I have recently acquired a 3D printer and have been experiencing all kinds of trouble with the fans used to cool the printer head, the parts and the controller board.

The controller is a smoothie-compatible custom board with a 24V PSU.

Some fans simply won't spin, some will spin only reluctantly ("wobbling" between two speeds for certain command values, spinning only when commanded at 75% or above), and all are far noisier when used in the printer than when plugged into a 12V PSU.

All this leads me to think these fans are rated for 12V and used with a pulsed 24V power supply.

I could not get the customer support to tell me the voltage these fans are supposed to be rated for. The fans have no label or visible identification. I finally sacrificed the one that would not spin anyway to try and get a clue about its rated voltage, but even the Hall sensor chip has no marking :

fan PCB

My knowledge of electronics is sketchy at best, I have nothing but a basic multimeter at hand and don't intend to buy an oscilloscope just to debug this pesky fan problem, but still I would be glad to understand what is happening there.

As far as I understand it, the controller board uses a PWM signal to achieve variable speed for the fans. The PWMs are generated using a sigma-delta algorithm with a 1ms base period. It means the fan power suppply is a 24v signal consisting of 1ms power pulses followed by the necessary number of 1ms no-power cycles to achieve an average power-up time matching the set value.

For some reason they added a 22µF capacitor between ground and the fan control signal (in the fan connectors located on the printer head), so I suppose the PWM must be somewhat smoothed.

According to the controller configuration, the PWM is limited to the 0-50% range, meaning the "average" voltage fed to a given fan never exceeds 12V (in reality, 1ms 24V pulses followed by 1ms power off, possibly smoothed by the capacitor).

Aapparently the smoothie designers explicitely warn about using 12V-rated fans on a 24V PSU, even using the PWM limitation trick to reduce the "average" voltage to 12V or less.

I've been scouring the Internet to try and understand what consequences overvolting these fans could have (on the fans themselves and possibly on the power supply), but my electronics skills are not sufficient to figure that out by myself.

So my questions are :

  • are these symptoms consistent with the use of 12V fans on a 24V PSU?
  • what are the consequences for the fans and the PSU?
  • what is the role of these mysterious capacitors? (the manufacturer claims they are just there to filter some noise, but I don't see what could need noise protection in a printer head consisting only of 3 fans and a heating resistor)
  • is there a way of getting these fans to work (e.g. by boosting the PWM signal above 50%) without damaging the fans and/or the PSU, or shall I replace them with proper 24V-rated components?
  • in case I put 24V-rated fans in my printer, shall I leave the capacitors in place or unsolder them (i.e. are these capacitors likely to hinder the fans, for instance making the spin-off more difficult) ?
  • 2
    \$\begingroup\$ I hate to tell you but it sounds like you bought a 3D printer kit that was designed by some folks that are clueless about how to properly design a system. The mere idea of trying to drive fans in the manner that you describe is total BS. Basic PWM for a fan should be up in the 24kHz frequency range. \$\endgroup\$ Oct 7, 2017 at 9:29
  • \$\begingroup\$ I think we are not talking about the same PWM here. The idea of the low frequency PWM is to have the fan spinning at full speed for a brief moment and slow down when the voltage cuts off, thus achieving an average speed somewhere between 0 and the 24V continuous speed. This is basically how all smoothie board-controlled 3D printer fans work, so I suppose the method is sound, provided you use 24V-rated components. This is not like the ESCs used in 3 phased motors, just a cheap way of modulating the speed of a monophased motor. Or did I miss something? \$\endgroup\$
    – kuroi neko
    Oct 7, 2017 at 9:45
  • \$\begingroup\$ How about just a 12 V zener diode in series? \$\endgroup\$
    – winny
    Oct 7, 2017 at 9:47
  • 1
    \$\begingroup\$ Well yes, or even a 12V regulator, but then I think replacing the fans would be an even simpler option (I'd rather not have soldered components dangling around in the already cramped printer head). 24V fans are dirt cheap anyway (something like $2 apiece on EBay) \$\endgroup\$
    – kuroi neko
    Oct 7, 2017 at 9:49
  • \$\begingroup\$ Where does that hall effect sensor connect to? Sounds like they are trying to drive it a simple brushed DC motor, which it obviously isn't. \$\endgroup\$
    – Trevor_G
    Oct 7, 2017 at 12:26

4 Answers 4


To answer your questions, these are my suspicions based on your observations: The symptoms you observe are very likely an attempt to provide "12V" to 12V fans, from a 24V supply. Such a theoretical attempt to provide 12V with high efficiency from a 24V supply probably deserves full marks for trying, but the designer IMHO perhaps did not think his or her way through all of the connotations of using mass-produced 12V fans. The only purpose I can think of for the 22uF capacitor is that it should be "pumped" by the 24V pulses so that it hovers around an average of 12V while being constantly bled by a theoretical, perfect, constant current draw which would be identical for all of the millions of 12V fans pouring off the fan-makers' assembly lines (real-world manufacturers will see the problem already).

Where the "12V" PWM circuit fails in practice, it has occurred (these are still my suspicions) from assuming an absolute belief in the supplied specifications of the ideal 12V fan from the fan manufacturers, and designing the efficient PWM supply accordingly, the faith in the specifications apparently leading to a prototype "12V" PSU design which is (and here is where we can all be sure) failing due to the simplistic PWM design.

Only ONE ideal fan with its constant, steady current, will discharge the 22uF capacitor just enough to keep it at a "ripply" charge state of approx.12V

Practical fans pouring off a mass-production assembly line, I don't believe for a minute are tested for extremely tight spec of constancy of current draw. That would never be conceived of as their prime requirement for their use.(my apologies, I am being wise after the event).

Thus my hypothesis for the natures of the observed failures in this case.

Now three suggested answers to the question: 1- If the fans as dismantled are modifiable to run on their originally rated 12V, then do so and put a 12V regulator in the 24V feedline to them. Use a series resistor with the regulator if necessary to keep the regulator within thermal spec. 2- You bought a 3D printer. Just a tiny fraction of a percent more cost at the local electronics shop will buy you a 12VDC supply to run all three fans. Wiring them in will be a breeze compared to what you have done already. 3- buy 24V fans as you suggest. Quick, simple, elegant. I would do that.

  • \$\begingroup\$ I should add, while retired now, significant portions of my career were to do with practical design of electronics for the commercial/industrial markets, prototyping, buying componentry, testing, fault-clearance & alignment of new product, both "hands-on" and management perspectives. Get 24V fans and dump the PWM feeders. \$\endgroup\$
    – Stan H
    Oct 7, 2017 at 10:00
  • \$\begingroup\$ Well thanks, that confirms my suspicions. I suppose it's off to EBay for a bunch of cheap 24V fans :) \$\endgroup\$
    – kuroi neko
    Oct 7, 2017 at 10:03
  • \$\begingroup\$ And about you being told by the manufacturer that the 22uF capacitors are there to reduce noise. Well, true, but only in the sense that any filter capacitor in any power supply will reduce the "noise" (or hum) from the raw, pulsed DC being fed into it from whatever source. Absolutely accurate information about the 22uF. You weren't being deceived per se, but if 24V fans had been used there would be no need for these capacitors at all and no need for the - in this case -poorly designed 24/12V supply. \$\endgroup\$
    – Stan H
    Oct 7, 2017 at 10:15
  • \$\begingroup\$ Yes, I think so too. I suspect they would tend to turn the 24V pulses into a variable continuous voltage, which would only make the 24V fans run at a suboptimal voltage. Better trust the smoothie board designers and believe this method of modulating the speed works best with pure 24V squarewaves. So I guess I should rather get rid of these capacitors. Or did I miss something? \$\endgroup\$
    – kuroi neko
    Oct 7, 2017 at 10:21
  • \$\begingroup\$ No, but perhaps I missed something. I perceived you had a real problem, and so I leaped on my "valiant white charger", to coin a hackneyed phrase, and leapt to your defence. The fan speed controller, I don't know why it is there for I don't know the exact needs of your 3D printer: is it important that the printer varies the speed automatically to close specs while printing is taking place? Or do you dial up a speed just to let air blow and keep things cool.? Or is the "variable" speed fixed and set in this case to output only 12V to the (assumed) 12V fans? \$\endgroup\$
    – Stan H
    Oct 7, 2017 at 10:44

I am unable to answer most of your questions but would point out that 50% duty cycle on 24 V supply will not result in 12 V average. We need to calculate the true RMS values of voltage and current.

If your 12 V supply results in a current of \$ i \$ then doubling the voltage to 24 V will result in a current of \$ 2i \$. Since power is given by \$ P = VI \$ then the instantaneous power goes from \$ 12 i \$ to \$ 24 \cdot 2i = 48i \$ or up by a factor of 4. At 50% duty cycle that will be double the power of a constant 12 V supply.

enter image description here

Figure 1. The RMS value of a squarewave is \$ V_P \frac {1}{\sqrt 2} \$. Source: Mastering Electronic Design.

  • \$\begingroup\$ Yes, I agree, this is why I added the quotes around "average" :) Overvolting these fans will feed them more power than they are supposed to handle. \$\endgroup\$
    – kuroi neko
    Oct 7, 2017 at 9:34

The 12V fans are brushless and have fragile SMD electronic components on a PCB.Consider the peak voltages of your proposed 24V 50% duty cycle scheme .25 years ago I experimented with running a 12V fan on halfwave rectified 24VAC from a 50HZ transformer winding .The fan ran for about 2 minutes and then expired ingloriously with a small puff of smoke .Fan teardown revealed burnt PCB.I always feed DC brushless fans with nice pure DC from a buck converter which gives the option of easily changing fan speed if needed .

  • 1
    \$\begingroup\$ Apparently the fans I got are strong enough to whitstand this overvoltage for more than 2 minutes (and I suppose the capacitors help mitigate the voltage jump), but that's probably seriously lowering their life expectancy. \$\endgroup\$
    – kuroi neko
    Oct 8, 2017 at 16:18

It's been an interesting discussion, and now, to help with answering the questions, It really is time to use an oscilloscope (I know you have not got one, but...) and begin some diagnostic checks of the smoothie controller board, the feed between the microcontroller and the power transistor gate/base, the input and output of the power transistor (checking with the oscilloscope, not your multimeter) and checking what is happening to the waveform and absolute voltage at their peaks, at the fan motors.

You will be wanting to know if the mark/space ratio, or duty cycle (however you want to name it) of the microcontroller output is being reflected in the actions at the output of the power transistor, and thence eventually to the fans.

You have been given plenty of information to work through from all of the valued contributors who have helped here with your problem, and yet there are still too many unknowns to give accurate answers to all of your questions, now that I have had a bit more time to think my way through it all.

And you will be doing those diagnostic checks so you can come back here and provide more background information to the questions you have asked. Good luck.

  • \$\begingroup\$ Yep, I sure need an oscilloscope to go further. At least it's a relief to know there is a real problem there and I was not just imagining things :) \$\endgroup\$
    – kuroi neko
    Oct 7, 2017 at 12:12
  • \$\begingroup\$ @StanH: It's not clear what's going on here. Is this a different answer to your previous one or are you trying to use the site as a discussion forum? The latter doesn't work as answers move up and down with votes and user sort preference. I suspect that this should be a comment or series of comments somewhere or put into your previous answer. \$\endgroup\$
    – Transistor
    Oct 7, 2017 at 23:30
  • 1
    \$\begingroup\$ Transistor, thanks. Yes, I am getting my head around the site rules. I take on board all of what you write. Just been here a couple of days and I imagine you are seeing my "learning curve" for the way the site rules operate. Yes, this answer, I now see, may have been better as a comment. Editors feel free to "tidy up" my incorrect placements, thanks. \$\endgroup\$
    – Stan H
    Oct 8, 2017 at 0:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.