CFW for KSGER/Quicko STM32 Soldering Stations

[illiac4]

Hi. So i did some testing and it looks like that with the original display named "Handskit_2.11.bin" display is bright. Then I have also tried "[HW2.1S]_[1.3OLED].hex" and the display was also bright.
After that I have flashed "[HW2.1S]_[0.96OLED].hex" and the display is not bright. It has the same brightness as your firmware.
In the picture below you can be see the brightness difference, but it is harder to see on the picture as it is in the real life.

Note: in your firmware display brightness on the picture is set to 10 - Max.

[DavidAlfa]

Thanks, now I have something to analyze and compare.

Edit: Try the attached file.
It turns out the previously tested SH1106 command was wrong, however I have no idea whether it will break compatibility with other displays.
I made two versions: One SH1106-only, and another executing both SSD1306 and SH1106.
Anyone else running Ksger v2, it would be great if you can try these too, specially if you have replaced the display.

[illiac4]

It is still not as bright the KSGER one but better. But funny is that obviously Handskit is using different parts display because one is brighter then other with the same firmware. I would say that on one Handskit the brightens is the same when one is in wake status and other in sleep.
Do not spend too much time since I doubt that it is worth troubleshooting.

Will test new file. The report above is for the first one you have attached.

Hmm. Unfortunately I do not see any difference. I think it are the displays that are not the same quality. Handskit also uses display that has not the same pin order as KSGER and other displays that are available in aliexpress.

TNX

[DavidAlfa]

Yes, displays vary a lot.
Last ones I got from AliExpress were much dimmer.
There's another thing, these oled displays lose brightness as you turn more pixels on.
This FW turns almost the entire display on when adjusting temperature, also font is much larger.
Compare brightness at the menu, the difference should be less noticeable.

[DavidAlfa]

Final test. This version uses exactly the same commands as the original.
If brightness isn't restored, then it's simply a display limitation caused by larger lit areas.
For example, the display also dims noticeable when the temperature plot goes full white!

[illiac4]

Hi. Have flashed many times and tested and I have a feeling that those last two are a little brighter than the official v1.04 is.
I think it is not placebo but those two in the latest attached zip file I do not see the difference between KSGERv2_SH1106_init.bin and KSGERv2_SH1106_same_order.bin but as I have said I think they are a little brighter than the latest official.

[DavidAlfa]

But did they made a difference between previous tests build?

[illiac4]

Hmmm I am not sure. The difference is really small. I would saj last two in the zip are the brightest but only marginal.

[DavidAlfa]

You'll notice the display flickers a lot more in the original firmware than in this one, that0s because the display refresh rate is slower, while thisone runs it as fast as possible for better view.
The benefits overcome the minimal brightness drop.

So the difference is just the SH1106 charge pump command. I still need people trying it out to ensure it doesn't break compatibility.

[ygi]

Not sure it helps but I replaced the original SSD1306 0.9" with SH1106 1.3" on my ksger v1.5 and didn't notice any loss in brightness. I can also visually confirm brightness increasing when going > 6 in settings menu. It gets less obvious above 7 but it still gets brighter with each step up to 10. To me, it looks like the software works as intended but hardware quality may vary by batch or manufacturer.

[DavidAlfa]

Okay, this would be the final version (This time for real).

I realized there's a command that can't be set as ksger does: VCOMH_Deselect_Level.
The parameter value goes up to 0x40 in SH1106, but only to 0x3C in SSD1306/SSD1309.
So it must stay at a value compatible with all.

Is the brightness reduced compared to last test version? If so, it needs to live like that...
And really, I need people with SSD1309/SSD1306 displays to testing it before I publish it!

[illiac4]

I would say that the previous one were brighter. This FINAL is the same as v1.04.

[bozo]

SSD1309 brightness from 6 to 10 has no change.

[DavidAlfa]

The brightness also depends on a external resistor, so it might reach the limit sooner.
Perhabs I'll add additional display settings to adjust Vcom voltage and few others parameters.
As I said, I can't set that setting like the original ksger fw because some displays are not compatible.

[DavidAlfa]

So, after few hours, I finally added a new screen for advanced display settings.
Had to figure out a proper way of editing the settings, 0-255 is definitely not the way for editing registers!
So I implemented a new hexadecimal editing widget.

I also discovered my display was SH1106-based (Not SSD1306).
The SSD1306 command "Turn charge pump off" would have no effect, but the SH1106 command did.

So now there's a new "ADVANCED" option in the display menu. Sorry, no translation for that!
Each setting has a different function for its higher and lower 4 bits (Each hex digit controls a different setting).
Please don't start making random changes and asking me to explain everything.
Instead, read the SH1106 / SSD1306 / SSD1309 datasheets.

CLK :   Set Display Clock Divide Ratio/Oscillator Frequency.
PRE:    Set Pre-charge Period.
VCOM: Set VCOMH Deselect Level.


I played for a while, I could definitely notice a brightness increase with this settings:
CLK :   0xF0 (Minimal diference, mainly affects the refresh rate and the flickering).
PRE:    0x0F
VCOM: 0xFF

KSGER original:
CLK :   0x80
PRE:    0x1F
VCOM: 0x40

Download: Release v1.05

[ricktendo]

Thanks David, BTW there is no KSGER_v3_SSD1306.zip

[DavidAlfa]

Huh? Did I miss a file when uploading? (Yes)
Uploaded!

[illiac4]

This was really the best option to implement. I like realtime change of settings. I can clearly see how refresh rate impacts the brightness. TNX

I know I am picky but would it be possible to add "default" under three settings.

[it9exm]

Any ideas to replace that with a regular STM32F103C8T6?

No. STC boards are completely different.

This is not STC, it's an XM1008K6T6, clone of STM32F030K6T6.

Maybe it's using external flash IC since 32k is quite small.

Bob Sava was suggesting it may be using an external flash, but I couldn't find any. Also the RAM is just 4KB.
Seems like there's no way to connect external ram to the STM32F0 mcu family.

I was thinking about ditching the original MCU and rewiring everything to a pico sized STM32F1 dev board, might it work despite being a lot eccentric?

[DavidAlfa]

would it be possible to add "default" under three settings.

Sure, didn't think about that.
Will update the release in a moment, let me take a morning coffee (It's holiday).

I was thinking about ditching the original MCU and rewiring everything to a pico sized STM32F1 dev board, might it work despite being a lot eccentric?

You could, following any of the ksger v2 schematics, take a bluepill board and wire everything in the same way.

[DavidAlfa]

Well, I frenzied a lil'bit and made a light interface redesign.
- Now only the editing field is highlighted.
- Sharper widget frame (Not round anymore).
- Added the advanced display reset option.
- Updated the Operation manual to match current features. Huh, it's getting huge! Sometimes I forget it has soo many options!

Lots of changes in the drawing code, I didn't notice any bugs, please report if you find anything strange.

Download Release v1.06.

[DavidAlfa]

Finally, I implemented a rudimentary smart load detection, called it "SmartActive".
Every 200ms it stores the temperature error in a circular buffer, keeping the track of the last 3 seconds, then sums it up and compares the restult with the threshold set in Iron settings.

In ideal conditions, the error average will be ~0, but when there's a small load, the error tends to be positive, still very small (It's compensated by the PID), but detectable after all.

A value of 4-5 seems to work nicely, but every station will perform differently.
Likely, it won't work properly with new tips (Very noisy), but otherwise it's doing great with my "tamed" ones, can even detect the heat loss caused by the air when moving the handle around.

Noisy stations/tips: Try tweaking the sensitivity, if still not working, just don't use it!

It's not a different wake source, but an a extra layer working on top of the timers and wake sources.

After setting up a mode, the functionality will be delayed for 10 seconds to give the tip some time to heat up.
When the smart detection is triggered, the shake icon blinks to provide some feeback.

Obviously, it only works in active modes (Boost or Run), smartActiv will keep the current mode active as long as it detects load.
Button and stand modes will keep working in the same ways as before.

Release v1.07.

[ygi]

Hmm, compiler spits out an error in settings.c line 279 when building for ksger v1.5. I stumbled upon it cause I uncommented #define HAS_BATTERY in board.h (that still built error free for 1.06 just a few days ago). I feel like I've seen this error being fixed before. Could it be when you deleted and restored the repository?

Code: [Select]

../Core/Src/settings.c: In function 'checkSettings':
../Core/Src/settings.c:279:7: warning: implicit declaration of function 'isIronInCalibrationMode'; did you mean 'getIronCalibrationMode'? [-Wimplicit-function-declaration]
  279 |   if(!isIronInCalibrationMode() && scr_index != screen_debug) // don't persist the temperature while calibration is in progress or in the debug screen
      |       ^~~~~~~~~~~~~~~~~~~~~~~
      |       getIronCalibrationMode
arm-none-eabi-gcc -o "STM32SolderingStation.elf" @"objects.list"   -mcpu=cortex-m3 -T"../_STM32F103CBTX_FLASH.ld" --specs=nosys.specs -Wl,-Map="STM32SolderingStation.map" -Wl,--gc-sections -static --specs=nano.specs -mfloat-abi=soft -mthumb -Wl,--start-group -lc -lm -Wl,--end-group
c:\program files\stmicroelectronics\stm32cube\stm32cubeide\plugins\com.st.stm32cube.ide.mcu.externaltools.gnu-tools-for-stm32.10.3-2021.10.win32_1.0.100.202210260954\tools\arm-none-eabi\bin\ld.exe: ./Core/Src/settings.o: in function `checkSettings':
settings.c:(.text.checkSettings+0x10a): undefined reference to `isIronInCalibrationMode'
collect2.exe: error: ld returned 1 exit status
make: *** [makefile:76: STM32SolderingStation.elf] Error 1
"make -j16 all" terminated with exit code 2. Build might be incomplete.

[DavidAlfa]

Probably some file was not updated in the commit, will check later.

Edit:
Ahh it was these options never use... Fixed!

[ygi]

Thanks for the quick fix.

The battery option/backup register thing's actually a nice reliability feature, plus I have to find use for my stockpiled CR2032 cells. Bought too many at once, and now they are discharging in a drawer