SAADC aquracy vs. VDD_GPIO

[uehans]

We noticed strange behavior when measuring voltages with the SAADC with our custom board.
When VDD_GPIO is at 1.8V (which is our target), the ADC measurement error is about 8%, which is quite high. If we increase VDD_GPIO to about 2.1V or above the ADC measurement error drops below 1%.We did also the same measurements with the DK 0.8.5 and the error there was below 1% with VDD_GPIO even at 1.8V.

We used this adc sample with a few changes:

• GAIN = 1/3
• Calibration bevor each measurement

• Oversampling = 7 (x128 for smother readings)

The DK 0.8.5 and our custom board have different HV revisions of the nrf960

DK 0.8.5:	INFO.VARIANT = 0x41414143	(AAAC)
Custom board:	INFO.VARIANT = 0x41414230	(AAB0)

Measurements with the custom board:

	VDDIO = 1.8V		VDDIO=2.1V
Input [V]	ADC value [V]	Error	ADC value [V]	Error
0.0	0.000		0.000
0.3	0.274	8.7%	0.300	0.0%
0.6	0.548	8.7%	0.600	0.0%
0.9	0.823	8.6%	0.900	0.0%
1.2	1.099	8.4%	1.201	-0.1%
1.5	1.374	8.4%	1.504	-0.3%
1.8	1.648	8.4%	1.798	0.1%

Measurements with the DK 0.8.5:

	VDDIO = 1.8V
Input [V]	ADC value [V]	Error
0.0	0.000
0.3	0.302	-0.7%
0.6	0.598	0.3%
0.9	0.899	0.1%
1.2	1.201	-0.1%
1.5	1.499	0.1%
1.8	1.794	0.3%

Is there an explanation for this strange behavior?

Best Regards,

Ueli

[Håkon Alseth]

Hi,

 

I am able to reproduce this behavior sporadically, but only just after the initial flashing of the device. It seems to be related to the offset calibration task that is called just before entering the loop, as this event seems to trigger an isr in the underlying adc wrapper.

I haven't been able to reproduce this behavior if I reset the device after programming. Could you check if this is the same on your end?

 

Kind regards,

Håkon

[uehans]

Hello Håkon

A reset (power-on and reset pin) after flashing has no effect on the ADC readings. I also removed the ADC calibration -> no effect.

This phenomenon affects not only one device we see this inaccuracy in all of our prototypes.

I have done some measurements with constant ADC input voltage of 1.5V and I modified VDD_GPIO and the ADC gain.

Best regards,

Ueli

[Håkon Alseth]

Hi,

 

Thanks for the detailed description. I checked with various designers and project managers to hear if this has been observed before, but this hasn't been seen.

Have you checked the voltage that goes into the VDD_GPIO pin to see if this is equal to what your power supply shows?

Have you gotten your hardware layout reviewed by Nordic?

 

Kind regards,

Håkon

[uehans]

Hi,

Yes, I checked VDD_GPIO with a multimeter. I also looked at the voltage with the scope to be sure that there was no excessive ripple.

We did our design according the Hardware Integration Guide. However, mistakes still can happen.

We have not done a review by Nordic yet. How would the procedure be for reviewing the layout?

Best regards,

Ueli

[Håkon Alseth]

Hi Ueli,

 

If you create a private case (ie: a case only shared with technical support), and ask for a layout review (upload gerbers / layout files), and give a reference to this case, a colleague of mine which has more experience with hardware will help you along.

 

Kind regards,

Håkon 

[uehans]

Hello Håkon

I will prepare the data and soon open a private case.

Thanks for your support!!

Best regards,

Ueli