Lishui "Open Source Firmware" project / KingMeter 5S
- Thread starter AndZab
- Start date
stancecoke
Minor legend
?! What is that? Please simply post a link to the branch/commit you are working with.
Sure.
I'm using this, it works with standard Bluetooth and with BLE, you can look at the live data and log the data to a file for later analysis.
Serial Bluetooth Terminal - Apps on Google Play
Terminal for serial devices connected with Bluetooth Classic / LE
Last edited:
Virtual EEProm @ page 62+63 · benly65/EBiCS_Firmware@2261029
Here is the latest commit that compiles and runs on my controller.
Here is the latest commit that compiles and runs on my controller.
stancecoke
Minor legend
?! This is very, very old. What is the problem with the latest commit of the new generation branch?! It works flawlessly for me, otherwise I wouldn't have published it without a hint...
One strange setting in your main.h:
Code:
#define R_TEMP_PULLUP 3000have you really connected a NTC and is the pullup 3kOhm really?!
The default value is zero, as most motors have no NTC installed...
EBiCS_Firmware/Inc/main.h at 58d4cded7e319627b46f5d684b39ddfc887c9ea6 · EBiCS/EBiCS_Firmware
Free FOC firmware for Lishui E-Bike-Controllers. Contribute to EBiCS/EBiCS_Firmware development by creating an account on GitHub.
github.com
The values for the torque offset and torque max are strange also, but may work...
Last edited:
I remember that we discussed this and I actually measured the resistance.
I definitely looked at this in debug mode via UART.
I'll now try to use ChatGpt to understand in detail what's going on in the firmware. I'm tired of poking around like a blind kitten. Then I'll measure everything again and check. Then my questions will be more meaningful.
Regarding the old version of the program.
Newer versions either don't compile on my computer or don't run on my controller. I tried all versions. I'll deal with that later.
Regarding the old version of the program.
Newer versions either don't compile on my computer or don't run on my controller. I tried all versions. I'll deal with that later.
set to STM32F103C8 · EBiCS/EBiCS_Firmware@46c13bd
Free FOC firmware for Lishui E-Bike-Controllers. Contribute to EBiCS/EBiCS_Firmware development by creating an account on GitHub.
github.com
stancecoke
Minor legend
Is the linked commit the last one that works, or the first that doesn't work? Which error message is thrown? Do you use the Java Tool for settings and compiling?
Edit; I just tried it with the Java Tool, it compiles without errors:
Last edited:
stancecoke
Minor legend
The next (relevant) commit changes the address of the virtual EEPROM. Have you run the autodetect again after flashing? This is essential, otherwise the hall angles are not available to turn the motor.
Commits · EBiCS/EBiCS_Firmware
Free FOC firmware for Lishui E-Bike-Controllers. Contribute to EBiCS/EBiCS_Firmware development by creating an account on GitHub.
github.com
Please try it with the latest commit at GitHub.
What says the STM32Utility, when connecting to the processor? It has to be "medium density", if it shows "low density" the processor is not the usual one for the "new generation"
Last edited:
Here is the result of working with non-human intelligence:
What exactly to change in the project (in practical terms)
1) Macro devices
Right now, your log shows that compilation is running with:-DSTM32F103xB
For C8, it is more correct to say:
-DSTM32F103x8
And (if available) select the appropriate startup file (not critical, but more accurate):
startup_stm32f103x8.s instead of startup_stm32f103xb.s (if both are present in the project)
2) Linker — you did the right thing.
Your option:FLASH (rx) : ORIGIN = 0x08000000, LENGTH = 0xF800 /* 62KB, до 0x0800F7FF */
EEPROM (rx): ORIGIN = 0x0800F800, LENGTH = 0x800 /* 2KB */
That's exactly "64KB minus 2KB for EEPROM".
3) EEPROM emulation — match addresses with the linker
If the EEPROM is located in the last 2KB (0x0800F800..0x0800FFFF), then eeprom.h should contain:start EEPROM_START_ADDRESS = ADDR_FLASH_PAGE_62
PAGE0 = page 62 (0x0800F800)
PAGE1 = page 63 (0x0800FC00)
It looks consistent to you now.
Why do you now have "both EE_Init and EE_ReadVariable already starting"?
Because you:- stopped hitting EEPROM emulation/code in "dangerous" areas (intersection with the program),
- and/or removed the conflict with incorrect pages,
- and the linker finally "forbade" the layout designer from occupying the last 2KB.
So yes, it seems you really did solve the problem architecturally.
I think the problem is that I'm compiling from a downloaded archive and using 128 instead of 64, whereas yours is compiling from an external source, where everything is written correctly. This is my amateur version.
Last edited:
stancecoke
Minor legend
I'm using 128 and it works flawlessly. But It's known that this pages may be corrupt. There are several reports on that:
So may be you have bad luck with your processor. But if it works for you with the virtual EEPROM on pages 62+63 everything is OK
Last edited:
I'm probably missing something, but why leave a ticking time bomb in the code if it's so easy to fix for someone who understands what's going on? I should note that I encountered this problem early on and asked why later versions wouldn't run on my controller. The answer is obvious but inaccessible. This should be prominently posted on the wiki or fixed everywhere. Or do you sometimes use more than 64 KB of memory during development, and this "kind of memory overclocking" is really convenient, so you don't have to worry about firmware size every time?
stancecoke
Minor legend
I've flashed a lot of controllers and many other users too. You are the first who reported issues with the virtual EEPROM on pages 126 and 127.
If we have this issue more often, we should think about shifting it to 62/63, but I guess you simply had bad luck with your processor.
Increasing risks without absolute necessity based on everyday observations and failing to document this danger is beyond me. Well, screw it, let's move on. I'll keep digging.
stancecoke
Minor legend
Feel free to edit the Wiki, anyone with a GitHub account can contribute there!
Wait, how quick. I'm currently learning how to use GitHub. So I can commit my code properly. When I get the hang of it and finally get it set up properly, I'll try to make a proper guide or a collection of jokes about what and where to break and tinker with.
stancecoke
Minor legend
It's faster to implement it, than to explain how to implement it
Automatic torque offset at startup · EBiCS/EBiCS_Firmware@7c981d0
Free FOC firmware for Lishui E-Bike-Controllers. Contribute to EBiCS/EBiCS_Firmware development by creating an account on GitHub.
github.com
