Significant packet loss at lower battery voltage

[Akhil]

We have conducted some form of RF performance test at various battery voltages (3V, 2.5V and 2V). And the results we got shows significant degradation in performance at lower battery voltage. Please review the results we got and let us know what is wrong here.

Test Setup:

A 50mAh coin cell battery was drained at constant 13mA (tx current of NRF) to get the target cell voltage. Our custom nrf51822 was running a program which advertises every 100 ms at 0dB tx gain. Texas Instrument's RFSniffer was used to capture the Broadcast Advertisements. Sniffing was done for 1 minute to capture a theoretical maximum of 600 packets. From this Average RSSI and % Error were calculated. This process was repeated at various distances.

I have attached the results. Please let us know your suggestions about this behaviour.

RF test @ 2V & 2.5V.xlsx,

RF test @ 3V.xlsx

[Øyvind Karlsen]

Hi,

First off I disagree with the method you are using to deliver your test case voltages to the board. The CR2032 coin cell battery will not have the same characteristics when the voltage drops, it is intended for use around 3.3 V, in fact 2 V output from a coin cell is practically breakdown. Instead use a power source with a current limiter feature.

We have never seen results like this internally. The results are also strike me as a bit weird, the RSSI values indicate that you should have no problems picking up packets.

It is likely that something is not working correctly on your board.

• Are you using internal LDO or DCDC for the nRF51822?
• What is the power supply like on your board, are you using any high ripple voltage transformers?
• Have you verified the behavior of your 16 MHz crystal?

You can also submit your layout for review on our closed support portal, MyPage.

Best regards,

Øyvind

[Akhil]

Hi, Øyvind Karlsen. Thank you for answering the question. Here we are trying to monitor the behaviour of the device while running on a battery. So it defeats the purpose if we use an external power supply. Also the battery's data sheet states that the rated battery capacity is the energy produced when it is drained up to 2V. This voltage drop happens due to the build up of the internal resistance as the capacity diminishes. I don't get why we cannot use the battery till 2V to extract the rated the capacity. Regarding your specific questions:

1. We are using the circuit designed for the DCDC. But DCDC was not enabled in firmware during the test.
2. The board power using a CR1616 battery which has a rated capacity of 50 mAH.
3. Could you clarify this point a bit more? Our board works fine at 3V as you can see from the attached 'RF test @ 3V.xlsx'.

Thanks again Akhil.

[Øyvind Karlsen]

Hi,

I have tried reproducing this using our sniffer and found it hard to reproduce. I did not see a difference in packet loss between the different voltages using a power source, and some light tests using batteries. The only way I was able to achieve the same amount of loss was by either using opposite polarisations in each end, covering the antenna with my hand, or walking inbetween the two PCBs. Be ware that obstacles in your environment will contribute a lot to the radio characteristics. I recommend that you submit your layout for review.

If you have a Nordic DevKit or Dongle you can use the sniffer that we supply, we also have a lot more experience debugging with this sniffer. The problem might be related to sniffer differences.

As for question 3 I would be sceptical to call the result good when you have instances where more than 50% of the packets are lost.

Best regards,

Øyvind

[Øyvind Karlsen]

Oh, and in order to use the development kit at 2 V you have to solder SB12 in order to bypass the backwards voltage diode, so that you do not lose voltage level. Then you can use the external voltage pins as before.

[Akhil]

Thank you for your effort, Øyvind. I will do a test with nRF sniffer and post the results here. regards, Akhil.