fix: Fixing timers on GD32

[ZeyHex]

Fixes the correct logic of the Klipper timer on GD32 series chips

[henrivdr]

@ZeyHex
I guess the CR10 SE and Ender KE would benefit from the timer fix, right?

They all have the same MCU:
https://github.com/CrealityOfficial/CR-10SE_Klipper/blob/b633196f0a9dfb010dd3fc2bd20dbeb37039b93a/config/F003/printer.cfg#L7
https://github.com/CrealityOfficial/Ender-3_V3_KE_Klipper/blob/a63fb1a71672b91b505e0bda68fae2ede49f1168/config/F005/printer.cfg#L5

[ZeyHex]

@ZeyHex I guess the CR10 SE and Ender KE would benefit from the timer fix, right?

If you open the logs (klippy.log) and see something like this, then this PR is definitely for you:

[omahena]

I would certainly like to see this fix merged and tested.

@ZeyHex This timer implementation brings the probe accuracy to about 0.002 and would dramatically improve if not fix all the first layer issues. I can't imagine anything more important than resolving this issue.

[henrivdr]

@ZeyHex I guess the CR10 SE and Ender KE would benefit from the timer fix, right?

If you open the logs (klippy.log) and see something like this, then this PR is definitely for you:

@ZeyHex would these warnings usually occur during normal printing, or only during probing only?
I saw a handful on the CR10 during normal use. Have not checked during probing yet.

[ZeyHex]

@henrivdr they always occur, even just during idle time, if your timers on the MCU are broken.

[henrivdr]

I would certainly like to see this fix merged and tested.

@ZeyHex This timer implementation brings the probe accuracy to about 0.002 and would dramatically improve if not fix all the first layer issues. I can't imagine anything more important than resolving this issue.

@omahena is this improvement seen with the stock prtouchv2? Or does it only work with the full bulk-ADC rework?

[omahena]

@omahena is this improvement seen with the stock prtouchv2? Or does it only work with the full bulk-ADC rework?

From what I understand this works with the stock PRTouch. @ZeyHex can confirm. I saw this issue debated on Klipper discourse group and am eagerly awaiting for it to be adopted to the official firmware.

This improves probing by an order of magnitude and makes the integrated probe fit for the advertised purpose.

[clrx25]

Hi! I'm interested in testing this fix on a Creality K1C as well.
Based on my research, the firmwares (MCU, Nozzle, Bed) for this model should be configured as follows:

Board	MCU Architecture	Processor Model	Bootloader Offset	Communication Interface	Baud Rate	Serial Device	GPIO pins to set at micro-controller startup
MCU (main)	GD32	GD32F303xE	12 KiB	Serial (USART1 PA2/PA3)	230400	/dev/ttyS7	PC7,PB0
Nozzle	GD32	GD32F303xB	12 KiB	Serial (USART1 PA2/PA3)	230400	/dev/ttyS1	!PB5,!PB6,!PB7
Bed	GD32	GD32E230x8	12 KiB	Serial (USART0 PA9/PA10)	230400	/dev/ttyS9

Could anyone who has already implemented this fix please verify or share the .config_mcu, .config_noz, and .config_bed files?
Alternatively, if the controllers and parameters are the same across the K1 series, and someone already has compiled versions of the three firmwares, that would work just fine as well.

@ZeyHex Would implementing only src/gd32/Makefile and src/gd32/gd32e23x_timer.c be sufficient, or are additional modifications required?

Thank you very much for the work that has been done!
Even though this is a relatively "old" topic, I think it is still very relevant to implement, since the issue of inconsistent Z probe measurements is well-known and frequently reported by many users.

In my search for a solution to this problem, I have already tried many ways to fine-tune the parameters in prtouch_v2 with limited results.
It is clear that the problem needs to be addressed elsewhere, and I believe this fix might be the final solution (reference).

In the meantime, I have a ticket open with Creality regarding this issue, and I am waiting for a response from their technical department.