The input range of the SAADC is not as expected.

[TML_YT]

Hellow

We are in trouble. Please tell me what to do.

The input range of the SAADC is not as expected.

I use:
・nrf52840
・VDD=3.3V by Regulator
・nRF5_SDK_17.1.0
・Sample program Unmodified　\examples\peripheral\saadc

Therefore:
・gain=1/6
・Resolution=10bit
・Reference internal =0.6V
-->The Voltage input range should be 3.6v.

Situation & The problem
Vin=1.0V->0x011e (ok)
Vin=1.5V->0x01ab (ok)
Vin=2.0V->0x0239 (ok)
Vin=2.2V->0x024f (NG)
Vin=2.3V->0x01e1(NG)
Vin=2.4V->0x017e(NG)

If the input voltage exceeds about 2.2V, the output code is not as
expected, and if the input voltage is higher than that, the output code
value becomes smaller.

What are the causes and what can be done about it?

Please help me.

[haakonsh]

That's strange. Are you using our DK, or are you using a custom board? 

What's you regulator configuration? 

[TML_YT]

Sorry it took me so long to get back to you. I accidentally clicked on
'verified answer' and couldn't reply.

I use DK of other manufacturers.

And then Regulator conditions in the DK board, as follows
5V→Regulator(3.3V)→nrf52840

As for the DK, it is made by another company, but the other functions
are normal and there does not seem to be any problem with it.
We are also concerned about the fact that similar questions have been
asked in the past and no clear answers have been given.

Is this an unusual situation?

[haakonsh]

TML_YT said:

We are also concerned about the fact that similar questions have been
asked in the past and no clear answers have been given.

Is this an unusual situation?

No, this is far outside of spec and not something that i've come across before. 


The first thing I would suspect was the root cause is some grounding issues in the PCBA. Can you replicate the issue on our DK? And can you share any info about the PCBA that you're using? 

[TML_YT]

The NORDIC DK does not seem to be readily available.
The module we are currently testing is the RAYTAC MDBT50Q (with nRF52840), which is further mounted on their MDBT50Q-DB evaluation board.

The USB 5V power supply is converted to 3.3V by a regulator on the evaluation board and fed to the MDBT50Q, and the GND seems to be common to all.

I also tried the following:
(a)When the gain is 1/4 and the reference supply is 4/VDD(VDD=3.3V), the input range should be the same as the gain 1/6 and the internal reference 0.6V, but the output code decreases after about 2.1V as well.

(b)Also, if I set the internal reference 0.6V and gain 1/3, the code is 2^10 at input 1.8V, and if I set the internal reference 0.6V and gain 1/2, the code is 2^10 at input 1.2V, which seems to be working properly.

(c)In addition, I tried to use differential mode and connect the neg pin to the same GND as the board, gain 1/6, internal reference 0.6V, but the same result as the first, the code decreases when the input voltage exceeds 2.1V.

In any case, the hardware is the most likely cause of the problem.
The information about the module MDBT50Q (with nRF52840) and the evaluation board MDBT50Q-DB of the company is certainly not enough, so I will ask RATAC about this matter.

[haakonsh]

I don't think there's a problem with the Raytac DK design at least, but it could be a QA issue.  

TML_YT said:

(a)When the gain is 1/4 and the reference supply is 4/VDD(VDD=3.3V), the input range should be the same as the gain 1/6 and the internal reference 0.6V, but the output code decreases after about 2.1V as well.

I don't understand your argument. The input range is always 0V to VDD+0.3V. You can scale the input range with the gain stage to fit the voltage reference. 

TML_YT said:

(b)Also, if I set the internal reference 0.6V and gain 1/3, the code is 2^10 at input 1.8V, and if I set the internal reference 0.6V and gain 1/2, the code is 2^10 at input 1.2V, which seems to be working properly.

That looks like the correct behavior. 

TML_YT said:

(c)In addition, I tried to use differential mode and connect the neg pin to the same GND as the board, gain 1/6, internal reference 0.6V, but the same result as the first, the code decreases when the input voltage exceeds 2.1V.

The SAADC is always operating in differential mode, when single-ended mode is selected the negative input is tied to GND internally, so you should get the exact same result when manually connecting the negative input to GND. 

I have a few questions: 

1. Do you use the high-voltage regulator (REG0), and if so, what is it's output voltage? 
   
   
2. Can you share your code, at least what's relevant to the SAADC configuration, sampling, and how you read the data?
   
   
3. Can you share some raw data from the SAADC? I need some 10-20 samples from when the SAADC is operating as expected and the same for when it is not operating as expected, and also the input voltage.