ADS1212: Self Calibration Command Results In Two Different Output Modes 150-200 Counts Apart

Chris Harthan said:

Part Number: ADS1212

Here it is with the images...

It looks like, possibly, a register is rolling over in the correction algorithm. The Offset and Gain terms generated during the self calibration are on a continum, yet the A/D output is bistable. I either get X as an output or X+150. Never anything in-between. This is persistent until another self calibration is run. I have a lot of parts doing this.

The configuration registers are set to Hex 42085457.

The input is 1.07 V.

This is one A/D reading for each of 500 self calibration cycles...

These are the corresponding self calibration gain terms...

Here is data from a normal part....

These results are very repeatable.

My system noise is generally around 30-40 counts. This was a four hour test, so there is a bit of thermal drift. I am not concerned with that.

Any insights would be appreciated.

Thanks,

Chris

Hi Bob,

Thanks for the reply.

I have checked a few of the references, and they were stable. I am not loading the internal reference, and have the recommended bypass capacitor.

Here are some statistics for sixteen devices run through 500 self cal cycles...

I can send you all the data in an Excel file, If you like.

A lot of the offsets went to 255. Is that the extent of it's range? If the gain and offset terms are related and the offset goes off scale that could cause problems with the gain term. (I don't know how the chip works, so I'm just spitballing.) Being a perfect power of two always seems suspicious.

I spoke with my software guy. He already has functions for reading and writing these registers, so we can do a marching bit test on the calibration registers tomorrow.

Thanks,

Chris

Edit:

The one thing that confuses me about this is that the self-cal gain terms don't have a big gap in the middle like the A/D readings do. If the registers had a stuck bit, wouldn't there would be a gap in the gain terms?