12 lead ecg system using nRF52832 - huge data loss when phone shake.

[Abhinav Cardea]

Dear experts.. we are trying to build a 12 lead ecg system which sends real time data to mobile phone for plotting using nRF52832. While most of the time it works ok, when we vigorously shake the phone, a huge data loss can be observed. the phone used here is Samsung with BLE 4.2.  I was wondering if that is a normal phenomenon? our sampling rate is 500 samples per second and roughly we are sending 11k byte per  sec.

[awneil]

Abhinav Cardea said:

we vigorously shake the phone

Who is "we" ?

In particular, is it the person who has the ECG attached?

It is a fundamental difficulty with ECGs to keep the electrodes in good skin contact while the subject is moving!

Have you tried with a wired comms link to see if the problem is actually in the comms, or elsewhere ?

Or have you tried just sending some known-good test data to see if it is actually the comms link that's the problem?

[Abhinav Cardea]

The signal getting plotted as seen in the video/pic is the test signal generated by the front end ECG chip and is not ECG signal obtained when put on a real person. In addition, if you see the video, the phone is moving, not the device. I want to understand if sending 11k byte per sec is the problem here or could be something else? Is packet drop acceptable? if yes, then how much?

[Simonr]

Hi

Have you seen this behavior from multiple phones, or just in this one specific Samsung phone? Could you please provide some more details on your test setup? It should not be a problem keeping a throughput of 11kBps per the BLE spec. or on the nRF side of things.

What Samsung phone is this? What Android version is it running? Are you using a specific app for the BLE communication on the phone side?

Are you using an nRF52 DK or custom design? How big is the packet loss you're seeing?

Best regards,

Simon

[Abhinav Cardea]

Hi Simon,

So we have been working to fix the issues and found some ways to optimise our codes and now the packet drops have drastically reduced. However, despite all efforts  we did notice 1-2 packet drops every 2-3 seconds. I was wondering if even that can be avoided to make it lossless transmission. Any thoughts on that?

[Simonr]

Hi

I don't know the specifics of your application, so it's hard to say what you should do to further lower the packet loss of your transmission or if it's even possible. First of all, you can try to set the scan interval and scan window equal to one another to ensure that the central is scanning continuously.

How many packets do you transmit per second, and how many percent (%) of the packets are successfully transmitted now? How is the environment you're testing in, is there a lot of other 2.4GHz interference that might cause some packet loss or not?

Best regards,
Simon