CFW for KSGER/Quicko STM32 Soldering Stations

[DavidAlfa]

That amplifier provides a gain of ~50, far away from  the 233 from the tip amplifier.
The adc resolution would be very low and catch a lot of noise, as with that low gain, it almost equals to 1bit/°C.

You would need to modify the gain by replacing amp resistors.
Why so much drama for cutting a trace than can be joined again very easily?
You can still use T12 this way, only needing to join the heater and the Jbc pins in the GX12 connector

[DavidAlfa]

edit: i have seen the update, working perfect.when in debug menu " Error In: graphics/gui/widgets.c" line:274

Hey, I'm migrating the graphic system from uGUI to u8g2. The latter is in active development and has much more options, fonts...
But the picture format is different (XBM). Can you upload the boot picture so I can convert it?

[jesusvallejo]

edit: i have seen the update, working perfect.when in debug menu " Error In: graphics/gui/widgets.c" line:274

Hey, I'm migrating the graphic system from uGUI to u8g2. The latter is in active development and has much more options, fonts...
But the picture format is different (XBM). Can you upload the boot picture so I can convert it?

here it is  , remove .c from the logo.kra.c to have the krita file in case you want to edit anything.


Also o another topic, the stm32m3discovery arrived and attemted the attack but i was not succesfull, i removed the stm32f07 from one of my quicko and replaced it with an stm32f103 , flashed last version and lock the flash with level 1, removed for the moment the cap on boot0 and the resistor on the reset, traced the tx and rx to one pin of the encoder and the other to a resistor that leads to the handle sensor, sonnected everything, followd instructions, but the serial terminal never appears, in the img he has the mcu alone , so i have bought a dip48 to pins(pins are very fragile to try soldering enamel wires)  to try it again, mb i have more luck  then , its aliexpress so it will take prob a whole month.

[DavidAlfa]

New u8g2 library fully implement in the firmware, replacing uGUI.
A brief review of the new gui

[DavidAlfa]

Actual stage is getting stable, and further customization easier every day due more widget changes.
Currently accepting any gui / widgets / fonts suggestions!

Fresh ideas and/or collaboration would be sweet.

Bought new C245 tips, and a pair of cheap T12 for testing.
There's a new C245 seller in Aliexpress, wasn't there last week, they're cheap and they look really nice. Bought the 3.

https://es.aliexpress.com/item/1005001953693376.html

I'm thinking on removing boost and standby modes, leaving only Sleep (power disabled) and Normal modes.
Never saw much sense to so many modes, specially keeping the iron partially turned on when it heats in 6 seconds from cold.
I might add additional options to disable that within the firmware.

Currently thinking on totally disabling the screen while in sleep mode and display a fancy sleep animation.

I'm pretty sure there's no other firmware with such level of customizable options... Actually so many that making a proper documentation is now a must!
The EMA/DEMA filtering looks good, I'm sure the problems were caused the low ADC delay used back then, so I added them back (already seen in the video).

[dreamcat4]

Nice C245 tips, thanks so much for sharing the link. Overall i am really impressed with your GUI is amazing. Really well done.

In terms of my own usage, I often find that nearly always I want to set ont of 2 temperatures:

* 350c for delicate control traces etc.
* 450c when the solder joint is connected to ground plane. Or +Vcc. Or for large connectors etc.

So if you take away boost mode, then can we have a way to select between presets instead? It is kindda the same thing, just not called 'boost'. I guess the real question is how many presets a user needs to access and save? 1, 2, 3. It is an open question. IDK.

The thing about 'boost' is i forget what it means! And I forget how to activate it. It would be easier (for me) to understand to just cycle between preset labels on the screen: T1, T2, T3. And draw a 1px box around the currently selected one. In fact I just do not tend to trust automatic sleep myself. And would prefer to manually switch the iron on and off by clicking a button. This is what I already do ATM. However the power switch is on my Lab PSU.... so then every time I turn the iron back on. I have to wait for the firmware to reboot. And go past the splash screen etc. Which is not so fast.

Rather it would be better to keep the controller booted, and just on/off the tip by using your firmware! Hope that makes sense. If you understand what I mean.

Was also wondering about the C210 handle. Because somebody said on discord they find it is smaller enough (than T245) to feel better at micro soldering. For a finer control under the microscope. However IDK if really worth it myself.

https://www.aliexpress.com/item/4001098984098.html

It is mostly just smaller distance between the fingers and the tip end. Of the handle Which T210 gives you.

[jesusvallejo]

the sleeping station    would be hilarious for the sleeping screen  . Also when we us mortals will be able to compile and flash, i cant wait.

[DavidAlfa]

The boost is enabled by long pressing over the setpoint widget, and stays active for the specified run time.
I never used such, if I need more power, just turn it higher.
It was just my personal liking. Do you guys think it's really useful? I will left it there then.

Jesus, still tons of bugs, why would I upload problematic and dirty code...
Do you like crashing every 2 minutes and funny widget bugs? 

The GUI isn't done either. Trying a lot of different fonts, screen organizations...
Today I found a nice font that fits great and has perfect readability for the iron temp.
The main problem with pretty fonts is that they usually look terrible on aliased, low-res displays.

[cosmin1]

Sorry to interrupt you guys.
I have a problem again with my Ksger board.
After a bad connection to JBC tip, i blewed the 4409 power transistor. I replaced it, now it's working again but the actual tempersture is smaller with 100-130 Celsius degree than the requested temp.
I replaced the opamp too but nothing changed.
What could be the problem???

[dreamcat4]

In a different scenario, i myself observed a similar result when grounding a thermocouple by accident. It interfered with the measurement and the reading became something like -100c lower than it was previously.

As for where in your damaged circuit that is happening... I really don't know! Maybe at least try a different tip. Hope it helps.

[DavidAlfa]

Are you sure it's correctly wired? Was it really measuring correctly before blowing up?

[cosmin1]

I am using the T12 handle with different T12 tips, all measure the same on that board. On the other board the measuring is right.
The wiring is correct, it was never been changed.
I got no idea what is happening... new MOSFET, new op amp... and still having this problem.

[floobydust]

So one board reads the wrong temps? Check the CJC reading from NTC, it can also make tip temperature seem wrong. To trouble shoot, sometimes I inject 10mV and see what reading I get at the handle.

The soldering station connectors and wire are not good for more than a few amps which is a problem at JBC levels. Direct soldering to the board is better. The handle wiring/cable is also thin copper and has two grounds - a power ground for the heater and signal ground for the TC, NTC and switches.

With the AO4409 I would add a fast recovery back-emf diode in parallel with the heater to protect the mosfet and TC op-amp. Check the two gate-drive resistors (1k, 2k) are not flipped. With 25VDC power, should give 10-15V drive. Max is 20V. The other mosfet seen TPC8107 has a gate zener for protection. But I think you might have -ve spike, wimp mosfet, or power supply is sagging too much- if it drops a lot with heat on, it starves the mosfet of gate drive and it will burn up.

[cosmin1]

First time i tried with TPC8107. And had this temp reading problem. I assumed it was not ok to use that Mosfet, so i ordered the propper one, AO4409. But problem was not the Mosfet type.
I think the board should work without modifications, if in the beginning was ok.
If i calibrate the tip to read the propper temp, i have another weird behavior:
https://youtu.be/DRw-xHCXBk8

[DavidAlfa]

I've always liked minimalist designs, keeping functionality, of course
I will definitely remove boost and sleep modes.

The boost mode itself is kinda useless.
If you really need the boost to solder big stuff, it will disable anyway after the timeout, and it will annoying having to set it again.
You can increase the time, but again you'll have to manually disable it when finished.

I hate overcomplicated stuff.
More power? Rotate clockwise.
Less power? Rotate anticlockwise.


To enter  settings, long press.
To change tip, rotate while pressed.
I have not assigned a simple click action yet.

And as I said before, no sense keeping the iron hot when it heats so fast.
So it will be running or in standby.

[floobydust]

First time i tried with TPC8107. And had this temp reading problem. I assumed it was not ok to use that Mosfet, so i ordered the propper one, AO4409. But problem was not the Mosfet type.
I think the board should work without modifications, if in the beginning was ok.
If i calibrate the tip to read the propper temp, i have another weird behavior:
https://youtu.be/DRw-xHCXBk8

In the video, looks like S/W issue at 0% PWM and it's always under the 300C setpoint. It's supposed to turn on the heat for more than a blink.
For wrong tip temp:
Check 3.3V rail is accurate, TC op-amp is not the 621K that seems to give poor results.
Cold Junction Temperature is correct, the handle NTC could be wired wrong, or just switchover to fixed value like 25C.

Firmware shouldn't read the NTC when the heater is on due to the voltage drop in the cable upsetting that analog.
Circuit gain is 250 (249k/1k), never seen 233 where is that from?

[DavidAlfa]

Quicko has 2-stage amp circuit, 16 and ~14.5 gain each.
So the total gain is around 232.

It's getting close to release a beta for testing! The SPI HAL crashes often and I have no clue why, I didn't change the hardware functions.
It just comes to a point where after a dma transfer, the HAL SPI gets stuck in busy state. All the bits checks are done, it's not the hardware being busy.
My guess is that it could be a bug in newer library versions.

I removed the debug screens from normal builds (can be enabled by setting a preprocessor define) as they took almost  1/3 of ram usage.
Now is a little over 9KB, allowing it to run on more basic devices.
The widget part seems to be ready, making everything self-aware of content changing was a little mess, there was always some situation where a widget wasn't refreshed, took old values or wasn't even drawn.
It's great to see the display completely idle while nothing is changed.
U8g2 has it's weird bugs too, certain drawing functions corrupt the whole display if certain parameters are exceed, they don't check them.
Now I'm thinking on doing a temperature graph or something like that.

[cosmin1]

Circuit gain is 250 (249k/1k), never seen 233 where is that from?

What do you mean by 233? It's 249k/1k.
3.3V rail is stable.
I will do more tests, even playing a bit with the op-amp gain.
Wiring is correct. Same handle works ok on other controller.

[DavidAlfa]

Oh c'mon, why's this thread so dead?
Nobody coming with ideas or anything!
It's very easy to sit down and wait for things to come...

Anyways, I'm decided to beat the crap out of these crippled firmwares
I have not set the labels font yet and they look ugly.




I also found the way to add handle type detection without modifying the hardware, by changing the wake/stand pin to analog.
This will also completely fix the iron detection.
No need of adding the op-amp resistor!



By changing the resistor value (potentiometer in the picture) we change the divider voltages.
We have only 2 possible voltages (switch on/off), they will tell us the handle status but also what handle type  is connected (T12, C210, C245...).
If no handle is connected, the open circuit will cause the reading to be very high, that will be no iron detection.

The way C210 and C245 are wired require different handles, so it's ok.

[dreamcat4]

I also found the way to add handle type detection without modifying the hardware, by changing the wake/stand pin to analog.
This will also completely fix the iron detection.
No need of adding the op-amp resistor!

By changing the resistor value (potentiometer in the picture) we change the divider voltages.
We have only 2 possible voltages (switch on/off), they will tell us the handle status but also what handle type  is connected (T12, C210, C245...).
If no handle is connected, the open circuit will cause the reading to be very high, that will be no iron detection.

The way C210 and C245 are wired require different handles, so it's ok.

Hey I really find this awesome idea. Thank you so much. We really needed to come up with a good solution for this.

2021 will be the year of KiCad. Because that software is getting some major improvements.... do you think we can publish these mods as schematics in KiCad? Because that would help for gathering them all together, and editing them in github. It would be really nice I think, to make them available for everybody in a way that is easy to edit.

I have not had time to really understand what you are doing here to get this to work. But I really like it! And will definitely want to try it out 'at some point'. Just I have these other unrelated stuff going on.

Also for this project is nice to hear from Jesus... about whether we still can use a regular MCU like the blue pill to read the protected flash. Or if we must order the special discovery board for that.

Anyhow I really appreciate what you guys are all doing here! Even without enough time personally to participate.

[Il_Marco]

I still can't experiment, still waiting for Aliexpress delivering my GX12 connectors and spare STMs. By the way, Is there a good reason to prefer a STM32F301C8 over a STM32F072C8 when it comes to change the original STM32F072C8 in my board with the new spare one? I'm also considering overwriting the original quicko firmware because,  as far as I can see, the great and hard work on this CWF by DavidAlfa is going to provide us a way better firmware and soldering station! I'm also starting to think that one day or another we will see Quicko soldering stations sold with this CWF installed 

[jesusvallejo]

for the moment the read attack is on standby, the attack does not what its meant to do with the original hardware stm32f3discovery neither with the blue pill port as its the direct "translation" from one bord to another, ive opened an isue on the h3 git but seems op is not willing to help, there is another issue opened a long time ago with no answerd, so when im done with exams ill go deeper into it. There are two problems, the attack consists in loading into ram a binary that enables uart, but in order to be able to read flash, protection needs to be disabled, hence the power attack, that should leave untouched the ram loaded code and give us access to flash, but the uart shell code does not open the uart and continues with the flsh bin execution(first problem),also i have tryed to flash shell bin into flash to see if it works but does not, has no uart output(problem 2). ive tryed also with the provided bins and self compiled , neither work and its strange as the self compiled is much faster , eventhough are supposed to be the same.

[dreamcat4]

I was wondering how are you powering the target mcu? Have you desoldered it, or lifted all the Vcc legs? And do you have a scope to check the power glitch is successful at pulling down the Vcc rail?

[DavidAlfa]

For sure, if it gets good enough, they will at least take part of the code, probably disabling the advanced settings.
Their developing is absolute crap, but, why even bother when it's already done and free?
They are like hienas, opportunists, makes me want to close-source it and request activations for a beer. F** them.

There's no really a big difference between these, else than higher speed and more memory.
Both are more than capable for the current fw, now more than ever, as it basically idles almost all the time, unless you board is crap and can't use hardware dma.
But why bother so much when you can buy few chips for little? The firmware itself is rubbish, specially the Quicko one.
Unless you do it for the challenge, which I totally understand, I've trashed countless hours on such things
If you want to power glitch it remember to remove or reduce the board capacitors, or they will filter the glitch out!

Dreamcat, that isn't a schematic itself. Just 4 resistors! haha.
But I agree to use a more standard schematic software.
The circuit is pretty simple, instead using the handle switch to short the sense pin to ground, it does it through a resistor, making a voltage divider with a known voltage.
If you put a parallel resistor, you will have always a voltage divider, but with a parallel resistor to it when the switch closes.
 Again, the voltages are known, so you can perfectly know what handle type is plugged in, and the switch status.

[dreamcat4]

yeah for the wiring between the board and the various different connectors. A schematic then gives the way to do the wiring. Regardless if the mod is on board or off board. Does not matter. Also kicad  can work with 3d models. Which might be helpful. As are other possible mods.

For the copy protection: The main benefit (to this project) is for those in the ex soviet states, on other forum such as Radiokot. We want to encourage them to participate in this project. But they are all prefering the chinese firmware over there. So (for them) being able to backup before flashing is really important. And we will do better to include that other group of people, they have very good knowledge and expertise. Much help is from Russians.

But even without, its still very useful for many other devices. Since STM32 is widely used in industry. And appears in many products. Therefore the ability to read and backup original firmware from other types of devices. That is a very useful capability. As a part of right to repair.