Accurate 4W measurement using ADC

hondgm · 2015-12-14T13:13:48+00:00

I'm working on embedded devices that measure the resistance of a temperatur e sensor. The resistance being measured is well under 100 ohms, and total measurement error cannot exceed 50 milliohms. The sensor is at the end of a cable which has conductors that can vary in resistance, which is the reas on for the 4 wire measurement strategy. Originally I was using a Microchip 24 bit ADC which works fine, but the acc uracy isn't quite where it needs to be. I figured that any ADC error would be consistent across channels and so because I'm performing 4W measurement , the error would cancel out. Apparently since this ADC has completely ind ependent converters the error is not the same across all inputs. So then I thought that it might be better to use an ADC that is multiplexed and contains only one actual converter. 12 bits may be plenty of resoluti on so it's possible that any micro with a 12-bit differential mux ADC may w ork. High sampling speed isn't all that important (10s/s probably ok). Does anyone have experience with this? Is it reasonable to expect a 24-bit delta-sigma with independent converters to be accurate enough for somethin g like this, or is it better to stick with a multiplexed converter? Any ot her advice? The way I'm doing the 4W measurement is by using a low TCR/hig h tolerance resistor to supply current to the sensor, then measuring the vo ltage across both that resistor, the sensor, and applying math magic in fir mware.

Dec 14, 2015 29 Replies

Rate this thread:

Bill Martin 10 years ago

Monumentally crappy...what page is that on in the spec sheet? :-)

Vote

Tim Wescott 10 years ago

I think it's usually in the absolute ratings section.

www.wescottdesign.com

Vote

bloggs.fredbloggs.fred 10 years ago

That's a lot of chip for a crummy RTD measurement, almost $11 at Mouser even.

Vote

John Larkin 10 years ago

We're paying $6.62 each, and they are very, very good. It's the heart of a multiplexed acquisition board that sells for around four kilobucks.

The fraction of our revenue that's due to parts cost has been declining over the years, partly because parts keep getting cheaper, partly because we have learned to use fewer of them.

Why are you always in such a bad mood? Is something wrong?

Vote

Chris Jones 10 years ago

On 15/12/2015 06:29, Tauno Voipio wrote: >[...]

Interesting. It has lower input current than the LTC2444 which was on the board that I had to modify. I had to add input buffer op-amps because the ADC input current did cause much larger errors than the 0.05 degree error of the RTD itself.

Chris

Vote

Chris Jones 10 years ago

Ok sorry, you knew what I meant. It does not reverse the thermocouple voltage but it reverses the effect on the resistance reading.

Chris

Vote

Chris Jones 10 years ago

Even un-trimmed resistors should have a typical ratio accuracy of about

0.1% or so, at least for decent sized poly resistors. The absolute value might be 20% off but that shouldn't affect the gain in any sensible design.

Chris

Vote

bloggs.fredbloggs.fred 10 years ago

Care to explain what that tidbit of information has to do with less-than-brilliant approach of 16-/24-bit digitization of a ratio of resistance /derived/ voltages??? It might make sense if Vref was derived from another RTD, anything else is a wish.

Vote