Fan monitors

  • Thread starter Thread starter Loren Pechtel
  • Start date Start date
L

Loren Pechtel

I'm looking for something that will simply tell me if the fans are
turning. I'm not after a fancy controller, just a set of LEDs.
Basically everything I'm finding is focused on controlling the speed,
monitoring is at best a secondary thought.

Any suggestions?
 
Loren said:
I'm looking for something that will simply tell me if the fans are
turning. I'm not after a fancy controller, just a set of LEDs.
Basically everything I'm finding is focused on controlling the speed,
monitoring is at best a secondary thought.

Any suggestions?
Click to expand...

If you cannot find a pre-built circuit, this is how you'd do it.

*******

If the fan has three wires (RPM output), look for simple circuits
with names such as "missing pulse detector" and using a 555 timer chip.

http://home.cogeco.ca/~rpaisley4/LM555MissingPulse1.GIF

Generally, you'd probably use two chips, one to do the basic missing
pulse or minimum frequency check, and a second chip to drive the LED.
Since the fan has to start, you also need a means to reset the circuit
for the first period of time (gate off the circuit while the fan
spins up, so the LED does not indicate falsely). An RC on a reset pin
somewhere can probably do that for you. I'd probably look for something
like a 4538 to work with (dual timer), as you can do RC tricks with them,
do AC coupling for the fan pulses, use a zener to clip off excursions
outside the logic range, and so on. The first timer does missing
pulse, and the second timer, you'd seek to catch a pulse from
the output of the first timer, and drive a LED at a constant level
with it.

http://www.jaycar.com.au/images_uploaded/CD4538BC.PDF

And this article claims to be able to do the job, using
nothing more than watching power consumption on a two wire fan.

http://www.micrel.com/_PDF/App-Notes/an-34.pdf

Another way to do it, would be with a frequency to voltage
converter. You could follow that with a simple thresholding
circuit, like an LM311 set to whatever level you want. The
LM311 comparator would drive the LED.

*******

Actually, the solution is simpler than that. You bought the
wrong kind of fan. There are fans available with "tachometer
output". Those are typical of computer usage, monitored by a
three pin fan header. But, if you buy a fan with a "locked rotor"
signal, that one doesn't pulse the third wire. It sends a
steady level on the third wire. One logic state for the
wire says "I'm spinning", while the other logic state says
"I'm not spinning". You can feed a locked rotor signal to
say, a transistor, and turn a LED on and off via "locked rotor".
The locked rotor signal may not be strong enough to drive
the LED directly, or if polarity inversion is needed, the
transistor can help with that.

A locked rotor output might be more typical of a fan within
an instrumentation device (expensive device with Panaflo
fans perhaps). So you have the possibility of building a
monitor, to look for missing pulses or minimum frequency,
or the nuisance of shopping for fans with optional "locked
rotor" output signal. A locked rotor fan still has three
wires, and the third wire has a steady logic level, rather
than pulsing twice per revolution.

http://www.coolerguys.com/lras.html

If you look at the locked rotor option here, they have
a sense of humor. A locked rotor fan may have
a true locked rotor signal, or, may have an output that
looks a lot like a tachometer output. I don't see the point
of the "case 2" output waveform. That isn't a locked rotor
indication. It's useless, unless followed by a
missing pulse circuit etc.

http://www.eminebea.com/content/html/en/engineering/fans/Sensor.shtml

Paul
 
If you cannot find a pre-built circuit, this is how you'd do it.

*******

If the fan has three wires (RPM output), look for simple circuits
with names such as "missing pulse detector" and using a 555 timer chip.

http://home.cogeco.ca/~rpaisley4/LM555MissingPulse1.GIF

Generally, you'd probably use two chips, one to do the basic missing
pulse or minimum frequency check, and a second chip to drive the LED.
Since the fan has to start, you also need a means to reset the circuit
for the first period of time (gate off the circuit while the fan
spins up, so the LED does not indicate falsely). An RC on a reset pin
somewhere can probably do that for you. I'd probably look for something
like a 4538 to work with (dual timer), as you can do RC tricks with them,
do AC coupling for the fan pulses, use a zener to clip off excursions
outside the logic range, and so on. The first timer does missing
pulse, and the second timer, you'd seek to catch a pulse from
the output of the first timer, and drive a LED at a constant level
with it.
Click to expand...

Deeper into electronics than I've ever gotten but I might just tackle
it as a learning project if I can't find any other answer.
Actually, the solution is simpler than that. You bought the
wrong kind of fan. There are fans available with "tachometer
output". Those are typical of computer usage, monitored by a
three pin fan header. But, if you buy a fan with a "locked rotor"
signal, that one doesn't pulse the third wire. It sends a
steady level on the third wire. One logic state for the
wire says "I'm spinning", while the other logic state says
"I'm not spinning". You can feed a locked rotor signal to
say, a transistor, and turn a LED on and off via "locked rotor".
The locked rotor signal may not be strong enough to drive
the LED directly, or if polarity inversion is needed, the
transistor can help with that.
Click to expand...

No--I haven't even attempted to monitor the wires yet. I was looking
for something on the market that would do it.
 
Get fans with LEDs.
Click to expand...

Wrong. You have to lick the back of your finger for a fan pulling air
into the computer to feel if a low RPM is there off a 100/120mm fan;-
those LED don't mean squat when it's not turning, except that it may
be a cheap fan with LEDs, when there are better fans with motors and
sleeves for their money without LEDs for the JedEye Mission.
 
Loren said:
I'm looking for something that will simply tell me if the fans are
turning. I'm not after a fancy controller, just a set of LEDs.
Basically everything I'm finding is focused on controlling the speed,
monitoring is at best a secondary thought.
Click to expand...

So why isn't software that monitors the fan speeds a sufficient
solution? Even software monitors that control fan speeds will report
what *is* the current fan speed.

I use Speedfan. Yes, it *can* control the fan speeds but that is NOT a
requirement of using it. You could just use it to monitor the speeds.
Your motherboard maker might also have their own monitor utility that
includes fan speed monitoring.
 
I'm looking for something that will simply tell me if the fans are
turning. I'm not after a fancy controller, just a set of LEDs.
Basically everything I'm finding is focused on controlling the speed,
monitoring is at best a secondary thought.

Any suggestions?
Click to expand...

SpeedFan. It's not a suggestion but The StandarD in terms of freeware
equivalence. In order to accomplish the JedEye Mission of attaining
The Force for your LEDs, you'll have to define an _Event_ with
Speedfan's temperature/fan_strobe signal monitoring. A critical
junction, should such an event occur, when your mission with be define
for Speedfan to spawn an external program resembling a Light Saber
LED, an elegant solution for older days when men were bolder. (I
suggest anything but the music Lucas's musical theme, as an
accompaniment it all too quickly becomes redundant, lacking impact;-
"Pop That Pussy," ***All Right reserved***, as composed, arranged and
performed by 2 Live Crew is presently the #1 billboard hit song for
America's premier Strip Clubs, among all sorts of choices, of course.
You must do you what you feel is right when in a den of villainousness
scum.)
 
So why isn't software that monitors the fan speeds a sufficient
solution? Even software monitors that control fan speeds will report
what *is* the current fan speed.

I use Speedfan. Yes, it *can* control the fan speeds but that is NOT a
requirement of using it. You could just use it to monitor the speeds.
Your motherboard maker might also have their own monitor utility that
includes fan speed monitoring.
Click to expand...

I've got a lot more fans in the case than headers on the motherboard.
 
I've got a lot more fans in the case than headers on the motherboard.
Click to expand...




http://geoffg.net/fancontroller.html


Try this. Supports up to 8 fans (though only 4 headers are shown, but
others can be added). and USB connectable in order to manage the
fans. IIRC once programmed by the USB, it can stay there, or continue
to operate "offline" for applications that don't have USB. The parts
cost shown is in AUD, it would likely be a lot less in the USA.
 
http://geoffg.net/fancontroller.html


Try this. Supports up to 8 fans (though only 4 headers are shown, but
others can be added). and USB connectable in order to manage the
fans. IIRC once programmed by the USB, it can stay there, or continue
to operate "offline" for applications that don't have USB. The parts
cost shown is in AUD, it would likely be a lot less in the USA.
Click to expand...

That's still about controlling their speed. The noise level is
acceptable, I don't care about that--I just want to make sure they're
actually spinning. If a fan dies you get part of the system with less
cooling but if it's not someplace with a sensor you get no warning.
 
If you cannot find a pre-built circuit, this is how you'd do it.

*******

If the fan has three wires (RPM output), look for simple circuits
with names such as "missing pulse detector" and using a 555 timer chip.

http://home.cogeco.ca/~rpaisley4/LM555MissingPulse1.GIF

Generally, you'd probably use two chips, one to do the basic missing
pulse or minimum frequency check, and a second chip to drive the LED.
Since the fan has to start, you also need a means to reset the circuit
for the first period of time (gate off the circuit while the fan
spins up, so the LED does not indicate falsely). An RC on a reset pin
somewhere can probably do that for you. I'd probably look for something
like a 4538 to work with (dual timer), as you can do RC tricks with them,
do AC coupling for the fan pulses, use a zener to clip off excursions
outside the logic range, and so on. The first timer does missing
pulse, and the second timer, you'd seek to catch a pulse from
the output of the first timer, and drive a LED at a constant level
with it.
RC on the reset (pin 4) would handle that.
Click to expand...


try say 10uF or so to ground and a 10k resistor to the supply.
If this is not long enough, try increasing these values.

This might need to be long, depending on how soon the motherboard
turns on the fans after it is powered up. If you have an audible alarm
it
could be quite annoying if you don't do this.

If you don't mind the "false alarm"
LED or other warning possibly being on during boot, then don't worry
about any of this.
Just put a 10k from pin 4 to the supply to keep it high


I would add a small capacitor in series with the input also,
because otherwise if the fan fails and stops in a position where the
tacho
signal is at 'high', (and therefore stays high) the missing pulse
detector will never detect it.
try say 1uf, 4.7 UF values. Put the positive side of the cap towards
the fan signal.

Finally, if you want to do a number of fans, the 556 has 2 555's in a
single package.
There is even a quad version !
http://www.datasheetcatalog.org/datasheet/philips/NE558N.pdf
 
I've got a lot more fans in the case than headers on the motherboard.
Click to expand...

They serve a purpose, an end. When the means is interrupted the end
is consequent. An end as an intended provision of the CPU, MB
chipset, GPU, and HD, also so monitored for assignment to
thermistors. Headers, if conditionally specified, is as well a
provisional intent to match for case design, which when exceeded may
involve drilling additional fan holes or adding a fan-control box into
the 5-1/4 drivebay, as feasible if not contradictory to excellent
designs economically implemented in today's garden variety desktop
size factors. Shoebox multimedia or rack server assemblies are not
within such means anymore than are flashing multicolored $300 game
cases;- they're contingencies, regardless of feasibility;- a brunt
being near five years now since engineering have pretty well nailed it
for heat issues, as a whole, from what we were dealing with at prior
Intel Pentium 3's and 4's, humping it over and past antiquated 2-1/2"
fan-cooled cases for 5/486's, the Athlon emergence and Cyrix/TI gear.
 
That's still about controlling their speed.  The noise level is
acceptable, I don't care about that--I just want to make sure they're
actually spinning.  If a fan dies you get part of the system with less
cooling but if it's not someplace with a sensor you get no warning.
Click to expand...



IIRC this device monitors up to 8 fans, and sends info on their speed,
and an alert to the PC if they fail.
As long as you have the software running on the PC, you will be
warned.
 
Franc said:
If you're using a 555, then it can easily drive a LED directly.

- Franc Zabkar
Click to expand...

The problem I was looking at, was the output of the retriggerable
monostable. If you're half way below the target frequency, the
output is a square wave, and the human eye integrates that
as a "half intensity" LED.

Having thought about it a bit, the best I could come up with,
was use the 4538, use both halves, the first to do the
missing pulse, fed to the second half. And the second half
would have a much longer time period. The output of the two
halves would be ORed together, with the second 4538 being
use to mask the pulsing activity of the first stage. The
4538 is edge sensitive, and edges from the output of the
first one trigger the second one. So if the first stage is
putting out a square wave (meaning, the fan is out of spec
and on the low side), the longer time constant of the second
stage, keeps the LED illuminated at 100% duty cycle (no blinking).

So it would take one chip (4538) plus a portion of a quad two input
gate to do the logical function on the two stages. Say, five
chips to monitor four fans.

I think that should give a good go-nogo indication. When the
system first starts, the LED comes on for a second or two
(as a lamp test). The LED would stay on, if the fan was jammed.
Since the 4538 is edge sensitive, there is no need to worry about
"stuck high" or "stuck low" behavior on the fan tacho output.
Only the edges matter.

There's still the matter of the input stage (components between
12V fan and 5V logic), but those can be cribbed from one of the
many SuperI/O data sheets. The CMOS 4538 is relatively high
input impedance, so that probably isn't a problem.

Paul
 
Back
Top