Op amp distortion via limiting supply voltage

[marcelomd]

What happens if diode-only sim has its input voltage swing set to 9 or 10Vpp?

Note that the "hardness" of a diode clipping stage depends on the current, which depends on the preceding impedances and voltages. It is not always a razor-like cut.

[MrStab]

I stand corrected, and am honestly a bit surprised that a higher clipping threshold isn't negated by a subsequent lower threshold. Does changing Vdd on the op-amp to 36V or so (with original input signal) produce the same curve as 9Vpp into raw diodes?

[Steben]

I do not want to be pedantic, but these signal forms are not what most oscilloscopes seem to show....
It all depends on the spice model of the diode I guess.

https://www.youtube.com/watch?v=bb_gmhgzLaA

check out when the gain changes. All diodes clip "quite harsh" early on.

[marcelomd]

I stand corrected, and am honestly a bit surprised that a higher clipping threshold isn't negated by a subsequent lower threshold. Does changing Vdd on the op-amp to 36V or so (with original input signal) produce the same curve as 9Vpp into raw diodes?

Like this?

[marcelomd]

I do not want to be pedantic, but these signal forms are not what most oscilloscopes seem to show....
It all depends on the spice model of the diode I guess.

https://www.youtube.com/watch?v=bb_gmhgzLaA

check out when the gain changes. All diodes clip "quite harsh" early on.

My point, again, is that two stages clip different from a single stage. And that's more because whatever is around the clippers than the clipping element transfer function itself.

Models are just models. They're there to give some direction.

Edit: typo

[MrStab]

I didn't consider earlier that -68 gain would still cause 1.84V to clip at the op-amp. My question was more, "What would the diodes produce if the gain element itself didn't clip at all?" which I guess would require schottkys and/or lower gain to ascertain.

Interesting stuff to think about in any case.

[Steben]

I didn't consider earlier that -68 gain would still cause 1.84V to clip at the op-amp. My question was more, "What would the diodes produce if the gain element itself didn't clip at all?" which I guess would require schottkys and/or lower gain to ascertain.

Interesting stuff to think about in any case.

Some diode to GND circuits actually add soft clipping diodes in the opamp loop to prevent opamp clipping. Odr1 for example.

[amptramp]

With some op amps, you can use the offset adjust and compensation inputs to limit voltages without changing the supply voltage.  This may not be characterized in the spec but that doesn't mean it can't be done.  All it means is that the factory does not guarantee the results or uniformity from batch to batch.  Just add diodes across the offset adjust or on the compensation pins and you might have something useful.

[Steben]

And.... inverting opamps with clipping loops do not "opamp clip"

[antonis]

Just add diodes on the compensation pins

Never crossed my mind, honestly..
(maybe due to almost exclusive engagement with dual op-amnps..)

[MrStab]

Just tried this myself in LTSpice, and I absolutely stand corrected. Whaddya know!

So it seems that beyond a point, the main benefit of the diodes is limiting the output voltage swing, whereas most of the edge-sharpening is done by the op-amp.

[teemuk]

...as op-amps tend not to clip in a pleasing way.

This phrase gets repeated often but how many have evaluated that in practice?

The Sansamp GT-2 is an exception, though - the TLC2262 it uses clips nicely. Or so I'm told.

For obvious reasons rail-to-rail opamps are often designed for decent clipping recovery characteristics. Then again, Behringer SansAmp clones used just generic opamps and they weren't at all bad sounding either. Tech 21 may have chosen rail-to-rails just due to their lower noise (which is virtue in high gain after all).

Opamp rail clipping will be very abrupt and jard edged, but on a contrary to popular belief, that may not be a bad thing or even sound that much different from softer clipping. Many opamps will also asymmetric headroom limits. The nasty point to focus on arecwhat happens beyond common mode input range; some opamps will invert their polarity, stability (or lack of it) and weird "rail sticking" characteristics. The unfortunate part is that datasheets rarely discuss operation beyond designed limits where circuit architecture differences start to matter and opamps stop being homogenous components.

[teemuk]

The LEDs are after the second stage, so you hear the LEDs clipping an already clipped signal. This sounds a bit different from pure opamp or pure LED clipping in a single stage.

The clipping functions become a composite and the hard clipping of opamp basically prevails. Softer clipping at lower threshold won't soften the knee of already hard clipped signal. Some designs exploit this feature and tune the clipping thresholds for a preferred touch sensitivity to blend from softer clipping to more aggressive tone of harder clipping.

[Steben]

Well fact check: many tube (power) amps with negative feedback clip much harsher than ideal diodes. They function not unlike an opamp.
Hard clipping is not bad or good. It is "a" flavour.
It explains partially why the "marshall" sound is easy to mimick.

Drawing a lot of current out a TLC2262 and adding a sag resistor to supply might even get you power sag.
A saggy Tweed amp or an AC30 is just a tad more different though. These have much more curved transients which give grainy wooly onset of distortion.
And dynamic cross-over distortion is a complete different story.

[MrStab]

Perhaps a separate debate covered ad-nauseum elsewhere, but TLC2262 in particular has one of the worst slew rates I've come across, with the 2272 being nearly 4 times higher.

Wholeheartedly confess this was just an assumption I'd read without looking into. Contradicted by the fact I've read plenty of people happily using no-diode mods on dirt pedals with zero consideration for op-amp selection.

[marcelomd]

Perhaps a separate debate covered ad-nauseum elsewhere, but TLC2262 in particular has one of the worst slew rates I've come across, with the 2272 being nearly 4 times higher.

Wholeheartedly confess this was just an assumption I'd read without looking into. Contradicted by the fact I've read plenty of people happily using no-diode mods on dirt pedals with zero consideration for op-amp selection.

In my experience people care a lot less about the specific mechanism of clipping than what you do with it. I.e. "if it sounds good, it sounds good". To me, and I think this is not controversial, what makes or breaks an overdrive/distortion circuit is the pre- and post clipping equalization.

[teemuk]

If an average non tech inclined person was shown an oscilloscope capture of "push-pull tube power amp distortion" with its
- symmetry
- hard clipping
- crossover distortion
- transient overshoots
they would think it must sound horrible because it practically breaks every cliche of "rules" of what kind of distortion sounds good.

I fully agree with notion that average diode clipping is much softer than clipping of an average tube power amp.

[amptramp]

One version of diode clipping that would be very hard is the use of an ideal diode as a clamp that prevents a signal from going beyond limits set as voltages"



Changing the value of R1 can change the hardness of the clipping but the clamp sections can have their own series resistors to soften the clamping.

[Steben]

If an average non tech inclined person was shown an oscilloscope capture of "push-pull tube power amp distortion" with its
- symmetry
- hard clipping
- crossover distortion
- transient overshoots
they would think it must sound horrible because it practically breaks every cliche of "rules" of what kind of distortion sounds good.

I fully agree with notion that average diode clipping is much softer than clipping of an average tube power amp.

Yes. Although "opamp clippers to GND" DO come close because of the opamp clipping. At least for symmetry and hard clipping. Inverting opamp clippers do clip hard as well given the way it works but the opamp itself does not. The guv'nor / drivemaster circuit might work with this as well for example. It would come close to a Bluesbreaker with no softening resistor. Non of these can simulate power sag though.
The onset of clipping is different with different types of tube amps.
The classic plexi / JCM800 amps have SS rectifiers so they do not even have sag. Besides the fact their preamp clipping section is dominant. These amps are known to be king in hard rock / 80ties where tight palm muting ruled.

I really really really dig a classic treble booster into the clean channel of a bandit 112 with high volume. With a strat very close to Rory's classic early sound.
Analog solid state but not that surprising: Rory was a rangemaster through an AC30.
AC30's have very little power sag because they are biased hot, leaving very little sway in the B class region. They do have cathode bias though. Power cathode bias shift makes for colder bias at large signals introducing.... cross-over. It gets "squeezed" below, a bit tightened in the throat where the "fixed bias power sag" rather gets pushed down from the top. The sizzle and slight loss of RMS power in the Bandit t dynamics power amp circuit mimicks exactly that cathode bias feeling.

[marcelomd]

I was watching mr Piera talking about the MXR Duke of Tone. Somewhere he mentions "clean mode", where all diodes are lifted and only the opamps distort. Reminded me of this thread.