Qiachip Smart Switch - BK7231N / CB2S - interior, programming

p.kaczmarek2 wrote:

So it's not even possible to reach the configuration page?

We have never experienced such an issue before. Safe mode is designed to work even with invalid settings.

Can you try reading the debug log output on UART 2 pin (TX pin)?

p.kaczmarek2 wrote:

This is an interesting crash, I have never seen it before. I would love to debug it, but it's hard to do that remotely.

The strange thing is it crashes ever before reading the config.

Have you used the correct file for the flashing? For N version it's QIO.

Can you do the following steps?
1. Do a full 2MB read from 0 offset on your device with hid_download_py and send it to me for analysis

p.kaczmarek2 wrote:

2. Do you have original flash (full 2MB) saved from Tuya? Can you try reflashing device again with original flash 2MB and then flash OBK again?

p.kaczmarek2 wrote:

I don't have time to check the dumps now, but I have found a full BK7231N flash dump from Smart Life Smart WiFi Led Bulb E27 with WBL2_M1 module (which is like WB2L but using BK7231N):

Please try flash it to your device and then reflash with OBK.

This can help if you really have something wrong in your OBK config, but I am not sure if that's the case. Your crash is very strange, it happens so early that I have no idea what could be wrong.

Have you tried configuring PWM pins before the crash?

p.kaczmarek2 wrote:

Any news? Can you redo configuration and tell me what breaks stuff?

I'd like to add some kind of failsafe, but I need to know what breaks.

But it's very, very strange, because after reboot pins are not started if more than 5 reboots fails (safe mode)

EDIT: maybe a string buffer overflow error? Have you set any lenghty names? I will double-check if all string sizes are kept at bay

p.kaczmarek2 wrote:

Hello, I will try my best to help you, but you need to provide more information. There can be multiple reasons for your flashing issues.
1. Are you doing a proper reboot while the programmer tool is waiting to get bus? You can do this by:
a) power off and power on the device (NOTE: too large capacitors will overload usb port so it's better to power from external 5V source or remove temporarily said capacitor from board)
*yes, I use an external power supply*

b) use CEN pin to RESET (NOTE: this is tricky because it works only if you ground CEN for a certain amount of time, you have to do this for a short while, it always takes me several tries to get right)
*I have tried in different ways*
2. Are you connecting to correct RX and TX? Are you connecting RX of USB converter to TX of Beken, and TX of USB to RX of Beken?
*I have checked*
3. Are you sure that RX and TX lines used for programming are not "busy"? By "busy" I mean, they could:
a) be used for TuyaMCU communication. They could have be connected to MCU. In this case you might need to temporary disconnect them, but not always
b) be used for something like a button with external pull up resistor, and it's the same in this case, you might need to disconnect them temporarily (cut the trace)
*I don't understand*

4. Have you tried the second flashing tool? We have bkWriter 1.60 for Windows and the python uartprogram for hid_download_py repository.
*i am using python uartprogram*
Please post some photos of your setup. Show how your device look like, how is connected, how you do a reset.

jony4t wrote:

*I have checked*

jony4t wrote:

p.kaczmarek2 wrote:

@00svd00 if you could provide information which pin exactly breaks then I would add a failsafe mechanism.

Regarding other questions....
- clearConfig is only in latest build, not in earlier ones
- you can switch On and Off by using the _N version of pin

- mqtt tries to connect but it does not hurt, albeit you might try settting mqtt host and names to empty string and tell me if it silences that

jony4t wrote:

second photo shows a CB2S (or WB2S, not sure) module connected to SOIC8 chip, most likely BL0937 or BL0942:

I am not sure now, but it might be on RX or TX pin and might be interfering with flashing.

How good are you soldering skills?

If you're good at soldering, it wouldn't hurt to check programming module separately.

See, I had this device:

And that didnt work for me:


So I did:

p.kaczmarek2 wrote:

@00svd00 if you could provide information which pin exactly breaks then I would add a failsafe mechanism.

p.kaczmarek2 wrote:

am asking because I can't verify myself whether it's just a fluke or my change really helped.

p.kaczmarek2 wrote:

I need information whether the behaviour changes with the 1.12.7 release. That one includes my latest change and on my side I can't reproduce the problem any more (but I am testing on T).

I will try to restart several times more and see what happens.

p.kaczmarek2 wrote:

Really? Then I might need to look myself at that on N device, because for me, on T, this page is perfect 100% times and I tried refreshing about 50 times

p.kaczmarek2 wrote:

Are you suggesting that the cache mechanism is responsible for me getting a better results?
I don't think so, because I also do PHP pages sometimes and I have cache disabled:

I will get N deviice (BK7231N) soon and test it as well

00svd00 wrote:

In safe mod controller will boot normally - bu we can't reset this setting, so it will never boot in normal mode. Only full-dump-reflash helps

p.kaczmarek2 wrote:

It seems you're doing a very extensive testing, thanks. No one seems to experienced those exact issues before. Or maybe something happened during one of last releases.

00svd00 wrote:

In safe mod controller will boot normally - bu we can't reset this setting, so it will never boot in normal mode. Only full-dump-reflash helps

Why only full-dump-reflash helps?

It should be possible to go to settings in AP mode and change back pin settings or startup values of pins.

Have you tried that?

bk_reboot
wdt reboot

V:BK7231N_1.0.1
REG:cpsr     spsr     r13      r14
SVC:000000D3          00401C1C 000033AC
IRQ:000000d2 00000010 00401e0c 0006d33c
FIR:000000d1 00000010 00401ffc 0006d33c
SYS:000000df          0040192c 00000158
ST:00010001
J 0x10000
bk_misc_init_start_type 1 10001
prvHeapInit-start addr:0x411330, size:126160
[Flash]id:0xeb6015
sctrl_sta_ps_init
cset:0 0 0 0
Entering init log...
Init log done!
Commands registered!
bandgap_calm_in_efuse=0x67
[load]bandgap_calm=0x67->0x27,vddig=4->5
[FUNC]rwnxl_init
[bk]tx_txdesc_flush

V:BK7231N_1.0.1
REG:cpsr     spsr     r13      r14
SVC:000000D3          00401C1C 000033AC
IRQ:000000d2 00000010 00401e0c 85fefcfc
FIR:000000d1 00000010 00401ffc 75d7bcae
SYS:000000df          0040192c 00000158
ST:00000000
J 0x10000
bk_misc_init_start_type 0 0
prvHeapInit-start addr:0x411330, size:126160
[Flash]id:0xeb6015
sctrl_sta_ps_init
cset:0 0 0 0
Entering init log...
Init log done!
Commands registered!
bandgap_calm_in_efuse=0x67
[load]bandgap_calm=0x67->0x27,vddig=4->5
[FUNC]rwnxl_init
[bk]tx_txdesc_flush
[FUNC]intc_init
[FUNC]calibration

jony4t wrote:

Yes. it is a BL0937.

I'm not very good at welding. But I have come to learn. thanks. I will try tonight. Let me tell you.

p.kaczmarek2 wrote:

@00svd00 if you could provide information which pin exactly breaks then I would add a failsafe mechanism.

Regarding other questions....
- clearConfig is only in latest build, not in earlier ones
- you can switch On and Off by using the _N version of pin

- mqtt tries to connect but it does not hurt, albeit you might try settting mqtt host and names to empty string and tell me if it silences that

jony4t wrote:

second photo shows a CB2S (or WB2S, not sure) module connected to SOIC8 chip, most likely BL0937 or BL0942:

I am not sure now, but it might be on RX or TX pin and might be interfering with flashing.

How good are you soldering skills?

If you're good at soldering, it wouldn't hurt to check programming module separately.

See, I had this device:

And that didnt work for me:


So I did: