Current Controlled Resistor

an

k to

ive

Can any other optoisolators be used in a similar way? No shortage of jellybean optos here.

NT

There's an IC for that; LM13700. Especially note figure #10:

it's good for nearly three orders of magnitude range of resistance.

can

IV

give

it

ll

ast

..

ut

ot,

The HF11F1, HF11F2 and HF11F3 are photoFETs rather than jellybean opto- isolators (which tend to rely on photo-bipolar-transistors) and I don't know of any real alternatives still in production.

FETS - as majority carrier devices - can look a lot more like resistors than bipolar phototransistors - which rely on minority carriers.

-- Bill Sloman, Nijmegen

Actually, that would be interesting, to plot the curves of a phototransistor type isolator down in the ohmic region... especially going negative. Maybe use a dual coupler and put the transistors back-to-back for +- symmetry? Spare-time project for someone.

The photofet is indeed ohmic, pretty linear at +-50 or 100 millivolts but getting worse for more swing. That would be handy for, say, stabilizing a low-distortion oscillator or things like that.

Building a real, floating, programmable resistor, with volts of swing, isn't easy.

I've played with using a BFT25 as a wideband voltage-controlled resistor, grounded emitter, with DC base current doing the control of collector resistance. It works up to about +-50 mV of signal swing, but DC offset gets up to 10s of millivolts, so it might be usable as a variable RF attenuator.

Small gaasfets behave like jfets, much faster but allow lower voltage swings before things get nonlinear. But no DC offset. NE3509 is nice, gets down to 4 ohms, and drain capacitance is something like 0.35 pF.

John

ou can

e IV

ut 5k to

to give

cuit

of

mall

least

-...

but

pot,

Ah, now I get why it behaves that way. Cheers

NT

can

give

That will work, but LDRs are pretty weird and drifty things, not very quantitative. A dual LDR would be better, so you could feed back on one side to sorta compensate the other side.

About the best variable-gain element around these days is a multiplying DAC. That, plus a few opamps, can make a wide-range variable resistor. We did one like that, with isolated power supplies and an isolated data link up to the floating resistor.

It turned out to be shockingly difficult, and took a couple of iterations to get it to be always stable, and to get the noise and zero offset down, not to mention keeping it from blowing up.

We played with the "PWM resistor" idea, but it had problems.

John

For RF, you can put a grounded variable resistor on the primary of a transformer, and use the secondary as a floating resistor.

Cheers

Phil Hobbs

Power PIN diode maybe, but not below some MHz.

A bunch of relays and a bunch of resistors is really a pretty good way to make a variable resistor. Except for the switching transients.

Our problem in the VME module was to simulate things like floating RTDs with serious accuracy and no switching glitches. And work with AC/bipolar signals. It was messy.

John

It's that BU influence... Larkin thinks he's a man ;-0 ...Jim Thompson

Denying? I said they were weird and drifty, which they are.

Crabby again, I see.

Use your imagination once in a while. You probably have one, hidden away and seldom used, but maybe still there.

The idea is simply to put a switch in series with a real resistor and PWM it. An input cap might help.

John

What in the world are you talking about? PIN diodes make excellent current-controlled resistors, as long as the period of the signal is faster than their recombination times, which is typically in the microsecond range.

I meant switching transients. Think about trying to make monotonic resistance steps using resistors and relays. If you're simulating an RTD, and you want to creep the simulated temperature up, it's not good to have erratic spikes above or below the desired temperature profile.

Mercury relays wouldn't help, even if they were practical. I think the toxicity thing is making them go away. SSRs make much more sense, but still don't solve the transient problem.

John

His weenie fixation getting in the way, maybe?

He has several fixations, but none have anything to do with electronics.

John