Extended (Periodic) Advertising - data handling

[axel_s]

Hi,

currently I'm testing several options to transfer data in broadcast manner, i.e. in fact extended advertising due to the amount of data to transmit. A few days ago, I found the max. data size I can transfer using periodic advertising. After that I focussed on extended advertising without the periodic option. All my testing examples are based on the periodic_adv or periodic_sync examples coming with ncs.

Today, I found my working object behaving differently from last week. That's why I did a step back, I tested my periodic example from last week. And I found it behaving different too.

I use a sniffer to control the transmitted data. Last week, I found 3 x ADV_EXT_IND, one AUX_ADV_IND, which contained one AD element with the complete local name and one AUX_SYNC_IND, containing the data (data size 249 Byte was ok for the advertiser, on reception side it was only possible to receive 245 byte, which still is ok). Today, I see 3 x ADV_EXT_IND, one AUX_ADV_IND, now containing the data (data size 228 was ok for the advertiser, I didn't test the maximum up to now, but 245 didn't work), and one AUX_CHAIN_IND containing an AD element with the complete local name.

WHY different types now as well as different assignment frame-data and different max. data size??? How often this will change in the future???

[Kenneth]

I am not sure I understand, the same code and kconfig options will always create the same result, at least for the same ncs tag release.

Kenneth

[axel_s]

Hi Kenneth,

 "same result" is not true. Code, which was running well a day before, did not in case of re-compilation next day. One day later, everything was fine again. The code comes from our git repo, there was no change in code or config options. I know, that sounds strange, but I had that issue twice within the last maybe 10 weeks, and I really don't know what happend.

Usually, programming of the Nordic dev kits runs without problems. But in case of strange behaviour one should think about such "funny" issues too and test a simple example, which is known to run well. In case of strange behaviour, switch the computer off that day and go out in the sun. Next day it will be ok again (hopefully).

[Kenneth]

If you have confirmed that the hex files are identical, and it still behave inconsistent I am thinking uninitalized variable or something floating externally, might be hard to find, but typically using freezing spray or similiar will change the symptoms and thereby confirm that is the problem.

Kenneth