CFW for KSGER/Quicko STM32 Soldering Stations

[DavidAlfa]

To know if it's still hot? Weren't german strong people? Just touch it with your hands and check! 
Maybe I can put it in between the voltage and ambient temperature. I'll make few tests.
It bounces for a reason, to not cause the oled burning. As I later added brightness dimming, that 's no longer an issue.
But I don't want to remove the bouncing thing, I like it! 

[hanzz]

Weren't german strong people?

absolutely

Maybe I can put it in between the voltage and ambient temperature. I'll make few tests.

good Idea

But I don't want to remove the bouncing thing, I like it! 

Why not let the actual Temp bounce...

[wickated]

How are these readings going?

full stock settings show this. on genuine wd16 tip.


maybe ill change transistors later - noticed that another station dont tick on pwm break event.  but im pretty satisfied now - tips now work fast enough to delay transition to jbc tips.

Why not let the actual Temp bounce...

there is only one true choice -(DVD) logo bounce!

[DavidAlfa]

Just as test, try really huge Delay time. 50... 100mS. Leave PWM to 200mS.
These spikes shouldn't exist. Maybe the transistor is getting a huge negative spike, making it to conduct slightly?
If you have oscilloscope,check the 3.3V rails for noise. Also the signal in the heater.
Think that only 13mV will completely saturate the amp send 3.3V to the ADC.
The TC outputs about 21uV/ºC. 200uV noise will be 10ºC noise!. 1mV will be 50ºC noise!

Some people put a schottky diode between GND and the heater to kill the negative spikes.
Ksger uses a led to clamp the voltage! I don't like that. I doubt a led response is close as a fast signal diode.

I also thought in the classic DVD screen saver a thousand times!

[DavidAlfa]

good Idea

However, it's not there for a reason. Usually the lowest reading will be 70-90ºC. With the tip completely cold.
There's nothing that can be done. Below 100ºC, the thermocouple output is almost null.
All you have then is the amp offset and noise being amplified.
And will differ between boards...

Maybe I can make make something like a warning,  "HOT", if the reading is above 120ºC.

[wickated]

Yep, i dont have oscil but i guess there is something wrong in stock transistors causing spikes.
power rail is solid , i placed dual ldo to negate noise.

tested on looong adc delay. result is a bit sloppier power up ramp compared to fast pwm cycle, and idle power % is higher, can go up to maximum power output even for small tips. but also is see advantages - fast max power output, and temp tip looks much more real - it can drop 100c down while in tea. no overshooting and fast switching to idle.
test run:
settings

idle run (power % is a bit higher than usual)

tea boiling - easy go to max power even for small tips

going back to idle - nice curve

[hanzz]

Maybe I can make make something like a warning,  "HOT", if the reading is above 120ºC.

This would be great and absolutly enough
Thank you!

[DavidAlfa]

You have to adjust the delay until the spikes start again.  Find the limit.
It's "real" because it's putting a lot less power on it and can pull heat out as it gets generated. Simple as that.
For 200mS PWM, every 10mS delay you rest 5% of power. 100mS delay will use only 50% of the power. So about 55W.
That's what I said about using it with 30V. The tip will get hot inside no matter what you do, it can't get rid of so much heat.
24-25V... doesn't make a difference. 72...78W. But 110...

The heater has thermal mass. Doesn't need high frequency PWM. The stock FW used 1KHZ PWM.
Why? 5Hz is perfectly fine.
The tip temp is read after every PWM cycle, so  the delay also applies.
if you set 50mS PWM, you'll be subtracting the delay to each 50mS cycle.
So don't touch that because is just useless. Only increase de delay in 10mS steps until you find it does ok.
Then you could make some fine tuning, but always leave 1-2mS extra.

[DavidAlfa]

What about this? Moreover, slightly bigger icons. It's a solder station, it's expected to be hot at any time 
I also added a new option to disable automatic dimming. If you want to burn your oleds its not my problem!

[wickated]

so you say that delay is after pwm cycle, adding idling time, resulting in huge power loss? i thought it was inside pwm duty cycle for better power control.
heater shouldnt overheat with proper pid settings, because termocouple is part of heater itself. even 110watts are not enough to overheat in 0,2 secs.
heater has thermal mass, sure, but thing about pid and so on is about algoritmically detect how much heat is drained from tip end and add power to avoid drops. thermodynamics says that thermal power is affected by temp delta. so its important to detect temp drops as fast as possible and build proper temp/power curve. t12 are definetly not best soldering tips, but are enough powerful and responsive. principles apllied on t12 tips are also apllied on better solder tips like c245. c245 just have better heater-copper alignment, more copper and dumb thin coating.
for example i use W series tips, which have additional 1mm copper around heater. it makes tip end much more responsive. also most of my tips have manually thinned coating - genuine are coated with 0,2mm iron, so i shred off 0,1mm from each. fake ones are usually coated with 0,4-2mm iron, so cant transfer heat at all when you try to solder big copper something.
and yes, i wanna tune my solder station to a metcal level where u just solder everything with one universal tip. its not impossible.

If you want to burn your oleds its not my problem!

its already burnt on mine in like a week cuz 255 brightness by default was too much, and i had no normal sleep - mercury switch is located at the highest end of solder iron so any wire touch actually wake it up.

p.s. retested. fake tips show great temp readouts with any delay, even with 1ms (pwm set to 200). genuine oscilate like crazy.

[DavidAlfa]

I ensure you that the heater reading will be lot if you psh too much power. I ran it at 30V for a year, burned like 10 tips, tried a lot of things.
Finally I started working in the firmware project. You won't believe the hours in testing I've done on it.
Say whatever you want, the heater is not perctecly bonded with the tip, neither is the thermocopuple.
If you push all the power with the tip under water, you'll see bubles coming. You turn off the power and still will be bubbling about 1/2-1second.
That is the residual power. If you still think you can put 110W on it under weater and read less than 100ºC...Whatever, but dont' ask me when things don't go well! 
Big tips do better, yes. K, KF, BC3 tips have thicker tips, but still read >150ºC usually in those conditions. Get a thin tip like ISL and you'll read >200ºC.
To the date, I only had one tip that had a nice thermal bond. I mean it woudl drop to 120ºC.  Maybe original Hakko does better.

Of course, it's done inside the PWM cycle. But the delay time is included in the PWM cycle itself.
So the max duty will be (PWM time - Delay time). You can't measure while the power is on.
You must turn off and wait for the switching noises to dissapear(That's the delay time).

I suspect hakko uses a heater coil with thicker nicrom wire, more turns, thus more inductance and bigger inductive spike.

[wickated]

Yes, it's done inside the PWM cycle. But the delay time is included in the PWM cycle itself.
So the max duty will be PWM time - Delay time.

Now it explains almost everything. oscilations are real, and genuine tips show real heater temp. fake ones have improper heater-couple thermal connection and thus dont oscilate, they show temp not of heater itself but of casing. also i never checked tip wire clamps in genuine tips - caps there are very hard to dismantle.

[DavidAlfa]

Search the thread , there are waveform pictures  of original Hakko vs chinesium.
T12 made a negative reading of almost 20mS before recovering.

[wickated]

Search the thread , there are waveform pictures  of original Hakko vs chinesium.
T12 made a negative reading of almost 20mS before recovering.

oh thanks. i backtracked only 10 pages but ppl actually noticed weird oscilations 20 pages back. on different boards, so OP or diode/mosfet gating isnt root of a problem. i have some guesstimate about genuine tips, need some testing

[DavidAlfa]

If you read carefully, you'll see that a lot of people just touch every setting without knowing how it works.
Also as this is development, sometimes something  sneaks in and causes new issues.
But as I say, if you put a lot power, you'll get oscillation more easily, you will need to reduce PID factors, because 10% of 70w is not the same as 10% of 110W.
You should know these things before starting modding. Lower P term to maybe 30, or 20. There's always a value.
Anyways, understand that I can't solve everyone's problem, this is a free fw, not a JBC customer service 

[mastershake]

trying to get back to work. i have another eye surgery now they need to do so i apologize if someone messages or tags me and i dont respond. but i did flash the newest ksger 3.0 tonight just to mess with it some. one thing is you can set it to shake to wake or stand to wake then there is a shake wake on or off? i would think if you set it to shake to wake then why would someone turn that off? so it never goes to sleep? just something curious. im not allowed to do any soldering for a couple weeks sadly they dont want the fumes to get in the eye.

a few asked me what they had to do and how i was doing (which is very kind thank you) they had to do radiation to my left eye with a plaque they sewed in / onto the sclera to kill a tumor. now there are some other surgeries to follow.

[DavidAlfa]

I will disable the shake wake entry in stand mode as it doesn't matter.
As I said, this only is to disable shake when already in sleep mode.
So you must manually use the encoder to wake up.

It doesn't affect in run mode, the shake input will keep resetting the sleep timeout, preventing it from sleeping.

About your surgery, man that sucks. I hope the treatement works and the tumor goes away.
Stay strong!

[wickated]

changed led to low capacity(10 pf) shottky. delay now stops oscilating on 15 ms instead of 80ms.
nvm. rebuilt all the scheme from the scratch. put wrong amp dividers, got red hot 900C iron  changed to nomial, recalibrated, no osciltaions on default settings. nothing more can be done without oscil, but i guess something wrong is with mosfets. why cant it just work

[DavidAlfa]

That's what you get from these chinesium things.
I put a very low voltage schottky (around 0.15V drop), didn't notice any better than the 1n4148 that was before. But the 1n4148 is very fast, with 4nS switching time. A led it's not.
Are you expecting zero delay? 15mS is pretty nice.
You can guess what the original firmware reads.
Maybe they pause for 100mS, but it's a really crappy solution.
Remember you are amplifying a very small signal where 100uV is 5°C.
You can't expect to suddenly turn it off and voilà! You have the heater and wires capacitance, inductance, also the power supply will for sure make a small spike due the sudden load release.
All that contribute to noise, and frankly speaking, it fascinates me that these boards get stable readings at some point with such careless design.

Just leave defaults, or don't complain!
You are not going to get anything better than that easily.

[wickated]

That's what you get from these chinesium things.

problem is about fake tips work with no troubles at all and genuine show strange behaviour. that just make me upset. by now i guess quicko boards with dual amp was good idea cuz it has filters. but my ksger board(3.3) is also same one newest quicko.
noticed that with iron disconnected i still see pulsations each pwm cycle and that ticking noise by end of pwm cycle. so ill grab some oscil and take at look at mosfets.

p.s. i think by this time i could already have built fully functional jbc board with their normal thermocouple control or even wire resistance controlled something no-need-for-pid-im-40khz-500w-glow-tip-in-2-sec

[DavidAlfa]

I think you need to stop for a moment, you're hearing bells from nowhere.
So you see pulses in the mosfet with no iron and guess it's a transistor problem.
Now the mosfets turn on themselves with fast pulses?
Have you seen the gate signal? Or the output from the stm32?
Pwm never turns off completely unless a critical error happens.
It stays at the minimum possible value, 1, equal to 10uS pulses for every PWM cycle.
That's totally intended. That's how the detection is achieved.
The small pulse charges the amp capacitor.
If there's no load, it will hold the charge when the ADC reads, making a very high reading and triggering the iron detection.
If there's load, it will read normal values again.

You can do whatever, but don't offtopic in this thread if you start doing something like that! Start a new one
This is only for the firmware and their supported boards.
And 40KHz... For heating a resistor...? At that frequency you can make an induction heater

[Il_Marco]

Hi to everybody,
At last I received my spare STMd and found some time work on it.
I'm experimenting on the same Quicko PCB as DavidAlfa.

1 - Changed original STM with Chinese STM32F103CBT6
2 - Flashed FW 7c4d613 for F103
3 - Removed C7, R11 and R15 from PCB. R23 is not there where you put it in the schematic, so removing C7, R12 and R15 completely isolate HANDLE_TEMP  (PIN 4 of Iron connector) from the rest of the circuit.
4 - cut the correct track near U7, above the PCB VIA, as in your MOD, because I'm planning to use C245 Tips.
5 - Wired C19 to C7/R15
Now HANDLE_TEMP (pin 4) has become TIP_SENSOR_INPUT, isolated from HEATER
6 - Soldered a single pin of a 2.54mm pitch strip into pin 5 (HANDLE_SWITCH) of the PCB Iron connector where I can in future connect a "C245 Stand".
7 - Wired pins HEATER (pin 1) and TIP_SENSOR_INPUT/HANDLE_TEMP (pin 4) inside T12 IRON Connector, so that it can work just like a normal T12 Iron bypassing C245 MODs above.

Now my observations, some of them obvious and some a little less:

1 - With no Iron connected, whether T12 or C245,  NO_IRON detection obviously doesn't work because there's no more pcb connection between HEATER and TIP_SENSOR_INPUT
2 -  It still doesn't work with a C245 IRON connected because it doesn't have that wire in the connector with no tip inserted
3 - It works fine with modified T12 Iron but only obviously able to detect missing TIP and no missing IRON.
4 - I tryed some calibration with two different T12 TIPS and while there's no issues in the procedure itself I find some with the Thermometer sensors that are way too fragile and brake easily (non fiscally but in wrong measurements, especially abusing 450°C measurements!). You can find yourself making a calibration and not understanding why 5 minutes before you had got good measures and 5 minutes later much lower measurements.
So my suggestion is to reduce at minimum contact with sensors, just to take your actual temperature, then put your Iron in the stand while adjusting the calibration data in the station. I'll Come to a suggestion later for you DavidAlfa (just a try, I know you have already done way much to the project with your loooong work on it).
Here my calibration for T12
BL TIP
1439
1951
2723

K TIP
1447
1968
2490

5 - I then tryed with C245 IRON...
My 72W power supply in the station can't stand their current requirements and the station keep restarting. This is with Power on option set to RUN, otherwise I found no way to start the station in SLEEP mode and then exit from SLEEP with some sort o Encoder button pushing or rotating. Maybe I'm missing something here? Is there a way to manually exit from SLEEP the same as entering it whit press and CCW rotate the encoder?  Pressing and rotating CW is for selecting TIPS, I see.
6 - I made a try with my PC power supply, 19V and 4,74A. That made it (but then I think I need a new PSU for my station anyway), I could make some C245 test.
6 - Tried to calibrate two identical ORIGINAL C245 TIP and, apart from Thermometer sensor as above, I could do it. Maybe in some way it was even easier.
TIP C245-944, Flat tip
1624
2199
2727
This values worked fine with the second identical C245 tip (good)
That is it for now. But as far as Tips calibration and Tips data saved I have some point to ask you and maybe to implement? (If possible, if you are like willing to do it and you think it usefull. And with many thanks).
Let's say add add some tips in the menu,
For T12 I add a K and a BL
For C245 I add -944 and maybe -xxx
Once you finish a calibration for each tip you get the calibration values
1111
2222
3333
You can see them just after the calibration.
Then why shouldn't I see these same data and shouldn't I be able to lightly modify them in the CALIBRATION, ADJUST menu for each Tip? I have a certain tip selected with it's own three calibration data, I think it would be much useful to be able to see and edit them every time I need it. Maybe to check them at the end if calibration if I have not taken note of them, or to fine adjust them from time to time without going through oll calibration process and so saving THERMOMETER sensors.
In CALIBRATION, ADJUST menu of each different TIP I always read the same generic T12 and C245 data, not much help.

Sorry for being so long and take my notes just as what they are, notes!
Once again, Thank You very much for you work. Thanks to everybody.

Here is my PCB after all MODs

[Il_Marco]

I Will add a 1Meg pullup resistor to TIP_SENSOR_AMP input (now pin 4 of Iron connector, ex HANDLE_TEMP) to solve every NO_IRON Issues for both T12 and C245. I forgot that some Ksger unit has it (910K for the one I own).
Then I should retry calibrations and see how it will affect its values.

[DavidAlfa]

Hey!
Of course, if you isolate that pin you lose detection  .
What I do is to join the pins inside the GX12 connector.
For the T12 I join heater+TC, and for the JBC I put a resistor.
That way you can change them on the fly.

But 1M is too high.
Since the minimum output is 10uS, you need to set a resistor that can charge the capacitor (was it 1nF?) in that time, otherwise it will start bouncing.
I don't remember right now what I had put there... I think it was around 100K?

About the features, I've thinking on such menu for a long time, but as the current firmware is using almost all the flash in 64KB devices, It'll be no easy.
I might end making a v2 firmware for 128KB devices only.

For manually waking up the station from sleep state, just enable button wake in the options. Then make a click to exit sleep.
(Entering menu or selecting tips won't do it).

Yes, the cheap K sensors are very brittle. When doing calibration don't cover the whole thing with solder.
That can short the wires and cause weird readings.

I usually fix the sensor in place with tape and put the handle on the table, touching the sensor.
Then add a little blob of solder and make the whole calibration without moving anything.

And yeah, asking 220w to a 72w supply usually won't work!
Try reducing the max power on the menu, first to something low like 20w, then slowly search the limit.
That will make shorter pulses and might avoid the power supply overcurrent detection (although it may, or not, go ka-boom! )
My Quicko has a 120w supply, the voltage ripples a lot with such loads, but everything works (goes down to 19V or so).
The stm32 supply stays clean so nothing resets.

I'm currently at the hospital for a surgery, so there won't be firmware updates for at least 2...3...4 weeks. It all depends on how it goes, wish me luck!

[mastershake]

hope it all goes great! i just finally can come back and you are heading in. we will see you soon enough im sure, good luck!