CFW for KSGER/Quicko STM32 Soldering Stations

[wickated]

I think I was already clear enough - I won't do any more work on these boards.
If default PID doesn't work, well, bad luck, tweak yourself. I won't make 50 tries until you get a good one. You can do the same.

well i dont even understand what u r talking about.

lateset firmware shuts down with at least 2 errors - in adc read and ntc read.

i dont know why are you talking about tweaking pid.

[DavidAlfa]

Don't you understand that a problem that only happens in a board that I can't test myself is a nightmare to deal with?
I repeat, I'm done wasting my time that way.

[DavidAlfa]

All I can say is go here and try each fw version, going back in timeline until you find one that works.
Avoid Jun 24-25 builds as they were bugged, pwm was not being shut down when going to sleep, burning the tip.

If you think that's a tedious work, then think how it's for me, coding, debugging and building each one of these builds.

https://github.com/deividAlfa/stm32_soldering_iron_controller/commits/master/BOARDS/KSGER/%5Bv2.x%5D/STM32SolderingStation.bin

I'm out again, the last working tip burned my handle socket down.

[wickated]

cmn chiil down. i only wanna say that fw hooked some bugs from some time. i flashed back may fw and its rock solid, except unworking tip menu adc settings.
i spent some time to reduce noise and now have no spikes at fast pwm. as i said before, with jbc tips fast pwm is a must cuz it can easily overshoot 150c or even more. and jbc tips are trend nowadays. chinese have shitty TC but heat up well.
i just said there ARE bugs, if u cant fix them by any reason, its ok, there is no way to do anything without bugs. i dunno how github work, so link for older builds is MUCH enough.

p.s. i see there is code, ill try to read it later

[DavidAlfa]

Don't get me wrong, I'm just tired.
Fixing bugs when you can't debug is just exhausting.
More than once I've spent an entire day trying to find a Ksger problem that doesn't happen with Quicko.
What I mean, simple and clear: I won't lose my time that way anymore.
If you want proper support, then each Ksger owner puts $1, I  buy the said board and  then we might start talking about bug fixing.
Otherwise, why would I spend my own money?
Development is almost over, everything works pretty nicely.
Yes, Ksger 2.1 seems to have some timing issues, but that is the hardest part to fix, no way when you can't see what it's doing.

[StephenR0]

Hey, even I would give a dollar to contribute.  And I'm a Quicko owner. 

[DavidAlfa]

It took another 4 hours, but I think I found the issue. I smell it appeared after the recent update of STM32 F1 library (v1.8.4).

I set my quicko as slowest as possible (8MHz), and software-mode display to slow it down even more (And it still did pretty nice! The graphics optimization gave its results  )

And the errors started to appear! Suddenly, the adc was triggering twice. But the program workflow was fine, nothing was calling the adc start again... was doing that by itself.
I suspected it was coming  from the adc stopping. Something going wrong causing to self-trigger again. That could only be the DMA?
Long, long time ago I had a issue when stopping the ADC, the library waited for the dma to finish, but it had already finished, so it would stall for long times, causing a lot of trouble.
I had to made my own detection method, and it worked fine all this time. I decided to remove the fix that previously worked, and directly call the adc dma stop.
Now it no longer did what happened months ago. A clean, proper stop, no stalling, and no more double adc triggering.
Why didn't happen with Quicko? Who knows. I really hope it wasn't a lucky shot and the error is still around, just hiding from me.

Try the atttached file
But what I said about ksger hardware remains unchanged.
Debugging/fixing things in hardware I don't have takes 10x more time. I can't/won't do that anymore.

**** you ST and your libraries!

[tatel]

Hey, even I would give a dollar to contribute.  And I'm a Quicko owner. 

+1

Also a Quicko owner.

I think people that get us out of vendor's claws hast to be supported. The amount of effort is not trivial and we all take profit of it. I find much more value in my solder station if I can fiddle with it's firmware. Please put a donation link and I will happily contribute.

And take it easy, man. Nobody here expects you to work non-stop for free nor wants you burn out.

No te encabrones, hazte el escandinavo.

Best wishes

[StephenR0]

And if Ksger owners don't want to contribute, we can always get him some tips. 

[DavidAlfa]

Thanks guys! I really appreciate it. I lost count of the tips I've killed for the benefit of the development. Easily 12...15...
It's not like I'm wanting money for it, but it would be nice to have some support for the project, buying new tips when they burn down, or properly supporting problematic boards. I added a link in my signature!
About the burn out, it's not the developing itself. I love new challenges, I like to support this project!
What I don't like is to waste my time. If there's an issue in Quicko boards, I set up the debugger and I ususally find the problem quickly. But with others, it means hours of trial and error.

Talking about errors, they seem to be gone, so... new builds!

Can't test PID settings until I buy new tips (The handle half-melted too ...  ) so meanwhile I reverted to older ones.

I made some work in the graphics driver, to allow simultaneous software and hardware i2c coexistence, and detect if the display is connected to the hardware, or use software mode instead.
It'll be specially good for the ksger 2.1, which has the display connected to normal pins, requiring software mode.
This will allow modifying them by connecting the display to the correct pins, and yet maintain compatibility with non-modded boards, without requiring any firmware modification.
This is still WIP, so not enabled by default. When/if the detection works correctly, it will be added to future builds.
Added this picture to github for modding instructions.

[DavidAlfa]

Thanks David,

CH32F devices aren't supported yet. I'm unable to debug them to find the issue. Get other clones or real ones!

I really agree with you!

Hey, I suspect the CH32 reacted a little different to the ADC bug.
Give it another try!

[wickated]

I made some work in the graphics driver, to allow simultaneous software and hardware i2c coexistence
This is still WIP, so not enabled by default. When/if the detection works correctly, it will be added to future builds.

i am awaiting for tests. as i read in code 2.1 boards are only ones using software i2c screen

**** you ST and your libraries!

well at least there ARE some IDE now, i took a look like 10 years ago and it was like programming a mainframe with perfocards.

[DavidAlfa]

First, why don't yiu try if the error still happens un these? Otherwise the display is useless!

[wickated]

First, why don't yiu try if the error still happens un these? Otherwise the display is useless!

as far as i guess display and adc shares processor time at most, so mod for dma i2c should offload plenty of cpu time. and my board is already wired for this.
as for latest fixed fw , station is currently busy working for food. i ll test fix a bit later.

[DavidAlfa]

That would be a terrible programming, maybe someone who just started learning would do that, after the first LED blinky program.

The display has absolutely zero priority in the system.
All the important functions are called using interrupts, and as the name suggests, that makes the CPU to stop whatever it was doing, run the priority task, and resume the previous one. That includes ADC, pwm, iron, PID, error/runaway detection...
That's why everything works perfectly even at only 8MHz and software display mode.
The display driver might be drawing a line, or refreshing the oled memory... it'll get interrupted 1000 times if necessary to do the important stuff.

When the system works correctly, only then new features might be added.
Otherwise, working on already broken code is like storing air your pockets, plainly stupid.

Anyways, the max attainable i2c speed depends a lot on the pullup resistors. The displays come with 10K, which limits the slew rate.
The firmware is already overclocking the i2c bus by default, with a peak speed of ~900KHz.
With 470R resistors I've been able to brutally rape the i2C standard,  only 50nS between data and clock, and 150nS clock high time, achieving 2.2MHz.

Stock: 48fps
Pullups + lowering delays: 141fps
Pullups + 64MHz: 211fps (Modifying code to maintain the same clock delay)

But of course this can't be done by default, it would break the compatibility with most v2.1 boards.

[Tugo]

David I did quite a lot of testing with new FW, old FW, different tips,...
At the end I found out that different tips mean different story...
I have:
- new tips
- old tips
- Hakko tips
- other tips (Quicko, Ksger, noname)

For the purpose of testing I used 1 noname "K tip" which is pretty new tip but with a burn in of at least a couple of hours but I didn't use it for work (just for this testing) and 1 pretty much the same "K tip" but should be a genuine Hakko (it has a logo on it) which also wasn't used for my work but was also stable in my Ksger soldering station.
Today after testing a lot of different new and old FWs I decided to try a tip that I used from time to time and it is a DL52 (from Ksger). Well... it works... I could set it to pretty much anything. Delay to 0.1ms, period from 50-200ms... temperature down to 180 C... you name it and it worked.
I also tested a Hakko JL02 which also worked ok as DL52. Then I tried different Hakko tips which I use the most and none of them worked (D12,  DL32) - I had the same problem as with first 2.
After that I tried 2 noname new tips (KU and D12) which had a lot of trouble even starting to work. New tips are pain in the a** to test because they need a brake in (well, we all know that).
I will be on vacation for the next 10 days or so but I will try to check everything if I can find something... anything... that could give me any kind of clue where to start searching for the trouble.
David, please don't get angry if I will ask you some questions in the next weeks because I would like to help and I need to understand a lot of things before I can give you something back. For the record, I am not dumb, I just don't have experience like you and somebody else but I am willing to try because I would like this to work as it should.
One thing I believe in is that if there are tips that work and some that not work no matter what and if Ksger has a code that works with both of them... I KNOW it could be done even in this FW! 

David, thank you! 

[DavidAlfa]

No problem, but you always forget one detail:
I show the real measured temperature.
Factory firmwares don't, unless the difference is huge.
So the tip might be oscillating 20°C up and down, and you simply won't notice.
So when you see a small 5°C drop and blame my FW, remember that. I could do the same, easy... But this is an advanced firmware, where you can tweak every little parameter.

To get all tips working, you must use the worst conditions.
Otherwise the behaviour will be random, some will do nice, others won't.
I don't want to explain again the delay thing, anyone tweaking the pwm settings will be at it's own!

The tip names are for your own reference, so you know what you're loading.
There isn't any kind of internal database, that would be impossible given the infinite providers of these tips.


For now, just try the latest firmware and confirm the ADC issues are gone

The rest is just PWM timing and PID adjusting.
PID will change a lot with the power, being applied, settings working at 24V will be probabbly a lot slower at 12V, because the power going into the tip is 1/4.But also the sampling interval affects a lot.

[Tugo]

Thank you for the pointers. 
I will first try to understand where things are and how are they put together.
After that I will try to do small changes and try to make my worst tip work 
In the mean time I will ask you some questions just to be sure I am not way off when I will find something. I still don't understand how a good tip can work at 180 C without a problem but a bad tip is almost working (at start it is heating at 100% and when it is getting to the set temperature it is losing power and even y graph on soldering station is showing temperature slowly getting to the set temperature but when temperature is just above the set temperature the temprature is shot to 220 C without putting any new power to the tip - pretty much the same when cooling down from 220 to 180 and when hitting just below 180 it shows about 140 C... not really make any sense and is pi...ing me off - I really think this is a software bug - David, did you have this problem any time or is this just me? Any other user also?).

Measured tempratures: from what I tested with your FW and Ksger I don't really see any different ways of showing temperature compared to a measured one. I don't know how is this on Quicko or on v3.x in Ksger since I have only v2.1. I will not work with oscilloscope in the next 2 weeks or so. Vacation is vacation and I will try to understand things first and maybe do some software changes to see what happens (I will have a controller, 20 V brick power supply with 1 tip and a FG-100 with me     Well enough to learn and try to understand).

[Tugo]

David, I forgot 2 more things:
- after I get home from vacation I can make a video to show you a comparison of your FW with Ksger in regarding of temperature with a good tip
- I have an original Hakko D16 tip that it got the same burn as yours. What I did was grind all of those black thing off and then I crimped it back together (not soldering just hold it firmly together). It still works and I use it a lot! Don't just throw it away.

[wickated]

tested new fw
tip menu works as intended both ways - manually setting adc or throyugh calibration menu
pwm/pid works very good and fast, but somehow it looks a bit wobbly like if adc readout shifts -+ random time. but anyway no critical bugs.

p.s. Chinese tip burnt in 1 "nominal" heatup at 250watt . no way to spend 10$ instead of 30$ for genuine one (in my country they cost 60€)
at least it have some copper

[DavidAlfa]

I still don't understand how a good tip can work at 180 C without a problem but a bad tip is almost working (at start it is heating at 100% and when it is getting to the set temperature it is losing power and even y graph on soldering station is showing temperature slowly getting to the set temperature but when temperature is just above the set temperature the temprature is shot to 220 C without putting any new power to the tip - pretty much the same when cooling down from 220 to 180 and when hitting just below 180 it shows about 140 C... not really make any sense and is pi...ing me off - I really think this is a software bug - David, did you have this problem any time or is this just me? Any other user also?).

Beacause of the thermal transfer. Because of this:
https://www.eevblog.com/forum/reviews/stm32-oled-digital-soldering-station-for-t12-handle/msg3594821/#msg3594821

You'll be very lucky if a tip has close to perfect heat transfer.
Doesn't matter the tip size or profile, the problem comes from inside, the heating element can't transfer the heat so fast.
You'll see how most of the problem is gone when powering from 12V. As you put a lot less power on it, it can effectively transfer that heat in time.
The only fix is to add delay. If you want to keep using 100uS delay, then please don't ask anymore about this issue.

[wickated]

well at least with proper tip internal configuration heat transfer is stable, so u can also refer to unsetled temp. but it need deep understanding of processes inside tip and stability of controller.

[DavidAlfa]

I call it "Don't complain, it's a $3 soldering tip, don't expect NASA engineering in it".
Genuine JBC work much better, but you must think if it's worth the money for a hobbist, ocassional use.
One cheap JBC tip costs the same as the whole T12 station.

wickated, don't buy chinese JBC copies, I already bought 5 or 6... They do much worse than T12s and cost 2-3x more.

[wickated]

wickated, don't buy chinese JBC copies, I already bought 5 or 6... They do much worse than T12s and cost 2-3x more.

sure i just bought it for tests, noticed improper TC voltage curve, and tested curved point tip, found it very useful but not suitable for my work

Genuine JBC work much better, but you must think if it's worth the money for a hobbist, ocassional use.
One cheap JBC tip costs the same as the whole T12 station.

yes but jbc tips work like if they are pure copper, no other plated tips can compete with pure copper.(except metcal, but they have another limits)
i stand with the price of tips but genuine stations like this combo are too much for a real hobby.

[DavidAlfa]

The pure copper thing is just **** , liquid tin slowly dissolves copper, after a week the tip will be gone.
So the tip needs to be iron-coated to prevent that (And the iron coating, chrome-coated where you don't want tin to stick)





The problem is not really the coating, but the contact between the tip and the heater.
Copies are badly pressed, you can see the air gap, so the heat tranfer is terrible.

Some years ago, I had a tip that became damaged after falling to the floor.
The tip end flattened, but I decided to "fix" it with a file.
After a while the tip end became this:



Because I was using "pure copper"