Ok, can you disconnect the screw terminal input, and try powering the board via the USB socket on the ESP32? Will need a good 2amp supply. Do you still the WIFI problems?

If you take a look at the fuse on the PCB (large black square near 5V input) does it say T300 or T185 ?

Maybe it is just easier to use a already made module for NTC resistors

16ch Temperature Modbus | Temperature Modbus Rtu | Temperature Collector | Ntc Modbus - Inverters & Converters - Aliexpress


ok, this explains the low readings i thought they are only two sensors on the balance board.

i am using one external temp sensor on t1 (showing 35C in my picture, this is a resistor dummy) without this my canbus does prevent charge or discarge.
i dont need temperature sensors on cells becaus i have an other dumb bms that take care of the protection.

i will look at the fuse and try to power the esp from usb


i have T185 fuses on my controllers

today i disconnected and disabled the shunt in the controller for testing all day

then i tested with just a usb powersupply original samsung 5v 2A
same problem it reboots if loosing wifi
i have measured the voltage on the esp it showed 4,75V

after this i connected power from the meanwell ps again an measured at the clamps
it showed 5,1V then measured at the esp it showed 5,03V (all the time no voltage drops)
but the same problem with reboot

i have an original espressif esp32
now i tested one 4.4 and two 4.5 controllers with faked and original esp32
4.4 and 4.9 cell modules with 5k and 10k communication, i have updated my wifi routor firmware…

but they behave all the same
somtimes they dont reboot on loosing wifi but then when it reconnects.

then i created a local hotspot on my phone and connected there to check if my wifi dlink router is faulty.
but no, the same sh… if i disable the hotspot the controller reboots instantly!

i also tested with custom ip and ip given over dhcp

this all happend after the firmwareupdate so it has to be the firmwareupdate, i cant go back because im now using 4.9 aio module.

what else can i test?

Can you get a high resolution photo of the esp32 board. Some of these have incorrectly rated 3.3V converters which only support 1A.

Im expecting that the esp22 is experiencing brown out when trying to find the WiFi point which has just vanished.

Do you have any fancy mesh network routers or anything like that?

i will make the pictures of the esp tomorrow.

no mesh network no repeater, only 2 wifi clients connected. no other data export like mqtt no sdcard no loging

but why it was running stable for about 250days with the older firmware, if there is a hardware problem?

With each new version of the firmware, the underlying ESP32 core libraries are included - these can change the behaviour of the in-built ESP32 hardware (often to fix bugs).

Can you get a copy of the USB serial output of the ESP32 when the WIFI crash happens?


ok here the hardware images and the serial log i have a textfile too

bms wifi crash.txt (1.7 KB)

Which one of the three different ESP32 devkits produces the crash? The crash log looks like something is accessing / using one of the flash pins of the ESP32, but I’d expect that to crash much earlier during startup. @stuart are you doing any operations from the WiFi event callback beyond a call to esp_wifi_connect()?

I’ve seen some people reporting odd issues with the AZ-Delivery ones at times. I’m not 100% certain what brand that first one is but it’s not an Espressif made modules on it. The third one kinda looks like a DevKit rev 4 PCB (it’s made to fit both WROOM-32 and WROVER-32 modules), I don’t think it is from Espressif but I could be wrong.

all of them, same problems. reboot if wifi signal lost

but where can i buy the original?
and why this Firmware_2023-02-21-08-55 had no problems?

In the code, I’m capturing the WIFI_EVENT_STA_DISCONNECTED event, which then stops the web server (httpd_stop), MQTT and MDNS services on the ESP32.

I’m also calling esp_wifi_connect() at this point which is where the WIFI connect quick retry message comes from. There is a retry counter so I timeout after 25 attempts to connect. I re-wrote this code about a year ago due to problems with crashing wifi.

I’m struggling to debug this problem because my ESP32 devices don’t behave the same as @voltmeter but I do have other users reporting very similar problems.

I (211503) diybms: WIFI connect quick retry 1
Guru Meditation Error: Core  0 panic'ed (InstrFetchProhibited). Exception was un      handled.

Perhaps instead of stopping / destroying of mqtt and other parts a test could be done to see if commenting line 1637 could help? I know I don’t restart all services that you are today and not seeing any reboots on loss of connection to AP.

It may be sufficient to call esp_mqtt_client_disconnect and call esp_mqtt_client_reconnect in the connected handler to force recycle connections. The http server and mDNS should both automatically recover upon reconnect.

The previous problem I was trying to solve was in relation to the MQTT library - that was crashing if the wifi connection disappeared, it would still try and connect and after a few attempts trigger a “guru”

i can send you my 3 esp32 those on the pictures
you can keep em for firmware tests in future

i have so much of them now, i can open a shop :upside_down_face:

Don’t worry, I’ll grab one from Amazon and see what I can do to fix this.

1 Like


today i had 3 inverter shutdowns and one was with under voltage warning, so the inverter has not started again. i have not touched my wifi, looks like it reboots random.

thankfully my diy victron ups saved my fridge, of course i was not at home when this happened…

all with Firmware_2023-12-27-12-02

i switched to my old dumb seplos bms till this reboot issue is solved.
now it is unusable for me.

I hear you - I’m still struggling to reproduce the error, I’ve turned my wifi off randomly several times, and the BMS just keeps ticking along - then auto-connects to the wifi when it powers back up.

I’m waiting for Amazon to deliver another ESP32 - hopefully the same hardware revision you have, so I can repeat.

1 Like

@voltmeter can you take a look at this link please - there is a new firmware attached to it to try out.

Please update the GITHUB issue with your findings.

1 Like

does the settings save and restore work on your controller?
i can restore settings from an older fw release but i cant restore the settings saved with the Firmware_2023-12-27-12-02

could you please check this

i have now running the experimental firmware so far no reboot even with forced switch off and on the wifi multiple times, lets see what happens in the next days.

Just checking in to see how long since my last post. I have been running very balanced with my 14s100p 18650 battery on v4.2? Attiny841 boards. Just upgraded to v4.61 controller (very nice upgrade!).

I have just received my new eve LifeP04 280ah batteries (16s) and in process of putting them together.

PLan to run the LifeP04s with VOLTAGE on Sol-Ark 8k at least to start (might go canbus later) with settings to accommodate the 18650 battery in parallel to bus bars. 60amp fuse on 18650 bank, 80amp on lifeP04, think the Sol-Ark is limited to 185amps charging/discharging, don’t plan on going over 100amps.

Waiting on shunt to come in so I can populate more tiles!

STL file for din rail mount (2 required) DIYBMSv4.61 board: