Grid stopper for phase inverter

[ThermionicScott]

I recently tried 470k on the grid of the cathodyne stage in my PR cloney thing, and really like the way it goes more smoothly into break-up now.

Although I rarely overdrive it, I've considered this mod for my 5E7. Have you noticed any high-end rolloff with that grid-stopper in place?

- Scott

[TransLucid]

Clarification: In my previous post I mentioned that the cathode of the PI is at a higher DC voltage. Of course, so is the grid, and the DC voltage difference between the grid and cathode is the same as in other stages. It can be more or less, depending on the bias operating point chosen.

With regard to Tubis' comment, the PI coupling cap is indeed often 22n (sometimes 10n), while the coupling caps to the output devices are usually .1m. This supports the theory that the output tubes are more prone to blocking distortion than the PI, unless reasonably-sized grid stoppers are used for the output tube grids.

[pdf64]

Sorry, I should have specified that there's no global negative feedback on the amp with a post phase splitter master volume - Pete.

[ThermionicScott]

With regard to Tubis' comment, the PI coupling cap is indeed often 22n (sometimes 10n), while the coupling caps to the output devices are usually .1m. This supports the theory that the output tubes are more prone to blocking distortion than the PI, unless reasonably-sized grid stoppers are used for the output tube grids.

The output tubes are most prone to blocking distortion because they are at the tail end of a series of gain stages. If you want maximum output from an amp, then you design it so that the output section clips (and draws grid current) first. (In other words, you generally wouldn't design an amp so that the PI clips first, because then it couldn't drive the output tubes to full power.)

BTW, coupling cap values aren't just chosen to prevent blocking distortion -- the prime consideration is low-frequency response, and because the LTP has a much higher input impedance than an output tube, you can use a smaller value coupling cap for the same bandwidth.

- Scott

[jimboyogi]

Yes, but also from a more general perspective. If avoiding grid conduction is important when overdriving a gain stage, isn't it also important when overdriving the PI?

I wanted to resurrect this thread because I am really wondering the same thing, and I don't see that it has been answered completely below.

For instance, i am building an amp with a LTPI, and a PPIMV. The LTPI can be hit with plenty of signal voltage to it's input to overdrive it, and theoretically to push it into grid conduction. So I have a situation where the grids of both triodes in the LTPI should be conducting, and a coupling cap in series with either grid (22nF for inverting input, 100nF for non-inverting/grounded). Why wouldn't I get blocking distortion?

Any insights greatly appreciated

[loudthud]

Later versions of the Traynor YBA-1 (a 5F6A inspired design) had a grid stopper on the input side of an LTPI. Some versions also had 47K (yes 47K not a typo) grid leak resistors. These changes were said to reduce undesireable distortion (I read somewhere). I have never experimented in this area.

[jimboyogi]

Interesting, thanks loudthud.

I was wondering where any grid stoppers might be beneficial, whether on the inverting input (usually with the lower value coupling cap) or the non-inverting input (usually with the larger value cap ~ 0.1mF). Or both!
Both inputs have the same value grid leak resistor (usually 470K - 1M), and in a cathode biased LTPI, bootstrapping makes the input impedance higher than the resistor value by at least a factor of 2, and even more if NFB is injected into the tail. Am I correct in thinking that the R in the RC time constant for the inputs is the input impedance value rather then the grid leak resistance? If so then the RC time constant is high, and should be susceptible to grid current blocking distortion when overdriven hard enough with low frequency notes.

Last night i actually tried the experiment. I put 2 * 220K grid stoppers (one for each LTPI input) on a switch near the valve socket, and tried overdriving the LTPI pretty hard both with and without the grid stoppers. To my ears there was a difference, but from my limited understanding of blocking distortion, I don't think that there was any blocking present either way. The difference seemed to me to just be a bit of HF roll-off when the grid stoppers were in the circuit, that gave a smoother overdrive sound, with less HF "fuzz".

[pdf64]

Am I correct in thinking that the R in the RC time constant for the inputs is the input impedance value rather then the grid leak resistance?

That's my understanding also.
I think 220k grid stoppers would roll off noticeable high end (-3dB @ 6kHz assuming 120pF miller capacitance).
I don't know why, but guitar amp LTPs seem to overdrive nicely without noticeable blocking distortion, even with a 0.022uF input cap; something to do with the tail resistor keeping the overall operating points in check maybe?