CFW for KSGER/Quicko STM32 Soldering Stations

[ygi]

I mean, after soldering for 15-20min, you can clearly feel a handpiece warming up. At this point, how is NTC inside the handle any better at reading ambient t° than the one on the controller board? This never made sens to me.

[dcore]

Yes, you might lose 10ºC in the worst case, is that so critical? Nope.

No not critical, it was what I observed in T12 test.
I wanted to experience it to learn.

Will it help having a NTC soldered to the pcb, inside the box, measuring ambient temp, compared to using cpu temp?

Did you calibrate the system? Otherwise you're losing your time, of course the readings will be wrong!

Regarding T245, I power on, T245 profile, NTC OFF, wakeup stand, startup sleep.
Station sleeping displays ~37C in tip, 36.7C in CPU and 25.3C in FG100. Ambient temp. in FG100 is 24.5C.

I started the T245 cold tip calibration, I noticed when the station is trying to stabilize 250, shoots at 800, 900C, but at same time tip is bellow 100C.
It stabilizes at 250 but FG100 measures 112C in tip.
I though something was wrong or I did something wrong and didn't proceed doing a manual calibration.

This in both stations, one I only cut R0, I soldered a short wire to PCB jbc pin (lower 6th pin in a row), then to T245 blue handle wire on the other side (with a crocodile).
T245 brown wire to pin 5 and green wire to pin 3 (using front GX12-5 socket).

-EDIT-
Proceeding with manual calibration:
cal 250 defaults to 900 and measures 110C in FG100.
Increasing the value to max 990 measures 120C in FG100.
I aborted the manual calibration and defaulted to 900.
I also ordered another tip.

Without calibration, waking up from sleep and going to the graph, it displays a lot of huge camel bumps, will take a photo, looks like noise? tip is touching nothing.
Going to the the home screen it displays a steady 320C, with a few occasional jumps to 315,326C, but tip is at 128C in FG100

[DavidAlfa]

Don't get fooled, we don't care about the ambient temperature, it's the cold side of the tip what we need.
Why?
Because that's where the Seebeck effect ends, at the connection of the T12 tip.
The probe voltage is generated based on the temperature gradient between the hottest part (Tip) and the coldest (Tip connector).
If you have 400 and 50°C, the probe generates voltage for 350°C.
If you have 400 and 10, it "reads" 390°C.
But the tip temperature is actually the same, it's the cold side what changes.
So that's why the NTC is ideally placed at the handle.
As the handle warms up, the tip sensor voltage drops, as the gradient is lower.
Reading the NTC compensates this, not greatly accurate, but does something.
Using the internal sensor might accidentally compensate enough as the board warms up, but that's pure luck, it might under or overcompensate.
Both handle and board will warm up, so using the internal temperature will be better than nothing.

[ygi]

Well I now know I'm thinking about it the wrong way. Thanks David

[DavidAlfa]

Dcore, just follow the User Manual, there's nothing else!
You're skipping calibration because, well, it reads lower?
What's calibration is for, then? Why would cal250 and cal400 values exist? 
Increase until reading 250 and 400°C respectively, you have room up to 4000.

As stated in the instructions, default calibration is usually pretty low to prevent tip overheating, there're a lot of controllers out there and some of them read a lot different than others, I don't want anyone getting the $30 JBC tip red hot, ruining it.

Also genuine JBC tips read about twice than cloned tips, which are basically expensive T12 tips.
You're using a chinesium JBC clone, right?

Of course it oscillates a bit, it's trying to control too small temperature changes, thinking it's 10°C when the variation is 5°C really, way too sensitive.
Will get better when properly calibrated.

[dcore]

I now see that I was being overcautious.
(Yes, I use china clone tips.)

I did the manual calibration, since auto was not possible.
I had to do manual calibration twice, since the first time, the max allowed value for cal250 was limited to 990.
In the second manual calibration, it allows cal250 values beyond 990, probably because I had set cal400 to 4000 before.

After the manual calibration was correct, I did auto calibration and it worked, values were only off by 4C.

Back to home screen, the temp is spot on 320C, 370 is +2 hurray!!!! 
I notice that cold tip, shows 39C, probably the CPU temp when I did zero set.

It's working very well now!!! 
Thank you so much for all your help and information!   

[DavidAlfa]

Yeah, repeat the zero set when really cold, like after 30 minutes off.
Take out the tip without your fingers (Or you will transfer some heat to it), turn on the station, go to calibration menu, insert the tip, enter calibration adjustment and capture the new Zero Set value.
Once done right, it usually stays pretty close, few degrees off depending on the board temperature (This slightly affects the amplifiers drift).
However remember this is not a thermometer and shouldn't be expected to work like one.
Few degrees off here and there are completely expected, specially at the center between 250 and 400, as the sensor output is not 100% linear but I do use a linear interpolation.
The error will be about 5°C in the worst case, for me it's just fine, not enough to bother with complex maths.
Most commercial products won't display anything below 100°C or more, also they filter all these small oscillations and you won't know they exist (But they do).

Enjoy!

[yeezy]

Which T12 (according to the technical base) is best to buy so that there are no problems with custom firmware? On 4pda I was advised a green diy kit ksger 2.1s, but it is quite expensive.
https://aliexpress.com/item/32933941125.html

[yelkvi]

https://aliexpress.ru/item/4000183089084.html

[wickated]

Which T12 (according to the technical base) is best to buy so that there are no problems with custom firmware?

https://aliexpress.ru/item/32788912011.html
this one

[DavidAlfa]

I'd like to fix the Quicko STM32F072 build, but I lost the chip long time ago.
Does anyone have that controller and some spare time to let me remotely debug it?

[koendv]

Firmware for soldering stations with MM32SPIN27: https://github.com/koendv/t12-958
This is Arduino, and easily adapted to your needs.

Basically I ported the relevant parts of Arduino to MM32SPIN27  https://github.com/koendv/MM32SPIN27-Arduino.
In case you are using MindMotion MM32 Cortex M0 processors in another project: In MM32SPIN27-Arduino/mm32/mm32-libgcc you'll find libgcc, patched to use the hardware division present in MM32 Cortex M0 processors.

[DavidAlfa]

Nice work!

[DavidAlfa]

Well, it seems they started to use relabeled chips.
Spent few hours remotely fighting with Cube IDE, discovered that anything never than JDK15 will break it in very strange ways...
Finally, CPU ID 0x13030410, definitely not a genuine STM32F103, Google says it's a GD32F.
So GD32F also has the same ADC issue as the CH32F.

[eevnoob]

hello, there is a newer KSERG T12 being sold on aliexpress that uses "STC Controller V2.3". one of the new "features" is that the chassis is now properly grounded and also the display appears to have a color interface. however, i suppose this means that the likehood of it using David's firmware is very low? AFAIK this is the first of the T12 clones to be properly grounded from the factory?
https://www.aliexpress.us/item/3256804499899529.html

[NathanBxr]

Well, it seems they started to using relabeled chips.
Spent few hours remotely fighting with Cube IDE, discovered that anything never than JDK15 will break it in very strange ways...
Finally, CPU ID 0x13030410, definitely not a genuine STM32F103, Google says it's a GD32F.
So GD32F also has the same ADC issue as the CH32F.

Did it look like this? I got this one couple weeks ago.

KSGER T12 OLED Soldering Station T12 Iron Tips STM32 DIY Assembled Kits ABS Plastic FX9501 Handle Electric Tools Welding Heating
https://www.aliexpress.com/item/4000008060128.html?spm=a2g0o.order_list.0.0.635e1802h1LEh9

[DavidAlfa]

STCs are not compatible. It's not a clone, but a completely different architecture.

Yeah it looked pretty similar.
After rubbing with some alcohol the label went away, original stm32 are labed with laser.

[ygi]

I thought every chinese faker got a laser engraving machine nowadays.

[DavidAlfa]

Probably got fake laser machines too 

[ygi]

Haha, that would explain it

[eevnoob]

just one other quick question that many of the experts in this thread can probably answer: what is primary difference between the 4 pin and 5 pin handles in the T12 clones?  i've read that 5 wire irons have an additional temperature sensor in the handle to allow for proper cold junction compensation. is that an important feature that was absent from the 4 pin versions? if so, are all 4 pin handles in some sense defective? thanks for help understanding.

[DavidAlfa]

Nothing really, all you need is 4:
- Gnd
- T12+
- Shake
- NTC

5-wire handles use two wires for gnd.

[DavidAlfa]

I made a KSGER v2 test build, enforcing Cortex-m0 instructions for MM32, but I have no way of trying.
File attached, let me know the results if you tested it.

[Spenoza]

Hello DavidAlfa. I am studying your sources with interest. I am not a programmer, and therefore I program a little. I saw a lot of new things in the source code for myself.
To facilitate the study, could you explain the principle of forming menu screens with widgets and displaying them on the screen?

[DavidAlfa]

Sorry, I can't really explain everything, that's a pretty long thing to do, specially if you don't know programming!
Start at gui.c, follow the program flow...
Most stuff happens at widgets.h/c, screen.h/c and the individual screens at "screens/**screen.c".
Mostly you can reuse existing blocks, function names are easy to understand and self-explanatory.

It's like asking a mechanic to explain how to rebuild an engine when your daily business is office work.
I'm not saying it's easy for me either, first time I got into it, it took me weeks to understand everything, few years had pased since last time I programmed.