Opamp bias for Asymmetrical clipping

[seedlings]

Normally an opamp is biased at 1/2 the circuit voltage.  Changing the bias introduces a DC offset to the wave.  Let's say a non-inverting opamp bias was 4V instead of 4.5V, with a circuit voltage of 9V.  Does this mean that once the output exceeds 4V p-p, the lower half of the output wave begins clipping, and at 6V the top half begins clipping?

Thanks,
CHAD

[amptramp]

To an extent this may happen but most op amps are not symmetrical about where they clip in the first place.  The workhorse TL072 appears to be close to symmetrical in peak output swing at ±12 volts for a ±15 volt supply but the LM324 swings from Vcc-1.5 volts to V- + 0.02 volts based on the output swing.  There is a separate from the input common mode range spec which tells how far the input can go without running into limitations.  For the TL072, the spec is a minimum of ±11 volts but is typically -12 to +15 volts operating from a ±15 volt supply and for the LM324 is Vcc-1.5 to V- volts.

Note that some op amps behave pathologically if you exceed the input common mode range including latchup and output inversion.  Check the common mode input voltage and the output swing.  If you have gain in the stage, the output swing may be the first limit you run into.  If you have unity gain or reduced gain, the input may be more critical and it also exhibits pathological behaviour with some device types.

[GibsonGM]

To an extent this may happen but most op amps are not symmetrical about where they clip in the first place.  The workhorse TL072 appears to be close to symmetrical in peak output swing at ±12 volts for a ±15 volt supply but the LM324 swings from Vcc-1.5 volts to V- + 0.02 volts based on the output swing.  There is a separate from the input common mode range spec which tells how far the input can go without running into limitations.  For the TL072, the spec is a minimum of ±11 volts but is typically -12 to +15 volts operating from a ±15 volt supply and for the LM324 is Vcc-1.5 to V- volts.

Note that some op amps behave pathologically if you exceed the input common mode range including latchup and output inversion.  Check the common mode input voltage and the output swing.  If you have gain in the stage, the output swing may be the first limit you run into.  If you have unity gain or reduced gain, the input may be more critical and it also exhibits pathological behaviour with some device types.

You raise some great issues, amptramp.  At some point soon I'll breadboard a couple of '72 gain stages and try to get them to clip assymetrically while watching them on the scope.   Due to the concerns you raise, we may see something a little different than we expect!   I'd think you can clip assymetrically to a point, then those limitations will raise their heads...

[ashcat_lt]

One question I've often had is "how assymetrical is enough?"  It must have something to do with the relative difference between the two swings, no?  I mean, an "assymetrical" square wave is really just a square wave with a DC offset, right?

Anyway, the LM324 is actually really cool because it can run off of very low rails and it has that assymetrical limit.  Set it up with a 3.5V supply and a 1.75V bias and you get something very similar to what you might get from a Rat or DS1 with a diode one way and an LED the other.  Course, somebody said something about crossover distortion...

[seedlings]

Thanks for the thoughts.  I'm going to breadboard tonight and check results with the scope (and ears).

Course, somebody said something about crossover distortion...

I thought opamps kind of reduced crossover distortion because of the feedback... but the bias DC offset will re-introduce?

CHAD

[Seljer]

I once breadboarded two opamp stages with off-centre bias. I was aiming to get the clipping asymmetry to mimic the clipping you'd get from a 12AX7.  I couldn't get the decay to sound natural, if you strummed a chord as it rung out you could precisely hear the points where the top and the bottom of the signal stopped clipping. I messed around with filtering and hybrid setups with one half clipping on the rail and the other half with diodes in the feedback loop bit but eventually gave up with it. The clipping from a simple diode going to ground sounded better to me :shrug:

Another idea that I've pondered about is instead of using a fixed reference voltage to mix in a bit of envelope taken form the input signal, so you have a moving bias point and clipping symmetry that varies with your playing.

[ashcat_lt]

Thanks for the thoughts.  I'm going to breadboard tonight and check results with the scope (and ears).

Course, somebody said something about crossover distortion...

I thought opamps kind of reduced crossover distortion because of the feedback... but the bias DC offset will re-introduce?

CHAD

Naw.  I haven't really looked into it, but I think it's PRR I've seen say that the LM324 is prone to crossover distortion.  I breadboarded a really nasty circuit which did essentially what I noted above.  Sounded pretty much the way I wanted, and if there's crossover distortion (never scoped it) it's not enough to come out sounding gated, and probably just adds to the fun.

[Johan]

Last stage in opamp big muff is biased ~20% offset...
J

[R.G.]

I thought opamps kind of reduced crossover distortion because of the feedback... but the bias DC offset will re-introduce?

(1) The situation with opamp outputs and crossover distortion is that only opamps with Class AB outputs have crossover distortion. Opamps with Class A outputs do not have crossover distortion. This is exactly like audio power amps, which may be thought of as giant opamps for all practical purposes.
(2) Opamp feedback does reduce crossover distortion, but it is remarkably hard to completely eliminate it entirely from a Class AB output stage.
(3) Bias DC offset doesn't change the crossover distortion, but it may change where you run into it. Crossover is, by definition, where output current stops being from one polarity output stage and "crosses over" to being supplied by the other polarity.
(4) Some opamps have a sliding bias arrangement that has less crossover at higher currents, so you can change things by loading them from one or the other power supply with a resistor to raise the current on one side or another. This is a subtlety beyond the common effects hacker to deal with, and may deliver only miniscule changes, if any.

I once breadboarded two opamp stages with off-centre bias. I was aiming to get the clipping asymmetry to mimic the clipping you'd get from a 12AX7.  I couldn't get the decay to sound natural, if you strummed a chord as it rung out you could precisely hear the points where the top and the bottom of the signal stopped clipping. I messed around with filtering and hybrid setups with one half clipping on the rail and the other half with diodes in the feedback loop bit but eventually gave up with it. The clipping from a simple diode going to ground sounded better to me :shrug:

This is exactly what one would expect. A tube's clipping is not only asymmetrical, it's softer, in that it has more rounded corners at the entry and exit to clipping. Opamps clip razor-sharply. So even if they clip at similar points, the opamp will sound harsher because it produces different harmonics by clipping more abruptly. In fact, it is hard to say that tubes "clip" as opposed to "squash".

Another idea that I've pondered about is instead of using a fixed reference voltage to mix in a bit of envelope taken form the input signal, so you have a moving bias point and clipping symmetry that varies with your playing.

It's a good idea to play with. I think it's been done a few times. The trick is in getting the signal size and clipping points synchronized well. Again, better to use some soft clipper than an opamp. Something like the Vox limiter would be handier. You will probably have issues with the need to DC couple this so that charging/discharging coupling caps doesn't create ugly lags and lapses.

[aron]

I did fool around with the bias voltage - not much there.

[seedlings]

Thanks all, RG... during the work day ideas pop into my mind ex nihilo, and usually they pop out of my mind to be forgotten.  This one stuck with me, so I posted a quick question.  I stay busy wondering about this and that, only to get home and run out of time to mess around with them.

I did fool around with the bias voltage - not much there.

Excellent.  Now I'll put the matter down, which gives me time to re-invent the ice cube.

CHAD

[PRR]

> only opamps with Class AB outputs have crossover distortion.

Good AB has minor crossover distortion. The Usual Chips won't be offensive.

Class _B_ exists. Is a "feature" of the LM10 (yeah, nobody uses the LM10). The LM324 shifts from class A for light loads to a pretty harsh class B on heavy loads. This was hinted in the old app-note, but never shouted-about.

Here's the '324 output:


For load currents to 50uA, the current-source keeps Q8 in class A. This could be 5V in 100K, or 0.5V in 10K. Yes, some systems stay class A.

For larger currents, point "A" must drop *1.8V* to turn-on Q13 and pull-down. Point "A" must hop 1.8V every time the load current changes direction.

[Johan]

I did fool around with the bias voltage - not much there.

Excellent.  Now I'll put the matter down, which gives me time to re-invent the ice cube.

CHAD

...the wheel could use a run over too. Sure, materials and colours have changed with fashion, but the basic shape has been the same since the beginning...

[amptramp]

> only opamps with Class AB outputs have crossover distortion.

Good AB has minor crossover distortion. The Usual Chips won't be offensive.

Class _B_ exists. Is a "feature" of the LM10 (yeah, nobody uses the LM10). The LM324 shifts from class A for light loads to a pretty harsh class B on heavy loads. This was hinted in the old app-note, but never shouted-about.

For load currents to 50uA, the current-source keeps Q8 in class A. This could be 5V in 100K, or 0.5V in 10K. Yes, some systems stay class A.

For larger currents, point "A" must drop *1.8V* to turn-on Q13 and pull-down. Point "A" must hop 1.8V every time the load current changes direction.

One of the mods that high fidelity designers use when using an op amp is to use a pulldown (or pullup) resistor at the output to keep the output pulling in the opposite direction for small signals, keeping the output in Class A.  For small signals like a guitar output with no boost, the output can be kept in Class A over the entire signal range.

[nocentelli]

I may be mistaken, but I think the Tech21 XXL pedal uses a pot to adjust the bias voltage going to the non-inverting input of the first opamp (it uses opamp clipping rather than any diodes). The control is labelled "warp" and has quite a profound effect on the distortion sound produced.

[seedlings]

I may be mistaken, but I think the Tech21 XXL pedal uses a pot to adjust the bias voltage going to the non-inverting input of the first opamp (it uses opamp clipping rather than any diodes). The control is labelled "warp" and has quite a profound effect on the distortion sound produced.

Well, now.. there is nothing new under the sun!  From the thread: http://www.diystompboxes.com/smfforum/index.php?topic=29094.0



Now I WILL try it on the breadboard.  Warp speed, Mr. Scott.

CHAD

[seedlings]

I tried varying the bias voltage by a couple of volts and it only seems to compress.  Meh.

CHAD