Weird (to me) 60 Hz pickup

Oh I should add that this is all open loop, I don't have the heater circuit energized. Though when it is energized I still see similar type pickup as the error signal moves across zero... and of course the idea of the loop is to keep the error signal near zero. :^)

GH

dl=0

Uh, that's weird. Have you looked at what's happening at the input to the _first_ instrumentation amp, and the output of that amp?

You're sure there's no dirt in any of your supplies?

Sure looks like reactively coupled Triac noise. I trust you have shielded twisted pair.

Well the power supply is not particularly "clean" but there's little sign of

George H.

The output of the first int amp is doing the same thing. ~10 mV p-p Looks like it's there on the diode too, but now it's 1mVp-p, and kinda lost in the noise. I uploaded more 'scope shots. (IA1-output-chan2. and diode-chan2.)

Changing to a regular "diode" did nothing. Maybe it's my current source?

One very frustrating thing, is that it seems intermittent. I move the error signal around zero and nothing... then it's back... grumble. Maybe I'll try making some of my own big 60 Hz fields... bringing out the Weller soldering gun!

George H.

Did you look at the power supplies? Where is Vref coming from? A variable supply with dry capacitors maybe?

Cheers

Phil Hobbs

Maybe somebody in the next room is using the microwave oven? If the supplies are really clean, it sounds like rectified RF to me, and microwaves put out a lot of nasty line-frequency-modulated stuff.

Cheers

Phil Hobbs

Hmm well no shielding nor twisted pair on the cable. (It's still in the disgusting looking prototype stage.)

George H.

Yup supplies are OK. Vref is from a 10 volt reference and 10 k pot. (REF102)

George H.

Hmm, being rectified in the diode? Dang interference has gone away... (The Weller gun just made bigger sine waves.. no spikey stuff.) If it comes back I'll try shielding the diode.

I think I can afford to run the diode at 100uA (~0.1mW max power dissipation, having the collector clamped to ground/hot plate should keep self heating to a minimum.)

George H.

George H.

Doesn't have to be a microwave. Electronic ballasts in celing mount fluorescents and compact fluorescents can be potent noise sources that are strongly modulated by (usually) 120 Hz in the US. And, don't trust anything else in the room, either. PCs and all sorts of other gear with switching power supplies can generate a lot of noise.

Jon

If it's the diode rectifying RF, then it would only be in one direction. And wouldn't a TIP-30 be slooooooow?

If it is RF, then a cap across the "diode", sized to not slow the measurement down too much, would be fix and diagnostic all in one.

There's probably some clever way of ignoring high-amplitude short pulses going into a low-pass filter (probably involving more diodes). I'd try to suggest something, but I currently have my head in Mathemagic Land and can't think practically -- I'd give you a suggestion with 27 diodes* that all had to be precisely matched, or something stupid like that.

• 8 x 10 color glossy diodes. I need to lay of the math, but it's what I'm getting paid for at the moment.

And why does the diode care whether the error voltage is close to zero? That's only happening at the input to the 2nd instrumentation amp.

I just had an odd thought -- could it be caused by that 2nd amp, but then either getting into the power supply leads or getting fed back through the I-amp? If it's happening fast enough it may not be getting attenuated by the usual "feedback and tons-o-gain" method of op-amps. It would still require that the instrumentation amp do something odd at zero common-mode voltage that would make it susceptible to weird things happening, though.

dl=0

What's the instrumentation amp? Your drawing makes it look like "AD670", but Analog Devices says that's a DAC.

Yep, that's the most likely. Initiating the current in a fluorescent lamp is just a little less abrupt than a spark gap, and (depending on warmup and other odd factors) can come and go into radiative modes.

Occasionally, a fluorescent tube starts to rectify. Sometimes, reversing the tube heals a bad light fixture, which means the whole ballast was starting to saturate... spike city when THAT happens.

I'll try a cap across the diode. Thanks George H.

Yeah, that may be my mistake, futzing around I found the spikey stuff at other levels... I guess I just notice it more near zero. :^) I have no idea why it's level dependent... I could make up some stories.

I've got 4 feet of antenna on the tip32, and get ~1mV * 2 ms (TC my diode-1uF cap low pass) ~= 2uJ of energy. (The spike rails, things w/o the LP)

The florescent lights over the bench make more at 25kHz. (CW)

Oh my, I've got a cluster "f" of electronics. (I'll take a picture, it's not pretty.) It started out OK, but then I wanted gain here, and something else over there,

George H.

AD620 ... GH

Hmmm I'll try some power supply filters on the current source. (Volt ref->R->out/opamp feedback)

George H.