nrf9160 SPI power consumption

[Tron]

Hi

On my custom board, I have an external flash connected via SPI. . The nrf runs on mfw 1.1.0. 

Without activating the flash, I get an average current consumption of around 45 µA. As soon as i do a spi_transcieve(...) call, the power consumtion jumps up till 1,4 mA, and will not go down again during sleep. This current corresponds exactly to what I see if serial communication is enabled. I don't know if this is relevant or not. 

The flash is powered down, and should in a ready state draw around 25µA.

Yours, 

Tron

[Simon]

Could you upload your application (I can make the ticket private if you would like to)? Then I can perform some tests myself and figure out what may cause this.

Best regards,

Simon

[Tron]

Please make sure you check out the develop branch. 

[Simon]

Tron said:

Is there an external flash on the DK?

 There is no external flash on the DK, only the 1MB flash integrated on the nRF9160 SiP.

Tron said:

How would you recommend me to do current measurements?

 Check out Using a current meter for current measurement on the infocenter. 

Best regards,

Simon

[Tron]

Measuring the described way, I see a current of around 1A

As far as I've uerstood, I need to use an external power as well. Which of the pins on the opposide side of the usb connector is +?

BTW, using usb as power-supply, did i burn the dk?

[Tron]

Hi

I've done voltage measurements on the custom card, and it seems I have left some pins high on the spi interface. These pins supply other spi devices with voltage, and that draws current to those components through those pins. 

Sorry for bothering you, as usual, the error is between chair and screen . Have a nice weekend. 

[Simon]

What version of the nRF91 DK do you have? If you have v0.7.0, v0.8.2, v0.8.3 or v0.8.4, you should follow the approach explained in this ticket.

• Have you remembered to cut SB44 (if you have v0.7.0, v0.8.2, v0.8.3 or v0.8.4)
• Measure the current with an ammeter by connection to pins 1 and 2, where pin 3 is closest to the USB connector (if you have v0.7.0, v0.8.2, v0.8.3 or v0.8.4)

Best regards,

Simon

[Tron]

After painfully debugging for a forthnite, I come to the following conclusion: 

The flash is not the issue, it's the SPI. 

In a test I wrote from the MOSI directly to the MISO, leaving the flash out of the equation, and it seems it's the SPI device that's causing the havoc. 

As long as I leave the SPI pins alone, the current stays low (around 20µA after some power-optimizations), but as soon as I write to the SPI using spi_transcieve(), the power consumption goes through the roof. 

Tron

[Tron]

Found the solution here, the source code set the polarity of the SPI clock to low on active, but it was the opposite. Reverting the polarity did the trick.