Homemade PWM smoothing harmful for the hardware?

FG

Hi,

I have a T3DP3D Kossel XL+ and use two 50x50x15 radial fans for part cooling and a 40 mm axial fan for hotend cooling.
With this setup I have an amazing part cooling power and I would like to reduce the fan speed to have a more quiet setup. But unfortunately due to the PWM noise the fans do not really become much quieter if I slow them down.
Hence I designed a small piece of circuit which I want to install in the wire going to the fans to smooth out the PWM signal.
My plan is to build this circuit (the resistance is a fan) (Sorry for the long link, it contains the full circuit https://www.falstad.com/circuit/circuitjs.html?cct=$+1+0.000001+10.20027730826997+50+5+43 R+96+160+48+160+0+2+10000+24+0+0+0.99 d+176+384+96+384+2+default d+176+384+176+160+2+default w+176+160+288+160+0 209+288+160+288+384+0+0.00009999999999999999+0.001+1 w+288+160+400+160+0 r+400+160+400+384+0+342.86 l+288+384+176+384+0+0.01+0 g+96+384+48+384+0 w+288+384+400+384+0 d+96+160+176+160+2+default o+6+64+0+4098+80+0.1+0+2+6+3 o+7+64+0+20481+5+0.4+1+2+7+3 o+0+64+0+20737+40+0.4+2+2+0+3 ) on a small board, solder it into the middle of the wire connecting the fans and - to avoid too heavy EMI with the smart effector, - fold a thin sheet (I think 0.5 mm could be foldable without a machine) of galvanized steel around it and connect both sides of it by soldering them together so that only the ends with the wires are still open.
What is more I have to set the PWM frequency for the fans to 10 kHz in order to let the 10 mH inductor work properly.
As you can see in the link above, I build in two diode at the power source and the ground to avoid voltage peaks (like the >40 V when switching the fan from zero or low power to full power) reaching the smart effector and the Duet board. My capacitor is rated 50 V and the diode 40 V. I think the fan should also have no problem with a millisecond peak of high voltage and current.

After all this description follows my actual question . Can I harm my Duet3D hardware with this circuit? And has maybe somebody with a better understanding of electronics (I needed some help to get explained which type of diodes (Schottky) and capacitor (electrolytic) I need) some idea of a possible mistake I have done in this circuit which will make it not working, destroying the fan, the printer, burning or something else? I would really like to hear your opinion before I test this circuit on my printer.

P.S. I worked some evenings on the burning aspect because with 500 Hz I would need a 100 mH inductor and they have a really low current rating before they heat up. The 10 mH inductor I have ordered now has a current rating of 90 mA and the equilibrium mA I have at 100% fan speed are only 66 mA. So I think the heat development should be ok(?).

Thank you for reading this long post

dc42

A smoothing circuit like that is good and will not harm the Duet, with the following provisos:

You don't need the first two diodes on the left hand side (the ones in series with the Duet fan outputs).
You only need the third diode (the flyback diode) if you are connecting it to a fan output on a Duet WiFi/Ethernet with PCB revision 1.0 or 1.01 or a Duet 06 or 085, or to a heater output on any Duet. Duet WiFi/Ethernet 1.02 and later, and all Duet Maestros, have built-in flyback diodes on the fan outputs.
You can increase the PWM frequency to 50kHz, then you can use a smaller inductor e.g. 2 or 3mH.
If you are concerned about EMI, locate the circuit close to the Duet so that the wires to the fan carry the smoothed PWM.

FG

Thank you for the info.

With 50 kHz an 1 mH inductor is also working in the online tool. Since I can get this with a 130 mA rating I will try it to avoid too much up-heating over time.

FG

Oh, I just found one problem. If I leave out the two extra diodes in the simulation tool I get a huge oscillation into the circuit which is not damped out even after 100 ms and has amplitudes of up to 40 V and 6 A. Current is running backwards through the fan outputs of the Duet.
Is that not a very bad thing?
EDIT: Reseted the circuit in the link
(The changed circuit: https://www.falstad.com/circuit/circuitjs.html?cct=$+1+1.0000000000000001e-7+25.510281670702206+38+5+43 R+176+160+48+160+0+2+50000+24+0+0+0.99 d+176+384+176+160+2+1N5711 w+176+160+288+160+0 209+288+160+288+384+0+0.00009999999999999999+0.001+1 w+288+160+400+160+0 r+400+160+400+384+0+342.86 l+288+384+176+384+0+0.001+0 g+176+384+48+384+0 w+288+384+400+384+0 o+5+64+0+4098+80+0.1+0+2+5+3 o+6+64+0+20481+5+6.4+1+2+6+3 o+0+64+0+20737+40+25.6+2+2+0+3 )

dc42

The circuit you are simulating needs to include the mosfet, which is effectively a switch. That's very different from applying a square wave from a low-impedance source.