Resistance Chip? 1

[JimMetalsCeramics]

Are there an off-the-shelf ICs which can measure resistance. I need to measure in the 0.1 to to 2 milli-W range.

 

[IRstuff]

> assume you meant milli-ohm and not milli-watt
> you'd need a 4-point Kelvin connection to eliminate the impact of lead and contact resistance, which pretty much precludes most IC's that I can think of.
> you'd need some fairly substantial current, e.g., on the order of tens of milliamps just to get voltage drop into the microvolt range, which is where what your accuracy requirements become important.
> i've seen lots of micro-ohmmeters, but nothing yet in an IC format.

TTFN

 

[JimMetalsCeramics]

Yes. I mean milli-ohm. The omega sign came out as a "W".

I am content with the device reading a starting resistance and merely giving values relative to that value afterwards.

 

[IRstuff]

As I mentioned, I don't think there are IC's for this job, so you'll either need to use a standard micro-ohmeter or build one using a stable current source, a gain stage and an analog-digital converter.

TTFN

 

[JimMetalsCeramics]

The circuit must be small (e. g., <2&quot; x 2&quot and cheap (e. g., <$2 in volumes of 5 million).

 

[IRstuff]

What resolution in resistance change?
How much power available?
What kind of output allowable?

TTFN

 

[Heydave]

Such low values as you seem to want are difficult because of low voltages that are plagued with thermal voltages from dissimilar metals and external noise. The approach I have used several times involve applying an AC signal through a known resistance and then measuring the in phase signal after extensive bandpass filtering. The AC eliminates the offsets and thermal voltages involved in DC measurements allowing much higher gain and lower stimulus power. All the filtering, level measurement, and calibration can be done using DSP algorithms. All this can be done in a single chip micro with a few external amps and descretes. Remote sense can improve results by eliminating variations in contact and probe resistance. Variations in probe and circuit temperature mostly results in DC offsets that are meaningless in this process.
By the way, inductance and capacitance appear +/- 90 degrees in the result so these parameters are measured if they are the dominant parameter.
David Stephens

http://www.inventgineering.com/

 

[JimMetalsCeramics]

Initially, it would be useful to have resolution of 0.1 milli-ohms. I am prepared to power it through either an RF signal and/or a thin film battery. One person said that 1 amp might be necessary.

I would like a digital output from the chip.

 

[IRstuff]

At the price you're talking about, you're stuck with an 8-bit A/D, which means that the LSB will be equivalent to about 8 mV, so to get 0.1 milli-ohms resolvable, you'll need current*gain product of 80 amps, which could be 1 amp with a gain of 80 or 0.1 amp with a gain of 800. The latter may have problems with op amp offset and drift.

That's why 1 amp sounds like the most plausible value, since it results in 100 uV of voltage change, which is well above the noise and drift floor.



TTFN