Understanding 3-pin fan drive

[bitwelder]

Hello,

as part of a project to replace a cooling fan with something less noisy, I began to check what kind of driving is applied to it.
This is the situation when the 12V fan is driven at full speed:

The yellow trace is the fan drive: above the average 12V drive there are periodic narrow 16V pulses.
Those pulses are always synchronous with the signal coming from the fan sense pin (blue trace).

I tried to 'hand brake' the fan to make it spin it slower and not only the period of the blue trace got longer (as expected), but the yellow pulses kept being synchronous with the edges of the sense signal.
As amplitude, when the fan is driven at reduced speed (i.e. about 9V) the voltage peaks are still reaching 16V.

Question is: where do those narrow pulses come from? What their cause or their purpose?

Thanks.


BTW: the snapshot was taken with RigolBildschirmkopie via LAN

[bktemp]

Those 16V spikes are unusual.
Is this a high power fan connected via a long cable or an inductor?
My guess would be the following:
Everytime the fan motor changes to the next winding, it switches off both windings for a short time. Therefore some inductance in front of the fan generates the voltage spike.

[RobertHolcombe]

Curiously the sense Vpp=2.88V is approximately the increase in the supply voltage you are seeing, and is in sync with the sense falling edge

[electr_peter]

Is it 12V PC style fan with 3 pins? GND + VCC + RPM out pins - you are seeing RPM (sort of) output from the fan. See simle description here: http://www.pcbheaven.com/wikipages/How_PC_Fans_Work/

[bitwelder]

Curiously the sense Vpp=2.88V is approximately the increase in the supply voltage you are seeing, and is in sync with the sense falling edge

I thought about that, but when the fan is driven at 9V the spikes are still 16V (with sense signal still around 3V)

[bitwelder]

Those 16V spikes are unusual.
Is this a high power fan connected via a long cable or an inductor?
My guess would be the following:
Everytime the fan motor changes to the next winding, it switches off both windings for a short time. Therefore some inductance in front of the fan generates the voltage spike.

It's a small 50mm fan (sorry, can't find its model right now), and it has a 20cm cable from fan to PCB connector (where I'm probing the voltages).
I tried to disconnect the fan, and plug it to a 12V power source, and in that case the voltage I see is a relatively smooth 12V, no spikes of any sort.

[SeanB]

Put a capacitor of between 10-100uF on the fan connector and try it again. These fans are notorious for injected noise into the supply, as there is no room on the tiny board inside to put any capacitors to decouple the supply rails, so the spikes are reflected onto the power rails as noise spikes. For lower noise on the supply side you really need either a LC filter on the supply to the fan as close to the fan as possible ( typically the connector to the fan) or a RC filter. Inductor value is not too critical, almost ant small SMD inductor rated for a little more than the fan current ( and preferably the lossiest one you can get, here the lower the Q the better, so typically the smallest one is right) and a capacitor ( typically a regular SM electrolytic, around 10-100uF, voltage about double the supply voltage, again can be very lossy with a high ESR up to about 5R) works to keep the supply line a lot quieter and gets a lot of the noise out.

[bitwelder]

Put a capacitor of between 10-100uF on the fan connector and try it again. These fans are notorious for injected noise into the supply,

Thanks for the detailed suggestion but... my main issue is the acoustic noise from the high speed fan, before the electrical one.    So the final plan is to actually replace the original fan with something else.

What can I conclude about those 16V spikes?
- Nobody suggested they are there on purpose for the fan drive. They must be some induced disturbance
- Hopefully they are too short in duration to overload and damage the fan
- However, they might have an effect on the driving circuit, so it would be a good idea to add a snubber capacitor

I'll make a few more tests, e.g. trying to disconnect the sense wire and see what changes.