Grid Current Emulation at Input to Power Amp

[loudthud]

Cool

Real scoping beats any "theory-without-practice" opinion any day of the week.

That said and for a fuller picture (literally ) , can you please also post each waveform besides its spectral "signature" as you showed above?
Thanks.

PS: and if available, also the schematics of "what" produced each

I thought JM's request deserved it own thread. I had to re-create the experiment, but here's everything.

On the left of each the scope photo is the output of a distortion analyzer. Signal is 1KHz at 1.7V RMS. The top waveform is what is across the dummy load resistor, (analyzer sets the gain so amplitude doesn't mean anything). The bottom trace is the distortion products with the fundamental removed by a very steep notch filter. Sweep speed is 200uS per division. On the right is a spectrum analyzer output from the top waveform. Deflection factor is 1KHz per division. Also posted is the original dual spectrum analyzer post that JM commented about.

The schematic is a slight adaptation of the single rail example on the TDA2040 data sheet. Note if you add the diode and switch to a dual rail power amp, you will get DC across the speaker (as Steve Connor pointed out) because of the non-symmetrical waveform. I've actually measured it with a Fluke DVM.

Despite the evidence, I'm not sure I can hear the difference when playing through an amp with this switch. You can hear a slight difference when you throw the switch, but I'm not sure which position the switch is in.

[Helmholtz]

By action of NFB the differential voltage between the +/- inputs of an op-amp is always close to zero. If you want to see some diode effect, its cathode should be connected to ground. I would also add some series resistor before the diode emulating a grid stopper.

BTW, which other thread does this relate to?

[nickb]

Correct me if I'm wrong but I think the idea is that the diode only comes into play when the power amp hit's clipping. Only then does it conduct.

That said, the effect is greater than I would have expected. I note the slew rate is horribly slow at about 3V/150us == 0.02V/us and that may have something to do with it. Also that would distort the spectrum from what would otherwise be expected.

[Helmholtz]

Correct me if I'm wrong but I think the idea is that the diode only comes into play when the power amp hit's clipping. Only then does it conduct.

It is clear that when the op-amp clips, NFB cuts out and differential input voltage increases.

But then I don't understand the whole concept. Normal asymmetrical grid current distortion of the first triode happens independent of volume setting and output clipping. Blocking distortion in a PP amp, however, is a symmetrical effect that cannot be emulated by a single diode.
Without using sufficient series resistance, the results with a low impedance signal source and a high impedance PU will be completely different anyway.

That said, the effect is greater than I would have expected.

What effect?

Please let me know the history of this thread. I don't have the time to read all threads.

[nickb]

It is clear that when the op-amp clips, NFB cuts out and differential input voltage increases.

But then I don't understand the whole concept. Normal asymmetrical grid current distortion of the first triode happens independent of volume setting and output clipping. Blocking distortion in a PP amp, however, is a symmetrical effect that cannot be emulated by a single diode.
Without using sufficient series resistance, the results with a low impedance signal source and a high impedance PU will be completely different anyway.

What effect?

It's asymmetrical and has even harmonics with the diode.

Please let me know the history of this thread. I don't have the time to read all threads.

Just use the search feature to find it. https://music-electronics-forum.com/...ad.php?t=48132

[Helmholtz]

Just use the search feature to find it. https://music-electronics-forum.com/...ad.php?t=48132

Thanks, it was not mentioned by the OP. I had actually watched that "limiter thread" but didn't recognize a relation to "grid current emulation". Still don't understand what is meant here with grid current emulation. Grid current in PP tube power amps causes a kind of more or less (depending on PI) symmetrical crossover distortion. Symmetry generally means low even harmonics.

Is this about emulating single ended behaviour in order to introduce even harmonics? It surely isn't a good limiter concept.

If even harmonics actually were desirable in power amp distortion, the design should provide for some asymmetry and power tube matching would be counterproductive regarding sound.

I think, it is sufficient to get some even harmonics from the preamp. Otherwise, why would a '59 BM, a JTM45 or an AC30 sound that good?

[J M Fahey]

I thought JM's request deserved it own thread. I had to re-create the experiment, but here's everything.

Thanks, itīs very kind of you

On the left of each the scope photo is the output of a distortion analyzer. Signal is 1KHz at 1.7V RMS

Yes, you need a loud signal, first to saturate the stage (pre or power) after it but also to forward bias the diode or "it is not there".

Note if you add the diode and switch to a dual rail power amp, you will get DC across the speaker (as Steve Connor pointed out) because of the non-symmetrical waveform.

Yes, non symmetrical waveforms have a DC component.
But in tube stuff you do block DC everywhere, either by coupling caps or output transformers, which in both cases shift the wonky waveform (still retaining most of its shape and all harmonics) so the average value is 0V DC .

Despite the evidence, I'm not sure I can hear the difference when playing through an amp with this switch. You can hear a slight difference when you throw the switch, but I'm not sure which position the switch is in.

Maybe subtle, but every single bit adds up generating what we know as "tube sound".

By action of NFB the differential voltage between the +/- inputs of an op-amp is always close to zero. If you want to see some diode effect, its cathode should be connected to ground. I would also add some series resistor before the diode emulating a grid stopper.

True.

But in that case, positive peak clipping starts above some 700mV , no matter what the power amp is doing.
The beauty of Loudthudīs approach is that it happens if and only if power amp clips ... what tube amps do.
The boring bland SS amp now is taking some flavour.

What effect?

It's asymmetrical and has even harmonics with the diode.

It isnīt exactly a triode clipping emulation nor PP tube amp one, if anything might approach a single ended tube amp clipping (which have tons of flavour on their own) but is "something interesting" ... definitely does not hurt.
I think it may be adapted to make it usable with split supply amps too, probably adding some extra cap between it and actual power amp input.

Is this about emulating single ended behaviour in order to introduce even harmonics?

As I see it, yes.
At least, true or not (it depends on each circuit), everybody and his brother touts "tube amps have even harmonics, SS ones have odd ones" ... which is at best a gross oversimplification, but if anything this circuit approaches that ideal.

It surely isn't a good limiter concept.

Donīt think thatīs the objective today.

[loudthud]

In a previous thread from back in 2014 I posted scope photos of a 50W Marshall power amp. Link: https://music-electronics-forum.com/...ad.php?t=37023 From that I concluded that the asymmetry comes primarily from grid current at the input to the phase inverter in amps that don't have a Master Volume. Also, I think speaker cone bobbing is just the subsonic component of asymmetric waveform clipping leaking through the output transformer in attenuated form. If you are feeding an already squared off waveform to a power amp, the diode to the feedback loop won't do much.

In the vast majority of tube amps there will be some asymmetry in the preamp. That may indeed be the reason I don't hear much difference when I throw the switch because all of my preamps in recent years have been JFET with unregulated power supplies. Perhaps I should build an opamp preamp and try that.

Hemholtz, it's a feature of the software that runs the forum that if you click the little icon next to the quoted persons name in the quote, you will be transported to the thread and post being quoted. See attachment.



Nickb, the edge speed is low because the drive is really pretty low. Some chips go crazy if you drive the inputs too hard. The power supply is 32V from a bench power supply. Clipping starts at 750mV RMS from the generator, the scope photos were taken with the generator set to 1.75 V RMS.

Back to the Limiter thread, I was pointing out that symmetrical clipping at the input to the Class D power amp would produce odd harmonic dominated distortion.

[J M Fahey]

From that I concluded that the asymmetry comes primarily from grid current at the input to the phase inverter in amps that don't have a Master Volume.

Yes, in a Tube amp thereīs a lot of unsymmetrical clipping all along the way.
It all adds up; thatīs why a single tube stage, such as in the typical Valvestate is not enough, while a full Tube strip is much better.
And in my point of view, the reason why high gain amps cascade so many gain stages ... with strong attenuators in between.
What makes me think is not just for the gain (they could drop the attenuators and save 1 tube stage, just do the Math), but to pile up flavour over flavour.

Also, I think speaker cone bobbing is just the subsonic component of asymmetric waveform clipping leaking through the output transformer in attenuated form.

Yes, definitely.

If you are feeding an already squared off waveform to a power amp, the diode to the feedback loop won't do much.

An on nthe contrary, a *very* unsymmetrical signal will be recognizable even if symmetrically clipped by the power amp.
In my view, one of the components in Hendrix sound.
He used the horrible cheesy Fuzz Face and then he clipped it to death by a Marshall power stage.
Of course that means playing **LOUD** , best but useless solution to modern sissy snowflake players.

In the vast majority of tube amps there will be some asymmetry in the preamp. That may indeed be the reason I don't hear much difference when I throw the switch because all of my preamps in recent years have been JFET with unregulated power supplies. Perhaps I should build an opamp preamp and try that.

I bet you will notice a difference then.

[Justin Thomas]

Of course that means playing **LOUD** , best but useless solution to modern sissy snowflake players...

I'm not sure what I want to remove in order to fit that into my Signature...
Decisions, decisions, and my apologies to who doesn't make it. It's not personal, it's just that this is so perfect!

Justin

[Helmholtz]

Hemholtz, it's a feature of the software that runs the forum that if you click the little icon next to the quoted persons name in the quote, you will be transported to the thread and post being quoted. See attachment.

Click image for larger version. Name: Click_1.GIF Views: 3 Size: 15.6 KB ID: 51888

Thanks a lot for this tip! Useful feature, indeed.

In a previous thread from back in 2014 I posted scope photos of a 50W Marshall power amp. Link: https://music-electronics-forum.com/...ad.php?t=37023 From that I concluded that the asymmetry comes primarily from grid current at the input to the phase inverter in amps that don't have a Master Volume. Also, I think speaker cone bobbing is just the subsonic component of asymmetric waveform clipping leaking through the output transformer in attenuated form. If you are feeding an already squared off waveform to a power amp, the diode to the feedback loop won't do much.

Thanks for clarification. After reading the old thread, the title of this one makes much more sense to me.
Yes, I think the described cone bobbing will contribute to the dynamic feel of the amp. The phenomenon observed is obviously asymmetric and dynamic, i.e. involves time constants. This will be partly due to delayed supply voltage sag but I assume that also charging times of coupling caps are involved. Did you particularly study the voltage-time behaviour at the PI input grid? What effect has a different value input coupling cap?

[nickb]

In a previous thread from back in 2014 I posted scope photos of a 50W Marshall power amp. Link: https://music-electronics-forum.com/...ad.php?t=37023
Nickb, the edge speed is low because the drive is really pretty low.

Yeah, I figured that out for myself. I had gone in, for no good reason whatsoever, with the assumption that it was being driven hard, then realized later that it wasn't.