remote-cutoff compression

[myboysherman]

I have a couple of 6ba6 remote-cutoff pentodes and naturally I wondered how they would sound. Seems like free compression and silly amounts of sustain are possible with these things.

A few issues come to mind. First is that they would sound "interesting" in the way that my mom finds Frank Zappa "interesting." So I would prefer to run it in parallel with a more conventional circuit and mix the two somewhere downstream.

Second, the compression is asymmetric. I think that would lead to 2nd harmonics which might be nice. But I am leaning toward using two of them in a PI configuration (Quad did this with a pair of EF86 in a see-saw thing which looks like fun.) If I do that, I will wind up with a diff pair. Since I would like to attenuate the output I will either have to use a gang pot and feed to opposite ends of the real PI (or even post PI) or I can remix the diff pair and stick it back wherever I want.

I would rather remix the pair before attenuation but I have no idea how to do this. Could I feed one end to the grid of a concertina and the other to it's cathode and then take the output from only the plate?

Third, I have been assuming that the remixing of the pairs would reduce the compression by half, but my brain started hurting, so my assumption may be wrong. If the mix eliminates the compression altogether, I will have to run the PI/compressor in Class AB . . . where's my aspirin . . .

I expect that lots of people have considered this, so is there a giant red flag that my beginner mind is missing?

Oh, I'm building a head in an apartment, so microphonics won't be much of an issue.

[Steve Conner]

No, what you call "remixing" the signal will not get rid of the compression. It cancels out the even harmonics and control voltage feedthrough, but the compression is not affected.

The classic vari-mu compressors like the Fairchild 660/670 were balanced push-pull throughout, and used a transformer similar to a tiny push-pull O.T. to combine the signal from the two vari-mu tubes. You can do it this way if you can find a good enough transformer at a price you can afford.

http://www.bavodekker.com/670.html

Otherwise, you can use a LTPI, which is after all a differential amplifier, to convert the push-pull signal back to single-ended.

Having said this, in guitar applications, even harmonics are good, and pumping from control voltage feedthrough isn't necessarily a problem either. Fred Nachbaur used single-ended vari-mu stages in his amp designs:

http://music-electronics-forum.com/s...ead.php?t=3024

[myboysherman]

Thanks Steve. I guess I should have searched for compression rather than "remote cutoff" before posting my question.

I still can't see how the signals will remain compressed after "remixing." It seems like the troughs would be exaggerated and the peaks attenuated and that inverting one and adding them together would result in an average that looks like the original signal. I'm not doubting you; the Fairchild proves you are correct. But it means I have to sharpen a pencil and disabuse myself of this illusion.

Since SE would add 2nd Harmonics I'll have to build it that way too. Fortunately, soldering irons and switches are made for such dilemmas.

As for the LTPI, I have only seen them output two differential signals. I can see how I would connect the compressor's signals to either side of an LTP, but do I grab the output from just one of the plates? If so, what do I do with the other?

I'm still building amps by cobbling together parts from other designs and sometimes making small adjustments to stages. I don't yet trust myself to layout my own circuit (although it's probably a good time to start.) If you have any tips on where I can find such a schematic (diff in/single out,) or if it's simple enough to just tell me what to do with the other plate, that would be great. Also would the single triode differential amp with inputs to cathode and grid work? That seems simpler and I think I will have a spare triode.

At USD30+ (currently about 15£, but by the time I post it may be >50£!) I don't think I will be purchasing a transformer for this experiment.

As for "control voltage feedthrough," I'll have to look that up, but it's nice to know it won't be much of a problem.

It looks like this sort of thing hasn't gained much interest since I don't see it in many designs apart from Fred's slick but smallish amp. So, I'm not optimistic about the funky factor. But after reading the thread you provided, I have decided to drill the holes and give it a go. (as a bonus, I learned that I may have a use for those whacky pentagrid on my bench!)

Much appreciated,
Craig

[Steve Conner]

Hi Craig

The LTPI is a differential amp, albeit not a very good one. At either tube plate, you get a signal that is more or less the amplified difference between the two inputs, though of course one of them is inverted. So you can just use one of the plate signals. Both plate resistors are still needed, though.

Classic compressor theory assumes a signal that is very small compared to the tube characteristic, and the gain of the vari-mu stage is basically the slope of the characteristic at the operating point, assumed to be linear.

So as you bias the grids more negative with your control voltage, your operating point rides down the characteristic to the left, where the slope gets shallower and shallower, thus decreasing the gain. In a push-pull pair, both tubes will vary their gain in the same way.

If you're thinking about stretched peaks and squashed troughs, you're thinking of non-linear operation with large signals, where the theory goes to hell, as does the sound of the compressor, probably ;-)

Oh, BTW, the single triode with inputs to cathode and grid is all very well, but the problem is that the grid has a high impedance whereas the cathode has a very low impedance. This would load the input signals unequally and spoil the balance. So let's add a cathode follower to drive the cathode of our first tube and buffer that low impedance away. We just reinvented the LTPI...

[myboysherman]

You're right, I misunderstood the theory. I was imagining a larger swing. Fred's compressor explanation makes much more sense now.

Promising or not, I still want want to try it the "wrong" way. I'm a lousy guitar player anyway, so it can't make me sound much worse. I was reluctant to drill a bunch of "bad idea" holes in my shiny new Hammond enclosure, but I'm over that now. I think I will add Fred's circuit as well.

Although I misunderstood the typical use of the remote-cutoff valves, my oddball intentions lead to different results and problems. That is, by keeping the valve at a stable operating point with a larger grid swing, the resulting waveform will be asymmetric. You can see why a push pull design has appeal and also why recombining the differential pair might reduce or eliminate the variable-mu effect -- at least with class A operation.

So, I'm thinking low-mu --> attenuation --> vari-mu (SE or PP --> SE) --> attenuation --> somewhere in main circuit

That allows me to adjust the "crush" and the level of the (probably bad sounding) effect.

As for the phase splitter/unsplitter: I totally missed the input impedance imbalance problem. I've seen plenty of warnings about the output impedance problems when using the circuit to split a single signal, but forgot the reverse of the problem. The real reason I prefer the single triode is that I don't really understand how the LTP works (what's the smiley for shame?) Impedance matching is also a dark science for me. Especially when there's a pot in between stages.

If nothing else, this project will force me to stop procrastinating an learn the harder stuff.

I appreciate your patience, you've helped a lot.

Craig

[Don Moose]

I still can't see how the signals will remain compressed after "remixing." It seems like the troughs would be exaggerated and the peaks attenuated and that inverting one and adding them together would result in an average that looks like the original signal.

I wonder if you're thinking signal waveform here and not signal _envelope_, which is what the compressor squeezes. (yes, the waveform gets squeezed, but the compressor's time constants are about the envelope.)

Hope this helps!

[myboysherman]

Don,

I'm pretty sure but not certain that I understand the terminology. If I now understand correctly how compressors work, then I was previously way off base.

I _was_ thinking that the the waveform, that is, each cycle would be "compressed" rather than the entire vaveform being pushed upward into a different, lower-mu, part of the curve (signal envelope?) at higher signal levels.

On a previous amp, I tried to feed the output of a high gain triode to the input of an EF86 in an attempt to get the pentode to squirm. Naturally, I did all my calculations with my soldering iron and butt rather than pencil and brain. It was a total failure. Rather than work it out, I rewired without the initial triode gain stage.

This time I'm going with a more methodical approach. But I suspect that the awfulness I heard with my EF86 fiasco is pretty much what I can expect from my remote-cutoff misconception/plan.

Not that I'm dissuaded from trying.

[Steve Conner]

Well, the EF86 is not a remote-cutoff tube, and it's not known for sounding particularly good when you overdrive it. I've mostly seen it used as the first stage of the amp, where it overdrives the next stage, and I've used it that way myself with good results. (well, I thought they were good...)

I expect an actual remote-cutoff tube would sound better.