@argo Update on this issue: using the curve smoothing solves the problem. Also, I made a copy of the original STL using solidworks, exported it as STL but using max quality for the STL and also the print came out PERFECT.
So all this issue is about the STL quality
Thanks all for the help

@chimaeragh Sure. For example, when I want to save the total uptime of the machine I execute count.g:
echo >"uptime.g" "global.ontime =",{global.ontime + state.upTime/60}

This saves in uptime.g the value of the global.ontime, and I get that value from the object model

Next time the machine restarts, in config.g I have a M98 P"uptime.g", so the value of the global variable is read again and keeps being incremented.

@Falcounet Today I tried again and this time connection was ok... weird. But problem is fixed. Thanks!

@blv there is no need to teach formulas or motion theroy, at least to me, Im a senior mechanical engineer, I guess I have those topics very clear. So Im pretty sure you can post your slicer profile, your config, and the STL so we can all see the sizes involved and it will be very simple to demostrate what I said is correct. But we can also do very super quick calulations looking the video: the size of the object is aprox 12 cms long. I has a soft cuve too. so lets say that segment is a real 15 cms as much it it is straight? If you print at 500 mm/s that means we should be watching your printhead cover those 15 cms in aprox 1/33 of a second without taking into consideration the acceleration. If we take into consideration the accel of 9000 you will reach something around 300 mm/s of max speed AND ONLY for a fraction of the path. Im dpoing all these calculation by my eye so I might be wrong in 20/50 mm/s max error.
And Im still not taking into account your jerk settings.
And this is the speed you get in your longest path, the rest of your STL model wont get even half of the calculated speed I have mentioned.
So...

And Im not fan of nobody. I use what I considere that covers my needs. I have used klipper. It is based on python, python uses an interpret to work. That means an extra layer between klipper and the processor. So it works very nive, but no software till today is as fast as a well C++ / C code is working directly with the processor... Or have you ever seen an OS written in python ?

The firmware duet is working on is really powerful, enough for me moving from marlin.

But hey... if you are fine with klipper, OK! keep using. You have to use whatever covers your needs man.

Also I think it would be usufull if you clear what you mean with "extra perfomance", it would be a good way to contibute here to the developers.

@davidewen @dc42 There are several post about this issue with heated chambers.
In my case the problem raised with version 3.3 and up. Something changed on that PID implementation. Before that, my heated chamber was using the old PID very good.

After I upgraded, the problem raised. So I tried to autotune again and I was getting the same errors like you. After playing a while, what I did was: I raised deadtime value A LOT. That way, autotune was able to finish.

BUT then it didnt worked: the values I got from autotune, should have worked but every time I tried to heat up the chamber, an error was there: temp rising too slowly.
Running autotune several times worked again, but then again, using the reported values always failed with the same error.

Then I did a trick: I ran autotune but using ONLY 80% of the PWM. Then to the reported values I changed the PWM to 100%. That is the only way I have been able to use PID on the heated chamber

I hope it helps

Im my case, the times this exact thing happened to me, was a bad connection on my thhermistor cables, a badly soldered crimp. Resoldering always solved my problem. This problem was never detected by eye, the graphics is the one that have indicated me the problem.

@chrishamm I solved the problem (so close the topic as solved please). I dont know the causes but this is how it is working now:

.- In config.txt I commented the line

#disable_fw_kms_setup=1

By doing this and rebooting, I can see the raspberry logo at boot but then again all goes white and stays there

So again in config.txt I made the change:
dtoverlay=vc4-kms-v3d to dtoverlay=vc4-fkms-v3d

And these 2 changes solved the problem. Everything in "normal " now, The camera is not working either but it is detected.

On the other printer I checked, and disable_fw_kms_setup=1 is not commented... Documentation says this uses kernel defaults. So my guess is since the difference between both raspi images is the date and I can see a difference on kernel vlaues, there is something internal that makes all this trouble to me.

Either way, it is solved. Thanks for the help

Sorry for the late update. In my case problem is solved. I used the spyglass plugin. I got it from here: https://plugins.duet3d.com/plugins/SpyglassWebcamServerPlugin.html

The build instructions... never used them. I just downloaded the pluging, intalled through DWC interface, on the general settings I completed the url http://printer2.local:8080/stream (later I noticed theat hostname was automatically replaced by the internal ip number)
And it works perfect.

The raspi image I have is a brand new one, nothing extra installed or removed but userland in order to have some commands to verify the cam that seems to not be installed by default:

sudo apt install cmake
git clone https://github.com/raspberrypi/userland
cd userland
./buildme
sudo cp build/bin/* /bin/
sudo reboot
and then tested camera with raspistill -o image.jpg

I really dont know if userland had any effect on camera installation but since bookworm removed any easy way to activate/deactivate the camera (to the level of knowledge I have at least) I installed userland in order to have some debugging on the camera. With raspistill at least I knew the camera was there and detected and then I just installed the pluging
I hope it helps

@fcwilt so far I got a tempral solution till I find a proper solution, I modified the start gcode to position the sensor out of printbed and so it doesnt get all the heat, I leave door open so chamber doesnt get hot. After probing I tape the pin to be sure it stays up, then close the door as soon as print start. LOL

I looked for info and I found it: https://reprap.org/forum/read.php?1,873463,873463 reports something very similar. There are other similar posts on other forums too. and the nice find is:
https://www.reddit.com/r/klippers/comments/102ycc3/bltouch_temperature_drift_over_time_seckit_cube/
and pictures of the process: https://imgur.com/a/bJwce1D/

Basically he reposts the same and his solution was to remove the electronics from bltouch body and take it to a cooler place doing a very surgical cabling work.
I have still 2 working bltouchs so I might give a try to this project.

Sorry for blameing duet electronics in the past!!!!!!

Here I have the latest data from a tunning try today: it is my heated chamber

I ran M303 H0 P0.88 S120

The P0.88 is because I need to lower the power input to the resistor in order to not burn it.

The tunning finished OK:

8/17/2022, 2:39:15 PM Auto tuning heater 0 completed after 3 idle and 14 tuning cycles in 8036 seconds. This heater needs the following M307 command:
M307 H0 R0.125 K0.058:0.000 D44.76 E1.35 S0.88 B0
8/17/2022, 1:10:13 PM Auto tune starting phase 3, measuring
8/17/2022, 12:47:41 PM Auto tune starting phase 2, settling
8/17/2022, 12:25:23 PM Auto tune starting phase 1, heating up

I saved the values, I restarted the board, I checked new values are being used.

Now I try to heat the chamber and I get:

Error: Heater 0 fault: temperature rising too slowly: expected -0.02°C/sec measured -0.03°C/sec

This has happened with every try to tune PID.
The PID was working on this same chamber , same resistors, same PWM in version 3.3

Actual version is 3.4.1, board duet3 in SBC mode

I have read this new feature on 3.6 version. Also read something about prusa doing this.
But I dont clearly see the difference beween phase steppiung and the coolstep feature of the trinamic drivers.
Can someone clarify please?

@fcwilt Thanks for your feedback @fcwilt . To answer both: I dindt knew that probe but... I have exactly the same design for a very hight temperature enclosed actively heatd chamber (it reaches 220 degrees celsius). As far as I can see it is the same (but body in my case is all cnc metal). Works very good but ccuracy is lower than bltouch because of temeprature effects on metal and other details but sill way better than anything else (we have an average error of 0.01mm, wich is way of what it can affect a first layer print ). A good example of a problem being solved in 2 places reaching to the same solution. In this case for the moment I would prefer not having to send a new "touchy" body cnc part (this is how we called this sensor internally) adn try to keep for some more time the blotuch, we have 10 of them deployed)

Regarding the other answer: we can debate if you want in the other topic, because doing it here is like repeaton all again, but trust me: thisis not a mecanical issue. The permanent magnet in no way can be affected by a 50 degrees temp in such way. I would agree if we would be talking of 70/80 degrees. Also they way it manifest is clearlo not a mecanical issue. And as I said in the original post where you can see numerical data, this is happenining on the electronics (blt side, not the duet)

@fcwilt yes. They have very good eclosures, no active heating but as I described on the originalk post, while printing ABS for example I can reach easily stable 65 egrees C, even printing PLA with bed at 50 degrees only, I can reach 45 degrees on the chamber.
I dont really blame bltouch, it is great to my opinion, it is a device that was designed in an era where enclosed printer didnt exist at the common user level.
I also had like 10 printer working with bltouch for years with no issues at all. This issue arised when using the enclosed ones.

In another post I reported some issues with bltouch. At the end conclusion is hard to dispute: bltouch is affected by temperature.
At the moment I really dont have the time to look for a suitable replacement (other bltouch like alternatives probably will have the same issues) inductive/capacitive sensors in wont work with my print beds materials or dont have at the mnoment access to them. So fo the moment, bltouch stays.
Temporal solution now is really ugly: after probing the mesh, I just put a piece of tape to avoid the clip to go down.
Today I did a final tests since I wanted to know if this "pin going down and up" event is a mecanical o electric/electronic.
I did 2 cld prints in order to not damage the pin, and in both prints I did have the event: pin goes down and inmediatly up randomly during print.
Then I repeated the same gcode, but these 2 times, I just disconected the bltouch conector. I monitored both print only for 1 hs (in the previous prints the event would for sure happen at least 1 time in the forst 30 minutes) and there was no event.
So im almost sure this problem is not mecanical, is coming from electronics. Sin ce as stated on the other post noise is discarded as a possible suspect (cable is twisted, also is shielded with special tape, and cable is also away from other cables as fas as possible). Electronics is not my field, but then my guess is something is happening on the bltouch electronics while printing and when electronics temperatures is around 50 degrees.
Hall effect is affected by temperature but I dont know why that would be triggering the pin down event.
So the other suspect is the servo signal being affected somehow??? this queston goes tothe electronics experts here.
So the other question (testing this would really take a lot of time so that is why I prefer to ask) would be: if servo signal could be the problem somehow because of temperature of the electronics, would be good to "shut down" bltouch using the config? Like disabling the pin for example? or disabling both .in and .out pins after msh probing?
Ideas?

Thanks in advance

@chrishamm That solution wont work completely, I tested it. By putting the files on read only mode, they are not protected from the user dwc. DWC (I might be wrong about whi is doing this ) open the files, reads them, and I have seen they are modified on their atributes.

What I did and worked @trulm (my situation is the same as you) is changing the sticky bits on the files;

sudo chattr +i /opt/dsf/sd/sys/*.g
sudo chattr -i /opt/dsf/sd/sys/config-override.g

This bit makes the files imposible to be modified. Do the same with the macros and the filaments directories.
Also, I have modified the html files on DWC deleting the links to the areas I didnt want my student even look, like for example I deleted the posibility to manually introduce a temperature for the hotend and I also deleted the fill area in the console . In that way students can modified temperature, can not enter gcode comands either ,)

@the_K Im not able to replicate the installation on another printer. The same as you it is happening to me: plugin just runs for 1 second. Looking at the logs I can see an error about a camera module, so I installed:

sudo apt -y install python3-picamera2

With this I can start the plugin and stays active. Still, I cant see the camera on DWC

Thanks for the help.

lets close the topic, It is not related really to the duet. I found the problem: if router goes down and then recovers power, at least on this particular router and internet provider, the router and the printer reconnect again, but the dhcp server on router just give the printer internal network config but ont the required config to get out to internet, and also something happens with the internal dns name ".local" wich is not founded again by other devices reconnecting.
Reconnection should happened only after the router reconnect to the ISP.
I added a crontab job every 10 minutes on the raspberry to handle these event, I post the info here just in case somebody else have this "problem":

sudo nano /home/pi/reconnect_wifi.sh :
#!/bin/bash
if ! ping -c 1 -W 1 8.8.8.8; then
sudo ip link set wlan0 down
sudo ip link set wlan0 up
fi

sudo chmod +x /home/pi/reconnect_wifi.sh
sudo crontab -e:

*/10 * * * * /home/pi/reconnect_wifi.sh

@Phaedrux version is 3.5.3

Yes, sorry for the tricky title LOL.

This ois the issue Im facing: power goes down, so printer goes down and also the wifi router goes down. Or sometimes, and this is worst I think, power of the wifi router goes down and so the printer looses its conection to the network.
When router is on again, the printer seems to not connect again.

My config is a duet3 with SBC (raspberry pi 3 B+)
So in this scenario, the printer continues to print but it is not conected to nwetwork so I cant access DWI.

I guess the reconection should be a task the SBC should do, but it is not doing it.
Should the SBC reconnect automatically when router is on again?

@fcwilt so far I got a tempral solution till I find a proper solution, I modified the start gcode to position the sensor out of printbed and so it doesnt get all the heat, I leave door open so chamber doesnt get hot. After probing I tape the pin to be sure it stays up, then close the door as soon as print start. LOL

I looked for info and I found it: https://reprap.org/forum/read.php?1,873463,873463 reports something very similar. There are other similar posts on other forums too. and the nice find is:
https://www.reddit.com/r/klippers/comments/102ycc3/bltouch_temperature_drift_over_time_seckit_cube/
and pictures of the process: https://imgur.com/a/bJwce1D/

Basically he reposts the same and his solution was to remove the electronics from bltouch body and take it to a cooler place doing a very surgical cabling work.
I have still 2 working bltouchs so I might give a try to this project.

Sorry for blameing duet electronics in the past!!!!!!