ADC Selection

[panfilero]

I'm trying to find an ADC for measuring the voltage across a 1mOhm Shunt, and a voltage divider (looking to measure current and voltage for power calculations)

I need 2 inputs, with the samples from these 2 inputs to be sampled within 100usec from each other
I would like to see full scale 150mV (150A) across the shunt, and be able to resolve 5mA or better.
1000 samples/sec
 
I started to look at 24bit Sigma-Deltas, with multiple inputs, but I'm not sure if they can sample between inputs fast enough (< 100usec)
I think 18bit SAR may be the way to go.

I was wondering if anyone had any advice on how to go about finding the right ADC for this type of design, as far as what to look out for/avoid, or maybe what to pay attention to for the front end design, I think with some of the 24bit Sigma Deltas, you could put the shunt right across the input, and not need a front end amplifier stage....  but I'm not sure.

much thanks!

[Rudane]

You can probably relax the 24 bit requirement if you use an instrumentation amplifier on the 1 mOhm shunt. Look into it.

[David Hess]

If I could not find an appropriate simultaneous sampling 2 channel converter, I would use two of them in parallel although something like an LTC2442 should be able to meet both your 1000 samples/second and your maximum 100 microsecond skew requirement.


I would try to stay with an integrating converter like a delta-sigma rather than a sampling converter because of noise considerations.

[panfilero]

Instrumentation amplifier you say? For some gain, buffering and perhaps filtering? I'll look into it.

I was actually looking at the LTC2442, but from the datasheet, I couldn't tell how close the samples from the MUX are to one another.  I'm not sure what parameter in the datasheet tells me that information, I think it's a fine part for 1ksps, but I don't know if the inputs are sampled within 100us of one another

http://www.linear.com/product/LTC2442

thanks!

[David Hess]

Ahhh, it does not matter.  The LTC2442 is not fast enough to get the channel skew below even 200 microseconds minimum.  I thought it would work handily but the details show otherwise.

What I was going to post is that using two converters in parallel is probably easier and definitely cheaper than using the LTC2442 anyway and that solves the simultaneous sampling problem.  TI and/or AD have some that might work with only one converter.

150 amps with a resolution of 5 milliamps is 30 thousand counts or about 15 bits so it should be possible at 1000 samples per second with an instrumentation style delta-sigma converter which I would still use instead of a sampling converter because of noise rejection.  Getting full accuracy at that resolution and speed will be tough though.

I think a pair of TI ADS1251s would be where I would start.

[lewis]

If this is a 50/60Hz application for measuring mains power, you might want to look into analogue front ends for energy metering applications. Microchip, amongst others, makes some AFEs which might suit your application: http://www.microchip.com/pagehandler/en-us/technology/energymetering/products/energy-metering-measurement-afe.html

They generally use programmable gain amplifiers prior to the ADCs to increase dynamic range and accuracy with two built-in 24-bit delta-sigma converters plus internal digital filtering.