How to silence an unneeded fan (but one that MUST keep running)

[velis]

Not exactly a standard scenario, but hope dies last as they say:

I have bought this active UPS unit and it came with the cheapest fans possible. They (2 of them) consume 6W and shriek like a couple of furies. Anyway, long story short, the unit sounds an alarm if at least one of the fans is not the original POS.

I have replaced the fans with quiter ones and placed them both on one header. The other header is used for one of the original fans, but the fan is just tucked away in the case. Still roars though. Please note that I tried to fool the UPS into thinking that it has original fans by adding 24 Ohm resistors, but that has had no effect.

So, exactly how could I make the original fan shut up without having it self-destruct from overheating?

 

[ClockHound]

Need more data.

What are the make and model of the fan? Are they 2pin, 3pin, 4pin?

Why didn't you put each new fan on its own header?

 

[aigomorla]

definitely need more data...

 

[velis]

Sorry, didn't realise fan specs would be important. Here goes:

HK FAN (dc brushless), model AB8025V12, rated at 0.50A, 2pin header, 2wires only, here's the only link I found for it.

It's an 80x25mm fan, standard configuration, 7 blades

My current solution is that I simply wrapped it in 3 cm of foam. Looks ridiculous but does dampen the sound somewhat.

@ClockHound here & here are the relevant posts explaining all the combinations I tried to fool the unit, all failed.

 

[Stuka87]

That part number must not be right though, as a google search only shows this post

But there are plenty of 80x25 fans to choose from out there. Some are higher quality, but they still need to draw close to 500mA to make the system happy. However, none of them will be super quiet when run wide open (no speed limiter). That size fan is just not known for being quiet as they run higher RPM. So the ultimate option is to add a resistor in line with the fan, to drop its speed some. The draw from the main board should still be enough to make it happy (as the resistor will be dissipating some of the power). The slower speed will hopefully still be enough to keep it cool. Unless that thing runs really hot?

 

[velis]

I tried that, but it didn't fool the unit. In fact, I added 2 24Ohm resistors (parallel to the fans, not serial). I also tried serial - on one fan only - but that didn't work either. Both definitely consumed enough juice, but the unit still somehow rejected the new fans.
It will definitely not run hot as it's a 3kVA unit, but highest planned load is somewhere around 500W. I bought it that big because I wanted the autonomy time.

 

[ClockHound]

Sounds like desperate noisy UPS calls for desperate measures. Do you have the ability/space to make an enclosure for the UPS? If so, could fabricate an enclosure (built using noise-isolation techniques - decoupled double wall etc..) with ducting to support 120/140mm intake/exhaust fans - then clip the wings (blades) of the little devils.

Such a high current devil too, if the specs are to be believed. (Have you put a meter on the fan header to see how much current is in use?) Almost 10 times the current draw compared to normal domestic 80mm fans. How fast does it spin? 4000rpm? 6000rpm? 240 ohm resistor ain't gonna cut it. Maybe some Heavy Duty Load resistors will fool the lame current draw detection method while quieter fans do the actual cooling work. Could also look to replace these hideous server-room fans with some not quite so hideous SanAce server room fans. Ugh.

As silence is a priority for my gear, this sounds like too much work to me. I'd sent this beast packing. And since it's a big battery, can use it's own fans to propel it back to the deaf cheapskate vendor.

 

[Mr Evil]

If it's complaining even when you used resistors to draw the same current as the old fans, then perhaps it is detecting the speed of the fans (which is possible even with 2-pin fans by detecting the changes in current that happen every revolution). You could test that by slowing one of the old fans down (by pressing on the hub) until its slower than the new fans and seeing if that makes it complain.

 

[velis]

@ClockHound yes, I considered extending the case 5 or 10 cm and simply adding 2x 20cm fans which would definitely pull more air than these two POSs, but I really don't see how that could help if the unit still complains about the fans used. Also, so far I have seen no signs of overheating by using the new fans (the unit has temp reporting). I guess they do enough work at my intended load. An alternative is to cut an opening on the side and mount a large fan there so that it blows directly on the sinks. This one's even easier.

@Mr Evil If what you say is true, I need a fan that spins as fast as these two do, and that would have to be pretty fast judging from the noise. Any fan would be loud at those speeds. I find myself considering @ClockHound's proposal to try and clip the wings. That would presumably increase the spin, but reduce the vibration and turbulence, making the fan quieter while ensuring any "greater than" speed comparion from the unit to still compute. Since I already accidently broke 3 wings off one of them, I'll just proceed cutting the rest of them seeing as I already voided my warranty. Maybe the hoax will work and just maybe the fan will be quiet without the blades, but I'm not holding my breath. Alternatively, if this works, I could look for a 30mm or 40mm fan that would be quiet, but still spin fast enough.

I just got my hope back. I'll try this

 

[velis]

Success!!!!!!

Cutting the wings didn't increase spin rate. Guess I don't really know how DC motors work. Note to self: more education

Anyway, without the blades, the motor is now very quiet. Remaining is a bearing whine, quite quiet, but still the loudest thing in the final solution by far. It doesn't quiet down even if I put the foam around the fan. I'd have to completely wrap it to get any effect I guess. I guess that while cutting the blades away I unbalanced the rotor. This is indicated by noticeable vibrations when placed on a hard surface. Not really sure how to solve this (Edit: some 120 grit sandpaper did the trick), but I may not even need to. The entire thing is quite quiet now.

 

[velis]

Eh, no success . The unit didn't start complaining after 2 minutes as usual, but after 15. So the only working solution remains an unclipped fan wrapped in foam.

Edit: the second attempt is now silent for 30 minutes. Hope it holds...
Edit2: 1.5 hr. Getting my hopes back up
Edit3: Second attempt resulted in fan being "detected" after 5 hrs. Now going for timing session #3

If the damn thing would just report fan failure instead of battery failure, I'd just disconnect the beeper by now.

I'm assuming that @Mr Evil 's guess was correct. However, bladeless the fan doesn't seem to spin any faster, so why would the unit detect fan failure now? Maybe because of the other two fans? The funny thing is: once it starts beeping, it doesn't stop. I just left it beeping for 5 minutes, but it never changed its mind. I guess it lacks any algorithm to stop the alarm if the conditions for failure are no longer met. It needs to be powered down / back up in order to stop beeping.

 

[ClockHound]

What a beeping pain in the ear canal.

I'm guessing the flightless bird fan is not pulling enough current...not sure how much honey to drop into the bearings to simulate the air resistance at speed, but it's the internet so someone probably knows and is willing to argue about it.

Do you really need the beeping warning? Isn't the sudden loss of lights, EM fields and lack of refrigeration enough of an indicator? It's your call, but I'd almost be willing to trade catastrophic data loss in exchange for some peace and quiet. ;-)

 

[velis]

The third timing attempt is now at 10.5 hours, but I can't rely on this. It may decide to start beeping at any time. For what I know, it might have done that even before my modding, but not much can be done about it now. Besides, my willingness to play has diminished.

So, the beeper is scheduled for some surgery, yes.

 

[velis]

My final solution is to simply stop trying to fool the unit: I unsoldered the beeper. Now the unit is silent, does its work and I just ignore any errors. Fortunately the software only reports the error once and never again, so I still have battery level reporting without any nagging. Don't really care about anything else at this time.