How to Fix Random Relay Switching on BL602 IoT Devices with SM-028_V1.3?

Good tutorial. It's a shame that manufacturer tries so hardly to optimize the costs that they don't actually even want to solder those two cheap components.

After all, these components cost from $0.10 with assembly, I do not understand this savings.

pavlenkoblok wrote:

To fix this, the components C6 0.1 mF and R1 10 Kom should be soldered out.

Maybe this should be corrected, as the sentence implies that these components should be removed, not added.

By the way, I didn't even know that you can flash BL602 from the command line:

I've always used BLDevCube.exe, and here's such a surprise. I wonder if blflash-windows-amd64.exe can also read the original batch?

So I'm wondering if soldering these components corresponds to flashing in any way? Does soldering alone solve the relay switching problem in the original "ecosystem"?
I have several switches under Tuya at home, and indeed some happen to change state on their own (especially to "off"). What are these two components responsible for?

speedy9 wrote:

I have several switches under Tuya at home and indeed some happen to change state by themselves (especially to "off").

But with the original firmware?

speedy9 wrote:

I wonder if soldering these components in any way corresponds to flashing?

Firmware can indeed affect the switch state detection to some extent. For instance, it is up to the firmware to determine whether the GPIO used has programmable pull-up resistors enabled, or perhaps pull-down resistors, or perhaps none at all. Also, the debouncing algorithm could be improved, etc. Perhaps this switch on the BL602 needs some kind of fix or change of approach in the firmware to make it work better.

I would have to have such a device for testing, but as bad as it is, even though I flash a relatively large number of devices, a smart switch on the BL602 hasn't hit me yet.

p.kaczmarek2 wrote:

Also, the debouncing algorithm could be improved

I don't think the ac is the problem. I need to take apart a switch that plays tricks on me sometimes. It's annoying insofar as it works as a pseudo-weather switch for the furnace. And it can turn itself off for no reason at all. It is only controlled by the Tuya app. I thought it was a problem with the software and control from the Tuya cloud, but maybe it's just a problem with the electronics?

>>20904710
Look how much the parts are for on jlcpcb. A $0.10 would be 40 cents by now. With all the middlemen along the way it would probably make a buck. Well no two parts cost that much.

By the way, this whole board is quite poor in components, even by Chinese standards. Only a fuse resistor protects the input, I don't see a varistor or anything else, and what's more, I don't see a second electrolytic capacitor and a filter choke either (there's only the one from the inverter).

Here, for comparison, is a schematic of a slightly more expensive product but with a similar construction idea:

Even in Tuya products I almost always see a circuit of C0, L2, C1 (from the above schematic) and on the PCB from the topic there is not even a place for it....

@speedy9 I wouldn't split it like that, probably it would be possible to algorithmically eliminate these incorrect on/off, worse that probably the algorithm was made for another device, then someone changed something in hardware, and now the algorithm works badly

p.kaczmarek2 wrote:

I wouldn't split it like that, it would probably be possible to algorithmically eliminate these incorrect on/off switches, worse that probably the algorithm was made for another device, then someone changed something in hardware, and now the algorithm malfunctions

Well, I don't know. These switches are rare and completely random in my opinion. It is difficult to talk about the influence of the algorithm here. On the other hand, an entry in the device log appears about the change of state.

pavlenkoblok wrote:

To fix this, solder elements C6 0.1 mF and R1 10 kOhm in the place indicated in the photo.

0.1 mF or maybe 0.1 uF ?

>>20904846
by supported commands - can read flash

I couldn't flash it using BLDevCube.exe, it throws an error and I couldn't do anything about it.

>>20904939
Yes, on aliexpress everyone also complains about random switching and installing these elements eliminates these shortcomings.
P.S. What unexpected activity in the topic I created! I even thought that this was not my topic

Added after 10 [minutes]:

>>20905059
This is really a very cheap module, I bought it for $2. I'm just interested in understanding different platforms.

>>20904492
this modification is not necessary by replacing the bl602 module with an esp 02s and installing tasmota.
I tested it and it works great

Just today I came across this thread (after last week's reprogramming of such a switch as in the thread) I took to testing with the swapped firmware on OBK.

Unfortunately, I did not get a response to shorting contacts S1 and S2. I changed the settings in the configuration to do so, but to no avail (I thought there was something corrupted). Unfortunately, I did not check the operation of contacts S1 and S2 beforehand - before reprogramming the circuit.

I found out that something is wrong on the way from the connector with pin S2 and pad R in the SM-028 board. I removed excess tin from the pad marked R, which caused the disconnection of this path from the board with BL602. I found the thread https://www.elektroda.pl/rtvforum/topic3945435.html helpful. Thanks to this, GPIO_3 - pin 4 (R) and, in addition, for testing, also GPIO_2 - pin 3 (W) - both signals began to work correctly - the circuit responded to shorting with GND. I made a bridge between the pads of the R board and the via at S2 (as a precaution I inserted an additional pull up 4k7) and cut the path so that there were no elements on the way that caused the above-mentioned problem - in the circuit from the main PCB. After this operation, the circuit responds correctly to the shorting of S1 and S2.

However, there were repeated random changes of state (transient) which caused frequent clicking and switching of the contactor (even a dozen times in a minute) - inserting a pull up resistor did not improve anything. Only the creation of a bridge between GND as in the first post of this thread solved the problem with the random change of state of the switch.

I did not detect a short circuit, the resistance of the path on this section close to zero - full transition (not counting the leads from the meter). As for the components at the top of the board - resistor and capacitor - in my case they are factory-installed such as in the photos in the thread - but their presence does not improve the stability of the circuit at least at the time I tested, only the bridge I wrote about to GND stabilized the operation of the circuit.

Drawing/view photo.
.

@Adinfo are you using the ToggleChannelOnToggle role here? Maybe it is indeed time to separate it into three versions (versions with pull up, with pull down and without pulls) and let users choose as required. Perhaps one of these versions would do better in the original configuration of this PCB.

But in general it's hard for me to say more, I have such modules on BK7231 alone and on them nothing switches on its own (and I know, because I even have a night light on such and somehow it doesn't turn on by itself, exactly from this topic https://www.elektroda.pl/rtvforum/topic3895572.html )

p.kaczmarek2 wrote:

@Adinfo are you using the ToggleChannelOnToggle role here? Maybe it is indeed time to separate it into three versions...

Yes, I am using ToggleChannelOnToggle - in this case the problem occurred, when I changed for testing, for example, to Btn then the problem with changing the state was not observed. As for the idea with versions of ToggleChannelOnToggle is great, by the way, this topic came up today during a conversation with a friend who has seen the behavior of this switch - he just happens to have experience with ESP and there he is sometimes resorted to pull up from the level of software and internal capabilities of a particular circuit. I will try to do more tests because I have 7 more similar switches to the one in the thread (different PCB versions, but most on BK7231N and one under ZigBee). The plan is to use some staircase solutions with two mechanical switches and the others under single switches wired to S1 and S2.

I also have BK7231N chip and its also chnage the state of S1 so it makes the relay on and off randomly. I removed the on board buttons but the issue is exist. I ll try to add a pull out resistor.

I made this external pulled up with 4.7k with 104 cap for low pass filter. It works 100%. Pcb down side button removed. My casing has no hole for downside button. BButton existence is not relevant to this problem . I think software pull up dosent seems to be worked.

Adinfo wrote:

only the bridge I wrote about to GND stabilised the circuit operation.

.

.
Gentlemen, explain to me what gives this bridge that the circuit does not trigger itself?