View Single Post
  #12  
Old January 20th 19, 12:40 AM posted to alt.windows7.general
Paul[_32_]
external usenet poster
 
Posts: 11,873
Default Case fan replacement OT

swalker wrote:
On Thu, 17 Jan 2019 11:51:18 -0600, swalker wrote:

I need to replace a 120mm case fan on my mid-size tower. Looking at
Amazon the prices are all over the place and I assume there might be
some differences, other than noise and air moved, between the top and
bottom price but I have no idea what those might be.

Opinion anyone?

Color doesn't matter.

Anybody like a particular brand?


Had the fan replaced by a local shop for $25 including a Noctua fan.
In the process found that another fan is in need of replacing and I
will do that one myself. The fan currently has a 3 wire connector so
that is what I will use. MB has only 3 pins where the connection is
currently made.

I use EVGA Precision XOC to monitor and control the fans that are
wired so I can. The one I am going to replace is one I would like to
monitor/control but because of the 3 wires I am guessing I will not be
able to.

Does that sound correct that only 4 wire fans can be monitored using
software?

Thanks again for the advice/help.

Jim


Four wire fan Three wire Two wire

+12V Red +12V +12V
GND Black GND GND
RPM (Yellow perhaps) RPM
PWM

adjust by PWM adjust by +12V adjust by +12V

On the fan, PWM is an input and RPM is an output.

You can place a four pin fan on a three pin header.
The "PWM" hangs off the end and doesn't touch any pin.

The PWM wire has a pullup resistor, so when it does not touch
anything, a solid logic 1 is on the wire. This also corresponds
to "max speed" on the fan itself.

To change the speed of the fan, the PWM signal is a square wave
where the duty cycle affects the fan speed.

If the PWM line was jammed to ground, that corresponds to a
"zero percent duty-cycle signal". But rather than the fan
stop spinning, the fan spins at "min-speed". When the signal
is logic 1 (+5V) all the time, that's 100% duty cycle and the
fan spins at "max-speed" in response. A 25KHz square wave with
a 50% duty cycle, would give a fan speed between the two
values. The only reason for selecting a high frequency
signal, is in the hope any side-effects don't alias into
the audible range. I don't have a scope here, and have not
verified the frequency used. But it's probably something at
least that high.

*******

The RPM signal has a couple definitions. On computers, it's
"2PPR" or two pulses per revolution, and this
can be derived from the motor windings and driver transistors.
In other words, they selected a "naturally occurring" signal
for that purpose.

In other industries, they prefer the "locked rotor" indication,
where the logic signal on the yellow wire is only activated when
the fan stops rotating. The fan part number might have "LR"
in the number, if locked rotor is used.

*******

When a fan has the small four pin connector, it's intended for
PWM control. You should not voltage modulate the +12V on a PWM
fan, because there is a MOSFET in the series path to the motor
windings. And the MOSFET needs enough voltage to operate properly
and not be "half-saturated" or something.

Intel wrote a spec for PWM fans, which can be used as a
reference document on the topic.

When a fan has three wires and no PWM (no internal MOSFET),
you can adjust the +12V input to at least as low as 7V or so.
A few fans continue to rotate down to around +5V or so.
Older motherboards had their own control scheme on the
motherboard, to deliver a desired voltage level. Since it
might cost a dollar per channel to do that, companies like Asus
only put one channel on the motherboard with that voltage control.

OEM HP machines, some of the motherboards in those had three
controlled fans. A sort of "deluxe version".

If you have a 2 wire fan, it's just +12V and GND and no
RPM at all.

if it was my system, I'd try to select a Noctua that has
the right airflow when the +12V rail is at +12V. And not
bother with adjusting it. Some of those fans are pretty
quiet at full speed. I have a CPU cooler made by Noctua
and the rear case fan is louder than the CPU fan. While
the CPU header is a four pin, I don't bother turning
down the fan and just leave it running full speed.
Which might only be 1200 to 1400 RPM or so.

I don't really want to be a brand promoter, but if
I visit my computer store now, there's really nothing
there that's worth buying. There aren't many fans where
you'd guess they will have a long life, or be reasonably
quiet. There's plenty of LED junk if that's what you
like.

The electronics store has Nidec and Delta fans, and
those are good workhorse fans with high air output
(and noisy). Those have ball bearings and good service
lives. Perfectly good for server room equipment. The
"quiet fan business" is an entirely different matter.
Good fan brands come and go (haven't seen a Panaflo
or a Vantec Stealth in a while).

Paul
Ads