Power consumption measurements of nRF52833

[adam_atmo]

Hi,

I have used a Joulescope DC energy analyser to measure the current and power consumption of the nRF52833 DK. I took measurements to see how different TXP powers and PHYs (Coded and 1M) change the power consumption. The first attached image shows the power and current when a connection is established but no data is sent.

1. Why are there periodic drops in current/power even though no data is being transmitted? I understand that the nRF52833 will still be drawing current but why is it periodic?

2. Is the peripheral communicating with the central to confirm that they are still connected?

The next image displays a zoomed-in look at the peak in current/power when data is transmitted. In this case, it is a temperature measurement that is transmitted every second.

1. As you zoom in on one peak, there appear to be more larger peaks around. Why are there so many peaks before and after an individual data transmission peak?

Kind Regards,

Adam

[Einar Thorsrud]

Hi,

1. Why are there periodic drops in current/power even though no data is being transmitted? I understand that the nRF52833 will still be drawing current but why is it periodic?

2. Is the peripheral communicating with the central to confirm that they are still connected?

As long as a connection is active, there are empty packets exchanged in order to maintain the connection. So based on your description that is probably what you are seeing here. If this is the case and you do not use slave latency, you should see this every connection interval.

1. As you zoom in on one peak, there appear to be more larger peaks around. Why are there so many peaks before and after an individual data transmission peak?

I am not able to see much from this plot. I suggest you zoom in with regards to time (showing a much smaller time window), so that you can see the "shape" of the current consumption. If this is for a connection event it will have a distinct shape that is similar to what you see in the online power profiler (when you select a connection).

Depending on your instrument it might also be able to show the current consumption peaks caused by the refresh mode (see this post).

[adam_atmo]

Thank you Einar,

I measured the energy consumption over the timespan of one transmission peak and I also measured the energy consumption over a 30 second period. Coded PHY is used and transmit power is 8 dBm.

As there are 30 signal transmissions during the 30 second period (data sent every second), I calculated the approximate total energy consumed from transmitting temperature signals alone as 2.82 mJ (30 x 93.97 uJ). Therefore, the energy consumption from transmitting temperature signals contributes approximately 4.2% of the total energy consumption (67.0 mJ) over the 30 second period.

Why is this such a small percentage? What is consuming the other 95% of power during connection?

Thanks,

Adam

[Einar Thorsrud]

Hi Adam,

I am not sure how you calculated this? Can you elaborate? You can alo refer to the online power profiler to see roughly what you would expect (though you must take into account that it uses 1 or 2 Mbps PHY so increase the data length to compensate).

Generally, the online power profile gives a good overview of what you could epxect, and the graph sows a current consumption pattern yo should recognize from your power measurements. If you do not, then there is probably a measurement problem 

adam_atmo said:

Why is this such a small percentage? What is consuming the other 95% of power during connection?

It is common to see a high idle current consumption before optimizing. Typically, that is because some peripherals are still in use, causing the high frequency clock to always be on, and use other resources. In general, you need to disable things like UART, other peripherals etc. when not in use. If not, they will continue to consume power. The key is to  disable everything you do not need as soon as possible. And in particular, remember to disable UART logging. I Mention this specifically as it is enable din almost all examples, and it will cause a high current consumption.