ADC signal outlier

[djqtsg]

We're using nRF52 to read two sensors data: one is an ADC chip using SPI; the other is a PPG sensor using I2C. MCU sampling rate is 250Hz, meaning every 4ms, MCU will poll the two sensor chips in sequence and transmit data out through BLE. The two sensors own sampling rates are higher.

The problem we face now is that there are outlier data points in the both data, though not often, once in a few thousand, and quite ramdon. Outlier in the ADC sensor is much more than that of PPG sensor.

Anyone has good experience or expertise on this issue? I wonder if it is normal to have that outlier point.

[Jørgen Holmefjord]

It is normal with some noice in sensor readings, but values should typically not get outlier data. Which model of ADC and sensor do you use? Have you checked with a logic analyzer if the outlier data is present on the bus as well, or if there is a problem with reading the data on the MCU?

[djqtsg]

The ADC chip is MCP3912 from Microchip, and the PPG is MAX30102 from Maxim.

What do you mean by "outlier data is present on the bus"? Do you mean SPI or I2C has some error? I think unlikely because for one SPI or I2C read, I read several data points; not all of them are outlier, only 1 or 2.

How can I check "if there is problem with reading the data on the MCU"?

[Jørgen Holmefjord]

I mean, if the error originates from the sensor or if it is some error on the MCU/code that outputs wrong data. This is hard to check, which is the reason why I want you to check if you can check if the data originates from the sensor.

[djqtsg]

We suspect the error is from the sensor; now looking for some signal simulator to generate known signal to test. Thx for your advice.