Op amp open-loop gain variance

[FiveseveN]

I've been working on a distortion that employs an op amp in open-loop configuration (à la Anderton Ultra Fuzz). While swapping different chips (TI TL062, same batch) I noticed a significant difference in gain between them. Fortunately too, else I might have run into issues down the line. Only other chips I had on hand were four RC4558s, which seem much more consistent (and unsurprisingly more gainy). Then again, not a very big sample.
So my question: is this a consequence of the JFET input? The ones with bipolar input seem to produce a more gated fuzz, which I'd rather avoid. Maybe I haven't found the right one?
I know it's not exactly a common issue but I'd appreciate any insight.

[antonis]

(although I didn't exactly get your query..)

Open loop is open loop, meaning no predictable gain, oscillation, hitting supply rails just with a sneeze, etc...

Also, the Gain-Bandwidth product is speaded even among Amps of same type/batch..

The main difference between Fet & BJT input amp is the greater input bias current of the second (lower input impedance) togheter with lower open-loop gain..

So, what exactly we are searching for..??

[FiveseveN]

So, what exactly we are searching for..??

I'm mostly wondering why my handful of 4558s are pretty much dead on while my handful of 062s are all over the place.

[reddesert]

Are you measuring the gain of the op-amps, or using "gainy" to describe the sound?  I've never built an Ultra Fuzz, but from the circuit description http://www.diystompboxes.com/smfforum/index.php?topic=56615.0 and using an op-amp open-loop as a comparator, I would expect it to be pretty gated sounding. It could be that the behavior is due to the input impedance loading the circuit in some way. Generally, op-amp gain is specified as "very high" but no guarantees of consistency. Running an op-amp open loop is not going to have predictable results in an AC circuit (in DC, it slams the output into the supply rail).

[PRR]

Did you notice the plan is for BI-polar supplies, not single +9V and Gnd?

If you power it single-supply, the at-zero inputs will be messed-up different for every internal topology. Some will "work", others will be "jammed". And since the thing is GROSS Distortion in any case, it may not be clear "why?".

[anotherjim]

If you leave the sensitivity control unchanged when swapping between JFET & Bipolar input amps, you would expect there to be a difference in the "gating" of the "comparator". The input bias current of the bipolar amps means a small, but significant voltage shift on the input resistors. If the resistance values were equal for + and - inputs, the input bias currents would cause roughly equal offsets, but the resistances are not equal. It might be less significant if that 100k on the + input was reduced in value which I think ought to reduce the offset voltage from a bipolar amp input. Or try adding 100k in series with the sensitivity pot wiper to get a similar offset on the - input.

[FiveseveN]

Don't focus too much on the Ultra Fuzz reference, I was just too lazy to show you what I'm actually working with, but I've come to my senses now, so here's the schematic



To answer reddesert: I am only going by ear but the difference is quite noticeable, and it only takes 5 seconds to swap chips.
I guess I'll have to try with 072s when I place a new order..

[anotherjim]

The effect of input bias current I discussed above still hold true with that scheme, although probably insignificant with the 1k switched in.
I had something similar going (in response to another experiment thread) - but no preamp - using CMOS amps that can sense input at 0v. It meant not needing any bias resistors at all so the reference input was connected directly to ground . The LM358 is a bipolar that can also sense to ground, but that was very gated until I added a resistor in the 0v reference input, so the bias offsets were more equal.

I don't know if slew rate limiting and recovery delay might be masking the effect of GBW running out between the two different breeds of amp you tried. At the upper frequencies it's handling, one kind may may act differently?

[R.G.]

All good replies, but there's a more fundamental truth here.

The open loop gain of an opamp is usually specified on its data sheet. But it's specified as a minimum in most cases, sometimes as a min-max. The makers of opamps think of open loop gain as something you have to have enough of, not a specific value. There is often a "maximum" open loop gain that's not specified, because if open loop gain gets too high, the internal compensation won't be able to keep it from oscillating.

Internally compensated opamps have frequency networks inside that start cutting AC gain so that the open loop gain is cut by one pole (-6db/octave) so the open loop gain is below unity by the time the internal device phase shifts hit 180 degrees. The higher the open loop gain, the low the frequency this compensation pole must be. For many opamps, this pole is down around 1Hz.

So yes, different opamps with different open loop design points will have different AC rolloffs when used open loop. but semiconductor manufacturing being what it is, this isn't going to be all that predictable from wafer to wafer. Get a batch of opamps all from the same wafer, they'll be really well matched. Get different wafers, less so. Different manufacturers, un-matched, different part numbers and manufacturers and you're luck if they'll work in the same PCB.

The whole point of opamps is that the high but not well-known open loop gain can be used with feedback to hide the quirks of the open loop amps under that feedback. Using them open loop means that there is no feedback to hide the quirks, and you may find nearly anything hiding under the rock when you turn it over.

[antonis]

there's a more fundamental truth here.

May we suppose that is you've well recovered from the surgery..??   

[R.G.]

Well - at least enough recovered to be ill-tempered about it not being  completely recovered. 

[tubegeek]

Well - at least enough recovered to be ill-tempered about it not being  completely recovered. 

So, back to "normal"? Get well soon!