NiceMCU XH-WB3S BK7238 Flashing, Testing, Pinout, Development, Porting

Question

NiceMCU seem to make a WB3S dev board, available on Ali and they're dirt cheap. Sparkle XH-WB3S marking on the module. Have ordered a couple. the expectation is that BK7238 is wrong and they are BK7231T

divadiow · Accepted Answer

the complete offline package as linked to on CSDN page lots of BK7231 mentions

insmod · Accepted Answer

>>21325471 Separate toolchain isn't really needed. It's just arm-none-eabi (at least for bk7238). edit: standard tuya ota keys apparently are enough. I even tried to change them, but bootloader somehow is able to chew through them as well. Though my manually generated rbl file fails to update, even though i specified app partition. Code: text Expand Select all Copy to clipboard[E/OTA] (rt_ota_upgrade:957) The partition (beken_onchip_crc) has no space (unless 1183840) for upgrade. [E/OTA] (ota_main:144) OTA upgrade failed!

insmod · Accepted Answer

>>21327466 But it is cheaper to get a devboard. I got a board for ~1.5$ and 2 modules for ~0.7$ each. And working with a devboard is much easier, unlike tr6260 module i got. I don't even need to reset it before flashing, ch340 on board does it itself. Btw, ota is working, i just otaized the wrong image (uart image instead of original one). Added after 6 [hours] 28 [minutes]: Tried porting openbeken - stuck at flashvars first boot: Code: text Expand Select all Copy to clipboardBK7238_1.0.14 REG:cpsr        spsr        r13         r14 SVC:0x000000D3              0x00401AD4  0x000011D4 IRQ:0x000000D2  0x00000010  0x00401E0C  0x000032EC   FIR:0x000000D1  0x00000010  0x00401FFC  0x11E03ACD   SYS:0x000000DF              0x0040192C  0x00000158   ST:0x00000000  Fal(V0.4.0)success  RT-Thread OTA package(V0.2.8-beken-1133282d-20220604) initialize success. go os_addr(0x10000)..........    %  ѵstart addr:0x415418, size:175080 [Flash]id:0x852015 --write status reg:4004,2-- [Flash]init over sctrl_sta_ps_init SDK Rev: 3.0.70 c825378 [THD]app:[tcb]417718  415710-417710:8192:5 [THD]extended_app:[tcb]417f88  417780-417f80:2048:4 [THD]idle:[tcb]4183f8  417ff0-4183f0:1024:0 [THD]timer_thd:[tcb]419180  418578-419178:3072:2 OSK Rev: F-3.0.44 c825378 cset:0 0 0 0 Entering initLog()... [THD]log_serial:[tcb]419a48  419240-419a40:2048:2 Debug:CMD:Adding command loglevel Debug:CMD:Adding command logfeature Debug:CMD:Adding command logtype Debug:CMD:Adding command logdelay Debug:CMD:Adding command logport Commands registered! initLog() done! Info:MAIN:Main_Init_Before_Delay Debug:CFG:flash vars not initialised - reading --write status reg:0,2-- --write status reg:7c,2-- Error:CFG:flash_vars_valid - not our magic, erase Debug:CFG:flash vars erase at offset 0 len 8192 --write status reg:0,2-- Debug:CFG:flash vars erase block at addr 0x1E3000 Debug:CFG:flash vars erase block at addr 0x1E4000 --write status reg:7c,2-- Debug:CFG:flash erased --write status reg:0,2-- --write status reg:7c,2-- Debug:CFG:header written Debug:CFG:header written 4 Debug:CFG:flash vars not initialised - reading --write status reg:0,2-- --write status reg:7c,2-- --write status reg:0,2-- --write status reg:7c,2-- Info:CFG:new flash vars --write status reg:0,2-- --write status reg:7c,2-- Debug:CFG:new offset 84, boot_count 0, success count 0 Debug:CFG:flash initialised --write status reg:0,2-- next Code: text Expand Select all Copy to clipboardBK7238_1.0.14 REG:cpsr        spsr        r13         r14 SVC:0x000000D3              0x00401C1C  0x000033A0 IRQ:0x000000D2  0x00000010  0x00401E0C  0xFAB087BD   FIR:0x000000D1  0x00000010  0x00401FFC  0x91E0AACF   SYS:0x000000DF              0x0040192C  0x00000158   ST:0x00000000  Fal(V0.4.0)success  RT-Thread OTA package(V0.2.8-beken-1133282d-20220604) initialize success. go os_addr(0x10000)..........    %  ѵstart addr:0x415418, size:175080 [Flash]id:0x852015 [Flash]init over sctrl_sta_ps_init SDK Rev: 3.0.70 c825378 [THD]app:[tcb]417718  415710-417710:8192:5 [THD]extended_app:[tcb]417f88  417780-417f80:2048:4 [THD]idle:[tcb]4183f8  417ff0-4183f0:1024:0 [THD]timer_thd:[tcb]419180  418578-419178:3072:2 OSK Rev: F-3.0.44 c825378 cset:0 0 0 0 Entering initLog()... [THD]log_serial:[tcb]419a48  419240-419a40:2048:2 Debug:CMD:Adding command loglevel Debug:CMD:Adding command logfeature Debug:CMD:Adding command logtype Debug:CMD:Adding command logdelay Debug:CMD:Adding command logport Commands registered! initLog() done! Info:MAIN:Main_Init_Before_Delay Debug:CFG:flash vars not initialised - reading --write status reg:0,2-- --write status reg:7c,2-- --write status reg:0,2-- After ~15 seconds it reboots stuck at Code: text Expand Select all Copy to clipboardddev_read(flash_hdl, (char*)data, len - 1, start_addr); Added after 1 [hours] 2 [minutes]: Works if flashvars are disabled. AP and STA works, setting hostname is not (default: wlan0). Everything else is using bk7231n functions. Disabled functions: drv_ir, drv_ir2, powersave. Added after 1 [minutes]:Test binary, flash at 0x0

divadiow · Accepted Answer

very nice I did notice this on TX1 on a constant roll Code: TextLog in, to see the code Added after 5 [minutes]:2nd sensor Added after 11 [hours] 38 [minutes]:XH-WB3S datasheet

divadiow · Answer

well, BK7238 was actually correct. boo Flash ID 852015 - 2Mb - Encryption key set is 84e97000 84e97000 84e97000 84e97000

p.kaczmarek2 · Answer

Very interesting, is it flashable with OBK?

divadiow · Answer

not tried yet, but same BL/chipset as though Code: TextLog in, to see the code Added after 5 [minutes]:

divadiow · Answer

maybe I should have started new thread

divadiow · Answer

Code: TextLog in, to see the code SparkleIoT just direct you to their Arduino guides when asking about an SDK Added after 12 [minutes]:can OBK be made with Arduino? I don't know what this would entail.

insmod · Answer

>>21302635 No need to do it with arduino, all the headers and libraries are there. You would need to create the makefile, as well as adjusting beken HALs. Btw, found an interesting function in the source, maybe it is portable to openbeken (for ds18b20/dht)? Code: text Expand Select all Copy to clipboard void  delayMicroseconds(uint32_t us) {     uint64_t m = (uint64_t)rtos_get_time_us();     if(us){         uint64_t e = (m + us);         if(m > e){ //overflow             while((uint64_t)rtos_get_time_us() > e){                __asm ( NOP );             }         }         while((uint64_t)rtos_get_time_us()

p.kaczmarek2 · Answer

We would also need to check the dependency - rtos_get_time_us; Can't seem to find that, so we would need to check their source of rtos_get_time_us as well?

DeDaMrAz · Answer

Or anything else with precise timing requirement that has been giving us trouble for a long time. Good find!

p.kaczmarek2 · Answer

The timings issues are due to the flash cache mechanism so I don't think the presented approach can work better. We would need to get ramfunc (execution of function from RAM) working in order to get precise bit-bat execution for things like WS2812B on any pin. Luckily, WS2812 are now handled through SPI DMA so they work reliably without the need for ramfunc.

insmod · Answer

Got a NiceMCU BK7238 board, my encryption key is different (84f67900 84f67900 84f67900 84f67900). Flash id from GUI flasher is 152085, which is not found in the database (err: Failed to find flash def for device MID!) BKFIL reads successfully, and flash id from there is different (Got FLashMID: 0x154285) Another XH-WB3S has (84fa7200 84fa7200 84fa7200 84fa7200) enc keys, and the same flash id as the previous one. So, each chip have a different encryption key. If openbeken is ported - everyone will have to build it from the source themselves.

divadiow · Answer

yes, I actually bought two and my second has key 84e46c00 84e46c00 84e46c00 84e46c00 I believe @piotrszulc1 has one too. attaching T1 toolchain from elsewhere on the webs. Shamelessly re-posting to Elektroda for people's references. Credit to user Cossid. Includes T1 (BK7238) only attaching T1 because the others are too large.

insmod · Answer

Just tried flashing unencrypted OTA file - bootloader blocked it. ([E/OTA] (ota_main:125) none encryption Not allow!) So, no ota. Also, available bootloaders from there aren't booting (bootloader_bk7238_uart1_v1.0.14.bin and bootloader_bk7238_uart2_v1.0.14.bin). Got it to compile in arduino, had to manually call beken_packager. But the resulting binary works, bootloader version is the same.

p.kaczmarek2 · Answer

Nice, it seems I need to get BK7238 as well. Was any BK7238 already found in a smart device?

insmod · Answer

GPIO and PWM are working, as is OTA.

divadiow · Answer

different SDK with BK7238 support it seems and 2023 datasheet added newer

divadiow · Answer

maybe development with that SDK could lead to BK7231U/S/Q support and BK7251/BK7252

insmod · Answer

Uploaded arduino sdk draft on github. Won't develop it further (as it lack certain things like xTimerPendFunctionCall etc. And while mqtt to lwip is not hard to add, freertos can't be simply recompiles, as it is a customized one). >>21328770 This sdk should also support N, and it contains rtos_get_time_us. delay_us, here we come. Wonder if bk7231 i makefile is bk7231t, so a possibility to migrate to it fully, while adding support to bk7231s/u and bk7238. Btw, why would we need bk7251/7252? I have only seen them in cameras.

divadiow · Answer

Yeh, no reason really, I was getting carried away. Added after 1 [minutes]: This is a newer version of BDK used by LT/EspHome (I think) which does support N, T, U etc (to varying degrees?)

insmod · Answer

Btw, should matter be developed? This sdk supports it, and beken matter source is publicly available ( The same sdk, but with history is available here Can be cloned without registration. Added after 2 [hours] 11 [minutes]: Several questions before porting. Several functions differ from tuya sdks, so they needs to be addressed. Assuming the sdk is more or less the same, as arduino bk7238, they should share the problems. Looking at the new sdk, several issues can already be solved - powersave, logging. Deep sleep is something different - it will have 2 modes - normal mode (600uA), with no need to reconnect to wifi, and idle mode (20uA), but wifi will need to reconnect and i am assuming it is how currently coded in obk. Allow selection via parameter in cmd? But then on obk boot it will try to reconnect in normal mode. Partitions differ. I assume they will have to be changed to the current ones? Flash vars will probably freeze too. SPI DMA drv_ir/drv_ir2 (as far as i remember they failed to compile due to something in i2c headers). And i haven't tested LFS on bk7238, so unsure if it works ok.

DeDaMrAz · Answer

@insmod drv_ir2 was a development test for N platform and it can be ignored pretty much (@pkaczmarek2 may share more on that). Late for this party but I just ordered 2 boards for 7238 for some testing.

p.kaczmarek2 · Answer

Very good progress @insmod , I will try to help soon once my board arrives: Sure, we can have that as a build option in obk_config.h We can add command argument for that, but from my experience, people don't want 600uA power save, they already say that devices drains battery fast. Depends on the advantages and disadvantages of each option, for BK7231T vs BK7231N we have two partition tables that are slightly different. Why do they freeze? I remember that we might have had similiar issue and we added a delayed init and it increased (?) the stability. Not top priority but good to have. The existing SPI code could also be improved... Are you sure it's I2C? IR requires only a quick timer in both cases. Like 50us one.

insmod · Answer

I encountered a very weird bug when porting beken sdk. Something is wrong with http, and i don't know what the problem is. For example the index page cuts off here And looking through the inspector is this Code: text Expand Select all Copy to clipboardBuilt on Dec  3 2024 18:22:46 version BK7238_Test With no futher data received. When trying to access the device through IP (without /index), chromium reports ERR_EMPTY_RESPONSE. Same is for the web app. At first, i thought it is something wrong with lwip (default is 2.0.2, and 2.1.2 is not connecting to wifi), but after i ported 2.1.3 from bk7231N sdk, the problem persists. Also, if opening some pages it looks almost alright, save Online for. The page is received fully. Code: text Expand Select all Copy to clipboard                         Built on Dec  3 2024 18:22:46 version BK7238_Test             Online for  -                          Device MAC: C8:47:8C:30:0E:ACShort name: obk8C300EAC, Chipset BK7238               Any ideas?

p.kaczmarek2 · Answer

My dev board is ordered but not yet shipped so my options are limited, but I remember having similiar issue in the past. Check the postany source code in the new_http.c , I remember we were trying to buffer TCP packets due to the similiar issue in the past: The following is not enabled on BL602, as it was not needed. Try using BL602 path, maybe? Compare settings of your LWIP and our LWIP: Assuming that you didn't port them... Does the problem still happen if you try to create a test only, simple TCP server and connect to it with Putty? I mean, is problem a generic, TCP-related one, or only occurs in our HTTP server?

insmod · Answer

>>21331248 BL602 method works, now pages are fully displayed and web app is working. As for lwip - there is nothing to compare, i did say i ported it from N sdk, so settings are the same.

NiceMCU XH-WB3S BK7238 Flashing, Testing, Pinout, Development, Porting

Post #1

Post #2

Post #3

Post #4

Post #5

Post #6

Post #7

Post #8

Post #9

Post #10

Post #11

Post #12

Post #13

Post #14

Post #15

Post #16

Post #17

Post #18

Post #19

Post #20

Post #21

Post #22

Post #23

Post #24

Post #25

Post #26

Post #27

Post #28

Post #29

Post #30

Topic summary