Connection drops

Thanks for the prompt response!

I2C has nothing to do with the issue, since so far I even do not issue any I2C command to the module. Also, it has no slave connected to it. The only module pins I use are P0_5 and P0_6 to connect it to external USB-UART converter.

I use Codeless SDK 6.380.10.4 with a free version of Keil v5.27.1.0 and commented out some AT commands in file user_at_command.h (see attached) to bring the codeless_stand_alone image under 32K limit. Following your advise I ran it in debug under Segger J-Link debugger. Right after starting the code I give to it AT+SLEEP=0 command, the module responded with OK and +AWAKE on the next line. Then I use Cypress CySmart tool along with their dongle as a central device to connect to the TINY module. Upon establishing the connection the module responded with +CONNECTED (in TeraTerm) and I can explore its services and attributes in CySmart used as BLE scanner tool. At this point everything works fine as expected - I can see the module's characteristics outlined in the manuals. Then I leave the module idling in the connection mode, that is issue no commands from TeraTerm and also no command over Bluetooth from CySmart. The connection is dropped after ca. 3 min as I described before. The debugger does not show me anyhing suspicious, definitely no NMI, actually it does not show anything, just runs the module code. The TeraTerm reports +READY and CySmart reports connection drop.

If you unaware with CySmart, I did the same experiment by establishing a Bluetooth connection with a pair of 531 TINY modules. One of them runs pre-compiled image (does not matter which one), the other modules runs my compiled image. The same connection drop happens after 3 min. However, if both modules run pre-compiled datapump or standalone images (by pre-compiled I mean the images provided in SmartBond Flash Programmer) tool, then no connection drop is experienced. But once one of the moduled is loaded with pre-compiled standlone set two, the connection drop happens. So, something is wrong with pre-compiled set-two image and with the SDK source code.

I wrote about I2C only because I need that option. So far I even did not try how it works, so the problem is definitely not related to I2C.

Here you go. The attached archive has a photo of my hardware (just DA14531 module and CP2104 USB-UART) and the BLE events Log. The Log shows that the 531 issued connection timeout 3 minutes after connection establishment (the last 3 records). I do not use SLEEP mode.

Yes, after dropping the connection the DA device starts advertizing again. You are right: the Link supervision is expired because DA14531 stopped responding to connection events.

I created a special app by using Silicon Labs Simplicity Studio and their Thunderboard Sense 2 board as Central that just connects to DA14531 and does nothing after that while keeping the connection active. In the attached archive there are two images provided by Simplicity Studio Network Analyzer. On those images the device 90:FD:9F:7B:86:16 is my Central and DA:E4:D9:A6:F6:E9 is DA14531. On image named Connection_parameters you can see that DA14531 requested to set Connection latency 5 and Supervision Timeout 1250ms, which were accepted by the client (see Events Detail window). Then on image Log2 you see that DA14531 works according to the Connection Latency by not responding to 5 consequtive connection intervals in series. But after 138sec in this case it was a longer series of not responded packets which caused Superivision timeout at Central. I also see this reason (Error 0x208) on Tera-Term output from my central device on the 3rd image.

从这些日志,它遵循that after some time (2-3min) DA14531 stops responding to connection events for no reason. So, Dialog - why don't you just fix your software. Such behavior is definitely a bug. You can easily convince yourself in that by compiling and loading your provided source code into two modules or dev boards.

好,对话框,我决定听从你的建议添加ing several break points in user_on_disconnect() function in file user_codeless.c. It turns out that your provided code does not go into this function after connection is dropped. Further investigation showed that the module resets itself after dropping a connection and if one adds a breakpoint in main(), the code stops there. So, I believe you need to create a ticket for your software development group. If you want me to do it, please give me instructions.

I ran it in a debugger again, the screenshot is attached. The code does not stuck anywhere, it just resets the module and the main() function starts over. You will see it on the screenshot which shows where the debugger stopped after dropping the connection. So, the module is ready to establish a new one, and this works if I keep running it. But the new connection is also dropped after ~3 minutes. Again, I do not communicate with the module after establishing a connection - just leave it idling. It looks to me like a Watchdog behavior.

我不使用任何(外部)flash,裸模ule connected to CP2104 USB converter, see the attached image, and am not sure how the device boots internally. I just compiled Codeless software (v6.380.10.4) image for codeless_531_set_two_standalone without any changes and loaded it into my two modules via SWD interface. So, it boots according to the algorithm in your software. And it works the same on each of the modules - drops connection after 3 min of inactivity.

The same connection drop behavior is observed by using SmartConsole app on Android (Samsung Galaxy S8) as client. Any ideas what else to check?

Added later (IMPORTANT):I just noticed that if I issue the ATI command to the module running under debugger, the response isCodeLess DA14531 v_6.380.10.4and this matches the SDK version that I use. However, if I now reboot the module with the debugger disconnected, the response to the ATI command becomesCodeLess DA14531 v_6.380.9.10(??!!!) Maybe this is the reason for the misbehavior?

OK, these days I learned that if one programs the module from within Keil (in degug mode) and externally with Dialog SmartBond Flash programmer, then the results are different. Namely, after flashing my compiled code with the Flash programmer it shows the right firmware image versionDA14531 v_6.380.10.4no matter how you boot it (I mean during the debug session in Keil and also with JLink debugger disconnected). So, the mismatch between the firmware images outlined in theAdded latersection of my post above is resolved. However, the module still drops the connection after 3 minutes. Am I the only one who experiences that?

Dialog team: if the connection is not dropping for you, could you please send me the HEX image forcodeless_531_set_two_standalone.hexthat works fine for you?

Dear Dialog Team! I have provided above already two BLE captures collected by different capturing software and hardware. One of them is collected by Cypress CySmart tool and their dongle CY5677 and the other one by Silicon Labs Network Analyser. Attached is yet another most detailed capture collected by Nordic Semiconductor BLE dongle PCA10031 and their Sniffer 3.0 software that works in tandem with Wireshark (v3.2.5). The attached capture involves two devices (my Galaxy S8 phone running SmartConsole and DA14531 module with MAC 48:23:35:00:07:AA) and it can be opened with Wireshark. In each of these cases I left the connection idling after establishing it. These are my all BLE sniffer tools.

I have replicated the connection drop issue on another another Galaxy S8 phone of my wife, on my older Galaxy S5 phone, and with at least 3 other non-phone BLE devices that I use as a master. DA14531 module just stopps responding on Master connection events and resets itself (I am not sure what comes first, though). Again, I experience this only if I load codeless_531_standalone_set_two image into the module. Connection with other two available pre-compiled images (codeless_531_datadump and codeless_531_standalone) seems working fine, i.e. without connection drops after 3 minutes, unless I compile them myself from the provided source. In the latter case connection drops appear no matter which image I compile.

The pre-compiled images that I used are the ones provided by Dialog SmartBond Flash Programmer. I asked you above to send me the image that you are using for testing. Please do it. Also, please try my attached hex on your hardware.

Thank you. I am looking forward for the issues to be resolved.

Thank you, Dialog Team. This issue is resolved. Waiting for resolving the I2C issues...