The imprecision SAADC data just after VDD voltage measurement

[Kiyoshi Iwai]

Hello all,

I have a question about the SAADC of nRF52832.

Now I use the SAADC to measure the some external sensor voltage and the internal VDD voltage.

The sensor connection is follows,

- differential mode (sensor =AIN1, and stable constant voltage=AIN0)

- oversampling x256

- GAIN 1/4

- internal reference 0.6V

- use CH[0]

meanwhile, the VDD is

- single-end mode

- oversampling x128

- GAIN 1/6

- internal reference 0.6V

- use CH[0]

I measure the sensor voltage in almost time and sometime measure the VDD.

The VDD source is coin battery (CR2032).

Finished the VDD and switched to the sensor , just 1st ADC data of sensor voltage is strange.

The 1st data is always 20-30LSB (14bit mode) higher than other normal data.

The sensor voltage is not change at this moment, but the data changed.

When switch from VDD to the sensor, I of course run the uninit() and the init() to reset.

But I use the same channel CH[0]  on both VDD and sensor.

Is that wrong?

Should I set the individual channels at each measurement sources?

Please let me know the correct way to use.

Best regards,

Kiyoshi Iwai

[Håkon Alseth]

Dear Kiyoshi Iwai,

 

For easier logic in your application, using two different channels would be preferable, but it should not make a difference, unless there's a timing issue somewhere.

For testing-purposes, it would be beneficial to see if the same issue occurs if they are on different channels.

What is your acquisition time set to? If it is less than 10 us, you might run into this errata: END event firing too early

 

Kind regards,

Håkon

[Kiyoshi Iwai]

Hello Håkon,

Thank you for information.

I changed the channel, but I still have the same issue.

The acquisition time is 3us. But I do not run CALIBRATEOFFSET at this time.

Can't I use the 3us acquisition time?

Best regards,

Kiyoshi Iwai

[Håkon Alseth]

Hello Kiyoshi Iwai,

 

Kiyoshi Iwai said:

Can't I use the 3us acquisition time?

If you do not use the CALIBRATEOFFSET task at this time, then it should not have an impact.

 

1. Could you this workaround and see if the problem still exists?

https://devzone.nordicsemi.com/f/nordic-q-a/39856/saadc-scan-burst-oversmaple/156103#156103

2. Is this issue present if you change the configuration, for instance the GAIN setting from 1/6 to 1/4?

 

Kind regards,

Håkon

[Kiyoshi Iwai]

Hello Håkon,

1.What do you mean the 0x40007FFC=0 or 1 ?

2.I can't set 1/4 for VDD because of 3.0V.

Best regards,

Kiyoshi Iwai

[Håkon Alseth]

Hello Kiyoshi Iwai,

 

Kiyoshi Iwai said:

1.What do you mean the 0x40007FFC=0 or 1 ?

You should write '0' to address 0x40007FFC after un-initializing the SAADC:

nrf_drv_saadc_uninit();

*((volatile uint32_t *)0x40007FFC) = 0;

When re-initing again, you should write '1' to address 0x40007FFC prior to initializing the SAADC:

*((volatile uint32_t *)0x40007FFC) = 1;

nrf_drv_saadc_init(...);

 

Could you see if this has any impact on this issue?

 

Kiyoshi Iwai said:

2.I can't set 1/4 for VDD because of 3.0V.

Ah, I understand. If the signal does not go over 3.0V, you can use "Gain1_5", to see if this also produces the same behavior.

Kind regards,

Håkon

[Kiyoshi Iwai]

Hello Håkon,

Thank you for rapid reply.

It looks some changes on this issue by using the 0x40007FFC.

Could you tell me the meaning of this register 0x40007FFC?

I can not use the uncertain registers because so scary.

I do want to understand it ,so please let me know.

Best regards, 

Kiyoshi Iwai

[Håkon Alseth]

Hello Kiyoshi Iwai,

 

Kiyoshi Iwai said:

It looks some changes on this issue by using the 0x40007FFC.

Could you tell me the meaning of this register 0x40007FFC?

The previous code will reset the peripheral.

Did you see any difference in performance after implementing this?

Kind regards,

Håkon

[Kiyoshi Iwai]

Hello Håkon,

"The peripheral" means the whole SAADC, right?

Do I need to have the wait time from 0x40007FFC=1 to the init(), or from the uninit() to 0x40007FFC=0 ?

By the way,  what is the field name of this bit on 0x40007FFC?

Does 0x40007FFC=1 mean that the SAADC is set? And 0x40007FFC=0 means the SAADC is reset,right?

Please let me know in detail.

Best regards,

Kiyoshi Iwai

[Håkon Alseth]

Hello Kiyoshi Iwai,

 

Kiyoshi Iwai said:

"The peripheral" means the whole SAADC, right?

Yes.

 

Kiyoshi Iwai said:

Do I need to have the wait time from 0x40007FFC=1 to the init(), or from the uninit() to 0x40007FFC=0 ?

No wait time needed.

Kiyoshi Iwai said:

By the way,  what is the field name of this bit on 0x40007FFC?

This specific address offset 0xFFC is similar to the register explanation in this errata, even though this errata is regarding other peripherals:

http://infocenter.nordicsemi.com/topic/com.nordic.infocenter.nrf52832.Rev2.errata/dita/errata/nRF52832/Rev2/latest/anomaly_832_89.html?cp=2_1_1_0_1_26

The text in "workaround" is explaining the functionality for the register offset for the SAADC peripheral as well.

Kiyoshi Iwai said:

Does 0x40007FFC=1 mean that the SAADC is set? And 0x40007FFC=0 means the SAADC is reset,right?

=1 means ON.

=0 means OFF.

 

Did you see any difference in performance by implementing this?

 

Kind regards,

Håkon

[Kiyoshi Iwai]

Hello Håkon,

Thank you for good information.

I understood.

By implementing this 0x40007FFC, I found the better operation, but it seems not enough.

There are sometime still wrong data.

I will add the dummy transaction to omit the wrong data.

Best regards,

Kiyoshi Iwai