CFW for KSGER/Quicko STM32 Soldering Stations

[DavidAlfa]

What's your controller?
What are you doing exactly in the adjustment menu?
If the temperature rises too much, then lower the values quickly, and start rising slowly.
For example, if cal250, with default value of 1400, already sets your tip way hotter than 250°C, what you should do is quickly set the value much lower, ex. 500, and wait for tip to cool down. Then slowly rise the value again.
When you go into 350 and 450 steps, do the same quickly.
If you tell me the calibration results I could set them as default for that controller, avoiding the issue.
The encoder changes the value, does the value change itself like crazy?

[empeka]

What's your controller?

ksger 2.1S, ksger 3.1.

What are you doing exactly in the adjustment menu?

adjusting value, constant ,fast rotations. for oscilloscope test it was half turn left, half turn right repeadedly.

For example, if cal250, with default value of 1400, already sets your tip way hotter than 250°C (...)

It is not the problem of default values being too high. In fact, they are too low, just as intended.
Default 1000ADC @250C gets tip temperature to ~130C, at 2000 for 450C, tip goes up to 240C, so there's no issue here.

The encoder changes the value, does the value change itself like crazy?

no, proper increments/decrements by 50.

If you tell me the calibration results I could set them as default for that controller, avoiding the issue.

2086, 2976, 3867 (ksger 2.1S, green PCB)
But I wouldn't trust it too much. I own 3 different KSGER controllers and none of them matches schematics found on the web, including opamp feedback resistor values.

[DavidAlfa]

Where's the problem then? If the calibration is too low, then adjust it.
I don't understand what's going wrong. How is the temperature going over 500ºC?
Some board pictures might help.

To flip the screen you'll need to recompile.
It's not hard, there are instructions and video, it only takes 5 minutes.

To flip the screen go to Drivers/graphics/ssd1306.c
Change these lines
   write_cmd(0xA0|0x01);      // Set Segment Re-Map
   write_cmd(0xC0|0x08);      // Set COM Output Scan Direction
to
   write_cmd(0xA0|0x00);      // Set Segment Re-Map
   write_cmd(0xC0|0x00);      // Set COM Output Scan Direction

Compile and done.

[empeka]

I don't understand what's going wrong.

Uhh... so, let's say I'm starting adjustment for 250C from default 1000. Real tip temp is ~130C. Everything ok so far.
I know that the value I need is 2086, so I'm turning encoder up to 2050. That's a lot of turning and during that, measured tip temp goes unexpectedly high, 400-500C.
Can't get that any clealer. But if nobody else experiencing this issue just forget it. I don't recalibrate too often.

[DavidAlfa]

Remember to push and turn for big steps.
Do it slowly.  Every step should increase the temperature slightly.
Try to find the point where it starts going crazy.

[wickated]

Hello David. im here to ask, could i swap from i2c bus screen to spi one on fw for 2.x boards? or should i use 3.x instead?

[DavidAlfa]

2.x screens have only 4 pins, can't be used in spi mode.

[wickated]

2.x screens have only 4 pins, can't be used in spi mode.

your answer is not for my question. ok, ill simplify it.

do fw 2.x have spi screeen support?

[DavidAlfa]

All fw are the same thing. Each board have their own adjustments in board.h
The screen interface is set at compile time, so setting i2c, spi etc will disable the unused code.
You need to compile your own. The whole process is highly documented and explained.

[wickated]

thanks! so if i want to make some minor board adjustments i have to recompile fw to fit it.

[DavidAlfa]

Exactly. Learn and do whatever you want.
By no way I'm starting to make customized firmwares for everyome, I have life

[StephenR0]

I thought I'd mention my experience so far.  I wanted to order a soldering station to run DavidAlfa's firmware specifically.  So I ordered this:

https://www.aliexpress.com/item/32994824865.html

even though it says they're using a compatible CKS microcontroller for the time being since they can't get the STM32.  I finally got it and tried out the original firmware.  It didn't seem too bad, but the fonts are pretty ugly.     So then I thought I'd try to back up the original firmware using the exploit documented here.

https://github.com/JohannesObermaier/f103-analysis/tree/master/d1-a/arm-cks32f

After wrestling with OpenOCD some, I managed to make it work.  I haven't tested it, but I did run it through "strings" and found some of the messages the firmware uses.  I don't really expect to go back to that firmware anyway.  So then I flashed DavidAlfa's firmware and ran into an odd problem with the flash being locked.  Again after some wrestling with OpenOCD, I got it unlocked and flashed it without problems.  I don't have much experience with it yet, but I've gone through the menus pretty well and run one of the tips some.  These tips seem to need some break in before they are very stable.  I'm going to do that before I try to calibrate them.  At this point, I don't see a problem with the CKS microcontroller.  Anyway, that's where things stand at the moment.

[DavidAlfa]

I accidentally burned my last working tip while making the Tip calibration values menu (To restore old calibrations), so I bought a 4 tip pack.
The tips are a lottery. I got issues with my last order.
They are a little unstable over 400ºC, anything over that and they'll start jumping +-10C.
It stays under control, it's justa little annoying. Neither a a big issue as I usually stay below that temps.
However one of them is specially bad, it keeps jumping between 400 and 700ºC no matter what.
I doubt it's a firmware thing, my last tips where perfectly stable at 480ºC. And I don't see anyone complaining ?

But yes, they usually need some burning time, normally no more than 15-20 minutes, but it depends.
I've had tips that never worked correctly, and others that were perfect out of the box.
Sometimes you can hear crackling noises as the moisture comes out, and the temps jumping like crazy, and after a while the temps get better.
I usually leave it around 300ºC for a while.

About the CKS: I think some stm32 clones had different flash layouts, so for running code there was no problem, but when writing to the flash within the firmware you could run into problems.
Thanfully that CKS seems ok!

[StephenR0]

@DavidAlfa, I've been burning in my tips at 300ºC as you suggested.  But when I tried to calibrate, it was pretty clear that any temp over about 350ºC would cause the temp to run away and error out with a no tip message.  Then it would recover and do it again.  After looking over some of the previous thread posts, I increased the ADC Delay from 20 ms to 40 ms.  That seemed to settle things down quite a bit and I was able to get a calibration.  So it seems that my power supply is more noisy than usual, I guess.  Do you have any advice about improving this?  Is this something that might improve over time?  Thanks.

[DavidAlfa]

That's a shame. I really like the actual firmware status, everything seems to work so smooth!

Can you attach some pictures of the board? From the listing pictures, it seems to be exactly the same as my Quicko board.
This is a dumb question, as nothing would work if you flashed the wrong FW. Are running the Quicko F103 FW?
Are you talking about real, measured 350ºC? Or the 350ºC calibration step?
Try something: Skip calibration. Attach the temp probe to the tip and slowly start rising the setpoint value, ex rise 20ºC, wait 10 seconds, and so on.
When the errors appear, note down the setpoint value and the real measured temp.

If the board is the same, then maybe the issue comes from the power supply.
My station came with the 108W version:


https://www.aliexpress.com/item/32834572016.html

[StephenR0]

Yes, I'm running the Quicko F103 version of the firmware.  And I get errors just running with too high a temperature or during calibration.

So, I did as you suggested.  I reset everything to default values and started at 300ºC measuring as I went.  Here is the result  The first number is the set point, the second is what I measured.

300    343
320    372
330    390
340    409
350    420
360    433
370    413  (I changed sensors after this.  I didn't trust it.)
390    448
420    464

I'm not finding these fg-100 clone sensors to last very long.  And I'm not sure how much I trust them.  While I was doing this, anything over 400ºC didn't hold the displayed temperature very well.  I jumped the set point up to 450ºC in steps.  As the set point increased, the fluctuations got wilder until it went up to 999ºC then said there was no iron.  I need to order some more sensors to see if I can get something that I have more faith in.

I've attached pictures of my board and power supply.  The power supply doesn't look like the same thing although I believe it was billed as 108W and is the same as the pictures of what I ordered.  Anyway, that's my experience.  Does any of this make sense?

[DavidAlfa]

Since it has a pot to adjust the gain, your board seems came with lower gain, resulting in higher real temps  than requested.
I think I'm goign to set the default calibration values much lower by default, to avoid more issues like these.
It's better to get 100ºC at Cal250 step, than 300.
Please try the attached FW. After flashing, make a full reset by pressing the button before turning on the station.
Then go into Calibration / Adjust and set a rough aproximation there.
Save and run Calibration / Start.

And yes, the AliExpress temp probes are absolutely crap.
The one that came with my Uni-T DMM did really well.
I lost it, the others have given a lot of headaches.
Touch, you get 400°C. Remove and touch again, you get 370.
Clean everything , new solder, touch again, 385.
To make it worse, my probes came with the last 5mm twisted.
That's ok when you put the whole thing inside a hot oven, but when you measure a small spot, it will short out the voltage and read something on the middle. I got 150°C less!
After de-twisting it got better. But it's extremely sensitive to contamination.
Ex. The tiniest burned flux will stick to it causing lower readings. I have to scrape the small metal ball, ensure it's clean and shiny, place it under the cold tip ensuring that it makes enough force, heat the tip until it melts solder and play a bit with the position. After that I can crank the temps up.
But if the probe comes out at 450°C,  putting it back will read 20-40°C less, requiring another exhaustive cleaning. Crap.
I think the probe wires are too thick, stealing heat from the tip of the probe, so any dirt will cause some isolation and wrong readings.

[DavidAlfa]

Anyways, after tearing your hair and clothes apart, the calibration is pretty good.
I've done it better, but getting these probes to work these showed to be a serious challenge.

[StephenR0]

Wow.  That worked much better.  I was able to get a quite accurate calibration, at least I think I did.  I did it twice.  The first time, I had to manually bump up 450°C three jumps to get it in the 50°C range.  But after that, it worked just fine.  So now I'm curious, could I have effectively done the same thing by manually adjusting all three settings lower?  Or did you do something in the firmware that can't be adjusted?

[DavidAlfa]

Didn't you adjust these settings before? That's why they are there!
In fact, checking your results: "350    420", that one should fail because it exceeds 50ºC over the setpoint.
And the error message should clearly tell you to go to calibration/adjust and adjust the values.
In that menu, if the default values for 250/350/450 steps were 1400/1900/2200, and you were getting higher temps, what you should have done was to lower them.

I've worked a lot on this project, the only big issue that remains is the default calibrations values. Some boards get -100ºC, while others +70ºC
That is because for now the calibration values are generic. I could change the code so each board has its own values, it only takes 2 minutes.
But it seems nobody cares, I've already asked few times to the users to report calibration results, because I really need to make a database for it.
That way people could get nice, closer temps out of the box without needing the annoying calibration process.
But it'll stay generic until then.

[StephenR0]

You're right.  I did have to adjust those settings to get the calibration that I got done before.  I just didn't realize that I could have adjusted all of them much lower and gotten past it.  Thanks for all your help.

[DavidAlfa]

I modified the readme, adding a FAQ section. Because some were really too frequently asked questions 

[Tugo]

Hi David,
I believe I have all the parts needed to modify and upgrade the controller.
I bought Ksger v2.1s and I already changed uC. Ksger version is that one with normal 3.3V regulator not a switch version.
Which modifications do you suggest?
Adding diode from MOS to ground?
Adding 1M ohm resistor from input opamp to Vcc as pullup?
Anything else?
I hope to assemble everything together and to start testing ASAP but lets take it slow so that I don't do anything stupid 
Any kind of advise is highly appreciated 
Again thank you for your effort in this project!

[wickated]

Adding diode from MOS to ground?

that could overload linear reg and gate transistor. i guess fet driver could be installed izi

[DavidAlfa]

How? We're talking about putting a diode in the source to kill negative spikes when the fet goes off.