Fan Yellow Wire...

[lupo666]

Hi all,

I have purchased a new front fan for my case and did the 7-volts trick,
using the supplied molex adapter and switching the cables around and
everything works fine.

The only problem I have is that I would like to use the speed sensor
(yellow wire). I have found an extra 3-pin connector from a broken fan
and plugged the yellow wire in the proper position, but it doesn't get
read from SpeedFan or ASUS PC Probe.

What would be the proper wiring to use, if doable at all?

Thanks, Lupo

 

[Paul]

Hi all,

I have purchased a new front fan for my case and did the 7-volts trick,
using the supplied molex adapter and switching the cables around and
everything works fine.

The only problem I have is that I would like to use the speed sensor
(yellow wire). I have found an extra 3-pin connector from a broken fan
and plugged the yellow wire in the proper position, but it doesn't get
read from SpeedFan or ASUS PC Probe.

What would be the proper wiring to use, if doable at all?

Thanks, Lupo

If you are using the 7 volt trick, I don't see a good reason
why the tacho output signal would work. The tacho output is
ground referenced, and when you connect the fan black and red
wires to +5V and +12V respectively, the tacho is referenced
to +5V, and can no longer pull to ground. That is bound to
upset the monitor chip, and it will not see a signal.

If you want to drop the voltage to the fan, and have the
tachometer output work also, try this:

| | | | | | |
+12V ---|>|--|>|--|>|--|>|--|>|--|>|--|>|------- to fan +12V
| | | | | | |

GND ------------------------------------------- to fan GND

The component in the picture, is a 1N4001 switching diode.
Any of the family of diodes would work, from 1N4001 through
1N4007, and they are rated for 1 ampere of current max.
Each diode will drop roughly 0.7V, (12 - 7*0.7) = 7.1 volts.
I have a string of those diodes in my gaming PC, for the main fan.

You don't need solder, and you can just twist the leads of the
diodes into a chain. My local electronics store sells these
diodes in bags of 25 for about $3 or so. Many other sources
sell them for a lot more, so shop around.

There is a band marking on one end of the diode, and in the
above picture, the band is on the right hand side of each
diode shown in the picture.

Make sure you insulate the leads on the diodes, so they
cannot touch anything in the computer. Mounting the diodes
on a piece of perfboard would be an alternative way to mount
them, and the perfboard will make it easier to come up
with a techique to mount the diodes inside the computer.

To make the fan run faster, use fewer diodes.

To make the fan run slower, use more diodes.

With the above diode technique, the tachometer signal will work.

You can also try running the fan from +5V and GND, which
will make the fan a bit slower than it is at 7V, but again,
the tachometer signal will pull to ground properly. Some fans
will not start with just 5V on them, so you'll have to test
it and see what happens.

Just a guess,
Paul

 

[UCLAN]

If you are using the 7 volt trick, I don't see a good reason
why the tacho output signal would work. The tacho output is
ground referenced, and when you connect the fan black and red
wires to +5V and +12V respectively, the tacho is referenced
to +5V, and can no longer pull to ground. That is bound to
upset the monitor chip, and it will not see a signal.

If you want to drop the voltage to the fan, and have the
tachometer output work also, try this:

| | | | | | |
+12V ---|>|--|>|--|>|--|>|--|>|--|>|--|>|------- to fan +12V
| | | | | | |

GND ------------------------------------------- to fan GND

[...snip]

Why not just use a 5.1v zener diode?

 

[Paul]

If you are using the 7 volt trick, I don't see a good reason
why the tacho output signal would work. The tacho output is
ground referenced, and when you connect the fan black and red
wires to +5V and +12V respectively, the tacho is referenced
to +5V, and can no longer pull to ground. That is bound to
upset the monitor chip, and it will not see a signal.

If you want to drop the voltage to the fan, and have the
tachometer output work also, try this:

| | | | | | |
+12V ---|>|--|>|--|>|--|>|--|>|--|>|--|>|------- to fan +12V
| | | | | | |

GND ------------------------------------------- to fan GND

[...snip]

Why not just use a 5.1v zener diode?

Power dissipation ? You wouldn't want to use one of those tiny
400mW ones. And a 5 watt zener is $1.20 at Digikey, and I think
I can get the 1N4000 series cheaper.

Even with the diode scheme, you shouldn't really run it all the
way up to an amp. A couple hundred milliamps probably won't hurt.

I was going to mention a three terminal regulator and a couple
0.22uF decoupling caps as another solution, but I don't know if
those are that easy to find. And that is too much of an
electronics project Next, people will want a pot on it,
so they can dial the voltage.

LM317 $0.75 and a better way to go...
For heavy currents, you'll need a heatsink.
http://www.national.com/ds.cgi/LM/LM117.pdf
http://www.digikey.com/scripts/DkSearch/dksus.dll?Detail?Ref=365239&Row=172496&Site=US

Paul

 

[kony]

If you are using the 7 volt trick, I don't see a good reason
why the tacho output signal would work. The tacho output is
ground referenced, and when you connect the fan black and red
wires to +5V and +12V respectively, the tacho is referenced
to +5V, and can no longer pull to ground. That is bound to
upset the monitor chip, and it will not see a signal.

If you want to drop the voltage to the fan, and have the
tachometer output work also, try this:

| | | | | | |
+12V ---|>|--|>|--|>|--|>|--|>|--|>|--|>|------- to fan +12V
| | | | | | |

GND ------------------------------------------- to fan GND

[...snip]

Why not just use a 5.1v zener diode?

More expensive and harder to come by (for someone who didn't
already know about & have one). Someplace like Radio Shack
might have 1W zeners but that's borderline and more common
are 500mW. Ideally, the control method will make the fan
subsystem more reliable instead of less-so, but does the
average person have > 1W zeners lying around? Doubtful,
most good designs don't use higher value zeners at all.

The optimal solution though is not a purely voltage limiting
control, rather current limiting. Something like an LM317
in current limiting configuration would work but far easier
and cheaper to just use a 2W resistor. A fan will run
quieter on a resistor than is possible with diodes or any
other means of voltage reduction. Typical value of resistor
depends on the fan spec, typically around 47 Ohm for a
higher speed fan or approaching 120 Ohm for lower speed
fans. Since resistors are a few cents each it's often
worthwhile to just buy a couple of the common values like
47, 68, 82, 100, 120 Ohms or at least every-other value.

 

[UCLAN]

If you are using the 7 volt trick, I don't see a good reason
why the tacho output signal would work. The tacho output is
ground referenced, and when you connect the fan black and red
wires to +5V and +12V respectively, the tacho is referenced
to +5V, and can no longer pull to ground. That is bound to
upset the monitor chip, and it will not see a signal.

If you want to drop the voltage to the fan, and have the
tachometer output work also, try this:

| | | | | | |
+12V ---|>|--|>|--|>|--|>|--|>|--|>|--|>|------- to fan +12V
| | | | | | |

GND ------------------------------------------- to fan GND

[...snip]

Why not just use a 5.1v zener diode?

Power dissipation ? You wouldn't want to use one of those tiny
400mW ones. And a 5 watt zener is $1.20 at Digikey, and I think
I can get the 1N4000 series cheaper.

My time is worth a lot more than the difference in price between
8- 1N4000 series diodes and ONE $1.20 zener. The installation is
not as messy, either. A chassis mount 100 ohm pot is the solution
*I* use on my machine.

 

[UCLAN]

If you want to drop the voltage to the fan, and have the
tachometer output work also, try this:

| | | | | | |
+12V ---|>|--|>|--|>|--|>|--|>|--|>|--|>|------- to fan +12V
| | | | | | |

GND ------------------------------------------- to fan GND

[...snip]

Why not just use a 5.1v zener diode?

More expensive and harder to come by (for someone who didn't
already know about & have one). Someplace like Radio Shack
might have 1W zeners but that's borderline and more common
are 500mW. Ideally, the control method will make the fan
subsystem more reliable instead of less-so, but does the
average person have > 1W zeners lying around? Doubtful,
most good designs don't use higher value zeners at all.

As I said, my time is worth a lot more than the difference in
price between 8 diodes and ONE 1W zener. Construction ease and
"neatness" is also better with the zener. If I want to make it
difficult, I'll use a 100 chassis mount pot.

 

[lupo666]

Hi, thanks for the replies, they were very helpful. The reason I chose
the 7V trick is because is clean (just switched the molex wires around)
and gives me more than 5V, which is a bit too low. I think, in the end,
I'll do w/o the fan monitoring, because is not the CPU or power fan and
just keep monitoring the HDs, MB and CPU temperatures for any
anormalities...

THX, Lupo

 

[eight]

If you want to drop the voltage to the fan, and have the
tachometer output work also, try this:

| | | | | | |
+12V ---|>|--|>|--|>|--|>|--|>|--|>|--|>|------- to fan +12V
| | | | | | |

GND ------------------------------------------- to fan GND

[...snip]

Why not just use a 5.1v zener diode?

More expensive and harder to come by (for someone who didn't
already know about & have one). Someplace like Radio Shack
might have 1W zeners but that's borderline and more common
are 500mW. Ideally, the control method will make the fan
subsystem more reliable instead of less-so, but does the
average person have > 1W zeners lying around? Doubtful,
most good designs don't use higher value zeners at all.

As I said, my time is worth a lot more than the difference in
price between 8 diodes and ONE 1W zener. Construction ease and
"neatness" is also better with the zener. If I want to make it
difficult, I'll use a 100 chassis mount pot.

It's still a sub-optimal solution.

Do you expect anyone to have these > 1W zeners?
Did you expect a local shop to have them?
Unlikely, you'd be special-ordering them at a price premium because
very few people use them... why would they when a typical 10 cent zener
and 50 cent Darlingtion can be made into a pass-transistor arrangement
if you want more current... Then it's still much lower than it's
current rating too.

A zener simply isn't optimal, nor are diodes. Current limiting is, and
is possible just as easily, cheaper, and with more readily available
parts when using resistor(s).

The value of your time is in special-ordering these zeners when it
takes far less time to find the resistors, or they're common enough
some might already have them. I keep random stock of misc. electronic
parts here, including zeners, diodes, resistors and quite a bit more.
Generic 400x diodes I have plenty of, 2W resistors I do too, but off
hand I recall no > 1W zeners at all. They're just not the best
solution thus nobody stocks them.

 

[UCLAN]

It's still a sub-optimal solution.

Do you expect anyone to have these > 1W zeners?

Do you expect the average reader of this NG would have a stash of
1N4000 series diodes?

Did you expect a local shop to have them?

My local Radio Shack has them. And all online parts outlets, such
as DigiKey and Newark, have them.

Unlikely, you'd be special-ordering them at a price premium because
very few people use them... why would they when a typical 10 cent zener
and 50 cent Darlingtion can be made into a pass-transistor arrangement
if you want more current... Then it's still much lower than it's
current rating too.

Right...and the average NG reader has *those* on hand. Uh huh...

A zener simply isn't optimal, nor are diodes. Current limiting is, and
is possible just as easily, cheaper, and with more readily available
parts when using resistor(s).

So you are in agreement with my 100 ohm chassis mount pot solution?

The value of your time is in special-ordering these zeners when it
takes far less time to find the resistors, or they're common enough
some might already have them. I keep random stock of misc. electronic
parts here, including zeners, diodes, resistors and quite a bit more.

It takes me just as long to order 1N4000 series diodes online as it does
to order the correct zener. I don't care what random stock you keep. The
average reader of this NG does not, and will likely end up ordering his
parts. The 100 ohm pot is available at Radio Shack as well. Use high fan
speed when you need it; slow it down when you don't.

 

[kony]

Do you expect the average reader of this NG would have a stash of
1N4000 series diodes?

Well... if they have the soldering iron, the basic ability to orient
the polarity of the (1N400x or zener) and heatshrink it (or whatever
covering), yes I do feel if they have these basic skills and any parts
lying around, 1N400x diodes are as common as dirt and almost as cheap,
as are many resistors. Look at all the electronics you have in your
home, I meant retail products. Count the # of > 1W zeners in them,
I'll bet the # is low if not zero.

My local Radio Shack has them. And all online parts outlets, such
as DigiKey and Newark, have them.

You are fortunate to have such a well stocked Radio Shack then, they
dont' even offer over 1W zeners online. Sure, Digikey and Newark, et
al do, but if you're going to order from a big electronics house like
those then it makes all the less sense to use a zener. Voltage
reduction is not the actual goal here, it is merely assumed (by some)
that it is the means towards the end of reducing fan speed (mostly to
combat noise). Current, not voltage, limiting is the more effective
way to do this.

Right...and the average NG reader has *those* on hand. Uh huh...

Well you ARE the one who brings up Radio Shack or Digikey/etc. If
you're going out to buy the things then a 15 cent 500mV zener and 1A
pass transistor is better than a 5W zener and costs less too.

So you are in agreement with my 100 ohm chassis mount pot solution?

Given one of enough current capablity to be called a rheostat, it'd
work, but is a personal decision and I can only give the reasons why I
wouldn't.

- More expensive
- I don't need to change the fan speed, it gets set right at first and
never fiddled with. Fine-tuning down to a dozen RPM one way or the
other is pretty insignificant, IMO, and might not be obtainable anyway
on the wire-wound types since there is an inherant granularity to the
adjustment when it hits each wrapping of the element wire.
- Another mechanical part subject to failure. Remember my ideal is 50
years (even if not really used that long), and it's quite clear that
pots have potential to degrade in far less time than that, especially
if a marginally-current-rated "pot" subject to thermal cycle wear as
well as the other more obvious mechanical issues.
- You need a place to mount it, and it can look a bit ghetto unless
you have a fancy knob and even then it can look out of place. The
simple inline resistor has unde 1 inch of non-flexible segment in the
fan lead and can be completely unobtrusive. I've used them for many
years and they work great.
- 100 Ohm can be a bit on the low side. I typically buy
manufacturers' "low" speed fans to begin with and they may need a
little higher than 100 ohm, particularly on larger than 80mm fans.
There are ways around that if one had the rheostat, having a couple
diodes in series too, but unless that adjustability is needed the time
to do it goes up, as well as cost, size. At lower cost than such a
rheostat is also a linear such as LM317 plus one resistor, in
current-limiting mode. That is a bit more involved though, and why I
seldom suggest it as it has no particular benefit over the higher
wattage resistor alone, actually less of a benefit because of the vdrop
of the regulator being part of the power loss to the extent that the
resistor used with the LM317 is a lesser current limiter as % of
incoming voltage opposed to using only the resistor. The difference is
only slight though, could depend on a specific fan whether it mattered
in practice.

It takes me just as long to order 1N4000 series diodes online as it does
to order the correct zener. I don't care what random stock you keep. The
average reader of this NG does not, and will likely end up ordering his
parts. The 100 ohm pot is available at Radio Shack as well. Use high fan
speed when you need it; slow it down when you don't.

I'll bet it doesn't, actually take as long. Type 1N4001 into one of
the surplus electronics 'sites (else if going with Digikey, you pay the
small-order fee and that $1.42 zener ends up costing $1.42 + $5 small
order fee + (at least) $3 shipping = $9.42. For about that cost you
can get a few hundred 1N4001 delivered from a surplus 'site or at least
faster since you don't have to pour over the specs as with the zener.
Keep in mind though, I"m not the one who advocated the string of
1N4001. They will work, but I advocate a power resistor. You can get
a bag of 100 for about $2 at one of the surplus 'sites, or just spend
the $1 for only a handfull elsewhere.

To put it to the test though, right now I'll go to Digikey...
Enter in search term "1N4001" and you get one page of hits, can just
buy the very first item that appears and do fine. If you feel like
scrolling down a dozen lines they're a couple cents cheaper.

Next at Digikey I enter "5.1V zener". Immediately the next page causes
the user to choose package type. Who, not knowing already to use a
zener and having read your reply, will know the packages they come in?
Regardless, let's support they skip that and choose at least 1.5W for
some reasonable margin. OOPS! Those are all surface mount parts.
Hopefully they are experienced enough to realize this, but maybe not,
and either way, this is certainly taking a lot longer than getting
those 1N4001. So we go back and choose 2W zener, get one choice in
small quantity at $1.42 each. Ok so far, except we now need to
recognize that it's DO-41, meaning it's heat density is higher than any
other alternative... hook it up to the wrong fan and you will have to
isolate it somehow, because it will get hot. They're often rated up to
the (n)W with the disclaimer than to handle this, their leads are
soldered into a board to 'sink away some of that heat. You may find
that a 2W zener isn't even enough, so do you know this ahead of time
(does someone else buying them), or do they now go back to Digikey and
spend at least another $X for the 3W size. Oops, those are all surface
mount too. This is taking a lot longer.

OK then, back one page at Digikey to the 5W zeners. Hmm, they don't
stock any!

Ok, now go to Mouser or Newark or (insert your fav. here), and
lather-rinse-repeat.

I didn't suggest the resistor just for the heck of it, it's an ideal
way to do this _unless_ one must have adjustability later. I find I
don't and have a TON of systems with the resistors... even having
zeners and pass transistors, 1N4001, LM317, et al here to choose from.
Even so, the other methods will work provided the details (like Zener
wattage rating) are kept in mind.

 

[UCLAN]

Well... if they have the soldering iron, the basic ability to orient
the polarity of the (1N400x or zener) and heatshrink it (or whatever
covering), yes I do feel if they have these basic skills and any parts
lying around, 1N400x diodes are as common as dirt and almost as cheap,
as are many resistors. Look at all the electronics you have in your
home, I meant retail products. Count the # of > 1W zeners in them,
I'll bet the # is low if not zero.

Again, you are answering a question I asked of (e-mail address removed).
Strange. But, I suspect that the percentage of readers of this NG
that have 1N4000 series diodes on hand is very small. What I or you
have on hand is irrelevant, really. And yes, I have a bag of zeners
on hand, many of the 1W or greater.

You are fortunate to have such a well stocked Radio Shack then, they
dont' even offer over 1W zeners online. Sure, Digikey and Newark, et
al do, but if you're going to order from a big electronics house like
those then it makes all the less sense to use a zener. Voltage
reduction is not the actual goal here, it is merely assumed (by some)
that it is the means towards the end of reducing fan speed (mostly to
combat noise). Current, not voltage, limiting is the more effective
way to do this.

The only way to reduce the current that DC computer case fans draw
is to reduce the voltage across them. A resistor in series drops
the voltage across the fan simply by dropping voltage itself. I
would *really* be interested in your solution of reducing fan speed
of a 12v fan while still giving it 12v.

Well you ARE the one who brings up Radio Shack or Digikey/etc. If
you're going out to buy the things then a 15 cent 500mV zener and 1A
pass transistor is better than a 5W zener and costs less too.

LOL! His ([email protected]) comment about Daringtons, etc., came
*before* I mentioned Digikey - in the *same* post that he (you?)
said that my choice of using a zener was dumb because the average
user wouldn't have them on hand. My point was (which you realize)
that if the user didn't have the zeners, he isn't likely to have
the Darlingtons.

Given one of enough current capablity to be called a rheostat, it'd
work, but is a personal decision and I can only give the reasons why I
wouldn't.

Current capability is not a factor in determining if a device is a
rheostat or a potentiometer. How the terminals of the device are
connected in a circuit is the determining factor. A potentiometer
with its center contact shorted to either end becomes a rheostat.
A low current potentiometer becomes a low current rheostat. A chassis
mounted potentiometer wired as a rheostat makes an excellent fan
speed control.

- More expensive

Nonsense. More expensive than what? The cheapest solution to variable
fan speed control available. Many fans now come with a small pot inline!

- I don't need to change the fan speed, it gets set right at first and
never fiddled with. Fine-tuning down to a dozen RPM one way or the
other is pretty insignificant, IMO, and might not be obtainable anyway
on the wire-wound types since there is an inherant granularity to the
adjustment when it hits each wrapping of the element wire.

Many people want high fan speed when it's needed, and lower fan speed
when the computer is on, but idling. Soldering ONE pot is certainly
easier than soldering seven or eight diodes.

- Another mechanical part subject to failure. Remember my ideal is 50
years (even if not really used that long), and it's quite clear that
pots have potential to degrade in far less time than that, especially
if a marginally-current-rated "pot" subject to thermal cycle wear as
well as the other more obvious mechanical issues.

Horse manure! You want a part that provides 50 years of life in a PC
that is upgraded every few years? And your thermal issues are without
merit. The entire case is a heatsink, though not needed in such a
low power application. You're grasping at straws here.

- You need a place to mount it, and it can look a bit ghetto unless
you have a fancy knob and even then it can look out of place.

An unused drive bay cover is my usual candidate. Or, if none available
(or if rear mounting is preferred), an expansion slot cover.

- 100 Ohm can be a bit on the low side. I typically buy
manufacturers' "low" speed fans to begin with and they may need a
little higher than 100 ohm, particularly on larger than 80mm fans.

Low speed fans are not the usual candidates for quieting, are they?
Even so, just 50mA still drops 5v across 100 ohms. 100 ohms provides
much better resolution than does a 250 ohm or 500 ohm pot, and is
large enough 99% of the time.

To put it to the test though, right now I'll go to Digikey...
Enter in search term "1N4001" and you get one page of hits, can just
buy the very first item that appears and do fine. If you feel like
scrolling down a dozen lines they're a couple cents cheaper.

Next at Digikey I enter "5.1V zener".

Ah, comparing exact part number of the diode to a general description
for the zener. Cute, but apples vs. oranges. Try entering the part
number of the zener. What? That would ruin your test?

I didn't suggest the resistor just for the heck of it, it's an ideal
way to do this _unless_ one must have adjustability later. I find I
don't and have a TON of systems with the resistors...

I've used resistors, too. My suggestion of the zener was in response
to the suggestion of using SEVEN or EIGHT diodes connected in series.
Too messy, and too much of a chance that Joe Sixpack would get even
ONE diode backwards. Then you'd have a REALLY quiet fan!

End of thread, as far a I'm concerned.

 

[kony]

The only way to reduce the current that DC computer case fans draw
is to reduce the voltage across them. A resistor in series drops
the voltage across the fan simply by dropping voltage itself. I
would *really* be interested in your solution of reducing fan speed
of a 12v fan while still giving it 12v.

It's not my problem if you can't understand the difference.
You'd have to consider how a fan draws current, it is not a steady
current draw. Limiting with the resistor does control this better, it
is at the core of what voltage limiting vs current limiting, is,
regardless of fans.

LOL! His ([email protected]) comment about Daringtons, etc., came
*before* I mentioned Digikey - in the *same* post that he (you?)
said that my choice of using a zener was dumb because the average
user wouldn't have them on hand. My point was (which you realize)
that if the user didn't have the zeners, he isn't likely to have
the Darlingtons.

It's pretty simple. It's unlikely most people have the >1W zener, so
(excepting you, apparently), they can't make do with parts they already
have. Since they have to order online to get these parts, they then
have a choice of ANYTHING, and it would be a poor choice to choose
that >1W zener when they had the other options at the electronics
houses.

Note this is not just my opinion, show us any manufacturer of fan
products that uses a single zener (alone) to control fan speed. Is a
zener a *secret* part? No, they choose not to use them.

Current capability is not a factor in determining if a device is a
rheostat or a potentiometer. How the terminals of the device are
connected in a circuit is the determining factor. A potentiometer
with its center contact shorted to either end becomes a rheostat.
A low current potentiometer becomes a low current rheostat. A chassis
mounted potentiometer wired as a rheostat makes an excellent fan
speed control.

In practice, you're either buying a lab-grade pot at extreme expense if
it can handle the current or it's called a rheostat. Not theory here,
I mean actually going and buying the thing.

Nonsense. More expensive than what? The cheapest solution to variable
fan speed control available. Many fans now come with a small pot inline!

You dont' understand how those work. Fan manufacturers (real
manufacturer, not the PC-parts relabeler), include a control wire out
from the PCB of the fan to vary the resistance on the circuit board.
It is a distinctly different way of doing it as it does not pass the
current, only providing bias (in many cases).

You cannot take one of those low (fraction of a) watt pots and use it
on a fan as a resistance to the load current, it'll burn up immediately
if not a few days later.

Many people want high fan speed when it's needed, and lower fan speed
when the computer is on, but idling. Soldering ONE pot is certainly
easier than soldering seven or eight diodes.

Actually they may assume it but don't really know what they need. A
properly designed chassis cooling system can handle full load WITHOUT
the fans running at high speed. IMO, it's just silly to reach down and
fiddle with a pot and have that higher noise when one can have it
quieter and never have to adjust it. You are too consumed with control
methods and not enough with fan selection. I actually DO THIS stuff
constantly, it works fine to have single speed fan, no overheating even
with overclocking. Even so, if it's your box and you want a rheostat,
have fun with it even though it's not necessary for cooling or noise
reasons unless the system owner made a bad case choice and refuses to
correct that mistake instead of trying a work-around.

Horse manure! You want a part that provides 50 years of life in a PC
that is upgraded every few years? And your thermal issues are without
merit. The entire case is a heatsink, though not needed in such a
low power application. You're grasping at straws here.

You are clueless, the entire case is not a heatsink.

An unused drive bay cover is my usual candidate. Or, if none available
(or if rear mounting is preferred), an expansion slot cover.

Those will work, go ahead and do it. It's not necessary though, more
expensive and less reliable. Seems the only reason to do it is if
prior case or fan choices were poor or if you are just to stubborn to
accept that there is a better way.

Low speed fans are not the usual candidates for quieting, are they?

yes, why on earth would someone who wants quiet fans, start out by
buying a higher speed fan? That would just be silly (Maybe stupid),
and offers lower ability to throttle the fan back to lower speed no
matter what the control method. In other words, a fan starting out
spec'd as low speed will typically run to a lower (reduced through a
rheostat or whatever other method) stable (and noiseless, lack of
pulsating) speed than one spec'd as a higher speed @12V but then
reduced in RPM.

If you don't realize this, I suspect you lack enough fan experience to
properly evaluate reduction methods. What you find works acceptibly on
one make of fan and one speed, isn't necessarily valid for a different
fan make, model or speed. YOur zener idea cuts margins pretty close
unless one finds 5W zeners or deals with 'sinking them as mentioned
previously. Diodes are more subject to failure with overload than
resistors too. So if you have a specific zener or pot and specific fan
that work, it might be important to specify these specific parts else
it may not work so well.

Even so, just 50mA still drops 5v across 100 ohms. 100 ohms provides
much better resolution than does a 250 ohm or 500 ohm pot, and is
large enough 99% of the time.

You dont' need much better resolution in most cases, merely to be able
to hit rougly 10 Ohm near enough to the target. If what you need for
very quiet (near if not entirely silent) operation is about 125 Ohms,
then 100 Ohms just won't cut it. It is in fact very common to need
about 120 (+-10) Ohm with optimal low speed fans from many of the major
manufacturers, unless the user wasn't going for such a low speed to
begin with and if they aren't going for such a low speed, they might as
well just buy that low speed fan to begin with and it might be what
they wanted, ie- about 2200 RPM. I aim lower than 2200 RPM though,
often under 1000 RPM is possible, particularly with chassis that accept
120mm fans.

Ah, comparing exact part number of the diode to a general description
for the zener. Cute, but apples vs. oranges. Try entering the part
number of the zener. What? That would ruin your test?

Nope, because if you kept reading you'd realize they don't have stock
at all of the 5W zeners.

So in conclusion, a 5W zener will work but there's no reason to use one
over other methods, the best of which being a power resistor if later
adjustment isn't necessary.

If adjustment is necessary and you want only a low-power-rated pot, you
want that pot in the voltage divider feedback of a (usually integrated)
regulator, NOT handling the load current. To handle the load current
you will have to spend a lot more for the rheostat or even rarer pot,
and it'll likely be wire-wound which has the granularity problem
inherant in most cases (else an even more expensive pot type, could
easily cost over $20 for that part alone w/o shipping or small order
fees, etc).

 

[UCLAN]

It's not my problem if you can't understand the difference.
You'd have to consider how a fan draws current, it is not a steady
current draw. Limiting with the resistor does control this better, it
is at the core of what voltage limiting vs current limiting, is,
regardless of fans.

I understand how a fan draws current just fine. Nice current pulses,
usually two pulses per revolution. I designed a amplifier for these
pulses that would sound an alarm if the fan stopped. Worked fine.
Your resistor just drops a varying amount of voltage depending on
the fan being used (different current demands.) The SAME zener can
be used with most fans and still provide the same 7v to the fan.

It's pretty simple. It's unlikely most people have the >1W zener, so
(excepting you, apparently), they can't make do with parts they already
have.

It's unlikely most users have ANY parts on hand, let alone the correct
value resistor for his fan. A zener doesn't care. You can use the
same one on a fan that draws 100mA or on a fan that draws 200mA.
Not so with a resistor. Remember, this problem was about quieting
an already existing system.

In practice, you're either buying a lab-grade pot at extreme expense if
it can handle the current or it's called a rheostat. Not theory here,
I mean actually going and buying the thing.

I'm talking definitions here. A three terminal variable resistor is
called a potentiometer. As soon as you connect the center terminal
(wiper) to one of the end terminals, it becomes a rheostat. Doesn't
matter what the current or power rating is. Even a 1/4W pot can be
used as a rheostat.

Actually they may assume it but don't really know what they need. A
properly designed chassis cooling system can handle full load WITHOUT
the fans running at high speed. IMO, it's just silly to reach down and
fiddle with a pot and have that higher noise when one can have it
quieter and never have to adjust it. You are too consumed with control
methods and not enough with fan selection.

But this thread was about quieting an EXISTING system using the EXISTING
fan, not about proper design for cooling and selecting the correct fan.

I actually DO THIS stuff
constantly, it works fine to have single speed fan, no overheating even
with overclocking. Even so, if it's your box and you want a rheostat,
have fun with it even though it's not necessary for cooling or noise
reasons unless the system owner made a bad case choice and refuses to
correct that mistake instead of trying a work-around.

Re-read the above. This is about someone who asks what he can do to
quiet his present fan, other than "the 7v Molex reversal" solution
(which the OP did.)

You are clueless, the entire case is not a heatsink.

If it's metal, and constructed with metal screws, it sure is. You're
not using plastic cases, are you? Any heat buildup in the chassis
mounted pot will be transfered to the chassis, just as the chassis
gets quite warm in the area of the PSU.

Those will work, go ahead and do it. It's not necessary though, more
expensive and less reliable. Seems the only reason to do it is if
prior case or fan choices were poor or if you are just to stubborn to
accept that there is a better way.

Sigh...we're assisting someone who wants to quiet his existing fan by
methods other than the connector reversal, remember?

yes, why on earth would someone who wants quiet fans, start out by
buying a higher speed fan?

[..snip]

Nice essay, but this is about ways to quiet an *existing* noisy
higher speed fan, remember? The OP did the "Molex reversal" thing.
We started giving him alternatives, remember?

You dont' need much better resolution in most cases, merely to be able
to hit rougly 10 Ohm near enough to the target. If what you need for
very quiet (near if not entirely silent) operation is about 125 Ohms,
then 100 Ohms just won't cut it. It is in fact very common to need
about 120 (+-10) Ohm with optimal low speed fans from many of the major
manufacturers, unless the user wasn't going for such a low speed to
begin with and if they aren't going for such a low speed, they might as
well just buy that low speed fan to begin with and it might be what
they wanted, ie- about 2200 RPM. I aim lower than 2200 RPM though,
often under 1000 RPM is possible, particularly with chassis that accept
120mm fans.

Again, this was about quieting an existing higher RPM fan, not fan
selection. But please let me know what fan you are using that draws
less than 50mA. 50mA would give a 5v drop on a 100 ohm pot. Any more
of a drop would put it in the unreliable startup range when it gets
old and dirty.

So in conclusion, a 5W zener will work but there's no reason to use one
over other methods, the best of which being a power resistor if later
adjustment isn't necessary.

What value resistor? The OP likely didn't want to experiment with
multiple resistors. The resistor needed for a 100mA fan would be
quite different that with a 200mA or 300mA fan. What fan does he
have? The 5.1v zener will work with all three fans.

If adjustment is necessary and you want only a low-power-rated pot, you
want that pot in the voltage divider feedback of a (usually integrated)
regulator, NOT handling the load current.

I know of no low power pots that are chassis mount.

To handle the load current
you will have to spend a lot more for the rheostat or even rarer pot,

Any run-of-the-mill chassis mount pot wired as a rheostat works
fine.