Brushless Fan INductive Flyback

[dknguyen]

I am using a transistor on the ground side to act as an on/off switch for a brushless cooling fan and am wondering if I should have a reverse paralle diode for inductive spikes. I am not familliar with the brushless electronics inside the fan, and I have heard the diode should be connected across the inductive load rather than the switch for protection (with the exception of H-bridges because of the way they work). So reverse parallel diode across the MOSFET? across the fan? no need because the fan already has flyback diodes (or equivelant)? I do not know since I am unfamilliar with brushless drivers.

 

[hjames]

Some interesting app notes
https://www.electro-tech-online.com/custompdfs/2006/08/00771b.pdf
-slow down the transition to avoid/reduce an inductive voltage spike

**broken link removed**
-it seems that most brushless fans have a diode in series with the power supply which will keep the inductive issues inside the fan controller circuitry.

http://www.mlxsemi.com/Asset.aspx?nID=4849&sCmd=download
-sample circuit also has series diode - with a note that it's meant for reverse voltage protection...

These data sheets also seem to specify that PWM'ing should operate at in the 30 Hz region - which makes sense if there's a series diode (and a local power reservoir capacitor).

 

[dknguyen]

Ah, thanks. I guess diode or not a capacitor would be best to ramp down the voltage. since the fans won't be turning on and off at insane rates (and if the temperature electronics are trying to force the fan to do so, then too bad- the capacitor will stop this from screwing around with the fan).

Although...the transistor I am using is a MOSFET and it would seem that the capacitor need only be very small due to low gate leakage...it's all good. I guess that would mean that a properly chosen small cap would make it so the fan would be faster to turn on than it would be to turn off (uC gate source current > gate leakage current) so the fan could turn on quickly to react to sudden increases in temperature and take longer to turn off, reducing spikes and extra cooling- nothing wrong with that.

 

[Styx]

A couple of questions
When you say it is a brushless DC machine do you have access to the winding's directly OR are you just trying to treat it like a brush-DC machine?

I ask because brushless-DC machine (even single-phase) need zone-infomation to ensure that the power electronics are fired at the right commutation zone.

IF this is one of those brushless-DC machine's that have built in de-coders & switch such that you just need to apply a DC voltage to the terminal AND they will just spin (90% of all brushless DC fans), then you CANNOT just use a chopper cct to controll it


2) for an inductive load a FREEWHEEL path is ALWAYS needed!, no if's buts or maybe's
The inductor equation actually states it is needed:
V = Ldi/dt, ie IF you instantaniously interupt the current (ie open the switch) then an infinite voltage must be produced.

The two golden rules of power-electronics (and well general electronics) are:
Thou shall not short-cct a charged capacitor
Thou shall not open-cct a charged inductor

IF you open-cct an inductive load with no free-wheel yr swithc would be dead instantaniously no if,buts or maybe's

What I think is happening is you have one of these "controlled" brushless-DC fans and you are trying to treat it like a brush-DC machine by chopping the voltage.

Unfortunetly you have coupled two switching cct's together (BAD!!!)
The brushless fan will already have a free-wheel diode for the stator winding's already in the housing (if it didn't yr switch would be dead now).

A capacitor across the FET (ie a snubber) is a "frig" really when a cct is layed out but switching overshoots (or switching losses) are higher then you can cope with (both of these ONLY occur if you have a freewheel, no freewheel == switch blown up) and all you can do is cosmetic fixes
Don't get me wrong snubbers have their place (less so with 3rd Gen IGBT's) BUT they are not replacements for diodes

Hence my thought that you already have some control in yr fan.
switching such a fan will just cause it to speed up and then slow down (or stop) at yr switching freq rate.


the fan will want to see a voltage source so give it one:
rather then switching the fan directly (bad). Put a cap across the brushless-DC fan (wont work for a brushed or non-controlled brushless) and then switch that, the capacitor will provide a local DC-link for the power-electronic cct within the fan and allow it to ride through the period's where you don't switch.

Sizing of the cap might be a bit tricky. You want it big enough such that at yr maximum load it holds the voltage fairly constand BUT you want it small enough so if you change the speed (via the PWM duty) the voltage changes at a decent rate (ie you don't wait 10min for it to slow down when you go from 90% duty to 10% duty)

--EDIT--
just bolding something

 

[Hero999]

DC brushless fans have built-in driver electronics which would be damaged by back EMF, therefore they normallyt have protection diodes built in.

 

[Styx]

DC brushless fans have built-in driver electronics which would be damaged by back EMF, therefore they normallyt have protection diodes built in.

isn't that what I said

 

[Hero999]

Probably, sorry, I'm not very good at reading long posts, it's a dyslexic thing.

 

[hjames]

Well, the fan's built in protection diodes, along with it's local capacitor is enough to asorb the inductive kickback. Putting another capacitor across the fan, before the built-in series diode won't hurt anything, but probably won't do too much. I'd assume that any low-wattage fans meant for computer cooling are all built this way...

The brute-force, overkill solution is to just put a full adjustable DC-DC converter in front of the fan. Charging a DC-link capacitor directly from a low impedance switch (without a series inductor) isn't exactly efficient...

 

[dknguyen]

THis is not for controlling the speed of the fan (the internal electronics of the fan are already taking care of that so not much I can do.). This is not PWM- no voltage chopping or anything. THis is just for a remote on/off switch for the fan. I am just switching the power on and off so the fan is running at full speed when cooling is needed or off when cooling is not needed. I am wondering if there is already a free wheeling path inside the commutation electronics, or anything I must do to protect the "relay" transistor.

I have no access to the internal windings or electronics of the fan at all- just the power lines leading into the motor and electronics.

 

[Styx]

It's a brushless cooling fan with built-in electronics. I am using the transistor as a relay to shut the fan off when cooling is not required. So it's not PWM or anything. I do not have access to the windings, just the power supply lines as a whole. I am not commutating the fan at all, it's being done internally. I am just switching the power on and off. I am wondering if there is already a free wheeling path inside the commutation electronics.

I am not chopping the voltage with PWM or anything, it's just a remote on/off switch for the fan.

yes

padding to satisfy 10char limit

 

[Hero999]

Which is stupid. I can't understand the purpose of the 10 minimum charicter limit, why should posts be any longer than they need to be?

 

[Nigel Goodwin]

Which is stupid. I can't understand the purpose of the 10 minimum charicter limit, why should posts be any longer than they need to be?

Presumably it's the default setting for VBulletin?, and it's purpose is obviously to avoid one word (or one letter) posts!. Ten letters is hardly any great strain to write?, and it stops the 'Yes' and 'No' single word replies.

Admittedly it can be a pain, because occasionally (VERY occasionally) a 'yes' or 'no' answer is really all that's required.

 

[dknguyen]

Use and
Affirmative
Noooooooo

 

[Hero999]

Perhaps it should be changed to three letters, so it'd be alright to just say "No.", which is a good enough response to a "plz do my prjt 4 me" post.