Limiting Op-amp Output

Depends on how sharp you want the clamp to be. A 2-resistor level shifter and a diode makes a decent clamp, and you can drive one end from a pot or DAC to set the clamp level. Or use a DPOT and a diode.

A zener is easy, if leakage is tolerable.

Something like this ought to work...

ftp://jjlarkin.lmi.net/Clamp.JPG

or, programmable,

ftp://jjlarkin.lmi.net/Clamp2.jpg

The two resistors could be a DPOT, and -Vset some fixed voltage. Or use a pot.

A really sharp clamp would need another opamp in the loop, fairly messy. Diodes are just too soft.

John

Google "clamping opamp" Something like AD8036

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

One way is current limiters in the supplies.

In the bad old days we often wanted to prevent "integrator windup" in analog PID controllers.. but I digress..

How about a TL431, a PNP BJT and three resistors? Total about 10 cents on the BOM.

A clampamp, the AD or HFA1130, will clamp the integrator output, but the summing node will wind up. So it won't back off the clamp nicely.

Clampamps work great for amplifiers.

John

=A0

I'm plain confused by that one. Would yo like to talk thru the operation of the feedback network?

NT

But that won't provide clean, immediate backoff from an integrator clamp.

One could use a r-r opamp, run from +5 and ground, and use back-to-back diodes from the summing point to ground. It wouldn't be perfect, since there would still be some windup.

John

Either I'm constrained to a specific amp (it needs mondo current capability), or I add yet another amp to an already-porky circuit, and one that the customer doesn't currently stock, to boot.

Good idea, though -- if I can come up with a decent way (and I'll look at your diode idea) to do it without changing the amp I'll present that, plus the rail-rail idea.

That's exactly what this is for -- just in a circuit that closes a loop at around 100kHz, instead of an electromechanical thing that works three to five orders of magnitude slower.

Drive the non-inverting input. Extra P gain could be accounted for.

VLV

When the output starts to go below ground, the upper diode conducts and shunts the feedback resistor (or cap, if an integrator.) The other diode cancels the diode drop of the first one, so the clamp is closed to zero.

When the output tries to creep above +5, the PNP turns on.

Hmmm, the zero clamp could have been a transistor, too.

John

The only "dumb" question is the one you don't ask.

Thanks for writing what you're trying to accomplish here. Are you constrained to using an opamp? Have you considered a (GASP!!) 555?

Cheers! Rich [1] Maybe Nomex would be more politically correct. ;-P

It can, if this is not an RF amp. Running the rails sans bypassing and controlling secondary transistors with those is one of the methods to turn a plain vanilla opamp into a high voltage amp.

Ain't those constraints always nice ... :-)

If it's thermal damage you are worried about and space is a problem, how about PTCs? I am not a fan of those but if too many added parts are a concern it might be an option.

If you can bound the length of time the output can spend railed, you can use a MOSFET to disconnect the input current to the summing junction when the output gets close to the rail.

There's also Widlar's trick for nice sharp breakpoint amps (National AN-4). You hang an emitter follower on the summing junction, with the collector connected to the clamp voltage. When the emitter hits the collector voltage, the transistor saturates and essentially connects the SJ to the emitter, plus or minus a diode drop.

Of course that's pretty leaky for use with an integrator, but it makes decently temperature compensated breakpoints if used with some care.

Cheers

Phil Hobbs

Ignore my previous posts, thought for some reason you wanted current limit. But you want a 5V output limit. So how 'bout this:

|\| -|-\ | >------' .------o--|+/ | | ' |/| z Zener voltage: / | A Negative supply plus 4.3V \ \| | / |----------. \

Well, I think you can forget about a TL431 if you're looking for that kind of response speed.

Maybe the right hand GND should be -V?

Feed back Zeners?

Jamie

Oops, sorry, yeah. Rev B:

|\| -|-\ | >------' .------o--|+/ | | ' |/| z Zener voltage: / | A Negative supply plus 4.3V \ \| | / |----------. \