baruch / b3603 Goto Github PK

To make everything work we need the OPT2 AFR0 to be enabled so that PWM will work on the Timer1 Channel1 and Timer2 Channel1. To make installation easier it would be nice if the application took care of that at startup rather than rely on the programmer who may have forgotten it.

At the very least we need to read it and alert if it is turned off.

Find how to make use of the buttons and integrate them into a UI that doesn't necessarily depend on the serial port.

Currently settings are temporal and not saved, they should be saved in EEPROM so that if the unit comes back online it comes with the last common used setting. This is especially important for auto-on semantics.

It is also critical for calibration (Issue #8) or otherwise calibration will be needed every time the unit is turned on which is rather tedious.

Hi,
I received this power supply just a few days ago and it's a great little first lab bench power supply. I quite also like the DPS style power supply but they have several points that I don't really like. (One encoder instead of multiple potentiometer, user interface, button etc)
As a first step for the recommended noobs project "building a power supply from scratch" I would like to focus on the software side (how to limit current, pid loop etc). I think that making a replacement sheild for my nice B3603 cloud be a great exercise.
I see that in the readme you are talking about making a replacement sheild, still an ongoing project ? Seems like not to me on the github insights.
So Yeah, I'm french, only knew how to code with arduino and basic python3. If anyone is interested to help say hello !

Is there any known protocol to control CC and CV? I bought a D3806 which has the same control board as the b3603 and hoped to be able to control it like described here: https://blog.benjames.io/2018/06/29/secret-uart-on-chinese-dcdc-converters/

but everytime I send something over the UART connection nothing comes back

Accurate will suffer considerably if the unit is fed less than 6V and it gets pretty inaccurate once you go below 5V. The device should alert about this, it should allow operation but give a warning to let the user know he is outside bounds.

When I plug in power source, only first display from the right shows numbers. Those I think are random, but it is to investigate.

The board responds correctly to setting voltage, current, etc., but there is almost no physical reaction. The only thing that changes is CC/CV/OUT LED state.
It happens not because power source, laboratory power supply is far better than cheap inverter and it still goes on with full input voltage on output.

I hope that this is the last time I submit an issue :)

When I tried to communicate with the unit through a script I had to put sleeps between characters being written or data was lost. The polling rate of the unit is too slow to handle the required speeds.

We need to either speed up the polling loop or slow down the uart rate to avoid such overruns.

I've programmed a fresh STM8S with provided code, but it looks like there is no response from board when it starts or when I write to it through serial port. CC (yellow) LED is always on.
I've tested my USB-SERIAL adapter and Intel Galileo serial port. Both working.
STM was programmed using stm8flash/STVP and ST-LINK/V2 mini.

To prevent the case that the unit is stuck and keeps on going we need to have a watchdog reboot the unit and reset it in case the MCU is stuck either due to a software or a hardware issue.

Need to calibrate:

• Vin ADC
• Vout ADC & Vout PWM
• Iout ADC & Iout PWM

For the Vin a single data point is needed (no offset), but maybe using two points would help find an offset in the ADC itself

For the Vout and Iout two points are a must since the basic formula has an offset in it for both sense and control.

Hello,
Can you provide some how to about the "byte options" to flash the the *.ihx hex?
I flash the STM8 with ST-LINK with ST Visual Programmer, but in the display only shows random digit from 6 to 8 in B3603 module.
I also try with your old *.ihx hex file with same results.
Please point me in the right path.
Thank you.

When using five 100ohm resistors in parallel I tried to set the voltage to 7V and the current to 0.01A, the voltage was at around 1V and the current at 0.04A (the PWM is not calibrated yet) but it is definitely a constant current situation. The CC led didn't turn on as expected.

The output pin means enabled at 0V, when the unit comes online its default for the output control pin is LOW 0V so the output is enabled until the MCU boots up sufficiently to turn the output off, assuming the default is to be off.

This is an inherent limitation of the hardware and there is very little that can be done about it. The best that can be done is to optimize the software such that this glitch is as small as possible and that the output pin is configured and set as per the definition as soon as possible.

This is needed to be able to drive the screen, we need to know both how the '595 is wired to the MCU and how it is wired to the 7-segment 4-digit display.

The top board can be inserted reversed and it seems to work. The ADC shows up saturated with a raw reading of 1023 and this can be detected and maybe some action taken to avoid damage in this case.

The unit was trying to feed out about 5V, nothing good can happen with that setting...

The CC/CV signal seems to come from the bottom board, how is it then relayed to the leds? How/who lights up the active led?

The approach is a different one: not reprogram the STM, but replace the board by a TTGO module on a adapter PCB, which provides a gorgeous display, Internet connectivity and thousands of features...

https://github.com/rin67630/Drok-Juntek-on-steroids

When the display is updated much faster than the persistence-of-vision rate the digits simply seem blurry especially if the numbers change back and forth. What is needed is to have a schedule to the updates and to not update the data to be displayed except on those boundaries.

The most responsive solution would be to have a timer run after an update to count 1/10th of a second and only change the data after that timer expires. This will allow changing at a maximum rate of 10 times per second but if there was no change for some time than a new change will take effect immediately.

After running make I get this:

CC main.c
CC display.c
CC uart.c
CC eeprom.c
CC outputs.c
CC config.c
CC fixedpoint.c
CC parse.c
CC adc.c
LINK b3603.ihx
Code is too large, it is 8570 bytes
Makefile:12: commands for 'check_size' object failed
make: *** [check_size] error 1