Musical Soft Clipping Circuits - Tri-to-Sine Waveshapers & the Differential Pair

[teemuk]

To add crossover distortion ?

To drive (possibly) low impedance loads effortlessly. IVP preamp, which was a sort of "high end" thing aside the lo-fi "tube voice" circuit, buffers each output (outputs and two effect loop sends) with this circuit. Clone of the effect probably features the circuit just because the IVP preamp had one.

[Rob Strand]

To drive (possibly) low impedance loads effortlessly

Like 600 ohm.

[PRR]

Yeah, that's a 600Ω buffer for wimp opamps.

Note 13k(?) pull-up/down resistors. The opamp sees like 6.5k, ok for '741 etc. The load sees this through hFE, or (at the limit) say 13k/100, or 130Ω; in series with a 600Ω still leaves 80% voltage swing, a heap better than a naked '741 in 600Ω. The 10Ω emitter resistors also work well at 600Ω.

Here, all this grunt is spoiled by weeny 10uFd cap and 50k pot. So it does look like a hasty re-work.

[Vivek]

reference the schematic of the full blown IVP preamp,

The interesting schematic of transistors at output of Opamp also appear a few times at internal stages in the circuit.

Even at places where it is driving fixed loads of about 40K ohms at one place and 100K ohms at another place.

If the reason was to drive low impedance devices, that construct would only appear right at the very end,

and it would not have a volume pot there after the transistors. I suppose it is more correct that the transistors would directly drive the low impedance device while the final volume control would have been moved back as a gain control earlier on in the schematic.


Hence I do believe that there is more to that construct than only driving low impedance lines.

[Rob Strand]

he interesting schematic of transistors at output of Opamp also appear a few times at internal stages in the circuit.

As I remember it's when the signal goes out of the box.   FX loop and main output.
All outputs exposed to the evil outside world have the power booster stage.

Yes, when it goes outside the box.  Actually it's two FX loops, and the main output,
https://www.diystompboxes.com/smfforum/index.php?topic=121884

Hence I do believe that there is more to that construct than only driving low impedance lines.

It's a *very* common circuit back then.    Strong outputs is something you would expect from pro-quality gear.

[PRR]

OH!! This is a 1981 TAPCO! That's what they did: '4558 opamps and relay-driver buffers. I had a whole mini-console like that back in the day, quickly stolen. Sweet for the time.

Here's owner/service info and much better schematics.
https://projectivp.files.wordpress.com/2013/06/ivp-ev-tapco-preamp-service-manual.pdf

[Vivek]

Thanks PRR and Rob, for explaining to me the intricacies of earlier designs.

First Occurrence in full AION version of IVP schematic:

It is possible that the jumper was actually a send/return on some variants.


Second and third Occurrence:


The second occurrence is internal, no connection to the evil outside world, feeding a high impedance. Why use this config here ?

Third is exposed to the outside hence the transistors could have a raison d'être.

[Vivek]

OH!! This is a 1981 TAPCO! That's what they did: '4558 opamps and relay-driver buffers. I had a whole mini-console like that back in the day, quickly stolen. Sweet for the time.

Here's owner/service info and much better schematics.
https://projectivp.files.wordpress.com/2013/06/ivp-ev-tapco-preamp-service-manual.pdf

Thank you sir,

Now I understand !!!

It's as Rob says and PRR shows :

Some Send/Returns were removed, but to avoid redesign, they let the buffers remain vestigially.

[teemuk]

First Occurrence...

Second and third..

Maybe you should refer to genuine factory schematics. In those the circuit, as said earlier, is always employed in buffering signal outputs.

3rd party circuit drafts, sketches and schematics of various "based on" -effects are not the best source of information about the real circuit. The IVP clone site also has several schematics that merely portray the author's progress of sketching the circuit to Eagle and the most up-to-date Eagle schematics features those effect loops at places where the earliest versions do not yet show them. The factory schematics have naturally always portrayed them.

[Vivek]

Thanks Teemu !

I didn't know these factory schematics existed.

That might have made job of tracing the preamp very easy / unnecessary.

[bowanderror]

The tube sound on the IVP preamps used to sound pretty good on bass.

Many bass cabs in that era were loaded with EV15L's which removed a lot of the fizzy high end.   With some of the cabs today you would probably want to add a low pass filter, somewhat more roll-off than anything from a small transformer.   Although shaving of above 9kHz never hurts anyway.

to waste current so that the power supply sags ?

You might be interested in the Lab series L5, L7 etc amps.   They fed the signal into the IABC pin of an OTA to get second order distortion.

Large schematic link at the bottom of the page,
http://www.netads.com/~meo/Guitar/Amps/Lab/schem.html

Great find! In the L5-series, the Ch1/Ch2 channel volumes control the amount of distortion generated by the OTA, which then feeds the Master output & Compressor circuit.

That part of the circuit always interested me, which left me wondering why they'd go through all the hassle of adding a specialty component just for overdrive. But as an open-loop OTA, I believe it has similar non-linear characteristics to the other circuits we've been talking about!

The Aion L5 Preamp clone build doc has an easier to read version of the schematic:


I'm not too familiar with the CA3094, but it looks like a CA3080 with an integrated buffer section (like the LM13700 has). I feel like I remember reading somewhere that the CA3080 had different clipping characteristics than an LM13700, so I do wonder whether an LM13700 adaptation (the only OTAs I currently have) would distort in quite the same way. Does anyone have any more info on why overdriving a 3080 might sound different from a 13700?

[teemuk]

OTA overdrive is sort of a Dan Pearce trademark. After Lab Series he also started to combine a cascade of diode clipping and OTA overdrive. Peavey used them too in the VT series amps. What you feed to the Iabc input can be used to control symmetry vs. asymmetry.

13700 has those "input compensation" diodes that pre-distort the input signal to counteract OTA's non-linearity. They can turn transfer characteristics more linear and distortion more abrupt. For softer clipping one must disable them.

[bowanderror]

OTA overdrive is sort of a Dan Pearce trademark. After Lab Series he also started to combine a cascade of diode clipping and OTA overdrive. Peavey used them too in the VT series amps. What you feed to the Iabc input can be used to control symmetry vs. asymmetry.

13700 has those "input compensation" diodes that pre-distort the input signal to counteract OTA's non-linearity. They can turn transfer characteristics more linear and distortion more abrupt. For softer clipping one must disable them.

How does the Iabc input control symmetry, is it based on DC offset or something?

That makes sense about the input compensation diodes, do you disable them by just leaving that pin unconnected?

[ElectricDruid]

do you disable them by just leaving that pin unconnected?

Yes, exactly.

[teemuk]

How does the Iabc input control symmetry, is it based on DC offset or something?

Fixed Iabc current: The OTA differential amp overdrives symmetrically.

Feed the input signal to Iabc to modulate the Iabc current. The gain increases at one lobe of the waveform, correspondingly decreases on the other. Asymmetric distortion.

Feed full-wave rectified input signal to Iabc: The gain decreases at peaks. This instantaneous gain control is another way to achieve soft clipping, which does not actually rely on overdriving the OTA. (Refer to e.g. Fender Rumble V3 series amps). Modulate amplitude of rectified positive and negative lobes to control symmetry.

[Vivek]

Back to the initial question :

Do all these different esoteric ways to create distortion really make an audible difference ?

Can the different types of distortion be identified by the lay guitarist or lay audience ?


I want to believe that even the difference between "Soft" and "Hard" clipping is fairly minute (except on a scope). Hence all other differences are even minuter.

( Tom S. made some Rockman versions with hard clipping and some with soft. I feel that with the compressor, chorus, delay, cab sim etc, any differences between clipping styles gets hidden deep down)

[teemuk]

You can easily audition it to find out.

Meanwhile...

This youtube clip is not made by me but demonstrates quite well the impact of different types of clipping. There's also very little other processing / EQing included besides plain peak clipping, so you can hear how clipping sounds unpolished and in its rawest. (It, IMO, always sounds rather bad as is and without any notable pre & post emphasis).
If you fast forward near the end of the video there's a pretty good demo to audition where the demonstrator just plays the same riff and switches between different types of clipping.

Summary:
- Threshold makes the biggest audible difference, and in that perceived loudness wrt. amplitude after clipping
- if volume is normalised the difference in clipping types is slight and you mostly hear higher threshold clipping (e.g. LEDs) sounding cleaner and lower threshold clipping ( e.g. silicon diodes) sounding more distorted, which makes perfect sense; low threshold clips/distorts more
- if you normalise both threshold and loudness the differences turn extremely subtle
- very hard clipping sounds a tad brighter than very soft (and actually quite good, IMO), very soft clipping sounds merely "mushy" (and, IMO, rather poor). Soft vs. very soft clipping has a negligible difference
- all clipping types have the same basic character, defined by overall voicing and EQing, which again makes sense. i.e. Fuzz style voicing makes all types of clipping from soft to hard sound like fuzz. Type of clipping just slightly alters the overall timbre.

[anotherjim]

No mention of CMOS inverter soft clip?
Tri-Sine...

Ok, not perfect, but it's easy (the triangle isn't perfect in the first place). Just a CD4069U inverter amp with very modest gain. x2 in fact.

[Steben]

think of around 2:16. That's something that makes a difference when comparing clipping and transfer curve. Not a metal power chord blast or alike, any blown TL072 will do for that.

Or this. Completely.
Rory and Ritchie are gods but don't tell me they would sound as good as that with a LED clipper through a mixing board.
These guys felt that guitar volume control was an instrument on its own.

You can see the amps in the Irish tour video, the camera pans around behind them.

He's plugged into a low power Tweed twin, daisy chained from the second input of the Twin into a Dallas Arbiter Rangemaster into the Normal channel of the AC30. The Tweed Twin has speakers with aluminum dust caps. I believe those are Altecs.

This is my favorite era from one of my favorite guitar players. I've chased this tone forever, I have those amps and the original Rangemaster, and I can duplicate that sound exactly, minus the magic touch.

Hogy

[dschwartz]

No mention of CMOS inverter soft clip?
Tri-Sine...

Ok, not perfect, but it's easy (the triangle isn't perfect in the first place). Just a CD4069U inverter amp with very modest gain. x2 in fact.

Wondering the same...an inverter does the same job.
And the video of the differential OD sounded very similar to a single 4049 stage distortion.
For low gains works well, but for high gains you don't want soft clipping..it sounds mushy and congested.