slow down an opamp (compensation cap.)

Do the math. Write your loop equations. ...Jim Thompson

Grin... OK. GH

=0irst

Design an amplifier to avoid AC hum pickup, one has to avoid radiated commo n mode input turning into a differential input. This means this input imped ance must be matched and balanced. Any references to ground must not be ind uctive loops with all signal connections using matched impedances coupling to stray hum E fields and ground.

=0irst

You might be. One the ways that high frequency interference gets into slow op amps is when the input pair spends time outside it's linear region - rou ghly +/-50mV for a bipolar pair. FETs have less transconductance, and can t olerate an order of magnitude bigger excursions.

Slowing down the feedback loop allows lower-frequency interference to do th is.

Why not hang a cap across the TIP32? If there's hum there, it gets rectified into a DC error, and you can't fix that downstream.

But why run 10 uA? That makes the transistor/diode impedance high, easier to pick up hum. This isn't a Lakeshore diode!

Right, I do that... mostly to get rid of the room lights.

Yeah 100 uA is fine, we could add a switch... as long as it's a transistor connected diode you can get the temperature to ~1% by measuring the voltage at decades of current. (1-100 uA is the sweet spot.)

George H.

to

=0irst

mon mode input turning into a differential input. This means this input imp edance must be matched and balanced. Any references to ground must not be i nductive loops with all signal connections using matched impedances couplin g to stray hum E fields and ground.

I've got a four foot antenna (wire) leading to a diode with ~2k ohm of impedance. At the moment the pickup problem has very little to do with the amp... (It's an instrument amp, which should have fairly balanced input impedance.) My (perhaps silly) philosophy, is that if I can get it to work and look nice in my lab bench environs then once it goes into a can with a shielded cable it will be fine.

George H.

I'm with John--1mA through your sensor "diode" and a capacitor!

Slowing the heater driver seems like a good track. If that doesn't help, it's time to hunt for feedback from wiring drops, e.g. from the heater current getting into in your sensor reading.

Cheers, James Arthur

le to

dl=0irst

ommon mode input turning into a differential input. This means this input i mpedance must be matched and balanced. Any references to ground must not be inductive loops with all signal connections using matched impedances coupl ing to stray hum E fields and ground.

Well 10k x 3.2 nF..(3.21) I tacked the 2.2 onto the existing caps.

But it switched much slower than that ~100ms for 10-20 volts. (the 60 Hz, became ripples on my RC ramp.)

I might have made a circuit F-up... it wouldn't be the first time.

George H.

Hi Jim, Well I don't really know how to deal with the FET. But I drew up this circuit.

|\ Vin---|+\ | >-+------+ +-|-/ | | | |/ C 50R | | | +------+-10kR-+ | 1R | GND

I started to write down the loop equations, but I stopped half way through. (I'll continue over the weekend when I have more time. I also haven't done much of this in a while, so it's easy to wander around in algebra land.. wasting time I don't have today.)

Anyway just looking at it you can see that the 10k feed back R will have something like 1/50 of the voltage driving it, and so I'l be unsurprised if the time constant turns out to be 50 times longer.

George H. )

There is a TO-220 version of LM35. But it has its own personality.

dl=0irst

Like JT says, do your homework.

Consider that for the purposes of the behavior of the op-amp, the 1k resistor from the source of the FET and the cap from the output of the op- amp may as well be shorted. In fact, if I were doing that circuit I'd use the 100pF that you chose for stabilization, and do the rest with a cap in parallel with that 1k resistor -- that way you're actually measuring the thing you care to control more directly.

Thanks Tim. (I didn't get to this on the weekend. Saturday was in the 50's (F) and there was this bright light in the sky spreading light and warmth.... Sunday similar, except for the bright light. :^(

I sent it through LTspice, (huh, rolls off as ~Rf*Cf) and then rebuilt the driver stage. I must have blown the output stage (or something) on the opamp... there is 50V floating around.

So I'm back to sharp edges.

You're suggesting a cap in parallel with my 1k gate resistor?? (sorry I've got two 1 k resistors on the dropbox link, the feedback resistor (from 1 ohm sense R) is now 10 k.) My primitive mind things of sticking LP filters everywhere. Is it silly to think of hanging more capacitance on the gate?

George H.

...ugh..............thinks... my primitive mind things too... but not in public. :^) GH