common mode choke

Imperfectly balanced input filtering can convert common-mode noise into differential noise. A common-mode choke (or single-ended filters) can help reduce RF interferance, but isn't practical to attenuate noise at, say, 60 Hz.

Is your signal source floating or grounded?

What are you conditioning: thermocouples, load cells, something else?

One way to avoid common-mode noise is to have no common mode. Specifically, float the input channel. We make thermocouple/voltage input and output gadgets that float every channel, including optocoupled delta-sigma ADCs or DACs. That works great.

Each channel has its own little floating-ground-plane patch, and the high-side input is EMI filtered against that channel ground, to keep out fast stuff. The integrating delta-sigma ADC does the low frequency (like 50/60 Hz) filtering.

It gets a cover to keep air flow from inducing low-frequency thermal noise, which can be a problem at microvolt levels.

measurements (~10 Hz)? I have measurement problem where I need to make me asurements in microvolt range but in an electrically noisy environment. Un fortunately there are not a lot of options for removing the noise without s pending thousands of dollars which my customer can't afford. I was hoping to use some instrumentation amplifiers with extremely high CMRR and use a c ommon mode choke to limit noise on the lines. Any recommendations dearly a ppreciated.

Thanks for suggestions! I am measuring a Type B thermocouple in a furnace which is producing the 60 Hz noise I am seeing (it is an on/off relay, when relay is on the noise is there, when the relay is off the noise is gone). The furnace is controlled by a separate metal sheath Type K thermocouple a nd I don't see noise there. If I had a lot of money I could buy a platinum sheathed thermocouple for my Type B measurements but that is quite pricey which is why I am trying to filter out the noise. Also, for other reasons I don't need to go into it is necessary to use a platinum Type B thermocoup le attached to the sample and I cannot use a Type K (platinum does not cont aminate the sample but Type K does).

These are relative measurements over a small temperature range so I don't c are about DC offsets or even absolute accuracy. I just need relative measu rements of thermoelectric voltage over a couple hours as I change the sampl e temperature. These measurements don't need to be grounded. What do you guys think? I have done measurements like this before but this furnace in particular seems to have a lot of noise.

Scot

cy measurements (~10 Hz)? I have measurement problem where I need to make measurements in microvolt range but in an electrically noisy environment. Unfortunately there are not a lot of options for removing the noise without spending thousands of dollars which my customer can't afford. I was hopin g to use some instrumentation amplifiers with extremely high CMRR and use a common mode choke to limit noise on the lines. Any recommendations dearly appreciated.

e which is producing the 60 Hz noise I am seeing (it is an on/off relay, wh en relay is on the noise is there, when the relay is off the noise is gone) . The furnace is controlled by a separate metal sheath Type K thermocouple and I don't see noise there. If I had a lot of money I could buy a platin um sheathed thermocouple for my Type B measurements but that is quite price y which is why I am trying to filter out the noise. Also, for other reason s I don't need to go into it is necessary to use a platinum Type B thermoco uple attached to the sample and I cannot use a Type K (platinum does not co ntaminate the sample but Type K does).

care about DC offsets or even absolute accuracy. I just need relative mea surements of thermoelectric voltage over a couple hours as I change the sam ple temperature. These measurements don't need to be grounded. What do yo u guys think? I have done measurements like this before but this furnace i n particular seems to have a lot of noise.

Sounds like there's a very easy approach: turn off the relay when taking a measurement every so often. But presumably there's a problem with that or y ou'd be considering it.

NT

If you're lucky, this is capacitive pickup and not inductive, in which case you should be able to fix it by bypassing, which is easy.

Stick a capacitor across the inputs of the amp, plus another big cap to ground from _one side only_. Start with 0.1 uF.

If that doesn't help, it's probably inductive pickup, which is best dealt with by minimizing the loop areas. (Magnetic coupling goes basically loop-to-loop rather than wire-to-wire like capacitance.)

Twist the TC leads, insulate them and the TC junction from ground, and twist the wires to the relay and the heaters.

My guess is that it's magnetic coupling directly from the heater elements to the TC via a ground loop. (I'm visualizing this as an electric kiln, with coiled resistance wire running round the inside--an ideal geometry for magnetic coupling.)

Taking measurements when the relay is off would be a good method.

Cheers

Phil Hobbs

If your t/c is electrically attached to the sample, that makes a giant antenna and/or big ground loops. If this is a one-off application, why not buy an indicator/controller from Omega?

Type B only makes a few uV per degree C. It will be tricky to condition that with a lot of common-mode noise.

Does it help at all to make measurements synchronous with the 60 Hz? I've used that trick for "fast" stuff... (say a few milli-second signal) in the presence of an external 60 Hz B- field.

George H.

You don't need to be synchronous; just integrate for some multiple of one line cycle. 0.1 seconds handles 50 and 60 Hz.

Most delta/sigma ADCs make it easy to average over one or more line cycles.

Fast line-synchronous sampling converts noise to DC offset!

Or if this is just measuring kiln temp, a simple old fashioned cap somewhere to kill the frequency response. Not across the TC of course.

NT

I sort of gather that the OP isn't doing this under computer control, so precise timing may be a bit difficult.

Cheers

Phil Hobbs