easy to experiment with a cooler bias. Judge the result by the sound and don't worry about the "70% rule."

Also remember that the basic circuit works fine and is loved by many players. There is just something either malfunctioning or working outside acceptable parameters in your build at this time. It can be fixed and the problem is not caused by a sensitive speaker as was speculated in an earlier post.

Cheers,
Tom

[Britt Rossman]

There is just something either malfunctioning or working outside acceptable parameters in your build at this time. It can be fixed and the problem is not caused by a sensitive speaker as was speculated in an earlier post.

This is exactly what I thought, but I was unconvinced that the bias was the only culprit. Maybe it was the root cause, but not the entire equation...

Your build is not an exact clone of the original circuit so it's likely that some adjustments are needed. If the tremolo depth is sufficient for your needs without turning the depth control all the way up then you may just need to limit the LFO drive to the bias... and don't worry about the "70% rule."

I may still need to limit the tremolo depth, as I've never heard a tremolo of this style (one device for gain) be so strong and rich! I lowered the bias by replacing the 110k resistor with a 100k, which set bias at ~ -48V and roughly 26.2mA for L and 27.0mA for the R. The amp sounds very clean and crisp - I really can't get over it! (The JBL K110's are probably contributing a good bit to this...)

As far as the previous issues, here's my thought... The bias of the amp was set to give ~16W per tube to give ~32W out - this gave me the weird tremolo speaker coupling and lots of distortion that I couldn't place (intermodulation?). NOW, I've set bias to yield about 22W for the pair (like a really how set of 6V6s...) and thing sounds wonderful and clean - it lacks the sputtery/fizzy kind of distortion that seemed to ride under all settings... Was this due to hysteresis in the OPT? The OPT was originally built for PP pair of 5881's in an old Conn organ, so it may not have been up to handling that kind of power. If the OPT were built for 25W max, I gave it ~30 before and the amp had a few issues - all of which are fixed by lowering bias and output - then maybe the iron is just tired..?

In short, thanks to Jazz, Tom, and Soul for helping nail down the particulars of my franken-amp! Thanks to additional input by JM Fahey and g1 on specifics of bias points and possible speaker issues! And thanks to Loudthud for all the info on the trem and PI interactions with the Power section!

As soon as I can get a decent demo with the speakers etc. all mounted, I'll try to take some recording and post up here so you can see the results with your own ears!

[g1]

The bias of the amp was set to give ~16W per tube to give ~32W out - this gave me the weird tremolo speaker coupling and lots of distortion that I couldn't place (intermodulation?). NOW, I've set bias to yield about 22W for the pair (like a really how set of 6V6s...) and thing sounds wonderful and clean - it lacks the sputtery/fizzy kind of distortion that seemed to ride under all settings...

You are still confusing idle dissipation wattage and output power wattage. Example: many Fender Twins are biased around 50% for 6L6's which are 30W tubes. The output power for the 4 6L6's is not 60watts.
There is no straight formula equating idle dissipation and output power, they are not directly related.

Glad to hear you cooled the bias and largely corrected the issue.

[Chuck H]

I was going to offer the same correction as g1, but decided not to in case I was misinterpreting something. g is right. More current at idle doesn't mean more power out of the amp in use. In fact it often means less power out (but I think you discovered that by your own ears).

I was pondering your problem. While it's not uncommon, it's annoying. I think it was alluded that imbalance in the output halves can exacerbate the problem. So you could try a bias balance circuit?

To reduce the intensity of the trem you can try increasing the value of the 220k resistor that feeds the .1 cap that feeds the intensity pot. You might also be able to reduce the intensity with a smaller plate load to the oscillator triode. I don't think it will reduce gain enough to stop oscillation and it may tame things.

EDIT: Another thing though. Since your amp has more than a hundred volts greater Vp than the schematic posted you can take that for a hint. Bias wiggle trems work best in amps that run lower Vp and are biased pretty hot. The reason is that a colder bias, as required with more Vp, can allow the trem to hold the tube in cutoff for a longer time. This translates to odd, garbled sounding distortions for all but the cleanest tones. I learned his the hard way. I guess my point is that you'll want to get the intensity and balance under control because the effect will likely work better if you can keep the bias a bit warm. Test by ear and use safe electrical limits as your guide.