Why am I seeing an "immediate" disconnect?

RE: Why am I seeing an "immediate" disconnect?

Hung Bui — Wed, 16 Dec 2015 12:51:39 GMT

Hi Jim, I have an update. We reproduced the issue here and contacted SciLab and got to know that they are aware of this issue and are working on this. I won't expect a quick fix from our side for this issue.

RE: Why am I seeing an "immediate" disconnect?

jimvert — Thu, 10 Dec 2015 14:41:12 GMT

PCA 10031 V 1.1.0 2015.38

RE: Why am I seeing an "immediate" disconnect?

Hung Bui — Thu, 10 Dec 2015 08:59:19 GMT

Hi Jim,

I'm reporting this to R&D, could you let me know the nRF51 dongle board number and board version ?

Note that what I have done in the test cases already covered what you request with the modification of connectivity_115k2_with_s130_1.0.0.hex. Not much we can do now, until we have a BGM111 in house for further testing. We will try to get one now, and also to contact Silab for further investigation.

RE: Why am I seeing an "immediate" disconnect?

jimvert — Wed, 09 Dec 2015 20:58:27 GMT

I've now added two new Frontline captures -- one for TestA and one for TestB. In general, TestB seems to work better than TestA in terms of connection duration. I had to explicitly kill TestB once; however, there are subsequent connections to the BGM where is disconnects quickly. The BGM address is 00:0b:57:00:d4:26. The nRF51 address is e6:c6:e6:89:b3:fd. In general, these traces are much better than the traces I was able to capture using your sniffer.

RE: Why am I seeing an "immediate" disconnect?

jimvert — Wed, 09 Dec 2015 16:33:28 GMT

The Ubuntu 14.04 box is running an Intel 7265 Wi-Fi/Bluetooth combo module.

RE: Why am I seeing an "immediate" disconnect?

Hung Bui — Wed, 09 Dec 2015 16:31:20 GMT

I'm sorry, I meant the Ubuntu 14.04 Linux box that you tested earlier.

RE: Why am I seeing an "immediate" disconnect?

jimvert — Wed, 09 Dec 2015 15:59:03 GMT

Btw could you check which Bluetooth chip is on the Beagerboard ?

The Beagleboard doesn't have a BT chip. That's why I'm using the nRF51 dongle on it.

RE: Why am I seeing an "immediate" disconnect?

Hung Bui — Wed, 09 Dec 2015 15:56:55 GMT

Hi Jim,

Thanks for the test. It seems that we got an interoperability issue here. (I assume that you have placed the central - sniffer - peripheral close (less than 30cm) to each other )

I will escalate the issue and check with R&D to figure out what could be wrong. It's a irony that SiliconLab's office is just a few blocks from our office but our devices couldn't talk to each other.

Btw could you check which Bluetooth chip is on the Beagerboard ?

RE: Why am I seeing an "immediate" disconnect?

jimvert — Wed, 09 Dec 2015 15:31:53 GMT

I was now able to connect using TestCaseA. I've uploaded the capture. I connected 3 times.

RE: Why am I seeing an "immediate" disconnect?

Hung Bui — Wed, 09 Dec 2015 15:16:40 GMT

Hi Jim,

It's very strange that you didn't capture any activity with TestcaseA. It's almost identical to the TestCaseB (except for one parameter in the connect request). It should be at least captured by the sniffer.

I would like to ask you for a favour to do the test again with TestCaseA.

You may need to reset the dongle if you don't see anything on the sniffer because there is a chance that it has connected to other device and that's why you saw nothing on the sniffer. Please try about 3 -4 times.

RE: Why am I seeing an "immediate" disconnect?

jimvert — Wed, 09 Dec 2015 15:09:41 GMT

Could you clarify that you couldn't capture any activity with TestcaseA ?

Correct.

And with TestCaseB you only see connection happens and then disconnect immediately ?

The connection occurs, and the two devices seem to stay connected for a second or two prior to a disconnect. This is different that before, where the disconnect happened instantly.

RE: Why am I seeing an "immediate" disconnect?

Hung Bui — Wed, 09 Dec 2015 15:03:50 GMT

Hi Jim,

Could you clarify that you couldn't capture any activity with TestcaseA ? And with TestCaseB you only see connection happens and then disconnect immediately ?

That what I found in the trace.

RE: Why am I seeing an "immediate" disconnect?

jimvert — Wed, 09 Dec 2015 14:44:16 GMT

I wasn't able to see any connections with TestCaseA. I was able to see connections with TestCaseB. I've uploaded the capture under the original question.

RE: Why am I seeing an "immediate" disconnect?

Hung Bui — Wed, 09 Dec 2015 12:28:40 GMT

Hi Jimvert,

To generate a modified connectivity_115k2_with_s130_1.0.0.hex for testing I will have to go through R&D as I don't have the source here. And it may take some time. I have prepared some test case that I generated from the SDK and you only have to flash to the nRF51 dongle and capture the sniffer trace when testing. Having the result from the test case will help us understand what to modify for the connectivity_115k2_with_s130_1.0.0.hex.

To test you would need to:

Flash the nRF51 with the S120 softdevice I provided here (using nRFGo Studio).
Flash the nRF51 with one of the test cases hex I provided.
Start the sniffer and let it track the peripheral under test
The nRF51 will try to connect to any BLE peripheral around, please make sure you don't have more than 1 peripheral around. If you have, please reset the central if it connects to the wrong one.
Save the sniffer trace
Repeat step 1, this time flash the other test case hex file on step 2 .

In the test cases, I basically modified the clock accuracy, the second one also enable the comparability mode in addition.

s120_nrf51_2.1.0_softdevice.hex

TestCaseA_S120.hex

TestCaseB_S120.hex

RE: Why am I seeing an "immediate" disconnect?

jimvert — Tue, 08 Dec 2015 15:59:53 GMT

At this point, I'm not ready to commit to the nRF51 enough to install the SDK and embark upon developing code for it. I understand the flexibility I get with such a solution, but I'm simply attempting to see if the nRF51 will suite our purposes in a relatively short timeframe. Using connectivity_115k2_with_s130_1.0.0.hex is the fastest path. I would appreciate a modified version to try. Thanks.

RE: Why am I seeing an "immediate" disconnect?

Hung Bui — Tue, 08 Dec 2015 14:58:29 GMT

Hi Jim,

If that's the case, I would suggest you to go with the test case #2 of testing with S120/S130 first. And then you can test with the ble_connectivity project in the SDK. Then you can use the embedded controller to control the nRF51 board via UART or SPI. You have more flexibility with the serialization solution than with the ble driver.

RE: Why am I seeing an "immediate" disconnect?

jimvert — Tue, 08 Dec 2015 14:03:23 GMT

I tested an nRF51 dongle as a central to an nRF51 dongle as a peripheral -- both running connectivity_115k2_with_s130_1.0.0.hex. This scenario works properly.

You still can make PC application with the Master Emulator instead of the ble driver.

I'm not running the dongle on a PC. It's running on an embedded controller. I'm looking at the viability of the nRF51 as a Bluetooth solution in this environment. We would ultimately put the nRF51 directly on the PCB but would end up running connectivity_115k2_with_s130_1.0.0.hex on that chip.

Can you send me a modified version of connectivity_115k2_with_s130_1.0.0.hex to try out?

RE: Why am I seeing an "immediate" disconnect?

Hung Bui — Tue, 08 Dec 2015 12:43:10 GMT

Hi Jimvert,

I noticed that in the connect request from the nRF51 had the master sleep clock accuracy of 0-20ppm. This is pretty strict and maybe hard for the slave to match. The iPhone only has 31-50ppm.

The connectivity_115k2_with_s130_1.0.0.hex firmware may have the clock accuracy fixed to 20ppm but what you can do for testing is:

Test the dongle (running connectivity_115k2_with_s130_1.0.0.hex) with our nRF51 as peer device instead of the BGM or the linux machine. This is just to verify the dongle still work properly.
Test with S120/S130 softdevice instead of using the connectivity_115k2_with_s130_1.0.0.hex on the nRF51 dongle. This will require you to let the dongle run as a standalone central (not controlled by PC). You can find central examples in our nRF51 SDK. In the source code you can modify to have more oscillator tolerance ( use NRF_CLOCK_LFCLKSRC_RC_250_PPM_250MS_CALIBRATION when init softdevice with SOFTDEVICE_HANDLER_INIT() )
Test with Master Control Panel (Master Emulator firmware) instead of the ble driver on the dongle. You still can make PC application with the Master Emulator instead of the ble driver.

RE: Why am I seeing an "immediate" disconnect?

jimvert — Mon, 07 Dec 2015 19:18:27 GMT

I changed the scenario a bit. I pulled the BGM out of the picture, and replaced it with my Ubuntu 14.04 Linux box. On the Linux box I did the following:

sudo hciconfig hci0 down sudo hciconfig hci0 up sudo hciconfig hci0 leadv

I then attempt to connect to the Linux box with the nRF51. This connection fails in a similar manner as connecting to the BGM. I then attempted a connection to the Linux box using an iPhone. That connection worked.

democapture3 records these sessions. Packets 1-20 are for the nRF51 attempt. Subsequent packets are for the iPhone.

RE: Why am I seeing an "immediate" disconnect?

jimvert — Mon, 07 Dec 2015 17:29:54 GMT

I added a third connection capture. It doesn't show much except that the connection process takes place during the time delay between packets 28 and 29.

RE: Why am I seeing an "immediate" disconnect?

jimvert — Mon, 07 Dec 2015 17:14:40 GMT

I should point out that I don't perform an explicit scan using sd_ble_gap_scan_start() for the BGM prior to attempting a connection. That is, I know the public address of the BGM and attempt a direct connection with that address using sd_ble_gap_connect().

RE: Why am I seeing an "immediate" disconnect?

jimvert — Mon, 07 Dec 2015 17:07:07 GMT

I updated the case with the capture. The nRF51-to-BGM connection takes place in the first 50-packets. This is followed by the iPhone connection. What's interesting is that during the nRF51 connection no traffic other than the advertising requests from the BGM are seen. The nRF51 is performing active scanning. Those don't even show up. Again, though, the application on both sides of the connection do get notifications of the connection.

RE: Why am I seeing an "immediate" disconnect?

Hung Bui — Mon, 07 Dec 2015 15:02:07 GMT

From the trace, I couldn't find any packet from the BGM111 (address 0x3402861dcbb6 ) back to the central captured by the sniffer. Actually I couldn't find any packet from the device with address 0x3402861dcbb6 and also any other advertising packet from any devices. This is strange.

I would suggest you to capture another sniffer trace with the iPhone connect to the device, and then our device trying to connect to the device ( in one trace).

Also please try with our sniffer, it could be more sensitive and can capture the advertising packets.

RE: Why am I seeing an "immediate" disconnect?

jimvert — Fri, 04 Dec 2015 15:47:19 GMT

The peripheral is based on the BGM111 chip running their latest Geko stack. The nRF51 is www.digikey.com/.../5022448. I'm running a custom profile there.

On the peripheral side, I do see the connection come in at the application level, so the central is recognized. I have no security set on the peripheral. The disconnect code I see on the peripheral size is 0x023e, which is "Connection Failed to be Established".

As I mentioned earlier, I see no problem connecting from an iPhone or BlueZ to this peripheral.

RE: Why am I seeing an "immediate" disconnect?

Hung Bui — Fri, 04 Dec 2015 15:10:23 GMT

Hi Jimvert,

Could you give us more information about the peripheral device ? Which chip, BLE stack running on it. Which nRF51 dongle are you using ?

From the trace you provided, what I can find is that the peripheral didn't catch the connect request, or catched but ignored it (whitelist for example) and the central tried to send 6 connect event packet until it gave up. That's why you saw disconnection almost immediately.