Tremolo and speaker pumping?

[Rhodesplyr]

General question: which types of guitar amp tremolo circuits are least likely to have the side-effect of pumping the speaker back and forth when turned on?

[J M Fahey]

Opto/LDR types.
Or FET modulated ones.

[Enzo]

Ones that don;t work directly on the power amp.

[loudthud]

There are three major types of tremolo. The opto or JFET which change signal amplitude without actually injecting the tremolo signal into the guitar signal. It works by variable attenuation of the guitar signal between stages. That has the lowest injection of LF into the signal. The next two types inject the tremolo signal into a tube along with the guitar signal. Since the bias on the tube is effectively changing, the gain of the tube changes.

The second type injects the tremolo signal into a push pull amplifying stage and relys on cancellation of the tremolo signal. Some work better than others. Injecting the signal into the power tubes (hopefully matched) can work but when they are driven by a Concertina phase inverter, the cancellation is not ideal. The impedance driving each power tube is different so cancellation is not balanced. Some amps inject the tremolo signal into the phase inverter. This is good in theory but care has to be taken to insure good balance and cancellation at the output transformer.

The third type injects the tremolo signal into a single ended preamp stage and then attempts fo filter it out. You will find a high pass network of perhaps 4 capacitors in series with resistors to ground between the caps. This type is found in some Vox and Marshall amps. In the Vibro-Champ and many other low powered amps there is no filtering except by the output transformer.

The Fender blonde era and Magnatone amps are really in catagories by themselves.

[jpfamps]

As Loudthud says the "pumping" is due to imbalance in the power stage in a push pull amp.

I assume that even if the amp is perfectly balanced at audio frequencies, its likely to be unbalanced at the sub sonic frequencies used for tremolo.

Presumably though it would be possible to have a trim pot to help cancel the LFO in the power amp.

[Chuck H]

The third type injects the tremolo signal into a single ended preamp stage and then attempts fo filter it out. You will find a high pass network of perhaps 4 capacitors in series with resistors to ground between the caps.

I always wondered what the point was in that circuit. I thought it might be some sort of phase correction circuit to align the two channels.

[Wombaticus]

?.. but when they are driven by a Concertina phase inverter, the cancellation is not ideal. The impedance driving each power tube is different so cancellation is not balanced.

This turns out not to be true. So long as one is careful not to draw grid current, and one ensures that the loads are otherwise equal, the balance of a cathodyne (with respect to voltage gain anyway - CMRR and PSRR are a different story) remains extremely good even at high frequencies. In fact, it's hard to do better. It's easy in this discussion to get bogged down in semantics, and it's true that in the strictly technical way that output impedance of individual nodes is measured that the output nodes have different output impedance. But in *actual operation*, both ouputs are active simultaneously, and this makes all the difference. What matters in practice is the differential output impedance (or half that, really) *between* the nodes. It's an old result that, for practical purposes, the "effectively seen" output impedance (here lurks a semantic danger, but close enough, eh?) from *each* node is both low (like CF low - several hundred ohms typically), and, counter-intuitively, equal. One can prove this with a lot of fancy algebra (see Preisman in the link, and also another classic paper by Jones). Or, as in the link below to my blog, by direct simulation. But really, if you really stop and think about the fact that Ra and Rk share a common current and are (ideally) the same size, then the result seems almost trivial, so long as the loads are also equal.

Wombat Amplification: On the output impedence of the cathodyne phase inverter

[loudthud]

You are neglecting the fact that whatever signal reaches the cathode of the Concertina becomes an input in a common grid configuration. That signal goes to the plate in phase and that reduces the effect that the impedance at the plate has on attentating the LFO signal there. Run a simulation on that.

Edit: Be sure to include the previous gain stage, as well as the grid leak and cathode resistor of the Concertina like a 5E3. They come into play because they anchor the grid. Global feedback from the OT back to the gain stage cathode should also have an effect.

[Wombaticus]

You are neglecting the fact that whatever signal reaches the cathode of the Concertina becomes an input in a common grid configuration. That signal goes to the plate in phase and that reduces the effect that the impedance at the plate has on attentating the LFO signal there. Run a simulation on that.

Edit: Be sure to include the previous gain stage, as well as the grid leak and cathode resistor of the Concertina like a 5E3. They come into play because they anchor the grid. Global feedback from the OT back to the gain stage cathode should also have an effect.

I think I'm missing the point of what you're driving at, could you try explaining another way?