3D Printer's Fan PWM Noise 100% Elimination 

I appreciate how at the start you explained why you need each component. And my goodness, how tech has moved on, no longer needing you to draw with a pen and soak in acid yourself to get a custom PCB board. I was thinking "how do you know which polarity?" and then you answered that too!

Several of Marlin's settings are effective in reducing PWM noise, which is a problem with this KP3S. For example: #define FAN_SOFT_PWM #define SOFT_PWM_SCALE 3 #define SOFT_PWM_DITHER (Present in Configuration.h)

Using a 120 ohm resistor instead of a 51 ohm resistor will result is more linear behavior at the cost of a lower max speed. If your intentions are to lower volume anyway, that might be a better alternative.

Couldn't you use an inductor, instead of a resistor so that you can still get Max fan speed, and maybe even get a linear response to the signal? All the while preventing an inrush current that could damage the control board?

Thanks for the video! Very informative and useful like always :)

Excellent vid again. You explain matters extremely well.

Thanks for the video!

Very interesting video. I stumbled upon this problem myself recently. I always thought, I had noisy fans until I realized, Klipper was to blame and I experimented with Cycle_Time command in Klipper. Thanx for your effort.

Собрал схему как у автора видео, только резистор использовал на 5w мощности, в klipper установил частоту шим 1кгц, и реально, вентили стали работать идеально, без рывков и в разы тише. Спасибо 👍

Thank you for this great tip !! I'm going crazy over the noise that my blower fan does above 54% pwm ....

This is excellent and explained amazingly. Also what program did you use to show the electrical and flow with resistor?

You saved my sanity. Thank you!

I built a T filter. I forget what value little inductors but 470u capacitor is way too generous, i should have gone way way down as it holds speed for much too long. I didn't put thought into it, i just grabbed the first nearest components I had at hand. It allowed me to tune in the fan behaviour that i wanted anyway besides being too slow to ramp down. I have a suspicion that the inductors might be around 100uH 500mA type with several Ohms of DC resistance. I should check whenever i run into them again. I kinda didn't really do it for noise, noise reduction was a side effect. I have a Sunon MagLev 5015 fan as PCF with Stall Guard which is horrible and without some smoothing could run very unreliably. I actually still wasnt happy with how the fan behaved, the bearing did make funny noises from externally introduced vibration, apparently the bearing is good if you leave it alone but gets weird once you knock it, which well not good mounted on a moving part, so i ended up packing the axle with grease from the back. Barbaric i know but whatever, I'm happy with it now. I used to have one of those $1 sleeve bearing fans but the bearing didnt live very long and also it never made quite as much air as the Sunon one. Maybe Delta next :D

Great Work! To still get the full RPM, I took 12V fans and reduced the voltage of the fan port by setting FAN_MAX_PWM in the Marlin configuration to 40.

What about instead od resistor use MOSFET that’s Will be driven by PWM and VIN will be from PSU? I currently don’t have time to try it, but I want to make PCB for it on a CNC mill to test it in future. I am using klipper for faster printing and cooling is necessary so this may help.

I know this is an odd question, but what about printers that share a 24v supply to the heater, parts cooling fan, and the hotend fan, but each has their own 0v by which the board controls the PWM signal. I would like to run the heater as it was, the parts cooling fans as 24v non PWM, and the hotend fans as a 12v non PWM. Is there a way you could do this?

Do you have full lessons about electronics? On udemy or other sites? You are so good, I wish to have a teacher like u in my school.

you can add a step up circut to up the voltage bu 5-10% shoudl fix your speed drop

Very interesting video, and here I leave you my personal opinion: I actually believe that there is no satisfactory solution to the problem due to the variables that are handled: First, the RPM of each cooler, second the type of operation of the shaft - whether self-lubricated bronze (cheap) or bearings (expensive) -, then by the design format in terms of size and number of blades and finally if it is the traditional or turbine type cooler (traditional less blades, more surface and less RPM, turbine, more blades, less surface and more RPM). In general, manufacturers do not make an effort to put the best on the market as they seek to reduce costs. All this, added to the fact that the lowest possible weight is necessary in the extruder head to reduce inertia, puts us in a quite complicated situation. In fact, printers with silent plates have significantly lowered the level of mechanical noise at the cost of losing a little speed, but the noise is still present mainly due to the coolers. Basically with these high RPM coolers the only way to reduce noise is by lowering them at the cost of losing efficiency or opting for a MAGLEV bearing system like the ones that NOCTUA Coolers come with, which multiply by 10 or 20 times the price of the coolers that come stock on the printers. In any case, we will always have the noise of air moving through the fins of the heatsink. Nothing is perfect!. A big greeting to you and thank you very much for sharing your work!

Why not use a jst connector terminals on the part and a patch so you don’t have to cut any wires and make it easier to take it on and off?

i never expected to see my noice canceler in a video :D

I found 1 way to lower the fan speed lost I have uped my printer voltage from 24V to 25.5V but I had to re-tune the printers heaters.

Although I wouldn't expect airflow to solve PWM issues, I'd be curious if the bellmouth/velocity stack attachment changes performance and/or noise unrelated to the PWM. Did you notice anything and/or test it when it was installed?

What I got: - If the freq of PWM is within human audible Range 20Hz-20k Hz it causes annoying noise. - soln: use higher Freq using firmware " #define FAST_PWM_FAN - high Freq could damage some boards that contains components that can't handle high frequency - Capacitors are like sponges draw insane amounts of current when powered ON, we can limit that using resistors

Would this work on the power supply fan? And the hotend cooling fan? Or is it only on the blower fans?

I did skim the video so maybe I missed it, but did you not see any startup problems? This is one of those classical things that PWM is very good at, the pulsing nature means that its easier to break that friction from a stand-still. This could actually be overcome easily with software (just always issue a higher level PWM pulse when going from 0 PWM to < x % PWM), but I have never heard of this being implemented.

I'm an idiot but I think you could limit the drawback of the resistor by using an NTC thermistor. High resistance when cold to limit in-rush current and its resistance will decrease as it heats up, reducing the wasted energy.

i dont know man, ive changed the pwm frequency on all my printers, you will update the firmware some time anyways, or with firmwares like klipper its just a text file you modify

Which software you used?

Why would current flow back into the voltage source? The voltage source consists of plus 12/24 volts. A mosfet switches on and off against minus. When the Mosfet is off where is the current supposed to flow back? With the R-C element you have a low-pass that can be calculated in terms of frequency. If the pulses come faster than the current flows, the voltage is averaged.

Why did you stick the airflow meter against the flat table? basically cutting off the airflow...

Hi @DIY PERSPECTIVE i have built your circuit, and work flwaless in my fans, but one fan (radial - 12cm) Sunon PMB121PLB2 (9.8w) i just cant put it working, any ideia why it is happening? Any solution? Anyway tk soo much for your work!!! This tutorial is AMAZING!!

Problems I didn't know I had

it's better to put a input diode in series with a coil. in the middle, another diode must be use as a freewhell for the discharge current (in parallel with the coil+load), then a capacitor must be place in parallel to the load and that's it. it will have almost a linear response. The power loss are in the diodes only (can be ignored compared to the fan consumption).

I only buy another fan, problem solved. Had the same issue as you. New more expensive fans, no noise

Hello, what software did you use to simulate the circuit.

wont this just change the biasing of the mosfet. Making low pwm values unusable?

Excellent video. A unique approach to silence fans

I have a question. I want to run two radial fans to one circuit. Each of the fans uses 0.12A and is 12V, therefore 1.44W each. Their total consumption of power would be 2.8W right? Is this an issue or limitation (the amount of power I'm drawing). For my choice of components I assume I should go with the 330uF capacitor and the 3W 25ohm resistor (octopus board v1.1, my power supply is 24v but my fan ports have Voltage selectors)... Do I have to make any changes to circuit or have different components to accommodate for what I'm trying to do ?

I've a question, adding capacitor will defy the need for PWM? as it turns it on and off capacitor will ensure it's on all the time working on constant voltage that may damage the fan?

you have to use ice liquid cool water tank . its a best solution of noise.

what software did you use to test the circuit?

I made this for an Ender 5 at 24 Volts. I used a 1N4007 diode, a 220uF 35V capacitor and 2x 100 Ohm in parallel to create a 50 Ohm resistor. My 4010 fan goes totally crazy with it, it keeps rattling and spinning up and down all the time. When i tried dual 5015 fans, it worked better, but the noise isn't really gone. Any ideas on how to tinker with the values to get a better result? I mean, if they are more silent at 100% than anything else, i need to improve something. ;) (All fans are rated 24V by the way).

why not move the resistor to just act upon the capacitor for less heat generation. may be slightly noisier but it would help with thermals

I am using clipper what do you recomend my cycle time be is currley 10ms which is 0.010

If I have two 1.2W fans should I connect each fan up to a board and then connect both boards to the Header on the printer motherboard

buying the correct type of fans would eliminate the noise all together :)

Hey, Nice work, i tried with my components but it does not change anything. I own an ender 3 and the pwm signal is like "bz bz bz bz bz bz" like an abs system to control wheel of a car while brake are in use. What is the name of the app you use on the computer?

Could you use an inductor instead of a resistor?

why not replace the resistor and make a low pass filter with an inductor

can't see the price, how much is it to get it from pcbway?

And I can make a gasoline car much more fuel efficient simply by removing the gas engine and installing batteries and electric drive motors, but it's still a gas car just very efficient! I get that what you did works, but really you just made a DAC and un-PWMd the whole thing and turned it into an analog circuit to the fan like what everybody did before PWM fans existed.

Hello, is there a way that this circuit could rise my output voltage? Because my motherboard gives me 8.8V and when I connect the circuit it gives something over 12V😮🤨

07:00 Resister values for 12/24v printers

Hmm, in your circuit I don't think the diode is actually needed, IMO. In a printer the mosfet rapidly toggles, yes. But the charge potential the capacitor accumulates, won't exceed the voltage rail of the fan's output rail(12v or 24v), therefor it will never really try to backfeed into the power rails across the parasitic body diode of the switching mosfet. In your simulation, you are using a supply that is suddenly sinking current, that's not a normal behavior of PWM outputs! Simulating with a DC rail and a switch that operates a FET type device instead, will show this difference. The only time backfeed is possible is the moment power is turned off, and it will simply dissipate into the rail safely as it shuts down. Otherwise the RC filter you've implemented makes sense, and thank you for the insight into the PWM noises.

Excellent! But why don't manufacturers make such a simple revision? Maybe they're just deaf? Anyway, thank you so much and keep it up!

7:52 happened to me my 7530 effs up like this

Buy 24v fans and run on 12v and you can't hear anything... Works great for hot end cooling, Doesn't work that well for part cooling though.

Why the resistor doesn't go only in series with the capacitor? I should only limit the corrent to the capacitor, the corrent to the fan can remain unchange

my printer control board doesn't use pwm.

I just connected my fan to the PSU and added a on off potentiometer.

why not use a coil.... you know, like a buck converter, one that actually converts voltage or just get a proper fan? like those cpu fans, 3 or 4 wires, hook them up with 12V, and attatch your fan output to a 1k with a 1uF low pass behind that to control the speed.... if you use a simple resistor and capacitor like in the video, the fan might fail at low voltages (low dutys) because it doesnt spin up, but it heats up

You're wrong. The fan is switched by MOSFET transistor on the board and no current flows back if the transistor is closed! You don't need a diode. Diode simple steal another ~1V of voltage. Fan needs maximal about 0,15-0,2A of current, if a transistor allows 2A of pulse current you can reduce resistor to 12 Ohm or even less, because transistor open channel has resistance, capacitor has ESR > 0 and wires. Such big current flows only if printer will change power on fan from lower to higher. In steady state you have much smaller current fluctuations.

LIthuania pog

That moment when you've never heard of a linear voltage regulator, so you make your own

Can't Marlin can do that via software?

No, the easyest is to change PWM freq to 7-8Hz, recompile, reflash

this is a good concept, but imo not a problem many people have especially nowadays with good 3d printer motherboards like those from BTT, Mellow, Fysetc, etc, it sounds more like a problem with the fan rather than the boar

Use a proper pwm driven fan with 4 wires. This also gives printer feedback that the fan is actually spinning.

@Rato Borrachudo

Without the diode you will get more linear fan speed charactericstic...

I don't think you really need the diode

Simply use a fan with an integrated controller

This should have been installed from factory😂

Just use a voltage regulator😂

Cat tested:-)

Maybe what you can do to eliminate the performance loss is, if your printer is designed for 24V fans, use a 12V fan instead, and then adjust the resistor value accordingly.

Why not using a buck converter instead?

noctua = problem solved

Just add grease

The program he used is called circuit simulator. It is a Java applet. I don’t know why people never mention the names of these programs You’re welcome

lmfao how is this legal in london?