Cathode input gain stage?

[Chuck H]

I'm working on an effects loop (toob) and I would like to NOT invert the signal. The send is not inverting because it operates from a cathode. I thought I might try a cathode input for the return. I'm not tech enough to figure gain for the circuit (or even idealize the design?) though. Below is my proposed circuit. I've attempted elevating the cathode to provide a reasonable input impedance for the effect output. I'm also wondering if the cathode input should be as shown or at the junction of the bias and tail resistors. I'm also wondering how much gain I should expect and what any of you might do differently.

TIA

P.S. Please don't trouble with the usual "You could add a single transistor as a..." This is already an all tube amp and I hope to keep it that way.

[Zozobra]

I simmed it in LTspice and 1Vpp@400Hz at the input produces 28Vpp swing. The frequency response looks horrible though.

[eschertron]

I simmed it in LTspice and 1Vpp@400Hz at the input produces 28Vpp swing. The frequency response looks horrible though.

Could C1 be bigger? I'd think any 25vdc ecap would work in this position. 10u? 22u?

[Chuck H]

Great! Thank you. I haven't had any luck downloading 12ax7 models for LTspice. Perhaps I should look harder. I thought the 47k elevation would somewhat mitigate the need for an enormous input cap, but I guess not. By your sim it looks like I can get a voltage gain of about 35 (+31dB all the way down bolow 80Hz) if I work out the input cap value/impedance issue. I think that'll do it.

[nickb]

Just be aware the input impedance is very low at about 1.8K. That's too low IMHO.

Here's the best LtSpice lib I know of in terms of accuracy.

[Chuck H]

Just be aware the input impedance is very low at about 1.8K. That's too low IMHO.

Here's the best LtSpice lib I know of in terms of accuracy.

I had hoped the 47k elevation would solve for this. Remember that I'm not studied proper in electronics which is why I missed. I'll guess it has to do with the elevation being the common low V reference? Lack of tech is also why I don't know how to use LTspice beyond using it to simulate tone shaping circuits. Having never used the program for studies and not having read about it's operations that go over my head I really don't know how to use the attached file. I've checked out a few poorly layed out narrations on *outube and read some of the 'help' files, but so far programing new components into LTspice is also beyond me. Somehow I thought it would be as simple as downloading a "12ax7 model" to the program. I can't even figure out how to plug in any model parameters I've found. The standard 'triode' model doesn't seem to allow that as far as I've been able to tell. I'm sure it's just that I haven't properly studied the software. And I may not. I'm not shooting for my degree, my livelihood and future doesn't depend on it and my wife would rather I did some chores around the house

[nickb]

I had hoped the 47k elevation would solve for this. Remember that I'm not studied proper in electronics which is why I missed. I'll guess it has to do with the elevation being the common low V reference? Lack of tech is also why I don't know how to use LTspice beyond using it to simulate tone shaping circuits. Having never used the program for studies and not having read about it's operations that go over my head I really don't know how to use the attached file. I've checked out a few poorly layed out narrations on *outube and read some of the 'help' files, but so far programing new components into LTspice is also beyond me. Somehow I thought it would be as simple as downloading a "12ax7 model" to the program. I can't even figure out how to plug in any model parameters I've found. The standard 'triode' model doesn't seem to allow that as far as I've been able to tell. I'm sure it's just that I haven't properly studied the software. And I may not. I'm not shooting for my degree, my livelihood and future doesn't depend on it and my wife would rather I did some chores around the house

...and it can get a little addictive. Still you can't beat the real-time hands-on 100%-accurate real thing


The 47k does help, but it appears in parallel with the 12AX7. Think of it like this. You are an energy source and you try to pull the voltage up on the cathode so the grid to cathode voltage increases. This Vgk increase reduces the current through the tube (by gm x change in Vgk) and so tries to lower the voltage on the cathode. Therefore you have to pump more current in to get the cathode to move it. In other words the input impedance has been lowered.

[potatofarmer]

Is the goal to keep the effected signal in phase with the clean/bypassed signal? You could just add a zero-gain inverting stage to the clean signal, like how paraphase inverters work.

Or I guess use a common-anode mixer for gain and mixing.

Both of those mean adding an extra triode though; not sure if that's feasible.

[teemuk]

I think the cathode input will be inherently low Z unless you devise some "bootstrapping" scheme - which likely requires another gain stage of some sort. This is why the low end response is also so attenuated: The 100nF cap in the input is way too low capacitance to couple low frequencies to low Z of the cathode circuit.

Gain of grounded grid amp is approximately plate resistance divided by cathode resistance, same as with common cathode amp, but grounded grid architecture will in practice have slightly higher gain than that.

You can use cathode or grid DC bias, and whatever scheme to accomplish that. In the end its the grid-cathode voltage difference that matters. Grid naturally needs to be grounded for AC signals.

IMO, without "extra" something (whether its solid-state or tubes) you will have some problems interfacing stages with high Z output to low Z input at the cathode.

[Chuck H]

Thanks Nick. Pete was schooling on that same issue in another thread regarding the effective knee frequency and other effects WRT preamp cathode bypass caps. I might have put the parallel together (pun intended) if I were paying attention.

Potato, yes, maintain phase. But not to integrate with a bypass. One preamp stage clips asymmetrically and the final tone of the amp is negatively affected when I invert the preamp phase. I'm not going to add a tube to the chassis and I don't want to include even discrete SS devices (which I expect to come next in the suggestions, though not from you necessarily).

The SEND for this loop is currently from the cathode of an existing gain stage that is partly responsible for the voicing of this particular amp. It happens to use an unbypassed cathode so it's possible to take a low output, low impedance signal from there. A previous incarnation of the loop took the send from the plate. Unfortunately the high impedance caused significant HF loss when the loop was used with even ten feet of cable. So, while I don't intend to alter that triodes circuit for tonal purposes, what would be the best way to derive the lowest possible impedance signal from the plate of that triode? What if I just skipped any series resistance and used only a load after the coupling cap. Doing it like this, using the tubes internal resistance against a load to derive only a volt and a half would mean a load resistance of only a couple of k. I assume this would be seen in parallel with the 100k plate load resistor to determine output impedance, right? I might be able to get "acceptable" results like this.

If your still reading along, thank you

[R.G.]

You're finding out that the input impedance of a cathode is equal to its output impedance. As noted in the replies, this is for the same reasons that a cathode follower output impedance is low - the action of the conduction in the tube to changes in the grid-cathode voltage.

There are a few alternatives.
1. Use the plate output on the tube that's "sending", so it goes out inverting, as mentioned. Even if you have to pad it down, that is a reasonable thing. If the tube is driving a 91K/10K divider to lower the signal level, the output impedance of that divider is on the order of 9K. Even with the plate impedance of a 12AX7 before it, you get something like 10K as an output impedance. You drop the plate signal level to less than 1/10 of its former self, but then you probably didn't need all that signal going out to a loop, anyway. This lets you re-invert by driving a grid coming back in.
2. That presumes you're using the plate output for something else. If you were using a whole triode for the loop output, put equal resistors in the plate and cathode and use the plate output directly. With a few notable exceptions, pedals have a 1M or bigger input impedance. Again, you get to re-invert coming back in to a grid.
3. Cheat #1: use a transformer. Triad is now making a $4 transformer that gives 20Hz-20kHz at 10K impedances. This is the TY-250P, available at Mouser. You can drive this with your cathode, inverting the output, or drive it from the pedal chain, again choosing to invert. In fact, using the transformer lets you use a switch to invert phase should that be necessary because of a rogue pedal.
4. Cheat #2: use a TO-92 MOSFET and hide it inside something. A big old paper-tube electro cap corpse would work nicely.

[Chuck H]

HA! I knew the MOSFET suggestion was coming!

Thanks R.G.

The transformer idea is interesting. I run this thing pretty high in gain to OD the power amp though and I can't imagine adding an inductor to the preamp wouldn't come with other noise problems. I mentioned that I had previously used the plate as the send for this circuit. I just did it poorly. I'll revisit that option again and do it better. It will almost surely still suffer some HF loss with anything more than tiny patch cables, but maybe I can get it to where it's insignificant.

[R.G.]

Remember that the transformer can be either an output transformer, driven from the ~ 2K output impedance of the cathode have now, or an input transformer, driven from the (usually) even lower output of the pedal chain. Terminating a transformer in a low impedance really helps with noise pickup issues.

The TY-250p ( http://www.mouser.com/ds/2/410/media-346070.pdf ) is physically small enough that you could wrap it in copper straps and maybe steel or mu-metal shells to cut any pickup a lot.

[J M Fahey]

P.S. Please don't trouble with the usual "You could add a single transistor as a..." This is already an all tube amp and I hope to keep it that way.

Ok, silicon hater, then you can add an extra triode so you regain normal phase.

You have 3 options:
1) add a separate cathode follower (non inverting) to drive the (non inverting) common grid gain stage
2) you can do same in a more elegant way: use a LTPI stage as Loop recovery , you have the option of inverted or non inverted signal, how´s that?
Of course, technically it´s the same as (1) : first triode works as a cathode follower directly coupled to and driving second triode cathode
3) do it the conceptually simple way: use 2 (inverting) standard triode stages, signal at the scond plate will be gain with the original phase.
Use attenuation between them to get rid of excess gain.
4) use a cathode follower as Loop return; if you play your cards well you do not *need* to have gain there, consider it a "buffered passive loop" .
Since passive loops are one common option, doing the same but buffered is certainly a possibility.
5) use a standard inverting triode stage as "send" ; attenuate as needed to have proper signal level, a resistive attenuator will also give you low impedance as a bonus (hint: a 220k/10k attenuator has 10 k output impedance, not bad at all); then standard inverting gain triode will return phase to normal.

Oooops, looks like I got carried away as usual

EDIT: I forgot:
6) you can use a cathode follower as "send", driving a line transformer.
its output, of course, can be wired with either phase, so if you wire it out of phase then the inverting recovery stage will straighten things up.

In a nutshell: if you do not want the simple practical sensible solution (SS or Op amps), then you can complicate as much as you wish to keep ritual Purity

[R.G.]

There are some cases where too much is not enough.