ADC: Incorrect readings are "corrected" by external probing.

Thread Starter

Abbas_BrainAlive

Joined Feb 21, 2018
113
Hello.

I am experiencing a very strange behaviour in an ADC built into one of ST's SOCs. I have verified it several times, all the registers are configured correctly. When I connect the input line to a known-voltage node, the outputs are within acceptable tolerances. However, when I connect the input to an analog sensor, the ADC starts behaving weirdly. And, as soon as I probe the ADC's anti-aliasing cap, either through an oscilloscope or through a simple male-to-male header, the ADC readings get back on track. I have tested the sensor in a standalone setup, the sensor is working fine. I even have several of the same sensor, which I did test with this ADC setup, but the ADCs isn't responding any differently for any of them. I even have multiple of those ADC boards, and even changing them does not help.

I have no clue what might be causing the ADC to give such erroneous values for the sensor input when it is working correctly for fixed voltages, or, why the ADC starts working fine when the input line is probed!

It would be nice if anyone could help me in solving this mystery.
 
Last edited:

Ian Rogers

Joined Dec 12, 2012
1,136
Input resistance is above recommended criteria... For instance.. The input resistance on a pic is normally 10k (ish) either load the pin a bit more or increase your TAD( acquisition time ) value.. Just placing probes is enough to correct so it can't be too far out!
 

Thread Starter

Abbas_BrainAlive

Joined Feb 21, 2018
113
So the probe is on the power supply? You are not probing the analog signal? Scope probe from Ground to Vcc?
Bread board? PCB? More information. Pictures. Schematic of that area.

OOPS!
Sorry!
I meant the anti-aliasing cap. I am probing the analog input line with respect to ground on a PCB.
The schematic is pretty simple. It's just a single analog input going into the ADC through a single-pole RC anti-aliasing filter.
 

Thread Starter

Abbas_BrainAlive

Joined Feb 21, 2018
113
Input resistance is above recommended criteria... For instance.. The input resistance on a pic is normally 10k (ish) either load the pin a bit more or increase your TAD( acquisition time ) value.. Just placing probes is enough to correct so it can't be too far out!

Thanks, Ian Rogers.

I reduced the series resistor from 1K to 100E. Now, the output is more erratic. And, this erratic behaviour is again corrected when the analog line is probed.
 

Ian Rogers

Joined Dec 12, 2012
1,136
So recap... solid DC voltage appears fine... analogue output from sensor is erratic... You'll need to show the circuit on the ADC pin AND their values..
 

Dodgydave

Joined Jun 22, 2012
11,284
Hello.

I am experiencing a very strange behaviour in an ADC built into one of ST's SOCs. I have verified it several times, all the registers are configured correctly. When I connect the input line to a known-voltage node, the outputs are within acceptable tolerances. However, when I connect the input to an analog sensor, the ADC starts behaving weirdly. And, as soon as I probe the ADC's anti-aliasing cap, either through an oscilloscope or through a simple male-to-male header, the ADC readings get back on track. I have tested the sensor in a standalone setup, the sensor is working fine. I even have several of the same sensor, which I did test with this ADC setup, but the ADCs isn't responding any differently for any of them. I even have multiple of those ADC boards, and even changing them does not help.

I have no clue what might be causing the ADC to give such erroneous values for the sensor input when it is working correctly for fixed voltages, or, why the ADC starts working fine when the input line is probed!

It would be nice if anyone could help me in solving this mystery.

I would start by replacing the sensor for fixed resistors or a preset, and see if the readings are stable,, then you can prove the sensor or the microcontroller...
 

Thread Starter

Abbas_BrainAlive

Joined Feb 21, 2018
113
I would start by replacing the sensor for fixed resistors or a preset, and see if the readings are stable,, then you can prove the sensor or the microcontroller...
Thanks for the suggestion, Dodgydave.

I already tried that. The sensor is working fine on other boards I have, but not on the board I want it to work with. The problem seems to be on the ADC's side.
 
Top