Clamping diodes to limit crossover distortion?

[nsubulysses]

Is this the function of the 1N5062 diodes in the attached schematic? They are labeled CR27, CR30, CR31, CR32 in the schematic.

can anyone explain what these diodes do and why these diodes are chosen for this task? Is this so you can bias the amp really cold and limit or eliminate crossover distortion?

[Enzo]

They connect the grid to cathode, essentially to ground. The grid is negative, so they are reverse biased usually. If the grids go positive more than the diode junction drop, they conduct. They prevent the grids from going positive.

[nsubulysses]

If grid voltage is under 100v why is an 800v diode chosen?

[nsubulysses]

What would be the scenario causing grids to go positive? Shorted coupling cap from PI?

I was too anxious to see what would happen so I clipped in some 1n4007s and my output signal looks exactly the same. I guess these are not here for the purpose of limiting crossover notch at clipping

I also just tested the amp while monitoring bias voltage and bias voltage does not change either with 1N4007 installed or not installed. Meaning, bias voltage is 51V at idle pretty much up until clipping, then going far into clipping it goes to about -61V.

I am curious because of an old email I am re-reading about the model T. The author says

Fender recommends 60mV per side, or 30mA cathode current per tube at the schematic B+ of 489V, which equates to a plate current of probably 27mA and idle dissipation of 44%! Interestingly enough, biasing this cold does NOT produce a crossover notch on the output. It's quite clean. There is a load of NFB in the power stage. And when the amp is driven into clipping, you don't see a bias shift notch develop because there are clamping diodes to ground from the power tube grids. Point is, the amp is NOT sensitive to bias.

I guess I should ask the person who actually said this, but I figured someone on here would know the answer. How is it so that this amp is not bias sensitive?

[Enzo]

Nothing to do with crossover distortion. What would cause the grids to go positive? Signal level exceeding the bias voltage. Bias at -40 and send a signal with 50v peaks and you have 10v of positive grid.

[g1]

Do you have an audible defect? Crossover distortion is usually noticeable at very low signal levels, like the decay of a note. If you are seeing a crossover notch on the scope at full power clipping, I'm not even sure you'd be able to hear it as such.

[pdf64]

If grid voltage is under 100v why is an 800v diode chosen?

Maybe a beefy diode has been used to reduce the risk of the diode becoming collaterol damage in the event of an internal tube short, eg g2 to g1?

Re

And when the amp is driven into clipping, you don't see a bias shift notch develop because there are clamping diodes to ground from the power tube grids

I don't think that makes sense, as the silicon diodes are just in parallel with the tube diodes formed by g1 being (momentarily) biased positive to its cathode; so they may make bias shift under overdrive a tiny bit worse, if anything.

...How is it so that this amp is not bias sensitive?

I can't see why it should be any less bias sensitive than other similar amps.

[nsubulysses]

Audible defect is slightly nasally raspy high end tone that is a little grating and unpleasant.

I am messing with a few things because my amp has a bit of crossover distortion beause the screen supply sags a lot at full volume. It's a long story, there's a few threads about it. I have 590V B+ so my screen supply has 1K/20W dropping resistor in series with it so the screens don't die.

Anyway, that's why I was researching clamping diode phenomenon to lessen crossover notch. The Fender Sunn model T came to mind. and the infamous Chuck H mod, but I believe it's only for cathode biased output. Anyway.

I been trying a few things

I tried 470pF cap paralleling V1 plate resistor but it made the tone a little too subdued, although also less shrill in the high end like I was desiring.

What really helped was increasing my NFB resistor from 150K to 200K and also decreasing my bias feed resistors from 220K to 150K. The sound is tighter since the bias drifts less and cleaner/clearer sounding with less high end weird nasty sounds, I guess mostly from increasing NFB resistor

I am messing with a few things. Perhaps the diode clamp thing is on the wrong path

I can post a pic or two of the output if anyone is interested to see

[g1]

Yes, pictures would be helpful.

[nsubulysses]

The names of the files are the RMS output into 4 ohm load, ie. 47.png means 47W output into 4 ohms

I would say the crossover notch starts to show at about 45w output.

I should also add that when in my last post I said how upping my NFB resistor and decreasing bias feed resistors helped, I meant it helped the sound. It did not change the look of the output on a scope.

[The Dude]

What is the actual bias current? I don't see it mentioned here? If I were seeing crossover distortion, I'd check that first. I see there are shared 1 ohm cathode resistors already installed.

[Tom Phillips]

Those scope traces are showing an inordinate amount of crossover distortion even for a full power output signal. In my opinion, it is much more than normal and I suspect that something basic needs to be corrected. I.e. this is not a candidate for any special crossover distortion fixing modification.

[g1]

Could you post your actual build schematic?
Also, did it look/sound like this before you dropped the screen voltage? (or were there other changes made at the same time?)

[nsubulysses]

Bias is 26mA at 584Va. However. my matched quad has two power tubes doing 22mA at idle and one doing 28mA and one doing 29mA. I have a strong and weak tube on each side of push pull.

I'll get back with more info soon

[J M Fahey]

I donīt see crossover distortion at all (which is a low level phenomenn) but bias shifting (not the same by a country mile) caused by halfwave rectification.
The diode involved is G-K and the peak voltage storage capacitor is the coupling cap.

A perfectly normal phenomenon, has been there since the Pyramids were built, IS PART OF THE TUBE SOUND, and nobody worried about it until armchair designers bought scopes, started hearing with their eyes, assimilated this dynamic waveform kinking to static underbiasing and tried to fight this perfectly normal (and important, and useful) phenomenon desperately, by all means.

Including nonsense "biasing tubes by dissipation", for Godīs sake.
Of course, I am in a Forum, what do I expect?
Iīm trying to convert nobody, just do as you please.

Itīs interesting to remember that designers of Tube Emulation circuits (think Pritchard, Peavey Transtube, GK and a few others, including yours truly), introduce this kind of dynamic waveform kink (and half a dozen other things) *on purpose*