Limiting Op-amp Output

___ o----o-|___|o----------o | | | | | | o----+->z--z+-----------+ -|+/ |/| (created by AACircuit v1.28.6 beta 04/19/05

Something like that?

Jamie

Cute, but hard to incorporate into an integrator.

John

It leaves the integrator part completely undisturbed. In case that's what Tim wanted. But not sure if he does.

If the summing node moves you get windup... actually you get charge on the internal compensation capacitor giving you the slo-o-o-ows during recovery. The common-base tranny should have worked. Post your schismatic ;-) ...Jim Thompson

mal'

to

hen

ap. =A0

st

Cheers, I've got my brain round it now.

NT

No, it's good old traditional integrator windup -- beyond a certain point, the increasing amplifier output does nothing to affect the downstream circuitry. But when that circuitry does finally reverse itself, a honkin' big charge on the integrator cap takes _time_ to bleed off, adding much delay and unfortunate response time to the circuit.

That's because the summing node lifted. Sometimes it can be as simple as bounding the summing node with diodes to ground. ...Jim Thompson

Nuh uh. The integrator could be perfectly in "op amp mode" without hitting a rail, but if it rises too much then the following circuit's input saturates. The op amp in question turns around and starts decreasing its voltage immediately -- but it still takes a while for the following circuit to get out of saturation. Hence, delay.

Then John Larkin's suggestion is probably closest. But I'd be leery allowing so much reverse Vbe. If the opamp output ever spikes negative it'll destroy the BCX71, it can only take 5V. Can be protected though.

The transistor would have to be across the integrator capacitor so it will be forced to stop charging. The younger stages would have to reverse fast enough though, else it won't help much.

If you have to throttle at younger stages to avoid "slack" you could use my method with the NPN plus zener and flip polarity as needed. But you'd have to add a wee hysteresis to it so it doesn't go instable on you while hovering at the 5V point.

Aha! So your problem is really about overloading a following stage? Post you common-base schematic. That should have worked. Hurry! I'm alrady 1/2 bottle of Pinot Noir ahead of the discussion ;-) ...Jim Thompson

Question: How can that get stuck? If you let it gradually "ease in" as a regulated feedback like in John's circuit I don't see how it can get stuck.

Here's the schematic. It looks like it should work, and it's been so long since I was doing it I can't remember how it wasn't working -- I just remember that it was spitting up somehow, and I had bigger fish to fry at the time, so I just took it out.

Now the really big fish are fried, and I need to get at this one...

.---------o-----. | | | | | | | | \| Vlim | --- |--------o | --- >| Vin ___ | |\ | | o--|___|---o--|-\ | | Vout | >---o-----o------------o .---|+/ | |/ | | === GND (created by AACircuit v1.28.6 beta 04/19/05

One thing I do see wrong with it is that it's going to be exceedingly easy to exceed the BE reverse voltage. But I don't think that's why it was messing up. Sigh. More things to think about.

That might work. It'd be a bit soft, and not adjustable, but it might work.

Try some resistance in series with the emitter, "stiff" limiters often give heartburn. ...Jim Thompson

Put a full-wave rectifier in the feedback loop, with "load" as an adjustable TL431 ;-) ...Jim Thompson

The Vbe problem could be fixed, and should be:

.---------o-----. | | | | | | | | \| ___ | --- |--------|___|------- Vlim | --- >| | Vin ___ | |\ | | V o--|___|---o--|-\ | | - | >---o-----o------ '-------------- Vout .---|+/ | |/ | | === GND (created by AACircuit v1.28.6 beta 04/19/05

I don't know why it should mess up. It does have a ton of gain in this new feedback though, maybe that did something?

Looks like European resistors symbols are making it across the pond :-)

I included a zero clamp, too. So the base is only back-biased 4.3 volts.

I don't know the typical B-E zener voltage of a BCX71... I suppose I should. I've generally found PNPs to be higher than NPNs.

John

I can't remember! I yanked the transistor out of the circuit over a month ago, because it was only sticking occasionally and I had a lot of other unknowns to work through. Now it's the biggest problem, and all I remember is looking at my circuit, thinking "damn, of course that's not going to work", and ripping it up.

Sigh.

... but mind the off-state leakage current of the TL431. Easy on the Pinot Noir there :-)