Temperature fluctuations impacting MCP3428 readings?

Hardware
1

Hello,

I’m using the MCP3428 4-Channel 4-20mA 16-Bit Current Receiver with IoT Interface board to read the 4-20mA flow meter signal measuring natural gas flow to a flare. Since the flow meter is installed near the flare, ambient temperatures can become quite hot at ~135 degF. In the shop, I’m able to very precisely calibrate and read the 4-20mA signal, but on one particular device in the field, I was seeing about +2mA error as the devices temperature increased over 105 degF.

So my questions is - Are there any temperature compensations or temperature error mitigation features baked into this board?

I was able to recreate this issue on new boards in the shop as well.

One interesting thing - While reading all 4 channels under heat, channel 1 and 4 generally read low, while channels 3 and 4 read high. The average of all 4 readings generally is accurate.

2

Hi,
There is on board temp comp on this unit. It’s a pulse analog to digital converter. You can look into MCP3428 datasheet, if there is any temp comp, it will be there.

Thanks

3

Thanks for the quick reply @Anil_Bhaskar !!

Any other thoughts to why temp can be affecting readings? This seems to be a very repeatable problem. Also interestingly, CH1 and CH2, and CH3 and CH4 seem to have in inverse relationship. If CH1 is reading low, CH2 is reading high.

Here’s a screenshot with an 8mA signal at ~ 140 degF.

image
image596×820 32.8 KB

At lower temps, the error between the CH1 and CH2, or CH3 and CH4 is smaller:

image
image658×810 33.7 KB

4

This ADC can be used as single-ended and as diff ADC. which means ch1 and ch2 two are internally linked, and ch3 and 4 are internally linked. That could cause a strange relationship between 1 and 2.

Change your resolution to 12bit and sample rate to lowest and see how it does.

5

Is there anyway to change the setting to select single-ended or as diff ADC? Switching to 12-bit and reducing the sampling rate seemingly helped a bit with temperature variations, but resulted in an unacceptable drop in precision.

6

you are already using in single ended mdoe.

7

Any further recommendations explaining the strange relationship between CH1 and CH2, and how to get accurate readings from this board over the rated temperature range?