ADS1278-HT: Offset/Gain Warmup Drift Response

Thanks for looking into this Keith.

The customer will try to run the ADC in low speed mode. If this helps they might do a redesign.

Unfortunately they have several design that run in the low power mode already in field and the mode is “HW-jumperd” means without an option to change this over a Firmware fix in the µC interfacing the ADC.

Below are the follow-up inquiries from the customer:

1) The customer wants to know how this effect scales up with respect to temperature? Is there a curve over temperature on how the offset and gain drifts during start-up?

2) The customer understand it the way that the normalized drift value at lower temperature in low power mode is approx. 245mW (at 102,4kHz ? ) and 455mW at higher temperatures results in factor of approx. 1,8 power dissipation increment. Does the offset and gain error also scale up with this ratio?

3) In their system, they don’t have an explicit "ADC-calibration" they perform a system calibration including sensors as well as the ADC-Frontend, SMPS, ADC etc. while everything is fully settled. The calibration is done from Room temperature up to 175°C. Each temperature step will be settled for around 30min, after this time all sub-circuits are "settled". Therefore this calibration is not valid during start-up because of non-stationary offsets and gains.

One measurement requirement is to provide a precise measurement quite fast after start-up approx. 3seconds later. This ADC does not have a self-calibration function. Can you provide a calibration table from room temperature up to 175°C for the device? That theycan implement within the µC prior to their temperature calibration? Are these drifts “equal” for all 8 channels?

4) Is there another solution like increasing the clock frequency to higher rates to increase the die “quicker” temperature during short times and then switching back to target clock frequency?

Hi Art,

1)  I do not think we have warm-up drift at elevated temperatures, but it will be higher due to higher power dissipation in the device.  The warm-up drift is due to self heating creating temperature gradients across the die.  

2)  I will need to follow-up on this question.  If the data was taken at room temperature, then the errors will likely be higher at elevated temperature due to higher power dissipation in the device.

3)  Each device will have different warm-up drift over time, due to differences from part to part.  There will be differences between the 8 channels as well, but these should be much less than part to part errors. Each device would need calibrated after a few seconds at the target temperature, similar to the 'fully settled' calibration that is now done to significantly reduce the initial error.

4)  Increasing the power dissipation by increasing clock frequency will only make the temperature gradients larger.  This will likely increase the warm-up time, not reduce it.

The only approach I can think of would be to implement a correction based on the typical data in the datasheet, or, typical data that the customer has measured at elevated temperature.  This may reduce the initial error across several devices enough to meet the overall system accuracy targets.

Please give me a few days to research this and provide an update.  You can expect a response by end of business on Wednesday of this week.

Thanks,
Keith

Thanks for looking into this Keith!

-Art