Paul <nospam@needed.com> wrote in news:fpracp$qph$1@aioe.org:

> Mike wrote:
>> The fans were getting real noisy so I decided to replace the two 80mm
>> 12V fans. I bought two 12V fans and replaced the old ones making sure
>> I hooked up the right colored wires together, black to black and red
>> to red. For some reason neither fans worked when I turned on the
>> computer. The computer turned on without any problems. I tested the
>> new fans with a 9 volt battery and they worked. I even tried
>> reversing the wiring (black to red, black to red) without any luck.
>> Anybody know how I can get these fans to work?
>>
>> thanks
>>
>
> Did you check the old fan with the 9V battery ?
> Since you say the old fan was still working, but noisy, it
> should pass the 9V test as well. Always be careful with the
> polarity when using the battery method.
>
> By reversing the wiring on the new fan (black to red), you may have
> damaged it.
>
> I don't see a particular reason a power supply would care, but fans
> come in several types. If you buy a "case fan", it typically has two
> wires. Red and black go to +12V and GND (assuming it is a 12V fan and
> not some other voltage). At least the black color should be the
> ground, leaving the other color as the more positive connection.
>
> The next type, is the three wire fan. The third wire is for tachometer
> readout. But a PSU doesn't particularly care about RPMs. I have one
> power supply here, where they include a "sensor" cable, so the RPM
> signal from the fan can be connected to the motherboard fan header
> (I can read out the power supply RPMs as a result.) But the power
> supply itself, is probably not monitoring RPMs (it costs too much
> to bother).
>
> Another kind of three wire fan, includes a different signal on the
> third wire. Instead of a constantly pulsing signal, they put a
> logic signal called "locked rotor". If the fan detects that it is
> below a certain number of RPMs, the signal changes state. So that
> flavor of fan, has a signal that is "go/no-go" for the connected
> equipment. Such fans are included on "classy" equipment, such as
> lab equipment perhaps. Locked rotor would normally be a non
> pulsing signal, and stays in one state most of the time.
>
> So you have some steps to carry out, as part of your repair work.
>
> One is, to put the original fan back, and test that the power
> feeding the fan is still present.
>
> You also need to retest the replacement fans, to see if they're
> damaged by the reverse polarity. The company making the fans,
> doesn't usually go out of their way, to include a polarity
> reversal protection diode. (I haven't tested my fans, by
> reversing the polarity, so I can't tell you whether they'd
> be damaged or not. There can be either a chip inside the
> fan hub, or at least a couple transistors, so in fact not
> all fans have exactly the same design or components inside
> the hub.)
>
> A power supply can be "bench tested", before being placed back
> in the computer. For example, get a copy of the ATX standards
> for power supplies. Connect "PS_ON#" to COM or ground. When
> the power supply is plugged in and the switch on the back
> of the supply is "ON", the missing step is to connect a
> paper clip or jumper wire, from PS_ON# to COM.
>
> Having the cover in place on the power supply, protects you
> in the event that the main cap exploded. (Not that this
> is likely to happen, but the cap does hold a lot of
> energy. It is also charged to a lethal voltage when
> operating.)
>
> In a previous post I made on the subject, someone suggested
> that it is wise to always have a slight load on the supply.
> If you have a couple old hard drives you are no longer using,
> you might consider connecting them to the 1x4 Molex hard
> drive connectors. That would provide a very small amount of
> loading, about 5V @ 1A and 12V @ 0.6A or less. A couple of
> drives would double that load, to 2A and 1.2A. Some supplies
> actually have a minimum loading spec, printed on the label,
> and if your "dummy load" is not meeting the minimum stated
> value, then you'd look for another way to meet the
> minimum load.
>
> I have a load box for testing power supplies, that I built
> myself. I bought about $50+ of power resistors from the only
> good electronics store in town. They also had Mini-Fit Jr.
> crimp pins, so I could make a connector assembly. I included
> a fan in my load box as well, to blow over the power
> resistors. The box might draw about 100W or so, which is pretty
> small as loads go. The load box has a switch, which connects
> PS_ON# to COM.
>
> Using a box like that, I can test a new supply, before using it
> on a new build. I can connect my multimeter and verify the
> voltages under that 100W load. At least I can determine whether
> the supply is way off, on one of the output rails, before
> connecting an expensive motherboard.
>
> I've only ever replaced one fan on a supply, and the only reason
> for doing it, was the supply was from an old Macintosh
> computer. Like yours, the fan was noisy. Fortunately, the fan
> had the standard three pin connector on the end, so removing
> the old and inserting the new, was easy. The three pin header inside
> the supply, had the standard keying scheme, but I also verified that
> the wire color scheme on the old fan, the voltage etc, also matched,
> before doing it.
>
> For PC supplies, I'd just buy a new one. But for that Mac, finding
> something with the right power connector pinout, would have been
> a hassle. Probably nothing but pulls would have been available.
>
> Some power supply specs can be found here. The oldest spec is for a
> 20 pin supply which still has -5V. The second is for 20 pin supply,
> where the -5V has been removed. The third, is for a 24 pin supply,
> where the extra pins provide more current flow for some of the
> rails. You can find the color and location of PS_ON# in these
> docs.
>
> http://web.archive.org/web/200304240...factors.org/de
> veloper/specs/atx/ATX_ATX12V_PS_1_1.pdf
>
> http://www.formfactors.org/developer...X12V_1_3dg.pdf
>
> http://www.formfactors.org/developer...2_public_br2.p
> df
>
> Working inside a power supply, should only be attempted if you know
> what all the potential risks are. In my case, since the old fan
> could be unplugged, I didn't need to take the supply apart, and
> expose any of the high voltage areas. Otherwise, I might have just
> bolted the whole thing back together, and lived with the fan noise.
>
> Did I mention I don't really like high voltage ? Or getting shocks ?
>
>:-)
>
> The purpose of giving warnings, to stay out of power supplies, is
> some day, we're going to read in the newspaper, where someone
> was killed, because of advice given. And we definitely don't want
> that. It is not like you'd be posting back, if you were killed.
>
> Another reason the replacement fan might not have worked, is the
> PSU can use variable voltage to feed the fan. When the PSU is
> cool, at startup, it might deliver 7V to the fan wires. As the
> internal component temperature rises, the voltage is increased, to
> make the fan spin faster. It could be, that the new fan and old fan
> have a difference in the minimum starting voltage. For example,
> the other day, I saw a fan spec, where 10.8V was listed as
> the minimum starting voltage. There are some other 12V fans,
> that will start with as little as 5V present.
>
> Fans also have different air moving capacities, measured in CFM.
> Roughly speaking, fans comes in low, medium, high, and ultra
> ratings. When replacing a fan, you want to try to match the
> characteristics of the original fan.

Useful information about pinouts, power supplies and building a load box
too. Thanks Paul. I was unaware that PC fans were sensitive to polarity but
on reflection it makes sense since I assume they use a brushless motor in
order to avoid arching.

To the OP: A Digital Multimeter will indicate true negative in it's display
regardless of probe orientation so there's no guessing about wire color in
relation to D.C. polarity.

As an alternative to building the load box Paul describes you could use a
PSU tester:
http://www.newegg.com/Product/Produc...82E16899705003
Such a tester doesn't provide provide a full load test but will provide a
quick and convient test for functionality.

--
pcbldrNinetyEight