CFW for KSGER/Quicko STM32 Soldering Stations

[DavidAlfa]

If it's a ch32f (not related to the cks) then yes, somebody already had issues with it.

Edit : Also, actually the CH32F103C8 has only 64KB flash, so it's not compatible.

[wickated]

Edit : Also, actually the CH32F103C8 has only 64KB flash, so it's not compatible.

but there are very old builds that fit in 64kb space ?

[DavidAlfa]

Yes, but the ch32f had such issues anyways, and the datasheet is chinese only.
Clearly something is wrong with the ADC and/or the DMA, but go figure.

[ygi]

He's better off swapping this counterfeit crap for a real STM32. They now sell for 20x what they're worth but all things considered it's a lot less hassle for "only" ~10$.

[DavidAlfa]

It's the CH32F which makes the trouble. The CKS works perfectly...

David, thanks. Now everything is fine, no temperature jumps occur. 

I saw your post at Radiokot about the new display. Try this, I relaxed a bit the I2C timings.
There're 2 files: One is the normal firmware, the other makes a FPS test to show the current screen throughput.
Another option would be to set lower I2C pullup resistors, like 2k2, to make the signal stronger.

Also, if you soldered wires to the display, the signal will have more issues, like crosstalk:
https://axotron.se/blog/crosstalk-problems-when-running-i2c-signals-in-a-cable/

Twisting a gnd wire with each SDA/SCL signal, or even simple interleaving (signal-gnd-signal-gnd...) will help:


But keep the wires shortest as possible.

Edit: File removed, please refer to this post

[Polf]

David, thanks! Didn't expect you to answer this. I used the firmware from the KSGER_V2_I2C_Timing_Test archive. The image has improved slightly. Then I installed pull-up resistors for SDA and SCK, with a nominal value of 2k2. In the display itself, 4k7 resistors were already installed, I did not remove them.
The image is now stable, but sometimes there are slight glitches. This can be seen in the video when the soldering iron cools down in sleep mode. For example, at a temperature of 179 and 100 degrees.
Video:
https://drive.google.com/file/d/10-XvzIZxIF8Xr-7NKQLV1ecxPNybR5xY/view?usp=sharing

[wickated]

is this screen so slow or just set to 0,1 fps?

[DavidAlfa]

Leave the standard resistors for now, I want to make it work without modifications.
Can you post pictures of the connections?

Wickated, really? You should already know this.
Standard PWM/ADC  is 5Hz, the display values won't update faster than the ADC.

[Polf]

Remove the 2k2 resistors (4k7 resistors will remain) and shoot a video of the display?
The display is connected without long wires.

[Polf]

The firmware KSGER_V2_I2C_Timing, the 2k2 resistors are removed, the 4k7 resistors remain.
Video:
https://drive.google.com/file/d/101dvsNCqSw53iQEUXo2pdfUXcV34vfnE/view?usp=sharing
or
https://disk.yandex.ru/i/x4AzTPtzD_qymA

[DavidAlfa]

Lets go. Made some additional tweaks and delay adjustments.
There're only a few versions to try  .

Code: [Select]

KSGER_V2_I2C_Test_H10_L01_140fps.bin
KSGER_V2_I2C_Test_H10_L04_128fps.bin
KSGER_V2_I2C_Test_H10_L08_92fps.bin
KSGER_V2_I2C_Test_H14_L01_102fps.bin
KSGER_V2_I2C_Test_H14_L04_92fps.bin
KSGER_V2_I2C_Test_H14_L08_83fps.bin
KSGER_V2_I2C_Test_H18_L01_91fps.bin
KSGER_V2_I2C_Test_H18_L04_83fps.bin
KSGER_V2_I2C_Test_H18_L08_76fps.bin
KSGER_V2_I2C_Test_H22_L01_82fps.bin
KSGER_V2_I2C_Test_H22_L04_76fps.bin
KSGER_V2_I2C_Test_H22_L08_70fps.bin
KSGER_V2_I2C_Test_H22_L12_65fps.bin

The versions are called H--_L--, where H is the signal rise delay (the slowest part on the I2C bus  because the resistor limits the slew rate), and L is the fall delay (Much faster because it's actively pulled down).

Start whith the higher: H22_L12. If that doesn't work, don't bother trying the rest.
If it works, try H22_L08. If it keeps working, then try in this order: H18_L08, H14_L08, H10_L08 (keeping the highest L).
Stop when it starts failing or there're no more files, note down the lowest "H" that works.
Now, strat trying the lower L versions, keeping the H version that worked.
Ex. If the lowest working H was L14, now try H18_L04 and H18_L01.
The lowest, the faster the display can be updated.

This not about refreshing the screen at 2000fps, but to refresh the screen when required using the lowest cpu time.

This firmwares only run the screen test, which updates the screen as fast as possible, so the errors will be detected easily.

I know wickated is going to try the 140fps version right away.. 

Edit: File removed, please refer to this post

[Polf]

Display does not work even with software H22_L12. 

[DavidAlfa]

Take a picture from the rear of the oled module and the current connection in main board.
Could be any little thing...

More tests:

Code: [Select]

KSGER_V2_I2C_Test_H30_L08_61fps.bin
KSGER_V2_I2C_Test_H30_L16_48fps.bin
KSGER_V2_I2C_Test_H40_L08_52fps.bin
KSGER_V2_I2C_Test_H40_L16_42fps.bin
KSGER_V2_I2C_Test_H50_L08_44fps.bin
KSGER_V2_I2C_Test_H50_L16_38fps.bin

It's like this one?

[Polf]

Attaching photo. The display is switched to the I2C mode:

• 0 Ohm resistor moved from position R8 to R11 (connects pull-up resistor R7 to SDA)
• installed resistors 4k7 in positions R9, R10, R12
• removed linear stabilizer Q1 3.3 Volts, replaced by a jumper
• installed tantalum capacitor 4.7 uF in position C7

[DavidAlfa]

What about the DC pin? Should be connected to gnd, not floating.
That display has very weird gnd traces. The gnd plane is connected only in the pads where the oled is hold, then everything else is routed to gnd using thin traces.

[Polf]

I tried all the firmware, the display does not work normally. But even the firmware from the github works with 2k2 resistors. DC is connected to the ground.

[DavidAlfa]

Do you have oscilloscope? Watching the signals would be nice.
Try also 1K or 470ohm resistors, it won't cause any harm.

Check this:

[Polf]

I didn't say they are connected to the VCC, they are actually connected to ground.

[Polf]

David, I watched the video, you gave the link to it. I noticed that all resistors that are installed in positions R9-R12 have a nominal value of 0 Ohm, and not 4k7, as I installed according to other information on the Internet. Comparing the two wiring diagrams of this display for SPI and I2C, I noticed that there are no R9-R11 resistors in the diagrams. But on the connector 24 for the I2C version there are jumpers on pins 5-6 and 14-15. This means that it is correct to use 0 ohm resistors and not 4k7. I replaced the 4k7 resistors with 0 ohms. The image became more stable (firmware 21-10-12a with github), but I still had to install 2k2 resistors for SDA and SCK. Now I do not see the image slowing down. If this happens, I will change the SDA and SCK resistors to a value of 1k. Thanks for the help!

[DavidAlfa]

Give another try to the tests to find if I can optimize the timings further.
So no image corruption with 2k2?

[M-Reimer]

Hi together,

I'm looking for some "cartridge based" soldering solution for quite some time already. Just had too much soldering jobs lately where my old Weller was not up to the task with big ground planes and non existent thermal relief.

After some searching I found these cheap STM32 based stations. Seemed to be a good choice for the price and should perform better, right?

To benchmark I used a big copper plated PCB like in one of the EEVBlog videos where Dave did the same test. My Weller on 300 °C can't really solder on this. I can just pile up some solder. But neither can the T12 clone... The display keeps telling me that I am still at 300 °C but I can barely get solder to melt.

So what could I do next? Does the custom firmware improve the regulator to a way that it performs better in challenging situations? Do the T12 clone tips suck and I should get legit Hakko ones? Or is it better to try a JBC handle on the STM32 station with the CFW?

[DavidAlfa]

Soldering ground planes at only 300°C...?  It's very low for most soldering stations.
Maybe a genuine C245 tip could do that.
Improve what? If the tip has bad thermals, there's no way to know what the real temperature is, the internal sensor reads 300°C.
These jobs require 400°C when using T12 tips. Most tips can't transfer enough heat, so you need to compensate by increasing the delta (difference between hot and cold).
The board won't see 400°C...
Also, It heavily depends on the tip model. Don't expect Soldering ground planes with a small or thin tip.

Edit: and if it's lead-free...much worse, as the melting point is around 235°C instead 180.

[Polf]

With 2k2 resistors in SDA and SCK, the following firmware works fine:
KSGER_V2_I2C_Test_H10_L08_92fps
KSGER_V2_I2C_Test_H10_L04_128fps
KSGER_V2_I2C_Test_H14_L08_83fps
KSGER_V2_I2C_Test_H14_L04_92fps
KSGER_V2_I2C_Test_H18_L08_76fps
KSGER_V2_I2C_Test_H18_L04_83fps
KSGER_V2_I2C_Test_H22_L12_65fps
KSGER_V2_I2C_Test_H22_L08_70fps
KSGER_V2_I2C_Test_H22_L04_76fps
KSGER_V2_I2C_Test_H30_L08_61fps
KSGER_V2_I2C_Test_H30_L16_48fps
KSGER_V2_I2C_Test_H40_L08_52fps
KSGER_V2_I2C_Test_H40_L16_42fps
KSGER_V2_I2C_Test_H50_L08_44fps
KSGER_V2_I2C_Test_H50_L16_38fps

firmware does not work (image is distorted):
KSGER_V2_I2C_Test_H22_L01_82fps
KSGER_V2_I2C_Test_H18_L01_91fps
KSGER_V2_I2C_Test_H14_L01_102fps
KSGER_V2_I2C_Test_H10_L01_140fps

[DavidAlfa]

Ok, one last try, slightly increasing H time to have some margin.
This might become the new timings, so other ksger v2 users, please try it...
For me it works with the standard 10K resistors, no modifications.

Edit: File removed, please refer to this post

[M-Reimer]

Soldering ground planes at only 300°C...?  It's very low for most soldering stations.

That's the video I was talking about: And Dave actually used even less temperature in his test.

Maybe a genuine C245 tip could do that.

Maybe. Currently I don't really know where to invest my money in. Completely give up with T12 tips? Get more of them? Probably genuine ones? So far I have the two tips that came with the station (T12-K, T12-BC2) and one T12-D08 I ordered from a different supplier. All "non-genuine" tips. And in my test the T12-D08, that just arrived, already performs somewhat better than the T12-BC2 I did the initial test with.

So far I used a Weller WS 81 station including several tips. And at least from my experience the idea of having a really short tip just sitting flat to the heating element with a pretty tiny contact surface seems to be not the best idea. In almost all other similar heat transfer jobs, you would put some thermal grease to such a joint but I think this will get very messy if it is used in this situation.

Improve what? If the tip has bad thermals, there's no way to know what the real temperature is, the internal sensor reads 300°C.

So far I have not changed my "china station" at all. It runs stock firmware. And it makes me a bit suspicious that it always shows 300°C no matter how big the "heatsink" is. So either the tips are that bad or the firmware "hides something". I expected at least a small dip in temperature and probably a small overshoot once the tip is removed from the heat sinking surface. Will replacing the stock firmware actually bring in a better temperature regulator or can I expect the same performance as with the stock firmware?

These jobs require 400°C when using T12 tips. Most tips can't transfer enough heat, so you need to compensate by increasing the delta (difference between hot and cold).

That's exactly what I really want to get rid of. Unfortunately the last board that was really difficult is properly mounted in a case, now, so I can't take a picture, but there I always had to vary the temperature. Somewhere above 400°C for the ground connecting pads and then back to 300°C to do the pads with proper thermal relief. And no, I did not always remember this process and once you touch a "non heat sinking" pad for a second too long you already damaged the board 

Also, It heavily depends on the tip model. Don't expect Soldering ground planes with a small or thin tip.

So we are back to the question about where to spend my money at. Get some wider T12 tips?

Edit: and if it's lead-free...much worse, as the melting point is around 235°C instead 180.

I use lead free solder with a pretty big amount of silver to get the melting point down.