CFW for KSGER/Quicko STM32 Soldering Stations

[Wirehead]

That seems to have done it (using the 500ms one). For reference, the 2.1 version that I have, is using a linear LDO + tant cap. Maybe power rises indeed a bit slower than with a switcher.

[DavidAlfa]

That's what I guessed.
I already updated the code, next releases should work fine.

[Wirehead]

Works great. Mine is powered with a floating SMPS, and the ksger 2.1 board technically couples tip earth to the negative supply. So I put the shaker on the NTC ground as well, but oriented so while soldering it doesn't trigger. I now have a nice separate, earthed tip, instead of floating, possibly damaging components. Only gives a small *blip* when the NTC is read at the same time (technically, it would be possible to sample these out in code I guess  , so shaker sampling does not occur when NTC is read).

[DavidAlfa]

I didn't understand it all.
The shaker must use the same ground as the controller, not the earth.
How are you going to damage anything by joining ground and earth?
And what's that blip you're talking about?
The Shake input is read at Systick rate (1KHZ), not with the ADC.
There's an exception, in stand mode it's read after every ADC conversion.
But ADC and Shake input are completely separate parts.

[Wirehead]

I didn't understand it all.
The shaker must use the same ground as the controller, not the earth.
How are you going to damage anything by joining ground and earth?
And what's that blip you're talking about?
The Shake input is read at Systick rate (1KHZ), not with the ADC.
There's an exception, in stand mode it's read after every ADC conversion.
But ADC and Shake input are completely separate parts.

So For the following - My switching power supply, supplying energy to the station, is a two-prong type. So with reference to actual earth, there's a 27V offset. This means, the tip; was also having a 27V offset in regards to earth. So I lifted the wire below; and coupled it via a resistor to real earth. 
[modification.jpg]

But the original handle pinout, had the following:
[original.jpg]
The shaker ground, was also connected to the tip ground (and that in turn to the negative voltage rail). So I had to move the shaker ground, to the same NTC ground. But when actually shaking, you see  some noise in the graph.

[DavidAlfa]

That handle wiring was either wrong or designed for a circuit that has earth and gnd connected together.
Of course you saw noise, it was connecting the earth through the stm32 and few resistors, bad idea if the smps is isolated.
How are you going to get 27V offset if you join negative and Earth? Once they're joined, both will become the smps ground reference.
New builds (21-10-06), oled should be working perfectly now.

[wickated]

offset from gnd ?
maybe there is just ac constant cuz in most SPSUs u have Y caps (afaik) that connect secondary gnd with ac gnd. i actually dunno real schematics, so can say wrong

[Wirehead]

That handle wiring was either wrong or designed for a circuit that has earth and gnd connected together.

Indeed, that's why I put the shaker on the negative wire instead of green "earth", and lifted tip ground, with a separate cable + resistor to a real earth. (see handle wiring in attachment)

Of course you saw noise, it was connecting the earth through the stm32 and few resistors, bad idea if the smps is isolated.

Well no, technically speaking, the negative of the smps is also the gnd of the stm32. Real Earth is completely separate now and only used for the tip. But when shaking, I do get some noise now in the ADC measurement.

[DavidAlfa]

First, stop the abstract descriptions and post pictures.
Clearly something affecting the analog circuitry.
Try putting a 1n4148 diode in series, so the shake sensor can only pull to ground.
That's a very bad circuit design anyways, no protection at all.

Try something:
Heat the tip to 400C, then set 180C setpoint.
As the tip is hotter than the setpoint, the pwm will be off.
Do the spikes still appear when shaking the handle?
It's really strange.
Again, if your mains earth is good (a lot of older houses have poor earth connection), there shouldn't be any offset.

[Polf]

Firmware 21-10-06. When turned on, it shows the real temperature of the NTC and the soldering iron, for example 22 degrees, then gradually increases to 30 degrees within a minute. I take out the tip, insert it again and now it shows the real temperature again.  This happens at two stations, one works with T12, the other with C245.

[ygi]

It's probably as good as it gets at such low temperature. I mean, the expected working range is 200-400°C , it's not a weather station.

[Polf]

I am not interested in the value of the ambient temperature. The strange thing is that change happens for no reason. In addition, this was not observed in previous firmware.

[DavidAlfa]

Enter the debug menu, insert the cold tip, wait some seconds and note down the AVG value.
Keep watching the value as it heats up.
Debug screen has 5 minutes timeout, so there's plenty of time.
Mine does perfectly, no issues at all. There were no changes on the measuring code.
Also, the tip detection sends small power pulses, it might increase the temperature some degrees, but not that much.
Could be a lot of things,ex. the internal amplifier causing a small drift with the temperature.
What happens when you remove the tip and plug it back?
Have you changed any filtering/pwm setting from default?

Ensure your description is accurate. Ex. 1 minute is not 10 seconds, It can mean a completely different thing.
Or just make some videos.
My station shows 23-24°C with 24.5°C ambient, all the time.
Ensure your calibrated zero cal value.

Anyways, not something I'd spend my time on.
My personal life has changed lately, and the project will receive much less time.

[Polf]

David, thanks for the answer. I shot several videos.
I looked at the AVG value, but I don't know what to do with it later.
Active det. is disabled, so there is no heating.
After removing and installing the tip, the temperature decreases by about 10 degrees.
I did not change the filter settings and power.
I completely erased the microcontroller and re-flashed it to exclude incorrect settings, but it did not help.
Returned to firmware 21-09-29, temperature is stable, removal and installation of the tip is not affected.
Video here:
https://drive.google.com/file/d/1-s2-wLOUAZt87tjqJ7ia5Y9WDzRMO9iA/view?usp=sharing

[DavidAlfa]

I can't access the file. Check the permissions, ensure it's public.

[Polf]

Sorry, fixed.
https://drive.google.com/file/d/1-s2-wLOUAZt87tjqJ7ia5Y9WDzRMO9iA/view?usp=sharing

[DavidAlfa]

It's not the tip what rises, it's the ambient temperature sensor.
It goes from 25ºC to 36ºC, seems it's disabling the external NTC for some reason, using the stm32 temperature instead.
When the handle or tip is removed, the internal sensor is never used, unless the NTC is disabled in the menu.
That can be quickly checked out,  remove the handle connector, the ambient should show -99.9ºC.
If it doesn't, the NTC is disabled. In that case the value should never change, regardless if the handle is connected or not.

Check your wiring. I can get similar results when I connect the handle only partially from one side, changes between 13 and 25ºC.
In your case, it seems like it isn't making a good ground connection, and travels back through the shake circuitry.

I checked the code and didn't find anything. Last weeks were minor updates, mainly to the user interface, translations, nothing to the sensors.
If you keep thinking it's a firmware thing, go to Github history and start trying builds, from newer to older.
Ensure it's not the wiring, test it thoroughly!


Is anyone else able to replicate that issue?

[Polf]

I mentioned above that this temperature change is observed in the same way at two different stations. The first one, which can be seen in the video, has a controller according to the KSGER HW2.1s scheme and works with the T12. The T12 handle is connected to the station.
The second one is assembled according to the same scheme and works in the C245 profile with a copy JBC handle.
I flashed both stations to the software on 21-09-29, now no temperature jumps occur.

[wickated]

Is anyone else able to replicate that issue?

definitely not possible.

[Polf]

Check your wiring. I can get similar results when I connect the handle only partially from one side, changes between 13 and 25ºC.

Thanks. I will check. I just double-checked the C245 station, it looks like it’s okay and the temperature doesn’t jump.
Tomorrow I will check the wiring of the T12 handle, maybe something is wrong there.

[DavidAlfa]

Keep trying newer versions, if you find exactly the one that causes the problem it will be much easier.ç
I have re-checked every change. Again, nothing related to the sensors.

21-09-29
21-10-01
21-10-02
21-10-02a
21-10-06

definitely not possible.

So a black capacitor can't be a capacitor anymore? I smell racism here  .
What did you do this time? Or you soldered a coil by mistake?

[wickated]

nope, thats just big black NTC i bought in store cuz black ones work better than chinese ones (no racism)

[ygi]

For what it's worth, temperature doesn't drift on my v1.5 using latest build (21-10-6) and it is also fairly accurate (error margin is less than 2°C with cold tip).

[Polf]

To avoid possible mistakes, I took another controller, assembled according to the same KSGER HW2.1s circuit and a different handle. First, I flashed the controller with software 21-10-01, I did not find any temperature jumps. Then I flashed the controller with firmware 21-10-06, the result is in the video. Profile T12. The jump in temperature is 7 degrees. Earlier firmwares do not give temperature jumps.
https://drive.google.com/file/d/1-soPKq4LCOnD7XktsQflZ7TuJuaTACkj/view?usp=sharing

[DavidAlfa]

What about 2 and 2a versions?