Killing high gain oscillation with tone stack lift

[lord preset]

I have a Small Special 50w that I seem to have been working on forever. I thought I finally had it done and was giving it one last test before I shipped it off to my friend and noticed that at extreme gain settings with the tone stack lift/boost engaged the amp produced dramatic low frequency helicopter sounds. Diming the gain and engaging boost is not something I or my friend would do in actual use but the amp should be stable at any setting.

This amp follows the attached layout except I omitted the deep switch (actually I tried it but didn't like it) and added a simple tone stack lift switch that lifts the ground to the 10k resistor on the bass pot. Since discovering the helicopter invasion I have played with different value resistors to ground rather than an outright ground lift. Anything between about 40k and 150k eliminates most but not all of the audible oscillation - i.e. the volume of the oscillation is reduced quite a bit and it only becomes noticeable when the volume pot is nearly dimed vs. about 3 o'clock with no resistor. I have also tried adding a 68k grid stopper to v1b but there was no change.

Any suggestions on how to eliminate rather than just reduce the oscillation?

[xtian]

I'm suspicious of lead dress. Have you chopsticked the wires around during oscillation?

[lord preset]

I'm suspicious of lead dress. Have you chopsticked the wires around during oscillation?

Indeed I have. No joy thus far.

[xtian]

How about trying a grid stopper on V1b?

[lord preset]

I did try a 68k grid stopper on V1b. No effect.
I may try exorcism.

[strelok]

Does moving your hand or a probe over the top of the board or near the input jack make any difference in the behavior of the oscillations?

[pdf64]

I can't see that grid stoppers is likely to have much benefit with motorboating?
Rather the low end gain should be reduced, eg lower value of cathode bypass caps, and / or the B+ decoupling increased, eg higher values, especially at the preamp nodes.
I hate layout only circuits - schematic too please!

[Stevem]

I would have to say that it's due to your buss grounding of all the power supply filters to that one spot!

Only your first two filters and the bias filter should be grounded where they are all currently grounded, the other two filters should pick up there ground over where the preamp section picks up all of there grounds!

A motor boating issue is always due to either the current flow in the grounding scheme, or gain stages that are not uncoupling enough from a latter gain stage due to not enough filtering and or not enough resistance between filter sections, the more gain you have, the more filtering and or resistance you need!

Of course you could have a poor performing ( high ESR ) filter trying to uncouple V1 but failing, for starters just jump more filter across that preamp section filter if changing the grounding layout FIRST does not releave the problem!

Also is that long grid wire off of the input Jack sheilded and ground at only the input Jack end?

[R.G.]

I'm with stevem and pdf64. Motorboating is a low frequency oscillation, and it comes from different causes than high frequency oscillation. It's from phase shift at the low frequency rolloffs.

Generally, it's from failure of the power supply decoupling network to actually decouple stages. Sometimes it's from grounding issues.

It happens when a later stage pulls some current and that lowers its power supply voltage. This pulls down on the power supply voltage of an earlier stage, and lowers its output.

The lower signal level transient is phase shifted through the low frequency rolloffs of the series-capacitor couplings of the stages between the two that are connected through the power supply, and the transient phase gets changed enough so that at some frequency, it reinforces the pulling-down on the stage that started this.

If at any frequency, the gain around the feedback loop is enough to reinforce the oscillation (that is , loop gain >= 1) and the phase is right to reinforce, it oscillates.

To stop it, you can either change the loop gain by cutting the signal passed through the feedback path - usually the power supply, sometimes the ground - or change the phase. Those series-R/shunt-Cs are in the power supply of tube amps to do precisely this.

Making the series-R bigger loses you voltage. Making the caps bigger does a better (and lower frequency) job of shunting the signal, so generally better/bigger caps are used. Another dodge is to split one R into two of half the value and install yet another shunt-C to ground. This adds both more attenuation and more phase shift, but usually the attenuation wins and the oscillation stops.

As an aside, I read that starting up a water-moderated nuclear reactor has a "motorboating" failure mode, as it oscillates due to the interaction of the neutron flux and the moderating effect of the hydrogen in the water. This is a bad place to let things get out of hand.

[Reeltarded]

Yikes.

[lord preset]

I can't see that grid stoppers is likely to have much benefit with motorboating?
Rather the low end gain should be reduced, eg lower value of cathode bypass caps, and / or the B+ decoupling increased, eg higher values, especially at the preamp nodes.
I hate layout only circuits - schematic too please!

I wish I had a schematic, but I snagged this from a TAG Small Special thread, and I never found a known good version.

[lord preset]

i will try separating the PS grounds. I usually do but in this case I just followed the layout. Now that I think of it, my previous Small Special had a different PS design due to a non CT transformer so I laid it out differently and did separate the grounds. OTOH if I had not added the Boost switch To this amp I never would have run across this issue. I'll report back. Thx

I'm with stevem and pdf64. Motorboating is a low frequency oscillation, and it comes from different causes than high frequency oscillation. It's from phase shift at the low frequency rolloffs.

Generally, it's from failure of the power supply decoupling network to actually decouple stages. Sometimes it's from grounding issues.

It happens when a later stage pulls some current and that lowers its power supply voltage. This pulls down on the power supply voltage of an earlier stage, and lowers its output.

The lower signal level transient is phase shifted through the low frequency rolloffs of the series-capacitor couplings of the stages between the two that are connected through the power supply, and the transient phase gets changed enough so that at some frequency, it reinforces the pulling-down on the stage that started this.

If at any frequency, the gain around the feedback loop is enough to reinforce the oscillation (that is , loop gain >= 1) and the phase is right to reinforce, it oscillates.

To stop it, you can either change the loop gain by cutting the signal passed through the feedback path - usually the power supply, sometimes the ground - or change the phase. Those series-R/shunt-Cs are in the power supply of tube amps to do precisely this.

Making the series-R bigger loses you voltage. Making the caps bigger does a better (and lower frequency) job of shunting the signal, so generally better/bigger caps are used. Another dodge is to split one R into two of half the value and install yet another shunt-C to ground. This adds both more attenuation and more phase shift, but usually the attenuation wins and the oscillation stops.

As an aside, I read that starting up a water-moderated nuclear reactor has a "motorboating" failure mode, as it oscillates due to the interaction of the neutron flux and the moderating effect of the hydrogen in the water. This is a bad place to let things get out of hand.

[lord preset]

You are all Super Geniuses. Separating the grounds fixed it. Thanks.

[R.G.]

Great! Good work.

Just as a bit of somewhat under appreciated knowledge, everything oscillates at some point if the gain is high enough.

[strelok]

Glad you got it working!

I'd even take that statement further to say that there is no difference, in terms of the class of governing equations, between an oscillator, a filter, and an amplifier. They're all the same, it is the differences in the constants/parameters of the system that determine which of the three we choose to call it. Another fun fact, since most any tube amp circuit is capable of undergoing chaotic oscillations, it is necessarily operating in three dimensions or more.