Disconnect Reason = HCI Error Code 0x2A(Updated)

[MoRan]

Hi,
i am using nrf51822 to develop our product. I'll try to describe more details to make you understand more clearly.

STEPS:

1. While testing my project, I had trouble with bluetooth disconnecting too much.

2. I turn on trace switch, and get the log that's attached.

log.txt

There I see DIFFERENT TRANSACTION COLLISION (0X2A) "The Different Transaction Collision error code indicates that an LMP transaction was started that collides with an ongoing transaction."

3. I try to change a couple of connection parameters, and there are the 2 different results:

1. Either the conn update event happens, the problem still exists (after random time)
2. Or the conn update event does not happens (The current conn interval is 25ms as default when connected, update to min=25ms max=1000ms. So conn update event does not come. And the problem does not come out during testing -- at least after one hour of testing.

4. Could it only be happening in my project? I also tested the examples

1. Nordic\nrf51822\Board\nrf6310\s110\ble_app_proximity -- NG (Not good)
2. an ANCS example got from FAE --NG

5. The devices I tested against:

1. Iphone5S(ios8.1.1) --NG(means met the disconnection for reason 0x2A)
2. iPhone 6 (iOS 8.1.1) --NG
3. iPhone 5 (iOS 8.1.1) --NG
4. iPhone 4S (iOS8.1.1) --OK(means never happen disconnect reason 0x2A during test)
5. iPad 4 (iOS7.1.2) --OK

Other information:

1. SDK v6.1.0
2. SoftDevice v7.0.0/v7.1.0
3. Keil 4.71.2.0

I hope I have explained my problem clearly. How does the HCI error happen? Help me to solve it, please?

Thanks!

**Update 2014.11.28 :

sd v7.0.0

1. Board\nrf6310\s110\ble_app_hids_keyboard -- modify conn_params to min_interval=100ms, max_interval=1000 ms, disconnect as hci_error 0x2A
2. Board\nrf6310\s110\ble_app_hids_mouse -- modify conn_params to min_interval=100ms, max_interval=1000 ms, disconnect as hci_error 0x2A

sd v6.0.0
i modified my project to sd v6.0.0. the problem doesnt' exist any more. i have monitored the test for 1.5 hour now, and it is still OK.

Have you received the report related with this problem from anybody else?**

[Ulrich Myhre]

This sounds serious, so I updated your post to be a bit more clear. Have I understood you correctly in that you will have to wait at least an hour for this problem to happen?

[MoRan]

Sorry to mistaken. Iphone4s didn't have this problem, according to my friends that have the device with connected to iphone4s in daily life , they said no disconnection happen except the phones is far away from the device. so i test the example last 1hour , didn't see the disconnection on ipad4, either, but on the iphone5s, it happens few minutes or few seconds, not fixed time. i also pass the problem to somebody to test it.

[Ulrich Myhre]

Okay, cleared that up in the question. It would be good to know the results of someone else testing this on another chip/board, because this is an unusual problem to see.

[MoRan]

the result comes out. the FAE of our supplier tested sd v6.0.0, v7.0.0 & v7.1.0, the sd v6.0.0 works well, the others turn out the same result as my test. He will upload the problem in mypage of their acount. So i will test sd v6.0.0 tomorrow, and update this problem if i make any progress on it ASAP.Thanks!

[run_ar]

Looks like you are getting a LL Procedure collision. This can happen sometimes when iOS sends a LL_PING_REQ to verify that the link/encryption is ok. In rare cases it will send a LL_CHANNEL_MAP_REQ during the LL_PING_REQ procedure. This is a spec violation and will cause the connection to be dropped. Please consider to file a bug report with apple. Read more about this in the Core spec, Volume 6, part B, chapter 5.3 (Not that it gives you any more information...) Anyway you can verify this by capturing an air trace. I assume I already got one from your distributor.