CFW for KSGER/Quicko STM32 Soldering Stations

[DavidAlfa]

And these are the real numbers!
So they can take 80W while heating up, but they can't transfer so much power!
The record is 28W for now
Testing a big tip like D52 will be interesting, also genuine Hakko vs "original" Ksger!

[maciek012]

And these are the real numbers!
So they can take 80W while heating up, but they can't transfer so much power!
The record is 28W for now
Testing a big tip like D52 will be interesting, also genuine Hakko vs "original" Ksger!

I didn't mention it but I had max power set to 120W as it's what was stated in specs for the station.
I changed it back to 80W and tested again with D24 tip, and the PWR goes up to around 50-51%.

[DavidAlfa]

Weird. The code only uses the limit when the max possible power is higher than you limit.
If you some math you'll see that 24²/8=72W, so you could set 80W or 500W, it wouldn't make a difference, the power limit would never be applied.

You can see this in the debug screen.
Enter the debug screen, set ex. 350ºC and plug a cold tip while watching the PWM value, it will max out when heating a cold tip, also PWR should be 100(%).
Try this with 80w and 120W, the values should be the same.
But if you set 60W, you'll notice that PWM is now lower.

My power supply outputs 25V, so the power is a little higher, 78W, I can see the limit working when setting 75W.

Are you letting the tip to fully heat up before doing the tests?
When the tip has just heated up, the tip body will still absorb heat for a while.
You can see this: The temperature will reach the setpoint, but the power will be still 15-20%, and slowly decreasing.
When the power stays at 5-10% and is no longer decreasing, that's when you should do it.

[maciek012]

Sorry, my bad, I didn't change the tip in the firmware back to D24. You are totally right heating cold -> 350 PWR 100 and PWM 40000 in both cases

[DavidAlfa]

Haha no worries.
But anyways, you should know that the power is calculated using the actual limit as base.

Let me explain:
If you set 500W, that's not the real limit, the firmware knows you're asking for more than the system can output.
Ex. for 24V, 8Ω, max 75W, if you set 500W, the limit will be still 75W.
However is you set 40W, the limits will apply.

So if you see 100% power, that's 100% of the actual limit, 75W, 40W, whatever.

You could set 5W, the power would be 100% all the time, yet the tip will take forever to heat, perhabs achieving 300ºC after 10 minutes.

[wickated]

Check your eyes! (Hot air is coming  )

The PWM timer uses 5us clock (each step is 5us). For example, for the default 200ms, the timer period would be 40.000 (40.000*5=200.000us=200ms)
PWM is the value compared against the timer counter that generates the duty cycle. So for 50% power, the PWM value would be 40.000/2.
The timer counts always up to 40.000, the pwm output is enabled as long as the counter is below the compare value.
So in this case (For 50% at 200ms), the pwm would be enabled until the counter reaches 20.000. In that moment the pwm is set low, and the timer keeps counting up until 40.000.
Then, it resets and a new cycle starts.

As you rise the PWM frequency this value will be lower, ex. 10ms would be 2.000, 50% would be 1.000.
You can esily see how the pwm loses precision as the frequency increases, although you have a broad adjustment range in any case, from 0.05% at 10ms,  to 0,0025% at 200ms.

ah that somehow explains why at 500+hz pwm is uncontrollable. its all about amount of digits available. lowfreq pwm bothered me cuz i know that proper mosfet can run up to 10khz without gate shutdown driver. anyway i was bothering only cuz mosfet prefer short pulses of load.

for my setup 10/0,1 x2 i got ~960-980 pwm numbers at 100% power

for wd52 as i remember i got up to 60w real heat drain
jbc 3mm high heat chisel can throw up to 80 watt in copper plate
jbc 10 mm shovel can drain up to 100 watt pure power in copper plate

[DavidAlfa]

Not only that, as the frequency increases, also does the noise, the filtering becomes less powerful because you're feeding the data much faster.

[hanzz]

(Hot air is coming  )

Really??? Tell us more! This would be great 

[DavidAlfa]

Noo  this was a joke someone told me here after I answered something "a bit" harsh (something like "You have the hot air without soldering station").

There's no way I can do that without heavily modifying the hardware, which is not the target of this project 

[DavidAlfa]

I started to modify the calibration to compensate the amplifier offset.
That should fix the temperature offset displayed in a cold tip and the setpoint accuracy below 250ºC

[wickated]

that was tlaked about. on ad8605 i have zero offset with calibrated 250/250 jbc tip
but i think that this i due to usage of shitty t12 tips that have huge difference in TC temp and tip end temp. ~40c offset. so i think that if ppl calibrate not the actuall tip end but heater outer case temp readings would be better.

[DavidAlfa]

Nope, you plug a completely cold tip and that's all, no measurement needed, it'll be the same as ambient temperature.
You know a TC can't generate any voltage if cold and hot sides are at the same temperature.
So the ADC value will be be offset from the amplifier (ex.360)
Knowing that, you map  the ADC values, ex. 360 will be used as the min. ADC value for calculation, being the ambient temperature. Not 0 as before.
It's done, working nice, needs more testing.
This gave a lot more work than I initially thought, as everything depends on everything.
Also I discovered few bugs here and there, mainly affecting performance and conditions I wasn't using before, few tweaks everywhere...

[levi]

You know a TC can't generate any voltage if cold and hot sides are at the same temperature.

Hi, it's not true,TC is always generating voltage, depending on temperature voltage will increase or decrease...
In turn Peltier elements behave like you described :-)

[DavidAlfa]

Peltiers behave exactly like I said, they are thermocouple devices
They generate voltage when there's a temperature gradient between the cold and hot sides.
Not just because they are hot ot cold. If both sides are at the same temperature, they don't generate anything.
I'm no longer wasting my time arguing these matters. 

[wickated]

there are some not bugs but ux features not impleneted like default temp step is 5c , if i change it (ex 10) in system it wont affect iron min/max/user temp steps, they will step in 5c. i know it require lot of linking so its not smth to bother.
cold tip offset detect would be awesome. this should fix low temp accuracy. also i suggest bigger time for stand check. cuz even with small RC it can run-stand when i accidentally catch cord.
and i still demand adjustable on/off feature for DISPLAY temp negation. if u select graph widget it shows everything, but normal temp display could be "sticked" cuz 1-0-9c jumps are a bit annoying. and yes this should be selectable so u should turn it on only when everything is tuned.

[wickated]

I'm no longer wasting my time arguing these matters. 

i think u definitely MUST not talk in such a rude way to yet another nobel prize winner who finally invented perpetuum mobile generating voltage out of pure nothing just by reading wikipedia

[DavidAlfa]

Not a bug, I'm completely aware of that behaviour, and it's how it's supposed to be.
500ms is plenty of time to debounce the stand, if it changes easily, then it's your handle not making proper contact with the stand...
You don't need any RC, the system ignores the input until it becomes stable for at least 500ms.
You could keep quickly inserting and removing the handle for hours, it would keep the current state.
I don't think I was rude, just what it is, people come from blinking 4 leds and think they're mastered electronics.
I just can't stand getting dumb advices, feels like telling a mechanic if he did check under the hood

You say completely turning off the screen when sleeping? Not hard to do.

I've been fighting the calibration algorithm, finally understood it (was the last part I never took much time to analyze) and spent the whole afternoon trying to find the best calibration method.
I always have small changes and little 4-5°C offsets, the lower temps keep losing accuracy even with the cold calibration.
I have yet to find out if this is caused by the NTC itself (you cannot perfectly compensate it in the way it's built), or what.

[levi]

i think u definitely MUST not talk in such a rude way to yet another nobel prize winner who finally invented perpetuum mobile generating voltage out of pure nothing just by reading wikipedia

I am glad for you that you have born already knowing everything, and do not need  Wikipedia. I belong to others, who learn from mistakes.

[levi]

I don't think I was rude, just what it is, people come from blinking 4 leds and think they're mastered electronics.
I just can't stand getting dumb advices, feels like telling a mechanic if he did check under the hood

Lots of people can't stand of different things, go drink some sedative to handle things. If you wrote two lines of code it does not allow you to treat and judge people or be prepared to be judged.
If i don't know something it does not allow you to speak with me like this, i can tell 1000 things that you don't know, but i do not treat you like you do.

[DavidAlfa]

I just said I'm gonna stop arguing when I see that kind of stuff, I see this too often, everywhere.
I understand your point ,sorry! But it was nothing personal against you dude! 
Otherwise ... let's say I'd like to have the same filters I use in the firmware

Added the option for shutting down the screen while sleeping! It's done automatically after dimming, can be enabled in the menu.
New build is close, maybe tomorrow, I'm still with the calibration.
It's pretty ok,  I just can't find why it always has a small drift after calibrating. Maybe I'm losing accuracy somewhere.
I already changed everything to use 0.1°C precision, which helped a bit, but still having 3-5°C off.

[wickated]

You say completely turning off the screen when sleeping? Not hard to do.

i ment actually thing that screen temp doesnt change unless it bounce out of ex.3c bracket, so digits dont jump(like in genuine jbc screen doesnt change temp shown until it jumps/drops 5c) (but everything else is calculated inside as usual). i know i can switch screen to 1fps mode, but it looks weird a bit. and also know that u added 0,5c for better rounding. that  might just help ppl be confident everything is fine.

also switchable screen off is cool too. i have things attached to station power supply so i definetely know its plugged in.

what about ntc - when i changed it to freshly bought 1% precision with known datasheet i think i have much better ambient reading.

[DavidAlfa]

I'm not that much ***hole.  Check this out! (Warning, terrible music! )

[DavidAlfa]

So I added that too. But I don't think I'll add an option for it.
If the temperature is within +-5ºC from target, it'll show the setpoint value.
The plot graph will behave as usual, showing the real temperature history.


I'm thinking in also adding dimming while in run mode, as the 90% of the time you'll be soldering, not watching the screen.
Edit: Added the option. In any case the screen will only turn off if in sleep mode and iron temp <100°C.

[wickated]

huge thanks, ill take a look as soon as possible, cuz currently i have 24/7 workload again for next week.
also i wanna say thanks again for great job for writing and adjusting all that filter/pid/everythingelse stuff, cuz i just built new station from scratch with unfiltered PSU, just wrote adc numbers and it work like a charm. and oversshots from cold start at 150w to 350c only for less than 10c. while i remeber SteveG vid about pace ads200 that overshooted like 50c at same condition lol.

[DavidAlfa]

Remember that the overshoot is not real, happens only at sensor level, so nothing to worry about.