High gainer, clipping, post PI MV, tone shaping, etc!

[cyclone]

a post PI MV and/or cut control is by definition a negative feedback loop. It is local degenerative feedback. I don't mix the two, no Nfb from the opt to pi input (sometimes described incorrectly as global) but YMMV. I personally like only post pi feedback loops, spongier, more mustard in technical termsand its variable.

get one of those cheap decade boxes(not really just common values) from Weber and change coupling and pi input cap values while playing, you will be underwhelmed at the effect through a range of capacitances and not worry so much about the math

[jimboyogi]

OK so I hooked up the signal generator and scope, and measured the response with a range of cap values. And now I am very baffled……

Here is the circuit -



And here are my measurements -
Signal is measured at speaker terminals, with MV at full, therefore NFB at full (~6dB).
In each case the signal input was adjusted so that the signal output was 10.0Vpp with VR at full 250K resistance.

C?=1nF, f=100Hz, VR=0K, signal=10.0Vpp
C?=1nF, f=1KHz, VR=0K, signal=9.76Vpp

C?=2nF, f=100Hz, VR=0K, signal=9.92Vpp
C?=2nF, f=1KHz, VR=0K, signal=9.44Vpp

C?=4.7nF, f=100Hz, VR=0K, signal=9.52Vpp
C?=4.7nF, f=1KHz, VR=0K, signal=8.64Vpp

C?=14.7nF, f=100Hz, VR=0K, signal=7.04Vpp
C?=14.7nF, f=1KHz, VR=0K, signal=6.64Vpp

C?=22nF, f=100Hz, VR=0K, signal=5.68Vpp
C?=22nF, f=1KHz, VR=0K, signal=5.60Vpp

With 1nF, there was very little attenuation at either 100Hz or 1KHz. With 22nF, there was nearly 50% attenuation at both 100Hz and 1KHz. At the cap values in between, there was not a great difference between the attenuation at 100Hz and at 10KHz.

From the theory, I expected attenuation at 6dB per octave. I'm not getting it. I'm confused

[J M Fahey]

Use a "cut" type tone control with an LTP type PI and it will work.
You have 2 gain stages (two triodes) where each:
a) has *defined* output impedance (plate load//Rp)
b) has a *defined* gain value , given roughly by the ratio of what each triode sees as a plate load and what each cathode "sees" : the actual cathode resistor used and its internal impedance (1/gm)
3) since both gain stages I mentioned are working in a similar way; both have, say, a 30X (or thereabouts) voltage gain, in a nutshell are doing the same, just that it happens to be out of phase, it is *useful* to add a cap , plate to plate, and use it to cut highs.
The cutoff frequency will be defined and can be calculated.
RC circuit is made out of the C value and the added output impedances of both stages.

Now to the can of worms:
4) the cathodyne is no such beast, but a klunky , goofy idea which just *happens* to work, sort of, wnd which lives to this day only because it´s cheap simple ... which dos not mean it´s "good".

Here we do not have 2 gain stages as before, but a single one, with a split load resistor, half at the plate, half at the cathode.
The working mechanism is absolutely different to the LTP PI I described above.

**pause** it´s still early here (around 6 AM) so I'll go sleep a little more, later will continue , but start thinking that there´s a problem between the cathodyne PI and a simple cut type control.

[Alan0354]

If you don't have NFB, you can play with post PI tone control. If you have NFB, you are going to get into big trouble. Tone stack by nature change the phase back and fore with all the poles and zeros. You affect the forward gain characteristics and you easily get into 2 poles 0dB crossover and the amp oscillates. This made worst because those poles and zeros walk around as you adjust the sound. You are going to make a good fork horn.

[jazbo8]

Here is the circuit -

From the theory, I expected attenuation at 6dB per octave. I'm not getting it. I'm confused

I have not seen a cross-line tone control used with a cathodyne, perhaps for the reason that JM hinted at, anyway, the cross-line filter is used to cancel the drive signal to the power tubes (the capacitor shunts both phase together), so it does get a bit messy. Also how is the NFB applied in your circuit?

[Dave H]

With 1nF, there was very little attenuation at either 100Hz or 1KHz. With 22nF, there was nearly 50% attenuation at both 100Hz and 1KHz. At the cap values in between, there was not a great difference between the attenuation at 100Hz and at 10KHz.

From the theory, I expected attenuation at 6dB per octave. I'm not getting it. I'm confused

As Juan said it won’t work with the cathodyne. The outputs of the cathodyne are low impedance (k ohms) so they can easily drive the impedance of even a 4n7 capacitor at 1kHz (35k ohms) without attenuation. I think the halving of the signal with 22u is because it is forming a capacitive divider with the 22u coupling cap.

[Tooboob]

If you reduce the gain of the power tubes/feedback path, The PI drives harder to compensate. That's why you can get PI distortion from the circuit even at low volume - tries harder than it can. This higher drive works against the post PI MV. Once the PI hits its drive limit and distorts, the loop is broken, but for the clean part of its range, the PI works hard to reduce the attenuation of the MV.

The result isn't unpleasant, or folks wouldn't use them. I think of them as more like a PI drive control than a volume control, though it has aspects of both. You definitely don't get the same thing when you turn it, only louder or softer, as I'd expect from a volume control. It's not a good solution for reducing volume of a clean clean channel, since it kills headroom, but there's usually at least a channel volume for that.

[jimboyogi]

Some great information, thanks everyone.

Although it seems that J M Fahey and Dave H have shattered my dreams here. I had read that the cathodyne has a very low output impedance, but now that you two have pointed out some consequences WRT the treble cut, it makes a lot of sense.

Jazbo8, the NFB is applied between the speaker tap and the cathode of the gain stage preceding the cathodyne.

Alan0354, I have the gain stage DC coupled to the cathodyne, which removes 1 pole, but there's still risks with getting too fancy with phase shifts inside the loop. I have not seen any instability so far with the trial cut control, the big problem was that it didn't cut treble!

If the cross line treble cut won't work, can anyone help me find something that will work?

To reiterate, I am trying to overdrive the gain stage-DC coupled-cathodyne quite heavily. I am therefore trying to variably boost treble (or cut bass) before this system, and then trying to variably boost bass (or cut treble) after this stage, and before the power tubes. I am trying to avoid using NFB based filters (eg. 'resonance' control) because I am using a PostPIMV, so my NFB is variable based on the MV setting, and this will alter the effect of any FB filter.

The treble boost early is easy. The bass boost?

[teemuk]

I'm wondering the best way to then boost LF and cut HF after the PI.

A loudspeaker.

A typical guitar loudspeaker is severly limited in its HF response and therefore is an imperative part in HF filtering the output signal. Ever listened to speaker signal directly without some sort of cabinet emulation? It usually sounds fizzy as hell. Seen many guitar amps with tweeter horns?

A typical loudspeaker also has a resonant frequency somewhere between 50 - 100 Hz so if you deliberately reduce feedback at such frequencies, or make the amp have a low damping factor in general, you achieve a notable LF boost.

So mostly it's easier to find the speaker system that sounds "right" than trying to implement some kind of an active filtering to the loop or passive filtering to the output and fighting with all the issues and compromises introduced by such architectures.

[Dave H]

If the cross line treble cut won't work, can anyone help me find something that will work?

It will work if you add a 47k resistor in series with each phase between the 22n coupling caps from the cathodyne and the 'cut' network.

[guitician]

Why not just add another cathodyne stage just for the purpose of achieving your overdrive, and leave the power section alone?

[Chuck H]

How about an attenuator!?! You can make a decent one cheap. A really good one requires some big inductors but still much cheaper than buying a crappy one.

[jimboyogi]

Again, thanks to all of you for the replies, and for your help!

teemuk, you make good points! I was looking for a variable bass boost/treble cut, and had discounted the 'resonance' control for it's NFB issues. Maybe I'll put it back on the table now after reading your post.

Dave H, this was very pertinent! I actually gave it try just now, by including 47K resistors in series with the PI output phases as you suggested. It definitely gives a freq. dependent filter now. I need to play more with cap values, and I was also finding the range and slope of the 250KA 'cut' pot a bit wrong. I think that I need to use a lower resistance pot, and maybe linear taper, to get better control. Will do some more experimenting.

guitician, it's a good idea. ATM my design idea is minimise gain stages, one of the reasons that I am using the PI to distort. I have the gain stage/cathodyne DC coupled to get some of the compression that you'd normally use a gain stage/cathode follower to generate. Thereby removing one twin triode from the circuit. But the problems that this approach is creating is what is driving this thread!

Chuck H, I have zero experience with attenuators. Do you usually use them as an alternative to master volumes (i.e. preamp distortion), or as an adjunct to MVs?

[Chuck H]

Chuck H, I have zero experience with attenuators. Do you usually use them as an alternative to master volumes (i.e. preamp distortion), or as an adjunct to MVs?

Often using a post PI MV is problematic in one or more ways. Remember that when an amp is clipping you aren't hearing all of the PI's influence because the power tubes are in cutoff half the time. Often, PI's are designed with this in mind and the normally unheard end of the waveform is a little weird and doesn't sound good. A post PI master volume allows you to hear it because the reduced signal doesn't drive the power tubes into cutoff before the PI's less than good clipping can be heard. An attenuator allows you to use the whole amp, power amp and all and keeps the normally hidden portion of the PI waveform hidden in the power tubes cutoff cycle. If you like the sound of an amp with a master volume at a certain setting then certainly an attenuator could be used in conjunction to reduce overall volume. More commonly an attenuator is used to get a more accurate representation of an amps overdriven character at lower volumes. Just think of it as a variable volume speaker and consider the uses for such a tool.

[Dave H]

Dave H, this was very pertinent! I actually gave it try just now, by including 47K resistors in series with the PI output phases as you suggested. It definitely gives a freq. dependent filter now. I need to play more with cap values, and I was also finding the range and slope of the 250KA 'cut' pot a bit wrong. I think that I need to use a lower resistance pot, and maybe linear taper, to get better control. Will do some more experimenting.

Do you have the 'cut' control wired ‘backwards’ (min treble when fully CW) like a Vox? That feels odd to me and cramps up all the action at the CW end especially if a log pot is used. I think it works better using a log pot wired like a normal treble tone control (max treble when fully CW). Just swap the CW and CCW connections. Wired like this a 250k log pot will only be putting about 25k in series with the cap in its mid position vs. 225k for the Vox wiring. It gives smoother control.