Hello! This is my first post to the forum. Today is the day that Math will save my ears! (long intro, scroll down for actual problem titled: 24 Volts -> 12 Volts)

My "desktop" pulls close to 1200 Watts (with all 5 monitors, laptops, and gaming rig under full load), according to the two UPS' I've burned out. So, I got my hands on a big powerful APC 2200VA unit allowing up to 1980 Watts (on a 30 Amp circuit). Since 1200 Watts is under the 15 Amps, I've installed a normal 15A plug on to it.

Powering the UPS unit on, I was greeted to the jet engines of the 24 Volt 6000rpm fans the unit has in it. Needless to say, this is unacceptable. Some searches on APC forums indicate that all APC Smart UPS units these days all have extremely loud fans in them, always running, even when powered off and unplugged from wall! This should be a bug! But, APC claims it's "normal". Considering my previous APC SmartUPS 2000VA I used in an old office didn't even have a fan running unless it was on battery, this incidated to me that there is something else going on (such as APC not caring about fan selection but wanting to keep the units from failing early).

Over the weekend, I dissassembled the unit and found the two 24V 80mm fans. I disconnected them, and closely monitored the temps inside the case as I stress-tested my equipment to around 1080 Watts. Turns out, my theory was correct: the fans aren't needed! The temps only rose 10C over the course of 2 hours of ~1000 Watts of power. I measured it in two places: next to the "heatsinks" that had 1 fan dedicated to, and the ambient air around the reducer coils on the other side of the unit. As a comparison, the heat being expelled from my desktop's radiators during gaming is 30C over room temp! Hot!

So, I am replacing the fans with some quality "quieter" rubber-isolated fans rated at about 2/3rds of the CFM the original unnecessarily-load fans are. Problem is, the fans are 12 Volts - not 24 Volts.


24 Volts - 12 Volts
And finally, I get to the point of my post...

First problem, I need to reduce the voltage from 24 Volts down to 12 Volts. Doing some searches online, I'm reminded of the Volt Divider rule. Simple math, but I remember the case where if I introduce a load in parallel to R2, then things change a bit...

Please check my math:

R1 = Resistor in series of the circuit.
RL = The resistance supplied by the load.
Rb = The bleeder resistance which is the resistance supplied by the resistor in parallel with RL.

Formulas:
RL = Volts / Amps
Rb = Volts / (Amps * .1)
It (Impendence Total) = Amps + (Amps * .1)

Rb is based on the theory that we want 10% of the load current to pass through it. Found an old post that mentioned that listed here.

Total volts in circuit: 24 V
RL is rated for 12 V at 0.17 Amps.

RL resistance = 12 V / 0.17 Amps = 70.59 Ω

Rb = 12 V / 0.017 Amps = 705.88 Ω (10% of 0.17 Amps)
Impendence Total = 0.17 Amps + 0.017 Amps = 0.187 Amps
Req = 12 V / 0.187 Amps = 64.17 Ω

R1 = 12 V (to be dropped) / 0.187 = 64.17 Ω

Checking my math:
0.187 Amps * (64.17 Ω + 64.17 Ω) = 24 V

Total Power = 24 V * 0.187 Amps = 4.488 Watts
R1 64 Ωsees 12 V, which is 12 V * 0.187 Amps = 2.244 Watts
Rb 705 Ω sees 12 V * 0.017 Amps (bleed) = 0.204 Watts

R1 = 64 Ω needs to be rated at 2.25 Watts
Rb = 705 Ω needs to be rated at 1/4 Watt


Combining resistors
Could you remind me: it is ok to place resistors in series to obtain the total reistance I need, correct?

I plan on combing five (5) 12 Ω rated at 3 Watts in series, plus three (3) 1.5 Ω 2 Watts to achieve the 64.5 Ω I need for R1. Though, I am calculating 2.25 Watts of load, the three smaller "Power" resistors are only rated at 2 Watts. It's all I can find...

I also plan on combining a 680 Ω with a 24 ohm to achieve 704 Ω at 1/4 watt for Rb, to be in parallel with the load (Fan).


Voltage Reducing
Finally, just need to confirm... The math still works if the APC unit lowers the voltage down to 16 Volts, correct? That seems to be what it idles at at all times (even when powered off, and unplugged from wall!).


Thank you advance! I know, it was a VERY long post... Let me know if I need to sum it up, and I'll remove the "bacl story" to the project.

 

DC fans have IC's inside the motors that do a number of things, regulating speed, protecting against over current if the blade is blocked, etc. Putting a resistor in series which limits peak current may not make them happy, especially at startup. I don't know if that is the case here, you might just use a cheapo regulator like an LM7812 which sells for about 20 cents. It would be dissipating about 2W so it would only need a tiny heatsink, maybe one square inch or so.

 

DC fans have IC's inside the motors that do a number of things, regulating speed, protecting against over current if the blade is blocked, etc. Putting a resistor in series which limits peak current may not make them happy, especially at startup. I don't know if that is the case here, you might just use a cheapo regulator like an LM7812 which sells for about 20 cents. It would be dissipating about 2W so it would only need a tiny heatsink, maybe one square inch or so.

Intriguing. Unfortunately, I don't know the first thing about regulators like that. Would you have a link where I could read more about them and how to calculate their usage?

As for 2 Watts, that's no biggie. The original fans were rated at 7 Watts total, which I am a far cry from. Also, 2.25 Watts would be used in my original plan above.

 

A 7812 is $2 or $3 at Radio Shack if you want to play with one, they are very simple 3 terminal devices, 1 is Vin (24V in your case), 1 is ground, 1 is 12V out. You will be dissipating 2.5-ish Watts if the fans draw 150-ishmA, so a decent sized heatsink in the airflow path would be needed.

Always design for a bit of overhead, for example, when the bearings get a bit worn or the blades get a coating of dust, current needs will rise.

 

A tiny aluminum stamped heatsink used with a TO-220 device can easily do 3W dissipation. You only need a thermal resistance of about 35C/W or less.

A 7812 probably does cost $2 at radio Shack so go to Fry's. They usually go for about 30 cents.

 

I think there is a much simpler way to solve this problem.

Consider this: If you had two identical 12 volt fans, you could just connect them in series across the 24 volt supply. However, if you connect two non-identical 12 volt fans in series across a 24 volt supply, there's the likelihood that the voltage across the fans won't be identical. The sum of the voltages across the two fans will have to be 24 volts, but they might divide as, for example, 15 volts across one fan and 9 volts across the other. This might not be good for the fans.

But, there's an easy fix for this problem. Get yourself a couple of 5 watt, 13 volt zener diodes, such as the 1N5350B (http://www.onsemi.com/pub/Collateral/1N5333B-D.PDF). You might measure your 24 volt supply and make sure that it's really 24 volts. You need to use a couple of 5 watt zeners whose breakdown voltage is slightly more than 1/2 of the actual voltage of the (nominal) 24 volt supply.

Connect one zener across each fan with such polarity so that if the voltage across a fan tries to rise above 12 volts, it will not be able to rise to more than 13 volts. The fan will be okay with one extra volt.

Normally, if a fan draws less current than the other fan which is in series, the voltage will rise across the fan which draws less current; a zener in parallel will carry the extra current needed by the other fan and cause the voltage across both fans to be nearly equal.

 

If the two motors are identical. Perhaps you can connect them in a series configuration

 

I think there is a much simpler way to solve this problem.

Consider this: If you had two identical 12 volt fans, you could just connect them in series across the 24 volt supply. However, if you connect two non-identical 12 volt fans in series across a 24 volt supply, there's the likelihood that the voltage across the fans won't be identical. The sum of the voltages across the two fans will have to be 24 volts, but they might divide as, for example, 15 volts across one fan and 9 volts across the other. This might not be good for the fans.

But, there's an easy fix for this problem. Get yourself a couple of 5 watt, 13 volt zener diodes, such as the 1N5350B (http://www.onsemi.com/pub/Collateral/1N5333B-D.PDF). You might measure your 24 volt supply and make sure that it's really 24 volts. You need to use a couple of 5 watt zeners whose breakdown voltage is slightly more than 1/2 of the actual voltage of the (nominal) 24 volt supply.

Connect one zener across each fan with such polarity so that if the voltage across a fan tries to rise above 12 volts, it will not be able to rise to more than 13 volts. The fan will be okay with one extra volt.

Normally, if a fan draws less current than the other fan which is in series, the voltage will rise across the fan which draws less current; a zener in parallel will carry the extra current needed by the other fan and cause the voltage across both fans to be nearly equal.

If the two motors are identical. Perhaps you can connect them in a series configuration

You know, you two are exactly right. I completely forgot about two equal loads in series divides the voltage across them equally. Yes, it was measured at 24.01 V, and yes they are 12 V fans, and equal sizing (the two I am replacing are equal). Therefore, wiring in series will produce a 12 V drop across each fan. And this works beautifully with the 16 Volts the UPS normally operates at, which will be 8 Volts to each motor.

That makes it extremely easy: series wiring of the two, nothing to buy! How come I didn't think of that... *doh*

I am glad I posted this. I now have two devices to read up about:

- Regulators
- Zener Diodes

As I can easily think of more uses for these.

Thank you all again! Series it is!

 

APC doesn't do anything unnecessary..
It typically takes a minimum of 2 hours for a product to reach thermal stability. During that time the temperatures of all components are slowly raising.
Measurements in only 2 points and not allowing thermal stability to be reached doesn't tell you squat.

If anything replace the fans with similar CFM units with a quieter noise rating. But I say just turn your speakers up

 

APC doesn't do anything unnecessary..
It typically takes a minimum of 2 hours for a product to reach thermal stability. During that time the temperatures of all components are slowly raising.
Measurements in only 2 points and not allowing thermal stability to be reached doesn't tell you squat.

Thanks. As mentioned in my original post, I measured it after 2 hours of use at the hottest points (I measured other places too, those two I talked about were the hot-spots). So yes, we are on the same page and I waited the proper time for the temps to stablize with a load for 2 hours (both with and without the fans running, total time about 8 hours over 2 days).

If anything replace the fans with similar CFM units with a quieter noise rating. But I say just turn your speakers up

Baby wouldn't like that (doubles as the baby room at night, and APC has elected to NEVER turn off the fans once the unit is plugged into the wall). The fans I have selected produce 80% of the CFM the original fans are rated at under full 24V load. Considering the fans never exceeded 16V from the UPS, even on battery and at 1000Watts load, tells me they won't even spin up to full speed (and thank goodness, they are VERY loud at full speed).

And, I lost my speakers - when the baby arrived. Besides, I love how quiet my huge gaming rig is... Don't want to ruin the experience by some thoughtless engineers that just wants to throw in some big loud fans to get the job done. Hence, why i outlined my max wattage (about 5/8ths of the max wattage rating of the unit), took my time to carefully measure the temps, with and with fans running and under load both times, and carefully planning "upgrades" for quiet operation for my "lower-than-threshold" use of the unit. Therefore, I feel I am in the right here.

 

I am glad I posted this. I now have two devices to read up about:

- Regulators
- Zener Diodes

As I can easily think of more uses for these.

Thank you all again! Series it is!

If you do connect the fans in series without zener diodes (which you will want to do because it's so easy!) be sure to check the voltage across each fan the first time you power up to make sure the voltage is dividing evenly. A slight unbalance won't hurt anything, but if it's too uneven you will be risking damage. There's also the problem of initial voltage division (at turn on) versus steady state condition. Also, as the fans age, things might change. You should include zeners if at all possible.

 

The problem with connecting 2 12V in series is that when one fails open, the other will stop running as well. Failing open is a common failure mode for when the bearings get hot, the internal temp fuse opens.

Failing short would make the other fan spin really fast, providing lots of CFM to make up for the failed fan, but I'm not sure how long it'd last at 100% overspeed.

 

If you do connect the fans in series without zener diodes (which you will want to do because it's so easy!) be sure to check the voltage across each fan the first time you power up to make sure the voltage is dividing evenly. A slight unbalance won't hurt anything, but if it's too uneven you will be risking damage. There's also the problem of initial voltage division (at turn on) versus steady state condition. Also, as the fans age, things might change. You should include zeners if at all possible.

I will read up on them and make a decision this weekend. I completely understand the risks (see next comment).

The problem with connecting 2 12V in series is that when one fails open, the other will stop running as well. Failing open is a common failure mode for when the bearings get hot, the internal temp fuse opens.

Failing short would make the other fan spin really fast, providing lots of CFM to make up for the failed fan, but I'm not sure how long it'd last at 100% overspeed.

You're exactly correct, and while I'll be looking into the "Zeners" and/or regulator approach.

I'd almost like to install some type of pot for fine adjustments, but that's just a wish list.

 

The problem with connecting 2 12V in series is that when one fails open, the other will stop running as well. Failing open is a common failure mode for when the bearings get hot, the internal temp fuse opens.

Installing zeners as I suggested takes care of a fan failing open; the other fan will continue to operate properly.

Failing short would make the other fan spin really fast, providing lots of CFM to make up for the failed fan, but I'm not sure how long it'd last at 100% overspeed.

Failing short is not common, but a fuse would take care of that eventuality. I suspect the original fans don't have a series fuse because these fans are commonly impedance protected.

 

Failing short is not common, but a fuse would take care of that eventuality. I suspect the original fans don't have a series fuse because these fans are commonly impedance protected.

FYI: I, uh, accidently "touched" the two leads of the volt meter as I was measuring the voltage for a fraction of a second... A whole bunch of relays flipped on/off, the green lights on the internal I/O board all turned RED and flashing, and the front display read "Code XX - Fan Fail".

So I would guess it, uh, has fan protection built in.