Is it a bad idea to underpower a DC fan?

[solderfreelead]

I was wondering if it's a bad idea to underpower a DC fan, basically driving it with a much lower voltage than the labeled operating voltage.
Or if there's even such a thing as "underpowering a fan"?
For example, I have several 12V 120mmx25mm PC fans. They just barely start at 5V but once stabilized they seem to run just fine without any weird noise or anything obviously wrong.

[Zero999]

The only concern is the fan might not reliably start, especially as it gets older, gathers dust and the bearing becomes stiffer. It might also be a problem at lower temperatures, although the fan might not be needed if it's cold, so that could be a non-issue.

[tkamiya]

A DC motor draws more current when it's not moving at too low of voltage than moving slowly.  Also, inside DC motor, there is a set of brush that alternates between winding.  So under normal circumstances, duty cycle for each winding is 50%.  When not moving, a winding spends 100%.  I have seen a motor under this condition overheat and emits smoke.

If it's rated for 12V and it's barely moving at 5V, then I'd go with 8V or so.

Another option is to use properly rated fan or power supply.  Even high quality quiet fans are pretty cheap these days.  As to under powering fans, there are commercial product like that....  Auxiliary PC fan with high/low switch.

[TheHolyHorse]

That's how the fans are controlled in a PC (unless you control it with pwm). So as long as it spins like other answers explained it will be fine.

[SparkyFX]

These are usually BLDC with external commutation built in. If the electronic starts the motor at 5V, there is no reason for them to have a problem. I also run some on a PWM output on a fan controller, which makes them occasionally emit some noise. An electrolytic cap parallel to the fan kind of solved the problem.

[David Hess]

I have never had a problem running a DC brushless fan at lower voltage even down to the point where it will not start.

[james_s]

Same here, I've run 12V fans on 5V for years without any issues. Many PCs modulate the voltage to the fans to control the speed, I've never heard of it causing any issues.

[andy3055]

I was wondering if it's a bad idea to underpower a DC fan, basically driving it with a much lower voltage than the labeled operating voltage.
Or if there's even such a thing as "underpowering a fan"?
For example, I have several 12V 120mmx25mm PC fans. They just barely start at 5V but once stabilized they seem to run just fine without any weird noise or anything obviously wrong.

In the first place, why would you want to do that? Did you consider the air displacement/cooling that will suffer due to the low speed?

[james_s]

To make them quieter. At least in my case I was not using them in the original equipment they came from. A large fan run at a lower speed can move as much air as a smaller fan at a much higher speed and make a lot less noise.

[atmfjstc]

I for one was just thinking of doing this very thing with the fan cooling my Raspberry Pi 4. It's very loud and annoying at night when I'm trying to get some sleep

I might try to switch it from the 5V to the 3.3V rail (have to check whether it can supply the 200mA) and see what happens. The Pi will run hotter, I'm sure, but surely not as hot as the 80 degrees Celsius (!!!) it gets to if you put it in a case without a fan.

Yeah... they really dropped the thermal management ball on that one.

[Circlotron]

I also run some on a PWM output on a fan controller, which makes them occasionally emit some noise. An electrolytic cap parallel to the fan kind of solved the problem.

A largish cap across a load that is being PWM'd is going to eventually break either the cap or the switching device. Large currents will flow at the start of every switching cycle. Much better to leave the cap in place and feed the combination via a series inductor and have a reverse biased catch diode between the switching device and + volts.

[SparkyFX]

A largish cap across a load that is being PWM'd is going to eventually break either the cap or the switching device. Large currents will flow at the start of every switching cycle. Much better to leave the cap in place and feed the combination via a series inductor and have a reverse biased catch diode between the switching device and + volts.

So far it worked out. I figured given that the fan controller needs to be able to handle a blocked rotor, starting currents and such it might be a smaller problem. When in operation the running rotor limits the current as well.

Might be different once the control is somehow closely matched to a certain fan (load), but it would be a very fragile system, should the fan once end up being clogged.

[Zero999]

He's right. Putting a capacitor on a PWM circuit is a bad idea. It might have worked for you, but it doesn't mean it's right and won't gain you anything efficiency wise over a linear supply. Every time the switch turns on, a large current flows, causing a lot of power to be dissipated. If the capacitor is large enough to smooth the ripple to a negligible amount, the efficiency will be exactly the same as a linear circuit, although the power will be dissipated in huge surges, rather than continuously in the linear circuit.

[james_s]

If you put a resistor before the capacitor then you have a low pass filter, and that will convert a PWM signal to a variable voltage. Probably not gonna be of any advantage when powering a fan though.

PC motherboards drive the fan directly with PWM and it works fine. I have had a few fans that squeal at very low PWM values but no other issues.

[Zero999]

Actually if you add a capacitor, it will try to keep the voltage close to the full supply voltage, because the R component of the low pass filter depends on the impedance of the switch which is very high, when it's off and extremely low, when it's on.

I've done a quick LTSpice simulation. The switch is modelled as a resistor which oscillates between 1M and 50m Ohms, at 50% duty cycle, 1kHz. Look at the huge current spikes.

[james_s]

If that box is the switch then you're missing the R component of the RC filter entirely, of course that's not going to work. Also 10R is quite a large load.

[SparkyFX]

It is just a 22µF electrolytic and surely not recommended to approach the same problem with larger loads.

PC motherboards drive the fan directly with PWM and it works fine. I have had a few fans that squeal at very low PWM values but no other issues.

I suspected this noise to come from the fan-internal BLDC controller suddenly driving the rotor reverse (or enough out of sync to make the rotor wiggle in it's bearing, or maybe just coil whine), because of restarts due to the supply being underpowered until it catches up. The point is the fan or the output might not be designed to handle the rush-in current at PWM frequency over long periods of time, next to the noise it makes.

This might be interesting in this regard and regarding the OPs question, at least for this example:

[tszaboo]

You can plug it between the +5V and +12V, to run it from 7V, and that is just fine in a modern PC.
Or buy a 2 EUR fan controller from aliexpress.

[David Hess]

In the first place, why would you want to do that? Did you consider the air displacement/cooling that will suffer due to the low speed?

It might be done for lower noise or longer operating life.  I have often done it for variable cooling with a temperature sensor and feedback loop.  One of my homemade power supplies is configured this way and at low power, the fan does not even run.

I also run some on a PWM output on a fan controller, which makes them occasionally emit some noise. An electrolytic cap parallel to the fan kind of solved the problem.

A largish cap across a load that is being PWM'd is going to eventually break either the cap or the switching device. Large currents will flow at the start of every switching cycle. Much better to leave the cap in place and feed the combination via a series inductor and have a reverse biased catch diode between the switching device and + volts.

Driving the fan directly with PWM is definitely a bad idea.  The better circuits include an LC filter in the form of a buck inductor and output capacitor so the fan just receives low ripple DC.

[Alfons]

A buck converter module costs one euro. Of these, I have five in my 3D printer, two for fans and three for other things. I have now installed one with a 12cm Sunon fan in my Infiniium, because this fan is much too strong. With that I could easily operate a cooker hood. Even regulated to 6V, now the highest temperature inside is just 30 degrees. Anyway, the converters have the advantage that you can regulate how much you need. The disadvantage is: they can break, so a source of error more.

[ebastler]

I for one was just thinking of doing this very thing with the fan cooling my Raspberry Pi 4. It's very loud and annoying at night when I'm trying to get some sleep

Have you considered switching it off? 

[I wanted a rude username]

You can plug it between the +5V and +12V, to run it from 7V, and that is just fine in a modern PC.

I did this in the Pentium II era. Felt so wrong ...



... but worked fine. As a bonus, when the (dual) CPUs were working hard, the 5 V line would dip, increasing the voltage across the fan, which would speed up. You could hear the computer straining. 

[Zero999]

If that box is the switch then you're missing the R component of the RC filter entirely, of course that's not going to work. Also 10R is quite a large load.

That was the point: he's not using a proper RC filter, just a capacitor in parallel with the fan, after the MOSFET.

If the speed doesn't need to vary, then it would be better to skip the PWM and add a resistor, with a small capacitor for decoupling the brushless motor driver inside the fan.

[pcmad]

you should be find with lower voltage but you and get lower volage fans as well also you can over voltage these fans as will they are quite hardy

[Nominal Animal]

There are two types of 12VDC PC fans of various sizes: 3-pin and 4-pin.

The 3-pin connector has +12 VDC, GND, and a +5V tachometer output (middle pin) with usually two pulses per revolution.  The +12 VDC is commonly reduced using diodes (e.g. Noctua), a voltage divider using a potentiometer (extremely common), or filtered PWM (most digital fan controllers), to reduce the fan speed.  Intelligent controllers use a higher starting voltage, to ensure the fan starts rotating, as the minimum voltage needed to start the fan varies; and essentially control the voltage to get the desired RPM (which can depend on the monitored temperature in more complex fan controllers).

The 4-pin connector has +12 VDC, GND, open drain tachometer output (controller expected to have a pull-up to 12VDC) with two pulses per revolution, and a 5V PWM control input at 25 kHz (21-28 kHz, less than 5mA).  For these, you use the duty cycle of the PWM control input to determine the speed.

I don't like the 4-pin ones, since I tend to "hear" the ultrasonic PWM whine, and prefer the 3-pin ones instead.   I've used e.g. Lamptronic FC5 V2 (I think), that controls four separate 3-pin fans to individually set RPMs (and if you use the same type of fans, you can put them in parallel, with only one providing the tach output), and has four thermocouples with Fahrenheit/Celsius display.  It's very useful when optimizing airflow and cooling for custom silent cases.

The widest variety of PC fans is currently in the 120mm×120mm×25mm category.  You have low-RPM quiet fans, high-RPM "hairdryer" fans (i.e., lots of airflow, but also a lot of noise), fans designed to work against a pressure differential (filters, restricted airflow), fans designed for unrestricted airflow, and so on.  At one point, I probably had three dozen different fans I'd tested, although only used about four of them in case builds...  The differences between them are surprisingly large, and it is very common that dropping the voltage (for the 3-pin fans) to 10-11 volts gives a significant reduction in the noise, with a comparatively smaller change in airflow.

Because larger fans need smaller RPMs for the same airflow, you should try and use the largest fan you can, to get the best airflow while keeping the noise to a minimum.  Baffles and filters are also very useful.