@krzaku:

…
How did you choose a 220nF capacity? Is there a formula of some sorts? I just found today that Sunon fans are picky about PWM, I bought 3 fans, and 1 of them is not working with PWM power.

For sure there are formulas, look for e.g. RC circuit or for Low-pass filter in Wikipedia.
I just did some simulations in the free LTSpice program to get a rough idea about where I would be with the currents and voltages and it seemed to work fine for the 220nF capacitor I found in my junk box

Since I did not add a series resistor I wanted to keep the capacitor relatively small in order not to increase the peak currents too much (the wires have some resistance as well as the switching FET. The downside of this is a very high ripple in the resulting DC but seems to work fine for at least the Sunon I'm using. With larger capacitors you get a less ripply signal, but then you should also add a series resistor (a few ohms should be ok) to protect the switching FET.

@Jorma:

The switching to ground means that on the Duet PWM the positive terminal is always at the system voltage and the negative terminal is modulated i.e. it is either floating or at ground potential depending on the phase of the PWM signal.

A small correction, on Duet Wifi the fan PWM can be powered from the system voltage or from the 5V rail or from an external supply. So you might also consider installing a small 12V supply for the fans.

You should be able to use your 12V fans but you need some extra components. You can't just connect two 12V fans in series, that does not work (with the current brushless fans) and not recommended for any older fan either as the load is rarely equal.

The simplest thing that you can do is to connect a 12V Zener diode on the plus side of the fan with the ring on the diode facing towards the positive terminal on the Duet PWM connector and the negative terminal to the negative side on the Duet PWM.
That way the Zener will drop the voltage by 12V but it will also dissipate the same amount of power as the fan, so if you have a 1W fan, I would recommend maybe a 2W Zener for some margin. You should check how warm your Zener gets and place it accordingly.
You may need to play a little with the PWM parameters in config.g as not all the fans like to play ball with PWM right away. You may start with M106 S0 B0.5 F25000 to check if the fan likes a high (25kHz) PWM frequency or M106 S0 B0.5 F150 to try a low 150Hz (or even lower as David suggests in another message). You may also try adding a small (e.g. 220nF) capacitor across the fan terminals and set the frequency to 25kHz, has worked for my 40mm Sunon MB40101V2-000U-A99 very well.
You can also use a series resistor and the Zener as a shunt across the fan. In this case you can use a lesser wattage zener if you dimension the resistor right but in this case the resistor is dissipating the power so it needs to be e.g. 2W for a 1W fan but the first option is simpler.
The switching to ground means that on the duet PWM the positive terminal is always at the system voltage and the negative terminal is modulated i.e. it is either floating or at ground potential depending on the phase of the PWM signal.

@Herrminator117:

Unfortunately I can't change the fan speed in the slicer because the fans I have running are only meant to run at 100% and they cant scale down sadly. At 4.5v they just twitch and don't spin.

Thanks for the replys tho! Typically I try to take the hardest route, so fresh ideas are welcome!

You may not have tried all the tricks to tame your fan for PWM control.
Without any hardware modifications you should try modifying the fan setup line in config.g as follows and see what happens (assuming you have a relatively current firmware):

M106 S0 F25000 B0.5

This sets the PWM frequency to a relatively high 25 kHz and gives a half second boost at full power at every start to make sure the fan starts spinning.
I have made my fan to behave more civilized by soldering a small capacitor ( 220 nF in my case) across the fan terminals. If you go higher in capacitance I would suggest you add a small (a few ohms) series resistor to protect the switching FET from high currents. With a small capacitor like mine the long and thin wires should give enough series resistance.

Just some of my experiences trying to find cheap and usable filaments, I have no affiliation whatsoever with the companies:

Hobbyking sells PA Nylon, POM and PETG filaments for very reasonable prices (little over 20$/20€ per a kilo roll).

I can't say anything about the POM or PA yet as I don't have a suitable print surface (nor an enclosed printer) but I really like their black and white PETG (solid colors).
It's not spooled so nicely, but the dimensional accuracy seems good and it prints very nicely and the parts are durable so it has replaced ABS for me almost completely. Hope they get more colors as well.

I print it on ABS juice on glass with no sticking issues whatsoever (80-90 C bed / 255C hotend). AT this temp it flows really well and the parts come out durable and with very good layer adhesion. At 240C the parts tended to get harder and more brittle.

I have also used PETG from a Czech seller (on Ebay.de) called Filament PM. Very nicely spooled and packaged and the dimensions seem good. To me it seems a tiny bit harder to print than the Hobbyking variety as it seems a bit more viscous so has to be printed a bit slower but they do have more colors. I have not made any strength comparisons but both produce definitely usable parts easily. The layer adhesion is much better than with ABS and no warping to talk about and no smell!

I did the upgrade to 1.19beta11 and with the first initial tries the problem did not manifest. This does not mean it's gone, however as it did not happen every time with 1.18.2 either. I will repeat the tests some more.

What did happen was that during one try the extruder came up as per the G1 X0 Y0 Z150 F15000 in bed.g and then paused right there as if there would have been a M226 in the end of the file and I had to Emergency Stop to get out.
I repeated the test with the M226 in the file and the 1.19beta11 also gets stuck there.

For now I will add a pause in the prologue for new files produced by the slicer, but the best solution would be if the web interface could be made to resume after a pause in bed.g and of course it would be good to find the root cause. How does the system know to go back to absolute extrusion after a pause retraction anyway? Maybe that is being missed and the relative extrusion set in pause.g is sometimes left in effect?

@dc42:

Are you using a mixing hot end with mixing enabled for that tool, or a single-extruder hot end?

What happens if you put a M226 command at the very end of bed.g? Does it pause, and if so, are you able to resume?

It would be helpful if you can try the 1.19beta11 firmware, to see if it fixes either of these issues.

I'm using a single extruder T3dP3d Kossel Mini.
The M226 pauses alright, but the only way to get out using the web interface is the Emergency Stop.
I'll load 1.19beta11 and report back.
Thanks!

@dc42:

Are you certain that it is the pause command that causes it to set the extrusion mode to relative? The behaviour you describe is not expected.

Obviously the pause.g sets it to relative to retract during the pause but this has not caused issues before. The absolute extrusion has continued after the resume right after where it left.
Something has changed regarding timing of things and there seems to be a race condition somewhere.
It used to work so that when I hit Pause right after the final calibration point, the hotend came up as per the final line in bed.g:
G1 X0 Y0 Z150 F15000 ; get the head out of the way of the bed
and then moved to the side as per the pause.g:
G1 X0 Y85 F5000 ; move head out of the way of the print
and when resumed, the print started normally.
Now it seems that the pause is somehow delayed and the hotend goes down ready to start printing and only after that moves to the side and it's a bit hit and miss what happens to the extruder. Mostly it goes berserk, so I guess the mode changes to relative (there is an M82 in the beginning of the gcode file as generated by the MatterControl slicer).

This is not a big deal, but for me the best solution would be if Resume could be made to work if there is a pause in the end of bed.g

I have a little bit different pause issue that surfaced after 1.16 -> 1.18.2 upgrade (Duet 0.85).

My mode of operation has been to let the calibration run (13 points in bed.g) and then hit Pause just before the print starts to clean the nozzle of all ooze.
With 1.16 and prior this has worked perfectly but now it seems the firmware looses track that I'm using absolute mode in extrusion and the extruder goes berserk right when the print starts.

I was hoping to sneak by this by putting a pause in the end of bed.g, but no suck luck. It pauses alright, but there is no way to get the job to continue. What I don't want to do is to edit all my existing gcode files to put the pause there, so is there any way to make the resume work if there is a pause in the bed.g, please?
Of course I can delay hitting the pause a little bit and let the print start first but it would be nicer to clean the nozzle when it's higher from the bed (like it is after the calibration).

Thanks for all the great work to David and Chris, lots of new features so you have been seriuosly busy!

Jorma