CFW for KSGER/Quicko STM32 Soldering Stations

[wickated]

first of all set adc delay to normal values like 10-20 ms

[illiac4]

If anyone is interesting there is a cheap option to get this soldering iron. It is STM32 based. It is 31€ including delivery and VAT. I see there are 1o available.
https://www.banggood.com/groupshopping-item-1911768_1764061.html?rmmds=category&cur_warehouse=CZ

A few notes:
Under about it says
HW Version: 2.1S
SW Version: 2.11
I have tried to switch handle from KSGER to this one and is says error the same in reverse. So wiring must be little different in the handle, but i think this will not make any difference.

[dark_hawk]

dark_hawk:
What are your settings for ADC time (or period), ADC delay, PWM multiplier and Filter (those 4 settings are in Iron menu)? Those settings are the first thing I had o set to "good" values so that I could even start anything else. Values that David told you are on my hardware (which is different then David's) the way to go.
100ms, 1ms (which could also be 0.5ms but for safety I set it to 1ms), 1x (but could also be 2x and not anything higher on my hardware) and I set filter to 2 (different values wasn't as good for me).

I didn't change any settings in the iron menu since I have 6 other tips that are working fine and regulating correctly with the default value, my problem is with 2 specific tips that I know have a physical problem but I want to run them anyway. So changing these values in the iron menu will mess up the 6 good tips for the sake of 2 duds.

[dark_hawk]

A question:
Form the theory of operation part of the operation guide:
"The voltage is basically proportional to the difference between tip temperature and room temperature. The controller uses PWM (Pulse Width Modulation) to control the tip temperature."
So the controller measures the room temperature through the temperature sensor in the handle? If that is correct, then if the handle gets hot, the temperature sensor in the handle also gets hot. For me when calibrating the reading of the sensor is around 35 degrees, but with extended periods of soldering that reading jumps to around 45. So that means that actual tip temperature will be 10 degrees above the calibrated temperature. I tried adding those Fiberglass Braided tubes around the sensor but it does not seem to help.
My question: wouldn't it be better to move that NTC from the handle to socket?

[bobbydazzler]

A question:
Form the theory of operation part of the operation guide:
"The voltage is basically proportional to the difference between tip temperature and room temperature. The controller uses PWM (Pulse Width Modulation) to control the tip temperature."
So the controller measures the room temperature through the temperature sensor in the handle? If that is correct, then if the handle gets hot, the temperature sensor in the handle also gets hot. For me when calibrating the reading of the sensor is around 35 degrees, but with extended periods of soldering that reading jumps to around 45. So that means that actual tip temperature will be 10 degrees above the calibrated temperature. I tried adding those Fiberglass Braided tubes around the sensor but it does not seem to help.
My question: wouldn't it be better to move that NTC from the handle to socket?

Yes the handle sensor heats up a little when you're using it, while using the cpu temp sensor is more stable after it's been turned on(though from when you first turn it on the cpu temp sensor is 17-18c lower than where it is after 10-15min, after 10-15min though the cpu temp sensor is more stable - on mine anyway).

[Tugo]

dark_hawk:
What are your settings for ADC time (or period), ADC delay, PWM multiplier and Filter (those 4 settings are in Iron menu)? Those settings are the first thing I had o set to "good" values so that I could even start anything else. Values that David told you are on my hardware (which is different then David's) the way to go.
100ms, 1ms (which could also be 0.5ms but for safety I set it to 1ms), 1x (but could also be 2x and not anything higher on my hardware) and I set filter to 2 (different values wasn't as good for me).

I didn't change any settings in the iron menu since I have 6 other tips that are working fine and regulating correctly with the default value, my problem is with 2 specific tips that I know have a physical problem but I want to run them anyway. So changing these values in the iron menu will mess up the 6 good tips for the sake of 2 duds.

After your worst tip is going to work, your normal tips will also work.
When I had problems I used the worst tip that I could and set everything that it started to work. Other tips were not a problem. David put a nice tutorial how to find a good PID setting for each tip and that was my main problem.
I had problems with oscilation on default settings. Those are Davids settings in which I never believed in but I always believed in Davids firmware. That is why I always insisted it had something that it didnt work on my hardware from the start. When he changed FW I told him he is on right path and that he did a perfect job. I also told him my HW doesnt work well with default settings. I could use delay as low as 0.5ms and period to about 50ms but for stability I set to 100 ms and 1 ms.
Another thing... set Kp to about 40 and Ki and Kd to 0. Then change period and delay to lower values. Use the worst tip you have and see how it behaves on 350C. When it should work, test it on 250 and 450 if it is ok. After that check if your other tips are working as they should. Oh, i have filter set to 2 and I dont know if it is set to highet values in newer version of FW but I also had some problems with it. I just forgot what was wrong with it. 2 is perfect for me.

[ricktendo]

I flashed the 2.x firmware 7-28 build on my ksger hw v2.1s and the rotary switch is reversed when set to the default "normal," I had to switch it to "reverse" to have it back to normal.

I also have a hw v1.5 and issue is NOT present on this one, rotary works normal.

[dark_hawk]

After your worst tip is going to work, your normal tips will also work.
When I had problems I used the worst tip that I could and set everything that it started to work. Other tips were not a problem. David put a nice tutorial how to find a good PID setting for each tip and that was my main problem.
I had problems with oscilation on default settings. Those are Davids settings in which I never believed in but I always believed in Davids firmware. That is why I always insisted it had something that it didnt work on my hardware from the start. When he changed FW I told him he is on right path and that he did a perfect job. I also told him my HW doesnt work well with default settings. I could use delay as low as 0.5ms and period to about 50ms but for stability I set to 100 ms and 1 ms.
Another thing... set Kp to about 40 and Ki and Kd to 0. Then change period and delay to lower values. Use the worst tip you have and see how it behaves on 350C. When it should work, test it on 250 and 450 if it is ok. After that check if your other tips are working as they should. Oh, i have filter set to 2 and I dont know if it is set to highet values in newer version of FW but I also had some problems with it. I just forgot what was wrong with it. 2 is perfect for me.

Thanks for the detailed reply.
I actually managed to get those 2 bad tips to stop oscillating with just tweaking the PID values and are working relatively fine now.
https://www.eevblog.com/forum/reviews/stm32-oled-digital-soldering-station-for-t12-handle/msg3617005/#msg3617005
The first one values were 35/15/25 and the second one were 30/15/25.

Yes the handle sensor heats up a little when you're using it, while using the cpu temp sensor is more stable after it's been turned on(though from when you first turn it on the cpu temp sensor is 17-18c lower than where it is after 10-15min, after 10-15min though the cpu temp sensor is more stable - on mine anyway).

I'm not aware of an option in the firmware to switch the temperature sensor from the handle's NTC to the controller's sensor.
I just moved the NTC from the handle to the socket that plugs into the unit, the ambient is now stable within 2-3 degrees compared to the 15-17 degrees when the ntc was in the handle.
But now I need to recalibrate all the tips again since the readings now are actually a bit higher since the NTC is reading lower.
I'm just waiting for David's reply to see if there is something in the code that compensates for the handle getting hot with use.

[DavidAlfa]

That usually happens with new tips. If the tip has already some use, the only explanation is to be defective.
Probably it can't handle the power. I mean, it can't tranfer the heat properly.
What happens when lowering the power to 20W or so?

[DavidAlfa]

I flashed the 2.x firmware 7-28 build on my ksger hw v2.1s and the rotary switch is reversed when set to the default "normal," I had to switch it to "reverse" to have it back to normal.

I also have a hw v1.5 and issue is NOT present on this one, rotary works normal.

That's the first report I have from the ksger v1.5. It was completely untested.
What probably happens is that actually the  reversed one is the v1.5!
Normal encoder behavior is to move down (and increase values) when rotating clockwise.
So what station is doing it wrong?



I'm almost done with the new version. All bugs are gone, now I'm fully testing it and thinking the interface layout.
The ram looks really good now.
5.35Kb are fixed (HAL, oled buffer, other variables).
Without dynamic allocation, it used 10KB (99.99%). Now in the worst condition (SYSTEM screen, which is huge) it uses 7.1KB.
The screen/widgets memory usage was reduced down from 4.7 to 1.7KB, which is a 64% reduction!

[Tugo]

I flashed the 2.x firmware 7-28 build on my ksger hw v2.1s and the rotary switch is reversed when set to the default "normal," I had to switch it to "reverse" to have it back to normal.

I also have a hw v1.5 and issue is NOT present on this one, rotary works normal.

That's the first report I have from the ksger v1.5. It was completely untested.
What probably happens is that actually the  reversed one is the v1.5!
Normal encoder behavior is to move down (and increase values) when rotating clockwise.
So what station is doing it wrong?

David, I also have Ksger 2.1s and had to reverse it to work normally. This is present (at least for me) form the start but I thought others have the same version and not have this problem 

[DavidAlfa]

And you say that now? How am I going to guess that? 
Fixing it in a breeze.

[wickated]

mine was reversed too, but i never bothered cuz just swapped encoder wiring since its not mounted onboard

[ricktendo]

1.5 encoder works perfectly, its 2.x that has the bug:

Clockwise it moves up
Counterclockwise moves down

[DavidAlfa]

You're terrible beta testers! 
Always complaining about little things or personal likes while missing the important! Learn from ricktendo 
Fixed, new builds will be ok from now.

Ricktendo, please test the v1.5 throughly.  The software is the same thing for all, don't try to find bugs there.
Just test the hardware functions: buzzer, oled, pwm, encoder, tip/ambient/supply readings...

[wickated]

You're terrible beta testers!

why? its always a lottery of what type encoder will chinese place onboard. they can even place 20 tick instead of 24 tick one.

[ricktendo]

Hi David, you think you can make the default offset on v1.5 to zero instead of two?

I do not think 1.5 has the same offset problem as the other boards (I could be wrong,) hopefully somebody else with a 1.5 can confirm this.

Note: I have a 1.5 with a yellow/blue color lcd, this is a smaller size screen than the 2.1s similar to (if not exactly) this one https://deathmarch.net/archives/1414

Edit: I will test the 1.5 more thoroughly when I fix my fake Hakko FG-100, buzzer, ambient and supply readings all seem fine.

[DavidAlfa]

they can even place 20 tick instead of 24 tick one.

The encoder resolution is something irrelevant. But the direction will be the same!

Hi David, you think you can make the default offset on v1.5 to zero instead of two?

Of course, that can be done for each specific profile. But nobody's telling these things! 

Added a new define in all board.h files to make it profile-specific. Took me 30 seconds!
What are the offsets you're getting in the different models?

[dark_hawk]

David:
What do you think about moving the NTC from the handle since it gets hot to the socket? Will it affect operation?

[DavidAlfa]

No! That's how it's supposed to be. You're not measuring the ambient temperature, but the thermocouple cold junction temperature.
A thermocouple generates voltage based on the temperature gradient between the hot side and the cold side.
It has 2 different metals. Those metals go all the way from the sensing tip until the end of the probe.
The temperature gradient ends where it reaches the connector (When the material changes).
Actually, it makes a second TC in series, luckly copper and solder generate little to nothing.
It gets really really complex when you dive into it!
Optimally that junction should be at 0ºC for best performance, no compensation would be required.
Since it's  impossible to achieve that in a lot of applications, you must compensate somehow... And that's why the NTC is there!

The TC will either output nothing or negative values if the hot side is colder than the cold side.
In this case, we can only read positive values. So if the junction is at 40ºC, any measurement will be over that temperature.
Of course, it will drif a bit. There are a lot of complex mathematics behind a compensation algorithm, which I'm not doing.
As I said before, this is a soldering station, not a a bio genetics laboratory. 5ºC off are completely acceptable.
Read a bit about thermocouple cold Junction compensation!

Otherwise, why the *** would the station care about the ambient temperature? Just a fancy reading? 

I modded my Quicko and moved the sensor to the handle to make exactly that!

I'm going to start saying a famous quote in the developing community, "use it or change it".
That means exactly that, when something doesn't cover your needs or likes, fork it and make your own!
Don't like the font? Don't like the interface? Can't live with a 3ºC error? Change it! Or contribute to the project!

[Tugo]

You're terrible beta testers! 
Always complaining about little things or personal likes while missing the important! Learn from ricktendo 
Fixed, new builds will be ok from now.

From my end: I AM A TERRIBLE BETA TESTER!
Now go live with it and make your god damn good firmware and stop  us 

When I made my first post you guys really had a debate for couple of months and nobody said anything so every time I just got into a settings and change that. Even when I did some tests and compile it I change that in the code so that wasn't a big deal and I really thought only my board is like that 
Well... I think I will have to review it from the start to see if I missed anything else 

EDIT: Regarding NTC: I set temperature to about 350 C and wait some time so that NTC is getting a little bit higher and then I do my tests with ADC at 250, 350 and 450. So when I will start using the tip the temperature will be a little bit less than what is should but will get to the desired after a while.

[DavidAlfa]

I'm running 20mS again.
I don't get nasty oscillations at all, but the power bar is going up and down by 20% all the time, which means the tip is not getting enough clearance to give a clean reading.
There are somw people here thinking the best is to have ultra fast pwm, ultra fast delay, ultra fast everything... when you must simply use what is correct.

I set up the D52 @ 400ºC, NTC @ 26ºC. The attached probe read 402ºC.
After 5 minutes, the NTC had heated to 38ºC. The probe read 401ºC.
So I don't see any issues at all.

[wickated]

to set things clear. ntc could be placed anywhere if tip and handle could withstand temp. its only a must if your handle get hot to 60c , what chinese certainly do in my casei use very high quality china copper wires, that get hotter than anything else and obviously placed ntc nearby, which results in pretty good readings.

There are somw people here thinking the best is to have ultra fast pwm, ultra fast delay, ultra fast everything... when you must simply use what is correct.

its pretty obvious that jbc tips require much faster everything. and im pretty happy with current 100hz adc readout.

[dark_hawk]

I'm just waiting for David's reply to see if there is something in the code that compensates for the handle getting hot with use.

Thanks for the reply.
I was not trying obviously to do something new or asking for work or code to be written, I'm trying to learn, hence my question about if there was compensation for the change in the NTC temperature and if it affects the readings of the tip not that I wanted it implemented.
I just did the same test you did with a D52 tip @ 400 C, and the difference between an NTC reading of 33 to 45 was about 4-5 degrees which is negligible imo, we are not working in aerospace applications after all.

Don't like the font? Don't like the interface? Can't live with a 3ºC error? Change it! Or contribute to the project!

I actually like the font and the interface a lot. I wish my scope of knowledge covered programing for the STM32 but it doesn't so all I can do for the project is report any bugs I find and suggest ideas for you to consider.

[dark_hawk]

What is the difference between "Reset Profile" and "Reset Profiles"?
Reset profiles does not seem to do anything when you click reset.
Also when you click "Reset Profile" I get the error message in the photo. And after I click the knob it resets the board and the first boot menu shows.