Wave shape of breakup in DC-coupled cathode follower

[jamesmafyew]

Just curious about a concept that I haven't found much about in the various tube-preamp design books.

Here are two scope traces, showing the signal at the grid of the DCCF stage (i.e. the plate of the gain stage that drives the CF) right before the tone stack, superimposed on the signal at the output of the CF (top of the 100k cathode R). The output of the CF has this triangular shape on the leading-edge of the negative-going half, which looks like slew-rate limiting, but I would assume isn't really. What causes this, and what is it called? Why would the CF stage, with unity gain, "break up" before the preceding gain stage?

This is a typical pre-EQ 12AX7 DCCF with the 100k/820R unbypassed gain stage into the CF with a 100k cathode resistor. It's driving a normal Fender-style tone stack with a 1-meg MV. B+ here is about 340V.

1kHz:


100Hz:

[nickb]

Cathode followers do not drive capacitive loads well as the only path to discharge them is through the cathode (100k) resistor. Not sure about a special name but yes, it's a kind of slew rate limiting. Looks like your load is about 2nF.

[jamesmafyew]

So it seems like this type of clipping would be pretty influential in creating that rippy-shreddy Marshall preamp distortion you get when you add a master volume before the phase inverter.

[jazbo8]

Did you try it with a 12AU7 or 12AT7? Should do a bit better (or worse depending on your perspective), Rk should be lowered for the AU/AT so you get more current through them...

Jaz

[nickb]

Certainly the MV gives more control over where clipping happens, but IMHO, you seem to be driving the CF pretty darn hard as the input is looking like a square wave at 100Hz. In fact this CF distortion is actually reducing the higher harmonic content. Now whether that is good or bad is subjective. jazbo8's suggestion is a very good one. Try one of these higher dissipation tubes so you can reduce the effect and let us know how it sounds.

[jamesmafyew]

Well, the signal generator was set to 100mV output. I adjusted the gain on the amp to push the signal to just the point where the slew-rate limiting appeared at each frequency. It was about 3 on the knob with the 1kHz signal, all the way to 10 with the 100Hz signal.

At the moment I'm just experimenting with high-gain preamps. Trying to understand what produces this or that type of sound. I think it would probably be easier to just attenuate before the gainstage/DCCF so that the previous stages' maximum output can't overdrive the CF.

In this case, I just don't think I noticed this behavior before, because in a cranked non-master-volume amp, it's totally swamped by the output stage and phase inverter clipping. But with a master added to that same amp right after the tone stack, its influence is huge, particularly if the input is boosted or if an extra gain stage is added.

[jazbo8]

In this case, I just don't think I noticed this behavior before, because in a cranked non-master-volume amp, it's totally swamped by the output stage and phase inverter clipping. But with a master added to that same amp right after the tone stack, its influence is huge, particularly if the input is boosted or if an extra gain stage is added.

I have not done a high-gain amp design yet, but I see what you are saying, usually the PA and PI are way into clipping before it gets to the CF with "normal" preamps, so I guess this effect is definitely something to watch out for, when designing HG (metal?) pre-amps. Love to see how different tubes fare under such condition, both on the scope as well as sound in real life...

[teemuk]

I have not done a high-gain amp design yet, but I see what you are saying, usually the PA and PI are way into clipping before it gets to the CF with "normal" preamps, so I guess this effect is definitely something to watch out for, when designing HG (metal?) pre-amps.

Indeed. In high gain design the preamp is usually well into clipping before power tubes and phase inverter's clipping tends to be a rather rare occurence to begin with.

So it seems like this type of clipping would be pretty influential in creating that rippy-shreddy Marshall preamp distortion you get when you add a master volume before the phase inverter.

Yes, I would say the cathode follower clipping is at least one of the main characteristics in classic Marshall tone (and all other tones inspired by it).

Just take a moment to think about the popularity of the said circuit architecture.

Certainly the MV gives more control over where clipping happens, but IMHO, you seem to be driving the CF pretty darn hard as the input is looking like a square wave at 100Hz.

Overdriving to "square wave" is what "high-gain" (in modern sense) is all about.

Cathode followers do not drive capacitive loads well as the only path to discharge them is through the cathode (100k) resistor.

This. Reducing the cathode resistor value considerably would have tremendous impact on reducing the said distortion. ...Yet folks still rather go with the "standard" 100K value. As said, the cathode follower in modern high gain designs is not there only to buffer things effectively. It's there to deliberately distort.

[daz]

This subject interests me in that while shaping the distortion in my build i am now wondering what i am hearing in the distortion that may be contributed by the CF. But i don't want to lose any gain in the preamp previous to the CF. Is there a way i can reduce the gain into the CF w/o reducing the gain of the previous stages ?

[jazbo8]

This. Reducing the cathode resistor value considerably would have tremendous impact on reducing the said distortion. ...Yet folks still rather go with the "standard" 100K value. As said, the cathode follower in modern high gain designs is not there only to buffer things effectively. It's there to deliberately distort.

So the modern high-gain designs use everything including the CF for their distortion? Once you have square wave, where else can you go? I have a fundamental issue with this approach (and I am sure others have brought it up before), in a typical tube amp, the power tubes are the most expensive (after the "irons"), so if you forego the power tube distortion, why not just skip them slap on a SS/chip amp at the end, and call it a day, it just does not make economic sense to pay for the power tubes and not use them... [mini-rant over]

[loudthud]

This subject interests me in that while shaping the distortion in my build i am now wondering what i am hearing in the distortion that may be contributed by the CF. But i don't want to lose any gain in the preamp previous to the CF. Is there a way i can reduce the gain into the CF w/o reducing the gain of the previous stages ?

Not sure what you want to do exactly, but you can increase the headroom of the CF by increasing the plate voltage of the CF only. Move it's plate lead to the PI B+ node.

[jamesmafyew]

So the modern high-gain designs use everything including the CF for their distortion? Once you have square wave, where else can you go?

Well, as you can see in the trace above, the slew-rate limiting distortion has a different shape entirely, and on only one half of the waveform. So that'll sound different from a pure square wave (which in its perfect form contains only odd-order harmonics). Not all square waves are created equal anyway, since duty-cycle variations will affect the harmonics enormously.

Then as for the power tubes, that's been discussed at length elsewhere. Sure, lots of metal guys are happy with solid state power amps. Then there are lots of metal guys (like me) who either 1) prefer the nonlinearities specific to tube power amps and output transformers to the available mass-market solid-state imitations, or 2) actually play their high-gain amp loudly enough that the power amp is clipping somewhat (I do!). There is a specific and very awesome sound that comes from a high-gain amp, where the preamp is saturating really hard, and the power amp is juuussst right on the edge, but not all the way into hard square-wave territory. It's glorious. It also makes some sound guys very unhappy...

Though at the same time there appears to be an increasing market for and appreciation of "sonic rock" styles where the audience actually wants to hear the amps cranked up, filling the small-to-medium room by themselves (instead of through a cheap 57 and a badly-adjusted house EQ), while the PA mix primarily concentrates on vocals and drums (see: Sunn0))), Boris, Jucifer, etc). These are music fans who have collections of new vinyl at home and use hi-fi earplugs whenever they go to shows. I actually tour with two 100W tube half stacks and a 300W solid state bass slave amp (<70Hz only), all for guitar. All that stuff is pretty much putting out maximum power during the loud/high-gain passages of the songs. At larger venues, sound guys mic it all up and make it work.

[daz]

Not sure what you want to do exactly, but you can increase the headroom of the CF by increasing the plate voltage of the CF only. Move it's plate lead to the PI B+ node.

Well, if thats the sound of the CF NOT distorting, YUCK ! Tried it....amp got sterile and bright. Lost all that is good. Gotta be more to it than that tho because i once had it set up with a switch to take the CF out of the circuit completely and had a plate fed stack and the tone was different but good.

[jamesmafyew]

Have you scoped it? You really need to look at the waveform to know for sure, comparing between adjacent stages. I actually like the sound of the CF stage running at a higher B+ and cutting off a little later. In the shots above, it was at 350V, which is basically the same as the PI node in a Marshall. If it were on the next node, it might be at 320V or something. It's not a huge difference, and it still clips.

[jazbo8]

Well, as you can see in the trace above, the slew-rate limiting distortion has a different shape entirely, and on only one half of the waveform. So that'll sound different from a pure square wave (which in its perfect form contains only odd-order harmonics). Not all square waves are created equal anyway, since duty-cycle variations will affect the harmonics enormously.

I did meant it tongue-in-cheek

Then as for the power tubes, that's been discussed at length elsewhere. Sure, lots of metal guys are happy with solid state power amps. Then there are lots of metal guys (like me) who either 1) prefer the nonlinearities specific to tube power amps and output transformers to the available mass-market solid-state imitations, or 2) actually play their high-gain amp loudly enough that the power amp is clipping somewhat (I do!). There is a specific and very awesome sound that comes from a high-gain amp, where the preamp is saturating really hard, and the power amp is juuussst right on the edge, but not all the way into hard square-wave territory. It's glorious. It also makes some sound guys very unhappy...

I'm clueless when it come to metal, so excuse the question, but why juuust on edge? Since the signal is already "prep'd" by the preamp, wouldn't clipping the PA just adds another layer of distortion on top? Or does that just become too mushy?

Though at the same time there appears to be an increasing market for and appreciation of "sonic rock" styles where the audience actually wants to hear the amps cranked up, filling the small-to-medium room by themselves (instead of through a cheap 57 and a badly-adjusted house EQ), while the PA mix primarily concentrates on vocals and drums (see: Sunn0))), Boris, Jucifer, etc). These are music fans who have collections of new vinyl at home and use hi-fi earplugs whenever they go to shows. I actually tour with two 100W tube half stacks and a 300W solid state bass slave amp (<70Hz only), all for guitar. All that stuff is pretty much putting out maximum power during the loud/high-gain passages of the songs. At larger venues, sound guys mic it all up and make it work.

That's just bloody awesome, all that fire-power! So the guitar amps are the PA for everything but the vocal and drums! What speakers do you run with them? Wall of 4x12s?! Very interesting, must try and attend one of these shows... look out grandpa coming through