Some time ago I bought a LED current and voltage meter (4 bit, 0-100V, 0-10A) on the Chinese website. According to the description, it was to provide values with a frequency of 3 times per second. It turned out differently, i.e. he gave them less than once per second, which was unacceptable for my applications.
The meter is built on the HC32F003 processor. On Chinese websites I found full documentation, SDK and utilities for this processor. I decided to play with him.
Attempting to connect with the SWD connector has failed. This should have been done using the serial port. It worked, but the flash could not be read - it was secure. So one thing remained - delete the flash and write the software from scratch, which I also did. By the way, I modified the circuit to measure 0-30V. Wiring diagram:
What difficulties did I encounter? A dedicated program did not detect my processor on any USB adapter - serial (I tried on PL2303 and CH340). The FlyMcu program detected the PL2303 chip, but I can't remember why I didn't use it to delete the chip (it didn't allow it?). Eventually, I deleted flash on a Linux computer with a Python program. Now I had full access to the CPU. I programmed and debugged it through the already active SWD port using J-Link (how to make a J-Link adapter with Blue Pill can be found in the network).
As a result of my scribble (I am not a programmer) I have a meter that gives values very quickly and does it more accurately than on the original software. The measurement frequency can be modified by changing the byte in the flash.bin file under the offset 0xbee (in the range 1-255).
Included software and flash batch.
The measurement is calibrated for the one I have and it gives the correct values throughout the voltage range (it was not possible to set it in the original software). I have one copy of the meter and I do not know how exactly it will be for others.
It is similar in my program - at 10 A, horizontal lines will appear. You would have to modify the system (voltage amplification from the measuring resistance) or supply it with 5 V (i.e. also modify it) and change the program.
This can be done by replacing the shunt because I always do it in order not to give two in series if the power / charging module has its own shunt. The only thing left to do is change / switch the range. As in the CC power supply, the shunt has a different value than in the AV module, I correct the LM321 / LM358 gain in the input stage by replacing one resistor and calibrating the existing peer. Well, but without modifying the program, the range is always limiting me, the charger gives me a maximum of 12A and above 10A the indications go out. Only one of the dozen tested by me does not have this blanking.
Even those that should show up to 20 A go out? You have a soft attached, which will turn off at 20 A. Anyway, the ADC will saturate earlier, but I think that 12 A will show. The current is given with two decimal places.
I wrote only about those in the 0-10A range, that only one in a dozen or so copies does not go out after exceeding the range. It is very difficult to buy VA with a current up to 20A at this price, usually they are with an external shunt and much more expensive.
Thank you for the batch, I will try it because I have the equipment for this, to be lucky, I need a modification that will increase the range without ADC supersaturation, i.e. replacing the shunt with one with twice less resistance, i.e. practically the second same shunt connected in parallel - in the program it means multiplication of the current indication by two and increase the range to 20A. I only have two VAs on such a processor, the rest are on the STM8F103.