Trying to smooth the abruptness of a classic OpAmp clamp circuit...
Any ideas?
Thanks! ...Jim Thompson
Trying to smooth the abruptness of a classic OpAmp clamp circuit...
Any ideas?
Thanks! ...Jim Thompson
-- | James E.Thompson | mens | | Analog Innovations | et |
Would resistance in the feedback path cause the diode to turn on slower? A few diodes in series? An LED?
(lotsa ideas, few good ones.)
George H.
Only if you restrict the gain of the OpAmp, then the final VCLAMP accuracy suffers... that's why I posted the question... I'm somewhat stymied :-(
...Jim Thompson
-- | James E.Thompson | mens | | Analog Innovations | et |
How accurate a clamping voltage? How smooth a corner? How many extra parts? How cleanly does it need to come off of the clamped voltage?
I thought you were a past master at piecewise-linear circuits with diodes and op-amps -- is there none of those you can use?
Triodes are supposed to have nice warm distortion characteristics -- what about using an op-amp with a 6J5 output stage to provide a smooth clamp. If you need to clamp top & bottom, invert that and feed another one. Or, give us enough information that it's less tempting to meet your specifications with something absurd.
I don't think it'll be easy to get a really accurate clamping voltage while getting a smooth transition to that clamping voltage, at least not without using lots of parts. But I'm much better at throwing lots of parts at a problem than finding a really elegant solution, so I won't be surprised to be surprised.
-- Tim Wescott Control systems, embedded software and circuit design
The simplest method is probably to reduce the effective open loop gain.
At the price of a bit of noise and inaccuracy, you can do the usual gain-reducing trick for running decompensated op amps at unity gain, inside the loop.
Resistor from Vout to SJ, resistor from SJ to +Vclamp, resistor from SJ to output.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant
Another approach would be to drop Vclamp a bit and use a diff pair to increase it again smoothly when Vin crosses Vclamp.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant
Oh how about two clamps, first one reduces the gain by 1/2 or something, and the second one clamps. Like the sqrt circuit Spehro mocked up here for me. (Then you get two sharp corners.)
George H.
Or use two low gain amps controlling a 4-diode sampling bridge that clamps Vout to Vclamp. That'll give nice smooth transitions.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant
Use a geranium diode. Or perhaps germanium + R // silicon. Make input impedance low. If it were for fun I might suggest copper oxide. Look at guitar fx, it's a classic question there.
NT
Or take it further with 4 diodes... silicon, schottky, germanium and HV silicon, 3 of those 4 with their own series R.
NT
You might take a look at the circuitry used in some of the classic analog function generators. They make sorta-sine-waves by generating a square wave, integrating it into a triangle wave, and then running it through a soft-clamping network which involves the use of multiple resistor-diode current suckers in parallel.
Either limit the op amp slew rate, or limit the abrupt turnon of the diode. So, putting an inductor in series with the diode would do... something nearly appropriate. So would using a low-power, low-slew op amp. The slew also goes with the forward voltage of the diode, so multiple diodes in series (or an LED) might be a productive approach.
That all presumes you are OK with time-domain trickery, which doesn't accomplish a low-speed characteristic curve correction.
Doesn't work too well at DC ;-) ...Jim Thompson
-- | James E.Thompson | mens | | Analog Innovations | et |
No. If you told us more...
NT
I'm pretending to be a leftist... you're supposed to guess the outcome
...Jim Thompson
-- | James E.Thompson | mens | | Analog Innovations | et |
IRL, or /in silico/? (I'd think the latter is a long-solved problem, given your toolkit of TANH's and whatnot, but checking just in case.)
The IRL manifestation is due to Vf knee divided by loop gain. Of course, Vf is temp sensitive, so you can't make the knee itself stable without adding many dimensions of compensation hackery (i.e., errors with respect to Vf/Vth, N, Is, and leakage).
Is that what you're after?
But then, Vclamp precision suffers, when gain is low. Are you looking to avoid that, too?
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design
In simulator-land ;-)
My TANH/COSH stuff is working great for merging to slopes, but I can't find a "soft" equivalent for clamping a node that is current driven.
I've solved the diode parameter variation... it's downright trivial... measure the diode drop and subtract it...
but "soft", when gain is low does exactly as you say... sloppy precision. ...Jim Thompson
-- | James E.Thompson | mens | | Analog Innovations | et |
Almost...
Finally sat down and did the math on paper rather than groping with the simulator :-[
Still not quite right, slope at origin should be RG, so back to revisiting the math ;-) ...Jim Thompson
-- | James E.Thompson | mens | | Analog Innovations | et |
You could always do the lightbulb & resistor trick. But with little info who knows if the resulting set of limitations suit you, probably not.
NT
You missed my post replying to Tim Williams that this is for a simulation model.
After I sat down and actually did the math properly, I came up with...
which gives me the independent degrees of freedom I need, slope, transition width, and clamping voltage... end use if for an OpAmp behavioral model... gain _and_ swing limit control... with _smooth_ derivatives. ...Jim Thompson
-- | James E.Thompson | mens | | Analog Innovations | et |
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.