Shunt ? (was Re: Ideas for inexpensive True-RMS Metering?)

om

I've seen shunts that were trimmed by holding 'em up to a grinding wheel. The problem is, mechanical stress raises the resistance (until heat or age anneals the metal). Chemical etching leaves a raw-metal surface, which will in time oxidize (changing the resistance). Laser trimming is fine for SMALL resistors, removing serious metal from a 200A shunt is a frightening prospect.

Make the shunt, anneal and passivate the surface, then use some other gain trim downstream.

We do anneal our shunts. Then dip the relevant parts in a little dilute HCl to clean them up for soldering.

John

The site I saw info on showed media where that would only be on the surface molecules that were proximal to the grinding.

Also, one can grind carefully and slowly so as to reduce thermo-mechanical stresses introduced.

They use a granular microstructure that also depletes and minimizes such effects.

One can also stress relieve the part after the grinding and before the next calibration reading is taken. Or finish the trimming with a file.

Also, a punch might be a way to shear off pieces without introducing time released stress modulus. Paper punch.

Fred Bartoli schrieb:

Hello,

look here:

Precision resistors for energy metering, but with 0.1 Milliohm and less.

Do you want 0.1/0.2 Ohm or 0.1/0.2 Milliohm?

Here are also the 100 Milliohm:

Au revoir

I find it amusing that manganin is also used to make pressure sensors... a nice linear piezoresistive response, and you can get a percent or three change in resistance with (large) pressure. I imagine you must have to consider the mounting method carefully to get ppm-level stability, considering the heat generated and connections required.

We glue the planar shunts to anodized aluminum blocks with thermal epoxy. A surface-mount pcb (amps, dacs) gets stuck to the block too, and a Minco or similar heater, and we stabilize the whole mess at 60C. It doesn't appear to me that the TCE differential between the block and the manganin is affecting the overall TC much.

The big problem turns out to be the eddy currents induced into the block, messing up the transient response of the shunt. You can see a tenth or two of a per cent voltage overshoot for a constant-current pulse, and it decays in a most ugly diffusion-type manner, out into milliseconds. We learned a lot of wasn't-obvious-to-us stuff here, some by accident.

You can see awful eddy effects in most heatsunk-type planar resistors, like some of the Vishay and Caddock parts.

John

Could you avoid that by using a ceramic block of some sort (perhaps natural marble) instead of aluminum? The thermal conductivity would be somewhat lower, but might be sufficient?

Yes, the eddy effects would go away but the thermal conductivity would be much worse. Self-heating changes shunt resistance and can cause transient thermocouple effects.

Aluminum nitride is really tempting. It's an insulating ceramic that's almost as good as aluminum thermally. Solid diamond would be even better.

John

Wouldn't an identical reference slab not under pressure, mounted (but isolated) from the test slab be the way to go? The stuff is pretty cheap so duplicating it wouldn't be costly.

What would be the purpose of the reference slab? Manganin doesn't have much in the way of tempco- I guess you could partially cancel out what there is, particularly if you made it from the same batch of alloy.

Best regards, Spehro Pefhany

Yes, made at the same time with the same geometry (having designed chips one starts to think this way ;).

Yeah, I was thinking about thermocouples. Nevermind.