CFW for KSGER/Quicko STM32 Soldering Stations

[DavidAlfa]

Both tips calibrated very close, but after the 10 minute burning time.
At first, the difference was huge, 3600 ADC at 400°C!
If you calibrate that way, once the tip settles, the real temp will be much higher!
Jesus are you still alive? How were your last calibrations?
Ksgers will be about 10% higher due the higher op-amp gain.

[DavidAlfa]

Okay, definitely the tips needs some hours to stabilize. Within 2h the temp had slowly drifted 30ºC, the calibration at 400ºC was now 2400.
I noticed that temps >400ºC were way off the setpoint (setpoint=480, measured=440), there's no calibration for that, it guesses the value based on the 300-400 calibration.
200ºC won't melt ROHS solder alloys (who solders at 200ºC anyway?), also a little error at lower temps don't represent a risk at damaging the tip or the electronics, which definitely can happen at >400ºC.
So I decided to change calibration values to 250, 350 and 450ºC.
The regulation is now better in the whole range, the biggest error I found was 4ºC

Remember to leave the station working for at least 20 minutes to heat the internals and calibrate in the best accuracy conditions.

I will leave the tip running overnight. It must settle at some point! And that will be the final calibration.

[mastershake]

wow mine never could get that stable and would bounce all over the place. ill re flash again tonight and test it out again. i know this tip is fine i was using i tested it with the stock firmware on both the fluke and the hakko thermocouplers. you say stock pid do you mean the ones you have as default or the ones you listed here in the thread as the stock values? thanks!

[DavidAlfa]

By stock I mean defaults. I'm also using default PWM settings.
Try EMA filtering and factor of 2, and tell me...

I didn't make new builds yet, although the code is uploaded.

I don't know if the hakkos do better, but this tips take about 10 seconds to actually get to the setpoint temp, after the TC has reached it.
Crappy heat transfer, but ok for small jobs.
Don't try with a ground plane or everyone will laugh 😂

I think I found even better pid... 25-42-15

[jesusvallejo]

yes, still reading you , been doing some exams. i will update with the lastest and report back.

[totalnoob]

200ºC won't melt ROHS solder alloys (who solders at 200ºC anyway?),

Maybe to weld plastics?  Burn wood?   

I don't think it matters, 200, 250, whatever, as long as the spread is even, the calibration should be spot on regardless of the starting and ending points. 

About the cheap T12 tips, perhaps the ones that came with my KSGER are pretty decent because I don't believe the break-in period was as long as 10 minutes.  5 minutes, tops, I'd say and they have been spot-on and don't drift up or down much, but that's with the stock FW which may not be as responsive (i.e. update as frequently) as your new FW.  Of course, I can only confirm within the limits of my FG100 clone.

[DavidAlfa]

It's very random. The BC3 is perfectly stable even only using AVG filtering.
But the D24 reads slots of spikes, can't be fixed even increasing the delay to 50mS.
You can hear the coil when the pwm turns off, like "tac-tac-tac-tac...". Luckly the EMA 2 is able to fix it.
I will try to see what's going on with the oscilloscope.

[jesusvallejo]

t12                 ,c200,c300,c400,m200,m300,m400
Conical big      ,1397,2091,2966,190c,306c,408c
Conical small  ,1577,2210,3150,200c,301c,404c
knife              ,1538,2184,2841,200c,302c,400c
curved conical,1591,2232,3457,197c,292c,not stable

Not stable means tip most probably needs to be burned in a bit more, the knife one wich i use the most is super stable.

[DavidAlfa]

My unstable tip has at least 12h burn time, and didn't get better.

[DavidAlfa]

I've set the EMA filtering by default. I think I've tested it enough, and gives a lot more of stability.
Ignore the display overshoot, it doesn't appear at the iron tip, in fact it needs 5-10 additional seconds to reach the temperature due the thermal impedance.

Changed the encoder behavior:
- Click and increase to change tip.
- Click and decrease to enter sleep.
- Click to wake (If enabled).

Changed build optimization to "O2" for faster code execution.
Updated builds.


I could replicate the KSGER issues (temp jumping around) by removing the input filtering capacitor at the amp input. But EMA filters it nicely.

Edit: deleted the mod pictures. No real improvement.

[mastershake]

i got the new oled white panels and they are very close size wise and exactly the same pin out (for ksger 3.0 / 3.1 6 pin) ill solder on in this weekend and report back but these can def work size wise.

is there a specific way the quickko is done vs the ksger then filtering wise? id love to stop it from jumping all over like that if i can.

[AndyC]

I found this thread a while back and got all excited. I got some KSGER boards, some dev boards, installed the CubeIDE...and then realized I have no clue about any of this. Any good resources for STM32/C noobs?
Eventually I was able to flash a prebuilt .bin to a blue pill and then moved the F103C8T6 to the KSGER board. I haven't seen this board anywhere else, but gave it a chance and the firmware (SW: r.c727393, HW:KSGER21F101) worked.
The temps were way off at first, but dead on after a calibration.
The only thing that didn't work was 'shake-wake'.
Some ADC numbers from a well used T12-K:
250C: 1826
350C: 2573
450C: 3320

I also have another T12 station from JCD, which at first looked like a KSGER variant, but the board is quite different. It does have an STM32F103, so I might tinker with that one too. The temps on the JCD station are quite stable as-is though. All these board designs seem very unfinished. Is it time to make our own?

Anyways, fantastic job with the firmware! I have more boards on the way (KSGER and Quicko), so more testing ahead.

[DavidAlfa]

Start with the basics, YouTube tutorials.
Read the datasheet and reference manual.
Read the HAL manual.
Open HAL code examples and try to understand how it works, try to implement in your own project.
Start with simple things, once at a time, until you have enough knowledge to start mixing things together.
There're some basic stm32 books in Internet.
Else than that, stm32 literature is poor.

C language is another different thing.
You must first learn its sintax, operators, variables, methods, structures...

It's a hard learning curve if you never used them.
Needs a lot of practice. Hours, everyday, for weeks, months... You never stop learning.
C can be very simple but also overwhelming.

There's a video in the github readme showing how to build the project from zero in 2 minutes.


I learned microchip PIC16/18/24 assembly and C programming by myself around 13 years ago.
Arduino didn't exist yet. And there where no libraries either, so it was hard, but you knew how every single bit worked, touching everything at low level.
And I did always keep in touch with that from time to time.
Still, it took me easily 2 months to get fluent with stm32 world and get the project running.
Mostly caused by the original fw issues, HAL bugs and lack of documentation. A lot of the time not knowing if it was my fault, neither what was causing the issues.
Learning a new architecture is no easy. And not to talk about the stupid HAL documentation...

About the shake issue. Doesn't a small wave icon appear at the bottom left when you shake the handle?

The stm32 version is very random, they mix 101,102,103... using what's cheaper at the moment.
As stated in the readme, the 101 build works with all of these.
I neither know that board,  but at least it's green and the traces can be seen easily.
At first it seems the same pinout is being used, but didn't check everything. The board labeling is very nicely done.

[mastershake]

the white oled screens worked perfectly! they do fit fine with the oem faceplate just a bit lower then the ksger ones but what i did is gently bend the header pins down a bit then straight out to raise the panel just a bit and its a perfect fit will take some more pics once i put it all back together.

 

[sergo351]

I try to flash the "STM32F102" file STM32SolderingStation.bin from the "BOARDS" folder and get this Screen, native firmware 2.11-2.12 is working / stitched, am I doing something wrong? 

[DavidAlfa]

The write protection warning might be the cause.
You have to go to option bytes and set all protection to disabled.
Something like this:



What board is it? Ksger 2.1s?

[sergo351]

What board is it? Ksger 2.1s?

Thanks, it helped))), Ksger Ve2.1S board

[DavidAlfa]

What board is it? Ksger 2.1s?

Thanks, it helped))), Ksger Ve2.1S board

It's working now, then?

Does the "shake" sensor work when moving the handle? (Icon appears in the bottom screen)

[AndyC]

There's a video in the github readme showing how to build the project from zero in 2 minutes.

Thanks, I tried it on Ubuntu first and just got lots of errors. Turns out I had forgotten about '.cproject' files being hidden. 
I've made some builds since.

About the shake issue. Doesn't a small wave icon appear at the bottom left when you shake the handle?

No, no icon at first. It turned out this board had no pull-up resistor for the handle switch. I set PA6 to 'pull up' but still nothing. Then I read about .ioc files in your readme. NVIC was not enabled in the .ioc I had. (Which was downloaded from github earlier this week). With these two changes the shake-wake works perfect.

[DavidAlfa]

As I said, I don't have these board so I can't test them thoroughly.
I see you are a fast learner, good job finding that!

I checked the KSGER 2.1 ioc files.
EXTI was enabled, however the interrupt didn't. Was that what you meant?

Fixed and updated the builds


Are you sure it doesn't have pullup? Seems strange to me, check this pic from other 2.1:

[sergo351]

What board is it? Ksger 2.1s?

Thanks, it helped))), Ksger Ve2.1S board

It's working now, then?

Does the "shake" sensor work when moving the handle? (Icon appears in the bottom screen)

  on the firmware dated March 6, the "Wave" did not appear, I have now flashed the version dated March 8, everything worked when you move the handle, an icon appears in the lower left corner)

[DavidAlfa]

Yes, AndyC found the problem and I quickly updated the fw
Thanks for reporting that the issue is now corrected!

[AndyC]

As I said, I don't have these board so I can't test them thoroughly.
I see you are a fast learner, good job finding that!

I checked the KSGER 2.1 ioc files.
EXTI was enabled, however the interrupt didn't. Was that what you meant?

Fixed and updated the builds

Beginner's luck
Checked the new .ioc, and yes, that was what I meant.

Are you sure it doesn't have pullup? Seems strange to me, check this pic from other 2.1:

Yes, traced and measured the board. There's no 'R12C' equivalent on my board. Like I said, I've not seen this board documented elsewhere.
The latest build works fine on the board though, with the small pull up modification.
I thought I would enable the debug screen and see what fun could be seen there, but it needs...debugging (Mismatch between cal temps)
Is there anything useful for me in the debug screen or should I forget about it?

[DavidAlfa]

Well, we should enable the pullup by default then
Oh, were you talking about the the JCD version?
Yes, it's a bit stupid to put resistors when you can use the internal pullups/puldowns and save some cents.

Debug screen is actually divided in 2,
The first shows internal pid values, ADC reading, pwm output.
And the second just lets you set the setpoint, but instead based on temperature, it's the desired ADC value. That's how it works internally.
Not really useful for most people and uses plenty of space, it will not fit in 10KB RAM devices.
If you want to enable it just go to screen.h, and uncomment the line that has something like:
"//#define ENABLE_DEBUGMENU"

I don't remember the exact name but it's seen clearly.

[AndyC]

If you want to enable it just go to screen.h, and uncomment the line that has something like:
"//define ENABLE_DEBUGMENU"

Yes, I did exactly so, but when building I get errors. It wants the old 'calADC_At_200' etc.
Just thought I would check it out, but I'll leave it as it is for now.