Understanding Power Amp Distortion

[SilverFox]

What are the differences between pre amp distortion and power amp distortion?

How does one determine the point at which power amp distortion begins to occur?

Any other information on power amp distortion would be most beneficial as I know very little about it.

Thank you,

silverfox.

[jazbo8]

Obviously, there is no one "preamp" or one "power amp" distortion, as different amplifiers employ a wide range of different topologies and tubes. But in general, preamp distortion is asymmetrical, while the power amp distortion is symmentrical, which means that the preamp distortion is primarily even order, while the power amp distortion is primarily odd order, also bear in mind that for PP amps, the even order harmonics of the power amp stage are cancelled by the PP operation, so its distortion is mostly odd order.

There is also a widely circulated mis-conception that "even order distortion sound good, while the odd order does not..." If anyone bothers to look at the output waveforms, such notion quickly goes down the drain. Most of the "great" guitar tones in rock n' roll contain huge amounts of even and odd harmonics, when someone is gushing about the great "overdriven" tone, he/she is likely referring to the odd harmonics rather than the even harmonics, since the odd harmonics are more prominant - precisely what one would expect from a near square wave...

As to when the power amp distortion happens, it occurs whenever the grid is driven beyond the bias voltage.

[Stevem]

Do a search for a paper called tubes vs transistors by Russel O Hamm,and in the section on distortion harmonics he covers you will better understand how distortion levels shape the tone and response we get from a amp when it clips!

[tubeswell]

And there's harmonic unclipped distortion vs wave-clipping distortion.

The typical operating region of preamp tubes (w.r.t. grid to cathode voltage) is such that the tube's transconductance* decreases as the grid voltage tends to get more negative. The outcome of this is that as the (unclipped) grid voltage swing gets bigger, the plate voltage swing gets more asymmetrical. This assymetry produces even-order harmonics (at the plate) of the fundamental note at the grid. This results in the characteristic 'chimyness' of a tube amp with the clean signal.

*change in plate current divided by the change in the grid-to-cathode voltage.

Whereas within the typical operating region of an output tube (where the grid voltage curves are much more evenly distributed), the transconductance is fairly much the same wherever one is on the load line (unless you get to the extreme edges of the load line, or under the knee of the grid curves). This means that there is less asymmetry (or conversely, more linearity) in the output tube's plate voltage swing (compared to it's signal grid). For push-pull output stages, any asymmetry from the pre-amp gets progressively annihilated as the stage goes further into Class AB operation (where the negative grid voltage swing produces tube cut-off on the affected side of the output stage). By then, the amp is so loud nobody cares about chimyness any more.

Grid-clipping distortion occurs when the grid voltage can't swing any more positive because of grid-current limiting. This means that the positive peak (at the grid) gets clipped off. Conversely, if the grid voltage swings so deeply into negative voltage territory that the tube current gets cut off, then the plate voltage flatlines, and you get clipping of the plate signal. In a tube, the entry and exit points of the clipped portions of the signal (in both cases) are gradual/rounded-off. This is due to the gradual onset and exit from grid current limiting (in the former case) and to the 'island effect' (in the case of plate cutoff clipping). The gradual onset and exit from clipping in either case produces a softer sounding clipping distortion than you get with a transistor.

[pdf64]

That 'Tubes vs Transistors' study seems a load of nonsense; does anyone take it seriously?

... In a tube, the entry and exit points of the clipped portions of the signal (in both cases) are gradual/rounded-off. This is due to the gradual onset and exit from grid current limiting (in the former case) and to the 'island effect' (in the case of plate cutoff clipping). The gradual onset and exit from clipping in either case produces a softer sounding clipping distortion than you get with a transistor.

I've long seen the above referred to but when scoping an overdriven p-p tube amp, I can't perceive this rounding off thing.
Amps with a cathode biased output stage and no global NFB loop, eg ac30, seem to transition smoothly into distortion as signal level is increased, but increase the signal level a bit more and the wave edges looks pretty sharp to me.
Fixed bias / global NFB tend to sharpen up the transition.
All above is with regard to a resistive load, as wave tend to get pretty wacky with a real speaker.
Can someone post scope shots and conditions for this rounded off clipping phenomena?

[tubeswell]

I've long seen the above referred to but when scoping an overdriven p-p tube amp, I can't perceive this rounding off thing.

You can hear the difference. A transistor amp cutoff is grainier distortion and sounds to my ear almost like ephemeral radio static buzzing mixed in with the signal. I've never heard this in a tube amp, but almost always in a transistor amp.

[SilverFox]

I'm sitting here listening to Novella, (according to a post on the ampgarage the guitarist used a Trainwreck), talk about a "Major Distraction" and yet, whatever amp is being used is totally making the point (I think), of pre and power amp distortion. I have no idea which is which in this tune for example: Fire in the house, https://www.youtube.com/watch?v=PXkXY8-Vpe8

Thanks for the input and I look forward to more great information. I've already down-loaded the tubes vs. transistors paper. Once I get situated I'll want to better understand how to tweak an amp to get the various grid states that lead to distortion without hitting the wall of what?, Blocking distortion? Cut Off? I'm sure it has to do with bias points too.

silverfox.

[martin manning]

That 'Tubes vs Transistors' study seems a load of nonsense; does anyone take it seriously?

I guess the Audio Engineering Society did in 1973. The journal is still being published, see: http://www.aes.org/journal/

[ToneMerc]

If we include the phase inverter as part of the power amp, how does one differentiate between PI distortion and power amp distortion? I'm certain that in many cases as what was perceived as power tube distortion was actually the PI.

TM

[RockinRocket]

If we include the phase inverter as part of the power amp, how does one differentiate between PI distortion and power amp distortion? I'm certain that in many cases as what was perceived as power tube distortion was actually the PI.

TM

PPIMV

[Smokebreak]

If we include the phase inverter as part of the power amp, how does one differentiate between PI distortion and power amp distortion? I'm certain that in many cases as what was perceived as power tube distortion was actually the PI.

TM

PPIMV

So PI distortion sounds awful ?
I guess that's not really fair because NFB is somewhat crippled as well.

[SilverFox]

Okay so I read through, not intricately, I read through the "Tubes vs Transistors" article. My initial conclusions are: I would expect the Material Sciences associated with transistor manufacture as well as audio specific OP Amps and other components such as JFets to have advanced significant orders of magnitude since the article was written in 1973. The appearance and quality of digital signal processing adds an immense amount of realism so the findings from the article in '73' are likely not as valid now as then.

Still, I was remarking just the other day to someone that, there is a certain generational sound associated with audio gear based in the differing technologies and that the current "sound" of digital electronics has a sort of boominess and lack of full measure. On the other hand, the current run of amp emulators sound very convincing, but, plug them into a tube based power amp and they sound even better. I suppose the determination should be conducted again. That would be interesting.

Perhaps, after all; Nature does not abhor a vacuum. It may very well be the case of: Things that can only occur in a vacuum vs what takes place in a solid state...

silverfox.

[Firestorm]

There's preamp tube distortion (which is clipping some portion(s) of the signal; there's output tube distortion (which is also clipping some portion of the signal); there's PI distortion (which varies with the type of PI); and there's OT distortion (some of which can be limitations of the OT and some is just the effect of two or more tubes fighting for the OT's attention ). Oh, did I mention feedback.

This is not a simple subject.

[Firestorm]

Okay so I read through, not intricately, I read through the "Tubes vs Transistors" article. My initial conclusions are: I would expect the Material Sciences associated with transistor manufacture as well as audio specific OP Amps and other components such as JFets to have advanced significant orders of magnitude since the article was written in 1973. The appearance and quality of digital signal processing adds an immense amount of realism so the findings from the article in '73' are likely not as valid now as then.

Still, I was remarking just the other day to someone that, there is a certain generational sound associated with audio gear based in the differing technologies and that the current "sound" of digital electronics has a sort of boominess and lack of full measure. On the other hand, the current run of amp emulators sound very convincing, but, plug them into a tube based power amp and they sound even better. I suppose the determination should be conducted again. That would be interesting.

Perhaps, after all; Nature does not abhor a vacuum. It may very well be the case of: Things that can only occur in a vacuum vs what takes place in a solid state...

silverfox.

Google history of the CD. The first thought was to throw lots of bits at it (problematic with early D/A and A/D converters). So there was a lot of arguing over bit depth. Then, somebody reasoned that sampling rate was key (analog having a theoretically infinite rate). It turns out, from the ear's perspective, that sampling at 256 times a second is just fine, permitting the use of 1-bit converters.

But that is about "accurate" reproduction of a signal. What if you want to color the signal with the gear. Counterintuitively, tubes have exceptional linear behavior if operated within their specifications. Push them, and things start to happen. Tubes misbehave in wonderful ways. Solid state, not so much.

[Stevem]

How did we get off the track of looking at how much and what frequency of harmonic distortion Triodes produce as compared to Pentodes?