CFW for KSGER/Quicko STM32 Soldering Stations

[wickated]

.SYSTEM_OledOff = "Oled power"
.SYSTEM_OledPwrOff = "Pwr off",
.SYSTEM_Run_Dimming = "Runmode dim"
stands for ?

[DavidAlfa]

Turning oled off in sleep / dimming the brightness while running.
It's not finished yet, 1st is duplicated, forgot to delete it.

[AndyC]

Swedish translation of the operation manual underway...this is taking some time.
Meanwhile, I tried the newest changes. Compiler complains about "CalNTC" not being declared etc. More work to be done I guess?
The ADC values for cold tip on my KSGER 2.x and 3.x are 0 and 2, so I don't know where 350 is coming from.

[DavidAlfa]

I haven't submitted all the code yet, tempsensors and calibration screen are pending, will be updated when ready.
The cold tip reading is provisional and needs feedback.
You can't properly see that value in the existing builds, the debug screen won't go down to 0°C (now it does), so a completely cold tip will get some power (50°C)
You can't trust an instant reading as there's a lot of filtering, so the proper way is set 0°C, insert a cold tip and wait few seconds for the average.

[wickated]

.SYSTEM_OledOff = "Oled power"
.SYSTEM_Run_Dimming = "Runmode dim"
stands for ?

if OFF stands for working display, ON stands for display off
SYSTEM_OledOff  = "Откл.Экран"
if ON stands for working display, OFF stands for display off
SYSTEM_OledOff  = "Пост.Экран"

.SYSTEM_Run_Dimming = "ЗатемнЭкр"

[pupkinv]

if ON stands for working display, OFF stands for display off
SYSTEM_OledOff  = "Пост.Экран"

Not entirely clear
Since it will be in the screen control menu, it seems to me to use "Пост. вкл"

[DavidAlfa]

Stop sneaking into the code you *** impatients!
I'm changing things back and forth all the time, searching the best layout. Those labels no longer exist!
I'll tell you when it's ready

Edit: Done with the screen dimming. Check these strings:

Code: [Select]

SYSTEM_Oled_Dim            = "Dimming",               -> Uses dimMode options
SYSTEM_Oled_Dim_SleepPower = " Slp pwr",              -> Screen power in sleep mode, uses existing Off/On options
dimMode                    = { "OFF", "SLP", "ALL" }, -> Dim mode: Off, only in sleep mode, always (I already translated this, ensure it's ok)


Also:

Code: [Select]

CAL_InsertColdTip = "Insert cold tip"
But I'm not sure if I'm keeping that option.
Using the cold value seems to break low temperature calibration in the opposite way, hotter than the setpoint, more testing needed 

[wickated]

Code: [Select]

SYSTEM_Oled_Dim            = "Dimming",               -> Uses dimMode options
SYSTEM_Oled_Dim_SleepPower = " Slp pwr",              -> Screen power in sleep mode, uses existing Off/On options
dimMode                    = { "OFF", "SLP", "ALL" }, -> Dim mode: Off, only in sleep mode, always (I already translated this, ensure it's ok)


i guess "Slp pwr" should be better named "scrnsaver" or smth with On means screen off in SLP state

[DavidAlfa]

I've been working with the calibration until now.
Definitely, the main error source is the compensation.
This controllers can't have a lot of accuracy due the design, in fact it seems that using the internal temperature is a little better than the NTC in the handle, which also have noticeable tolerances.

After hundreds of tests, I think I've made a decision.
I will use only 2 point calibration: 250 and 400°C, which are the range of temperatures a soldering iron usually works at.
Below 250°C, I don't care to much.
I removed the cold calibration in the end, it breaks more than it fixes.
The sensor response isn't exactly linear, it's slightly curved, but the calculation is linear,  so at the middle it has a little error. The curve changes slightly between tips, too.
I don't think it makes any sense making a complex and painful calibration for what it is.
You'll set 350, and if it's actually 355 or 345,  it won't make any difference.
There's a limit beyond the practicity and I think this was it.
Next builds will tell if this is a good idea or not!

[ricktendo]

I could help with Spanish translations, but Spanish is a second language to me and I would not know the technical definition to some stuff.

[DavidAlfa]

2-point calibration makes the calibration much easier and faster... I like it.

Calibration test :

Target      Measured
175         165
200         194
225         223
250         251
275         277
300         303
325         328
350         353
375         378
400         402
425         426
450         449
475         472

Below 200ºC the curve changes a lot, as expected.
As the handle heats up or cools down, the temperature drifts 3-5ºC, regardless of the NTC compensation.
I think 3-4ºC error in the whole usable range (250-450ºC) is more than ok, I doubt professional soldering stations have so much accuracy either.

ricktendo, I could do it easily, I've already tried, but spanish words are too long.
I experimented with sliding text, but I wasn't convinced about the result.

[DavidAlfa]

Code: [Select]

SYSTEM_Oled_Dim            = "Dimming",               -> Uses dimMode options
SYSTEM_Oled_Dim_SleepPower = " Slp pwr",              -> Screen power in sleep mode, uses existing Off/On options
dimMode                    = { "OFF", "SLP", "ALL" }, -> Dim mode: Off, only in sleep mode, always (I already translated this, ensure it's ok)


i guess "Slp pwr" should be better named "scrnsaver" or smth with On means screen off in SLP state

I changed it to " In sleep". So it appears like this:

Dimmer           [OFF, SLP, ALL]
    Delay          [10s]
    In Sleep      [OFF, ON]

[DavidAlfa]

Definitely, the calibration gremlins are my station's fault.
The calculation is correct, but little NTC changes affect a lot.
Not only that, when the cartridge contacts heat up and the NTC didn't notice yet, it will also change a lot.

So, yeah, end of this calibration drama. These things are what  they are, can't do magic.
I reworked few very delicate smd connectors few hours ago, set at 340ºC, but guess what, it worked great at 343ºC!

[wickated]

too many calibration points 
350c is only one anyone will need. 

p.s. next step - zero calibration step calibration 

"press "hot" when hot"

jokes away - actually all this martyr moves show that u dont really need ultraprecise ntc directly in handle like some ppl do think.

[DavidAlfa]

I'm broken! Too much calibration stuff/adding/removing/fixing code   .

Finally I found the way to make everything work together.
In the end I returned the zero offfset calibration option.
It's applied to the global profile, not tip-specific, as the error is mostly caused by the amplifier.

How to calibrate:

Now the calibration menu removes the power from the tip, and suppress any errors to let you remove/insert the tip.
The error detection is still active while calibrating (START) and adjusting the SETTINGS.

- If you already ran the initial zero/cal values adjustments, do the same as always: -> CALIBRATION / START.
  Now there're only 2 steps: 250°C and 400ºC. Simple and easy.

-If the controller has never been calibrated before: ->CALIBRATION / SETTINGS
 Insert a completely cold tip before entering SETTINGS!

 The options are:
   - Zero set: This widget behaves differently than others, when clicking it will switch between modes:
         - Zero set: This is the current system zero calibration value. Click to start sampling the ADC.
         - Sampling: Shows the ADC value in real time. When the tip is completely cold, the reading will be stable at a low value, click to capture.
         - Captured: Show the captured ADC value. This value is now being used for Cal250 and Cal400 calibration steps.
         - Click again to discard the capured value and revert to system stored value.
   - Cal 250: Adjust until you get 250ºC. It will be affected by zero set value.
   - Cal 400: Adjust until you get 400ºC. It will be affected by cal 250 value.

So, to run the initial calibration, follow in this this order: Zero set(Sampling,Captured) -> Cal250 -> Cal400 -> Save.

The current zero calibration method seems to bring more benefits than issues.
The calibration below <~200ºC will be a little higher, ex. 180°C->187ºC, but allows to track the tip down to ambient temperatures.
Now it's so sensitive that it will detect if you touch a cold tip with your hands.

Please, don't bother me with the translations and such stuff, this is still testing stage.
I'll make test builds in a while and post them here, I prefer to leave the current Git builds as they are.

Edit: Uploaded. This build will wipe your settings!

I also changed the default NTC beta to 3450, was what I was using some months ago.

Now, as always, I need reports with:
- Cold tip readings to find out the default values (Can also be seen in debug menu, set 0°C, check "AVG")
- Calibration accuracy at 200, 300, 400 and 450 (Or more steps if you wish).

[wickated]

I hope one day David will say that my ksger isnt pure shit and i never ranted w/o reason

[DavidAlfa]

KS  GER?
The day I saw a LED as a clamping device for the amplifier, or only one capacitor for the whole stm32+buck regulator, they lost my respect!

[AndyC]

I like the new calibration so far. Both the procedure and the results. Made some quick testing this morning, here are the results:

Code: [Select]

KSGER Calibration values and comparative measurements 210908
Temperatures measured with "KDRST 191"-thermometer

KSGER 2.1 Calibration:
TIP  COLD    250C    400C
-------------------------
K    0       1530    2524
D08  0       1428    2808

KSGER 2.1 D08:
SET    MEAS    AVG
-------------------
200    190    980
300    299    1615
350    352    2067
400    400    2520
450    433    2975
-------------------
KSGER 2.1 K:
SET    MEAS    AVG
-------------------
200    190    1030
300    293    1647
350    347    1974
400    400    2302
450    449    2628
-------------------

KSGER 3.1 Calibration:
TIP  COLD    250C    400C
-------------------------
K    2       1600    2553
D08: 2       1549    2870

KSGER 3.1 D08:
SET    MEAS    AVG
-------------------
200    218    1070
300    308    1737
400    403    2608
450    437    3043
-------------------
KSGER 3.1 K:
SET    MEAS    AVG
-------------------
200    194    1085
300    296    1725
350    345    2040
400    394    2354
450    440    2670
-------------------
Looking at these numbers I'm not sure I got the "Zero" right when I calibrated...

[DavidAlfa]

And what was the zero offset value?
The zero calibration is only used for temps below 250°C.
Well, not exactly, the calculation removes the ambient temperature first.
So if you set 250°C and the NTC reads 40°C, your setpoint is actually 210°C.
Between 250 and 400, the zero offset is irrelevant.

[AndyC]

And what was the zero offset value?

0 and 2. Don't know if I captured that correctly. Does the firmware use 0 as default now? If so, the above table should be OK.

[DavidAlfa]

0 and 2? 0 or 2? Can't be both. Yes, default is 0.
So that amp is able to go to absolute zero? Thats good!
You must leave it in "captured", then do the other calibrations.
If you set it back to zero set, you're resetting the captured data.
My Quicko reads 360 with a cold tip...

You posted 2 temperature tables. What's each?

Edit: I see you edited them.
Definitely, each tip has a different temperature curve.
Was the older calibration procedure more accurate?
This is all about the sensor non-linearity.
If it changes so much between tips, the only way would be to add a lot of calibration points.

[AndyC]

0 and 2? 0 or 2? Can't be both.
So that amp is able to go to absolute zero? Thats good!
Yes, default is 0.

Yes 0 AND 2, but not in the same controller It was 0 for the 2.1 controller and 2 for the 3.1 controller.

[DavidAlfa]

Now we talk, you confused me .

The temperature curve is similar to this plot:


If you take 2 points and make a linear interpolation, you get exactly what's happenning.
You get hotter temperatures when you go under the lower calibration point, also as you approach the mid point, and colder when you surpass the upper point.
I could add a coeficient, but frankly,  having such differences between tips and controllers, it seems mission impossible to me.
I think the point on all this is just to have close-enough temperatures, so when you set 320, you get something close, but not 300 or 350.

The cold calibration is a different thing, my Quicko reads 60-80°C with a zero set of 0, so it took a long time to cool down below 100°C, causing the screen to stay on all that time.
I don't mind taking 3 point calibration back, but I first want the the real differences against 2 point calibration.

[levi]

So my unit is indeed messed up...
Cold -- > 389
Cal 250 --> 2220
Cal 400 --> 3360

Idle temperature now seems OK, 30 on tip, 30.6 on NTC and 27 on fluke.
New dimming with option to completely turn the display off is awesome!

[StephenR0]

I installed the test firmware.  Then I went into the calibration settings menu and inserted a cold tip.  The captured ADC sampling was 51.  Then I set Cal 250 to 1200 and Cal 400 to 1900.  This turned out to be pretty close.  When I went to calibrate, Cal 250 was 1207 and Cal 400 was 1898.  Everything seemed to measure pretty close to that.  This was all with default values for everything.  I still have to tweak things to find the smoothest configuration, so there is some cycling of the temperature that does show up when measuring.  I would say about 3 degrees above and below that set point.  This was with my D24 tip.