RC Low-Pass Filter for Blower Fan

wilriker

@iamturbo1978 First of all a big thanks for this very detailed explanation.
I knew parts of it but there was some very interesting news in it.

Especially that the PWM high-low-cycles are always within a cycle that is defined by the frequency. Up until now I always thought that it would combine the latter type of cycles to continues high and low phases. But the way you describe it makes a lot more sense because otherwise you would loose a lot of fine-control at very low frequencies.

Also finally I understand cut-off frequency of the RC-filter.
With one open question though: you say if we have 10ms cycles and a 106Hz filter will only see 9ms of these cycles - what happens with the remaining 1ms? A friend of mine just proposed that it will "stretch" what it sees to match the "outer" cycle, i.e. it will create a 10ms cycle out of what it sees but using the proportions it sees in 9ms.

I guess I finally have to buy a book on electronics.

dc42

There is a complication which makes the R-C filer much worse in this application. The output of a Fan connector is not a square wave with a variable mark-space ratio. It is a switch (the mosfet) that alternately connects the load to V_Fan and leaves it unconnected (save for the flyback diode on later Duets) with a variable mark-space ratio. The difference is that when the switch is off, the capacitor does not discharge through the resistor. It discharges only through the load. This means that the voltage at the output of the R-C filter will always be higher than the mark-space ratio implies, after allowing for the loss in the resistor. Whereas with an inductor (of a sufficient value for the PWM frequency being used) instead of the resistor the flyback diode completes the circuit when the switch is off, so you do get an output voltage that is proportional to the mark-space ratio.

Iamturbo1978

@wilriker I'm glad it helped you to understand whats going on.

With your open question; the PWM cycle does not reset, so the remaining 1ms after the filter would be added to the next cycle.

cycle 1)
high for 5ms, low for 4ms

cycle 2)
low for 1ms, high for 5ms, low for 3ms

cycle 3)
low for 2ms, high for 5ms, low for 2ms

I agree with @dc42, to get the best output is to use a L-C filter.

wilriker

Thanks @dc42 and @Iamturbo1978 for all your patience with me so far and for your very understandable explanations in this topic!

Just to make sure I understood everything correct here is a drawing of what it should look like:

Of course power-source, MOSFET and flyback diode are already on the board and I would have to add the right half of the diagram. Right?

dc42

Yes, that's correct.

wilriker

@dc42 Yesterday I connected the tacho output of the fan to my Duet. Although this fan is obviously not a 4-wire PWM fan it gives me quite stable RPM readings when I set my fan to a specific speed.

You said it might be that still this would give me bogus values and I found something really strange so maybe you could tell me if that could fall under what you warned me about:

My fan according to its specs (TFD-B5015M12B) has a top speed of 4500 RPM@12V. If I run it with M106 S1 it will get full Vin which in my case is 13.6V so of course it will run faster. But tacho readings tell me it runs at 6500 RPM. If I then run it with M106 S0.88 it should get very close to 12V and the tacho reads 5500 RPM. Interestingly there is another model of the same family that would have exactly this speed @12V but I checked my order as well as the type label on my fan to not be the faster model.

Of course there is a chance that my fan is labelled wrong or that they have changed the specs since this specific fan was produced. But I just wonder would it be possible for the RPM reading to be off at around 1000 RPM and scale according to this?

JoergS5

@wilriker so this could be the H type with 5500. If you measure the current also, and it is 0.18 A instead of 0.14, then you have the other type.

wilriker

@joergs5 To measure the current I would have to put my multimeter in series, right? Never done that so far.

JoergS5

@wilriker directly into one of the main lines of the blower. Disconnecting the line and putting the current meter at those two open contacts. Be aware to set the maximum measure value of the current meter high first (above the maximum ampere, e. g. 1 A).

If it's a Molex connection, you can turn current off, draw out one of the cables (the pin have a barb, by pressing down you can remove the cable), connect to amperemeter at both sides, turn on and then measure.

wilriker

@joergs5 OK, thanks for confirming. As I already cut my wires to the fan I think I can quite easily insert the multimeter in this circuit using some screw terminals... I need to put ferrules on the wire ends anyway, so that's up next.

Then again it would be really strange to have the H type when buying the M type that is also labelled as being the M type (including the amperage). I would expect this on a super-cheapo Chinese fan but not from a brand. But let's wait and see.

JoergS5

@wilriker maybe labeled by someone who needs better glasses, M and H are similar

I thought it may be the H type because your 5500 measuring fits exactly to this higher model.

wilriker

@joergs5 I know it would fit perfectly. Also about 6500 RPM @13.6V would match that quite well.

SnowCrash

@phaedrux said in RC Low-Pass Filter for Blower Fan:

https://www.digikey.ca/short/jrmcdf

I was reading through this thread and was hoping someone could clarify the value of the inductor in the diagram above? I'm familiar with 1uH (micro-Henry) and 1nH (nano-Henry), but what does 1mH stand for? mili-Henry (1000uH)? for such a small amount of current?

wilriker

@snowcrash said in RC Low-Pass Filter for Blower Fan:

I was reading through this thread and was hoping someone could clarify the value of the inductor in the diagram above? I'm familiar with 1uH (micro-Henry) and 1nH (nano-Henry), but what does 1mH stand for? mili-Henry (1000uH)? for such a small amount of current?

Yes, mH is milli-Henry.

SnowCrash

Thanks @wilriker!

But that's HUGE just for a fan that draws maybe 0.15mA, no? (ok, the fan will draw more current at startup but still...). Is there some calculation behind this value?

wilriker

@snowcrash Actually I have no clue about all this stuff. I just took those numbers for granted (even though I would also like to learn about where they come from).

But just as an update: I dropped all this anyway since connecting the signal line showed that the fan actually turns at the speed I set it to, it just feels/sounds a lot lower - but then my ears and fingers are probably the worst way of measuring fan speed in the first place.

SnowCrash

@wilriker happy you got your fan working quietly (though perhaps not in the way you were hoping to achieve this )

Anyway, at another discussion, @dc42 insisted that an inductor-capacitor circuit is necessary for preventing a non-PWM fan from potentially being damaged while being run with a high frequency PWM signal, and just I started looking into the idea of making a little add-on module that would provide this circuitry for all 3 PWM outputs on the Duet (plus the ability to choose between 12V and 24V for each fan which is the main reason for the module).

So hopefully @dc42 or someone else familiar with this subject would chime in and clarify this point about the inductor's value.

dc42

@snowcrash said in RC Low-Pass Filter for Blower Fan:

Thanks @wilriker!

But that's HUGE just for a fan that draws maybe 0.15mA, no? (ok, the fan will draw more current at startup but still...). Is there some calculation behind this value?

Yes 1mH. The fan draws 0.15A (150mA), not 0.15mA.

SnowCrash

Thanks for the confirmation, @dc42! But what's the calculation behind it?

dc42

The calculation is L = V * t/I where V is the voltage across the inductor, I is the maximum ripple current current you want to allow, and t is the time involved. Or, turning it around, I = V * t/L. If we take the case of a 24V fan running at 50% PWM and 50kHz PWM frequency, then for L=1mH and 12V across the fan, we have I = 12V * 10us/1mh = 120mA. This is around the same as the fan current, so a capacitor is needed after the inductor to reduce the ripple current in the fan. For the capacitor, C=I*t/V where V is the allowed peak ripple voltage. So V=I * t/C and 1uF will give 120mA * 10us/1uF = 1.2V. This is 10% of the 12V average voltage at 50% PWM, so it is likely to be low enough.