CPU Install question

B

Bill

I'm doing a new build and installing a new Corsair H-105 cooler over a
new Intel CPU. I have 2 questions.

1. I assume it's okay if the "water tubes" come around over on the RAM
side. When I was was thinking about it the first time, I was thinking I
didn't want to block the output fan in the back of the case (so that
was the way I went). They will be at least an inch over my Corsair
Vengence RAM, but I don't want pre-mature failure (due to the heat
drying out the tubes?)...

2. The Corsair cooler has thermal paste on it, of course. Will cleaning
the CPU with 99% alcohol be proper "prepping" before I attach the cooler?

Thank you for your kind assistance!

Bill
 
P

Paul

Bill said:
I'm doing a new build and installing a new Corsair H-105 cooler over a
new Intel CPU. I have 2 questions.

1. I assume it's okay if the "water tubes" come around over on the RAM
side. When I was was thinking about it the first time, I was thinking I
didn't want to block the output fan in the back of the case (so that
was the way I went). They will be at least an inch over my Corsair
Vengence RAM, but I don't want pre-mature failure (due to the heat
drying out the tubes?)...

2. The Corsair cooler has thermal paste on it, of course. Will cleaning
the CPU with 99% alcohol be proper "prepping" before I attach the cooler?

Thank you for your kind assistance!

Bill

But the water tubes have cooling fluid moving through them. How
high a surface temperature could the tubes manage, if cooling water
is running inside the tube ? Only if you had an open wood fire,
with intense infrared pouring out, could you punish the outer
surface of the tubes.

The plasticizer used for the tubing does not last forever.
But I don't think a little 35C case air is going to make that
much difference, when the tube has the thermal mass of water behind it.

*******

On the CPU surface properties, you want to clean off any
amount of paste which is hard and uneven, and causes the
new cooler to "wobble" when placed on the surface. Some
Intel phase-change materials are candidates for removal.

If you need more trivia about paste products, they
have instructions here you can examine.

http://www.arcticsilver.com/instructions.htm

The biggest liability on a certain range of years
of Intel processors, is the surface shape of the
processor top. It's convex. This was done for
mechanical stress reasons. The processors where the
lid is soldered to the CPU die with low-temperature
solder, there are extreme stresses involved when the
solder cools. The curved surface of the processor,
has something to do with controlling the stress
(pulling forces) when the solder cools. Now, those
CPUs, the heatsink sits on the center of the CPU,
whereas out on the edge, the thermal paste has to
fill the air gap. It means the edge of the CPU
doesn't conduct heat quite as well (as a thick layer
of paste is a thermal insulator).

By comparison, lamenting over the cleanliness of the
"grooves" in the metal surface, is a secondary effect.
If your paste product was a good one, and it has
"tinted" the metal surface, you could just leave that,
as thermally conductive material now fills the grooves.
So I would not necessarily let my "cleaning fetish"
take over in that case.

The only material I thoroughly remove, is the Intel
phase change stuff. It's hard as a rock. And the solvent
isn't going to make much of a dent on it. If a proper
paste product was used, it's not going to get in the
way of doing a good job.

The danger of using a solvent, is some of the solvent
could get left in the groove. And take the place of
some paste. Even a cleaning rag can leave enough
debris in the grooves, to negate all the work you
spent "cleaning". So the surface conditions
are kinda tricky, if you worry about such things.
When the metal was new and paste-free, that's your
one good chance for an "impurity free tinting" process.

Your priorities:

1) Remove obvious surface imperfections (mountains of
Intel rock-hard phase change material, causing wobbly
fit).
2) As thin a layer of paste as possible, consistent with
displacing air and forming a conduction path. Do not
build an Oreo cookie (the way we did with one of my
engineering projects at work :) )
3) Work on your "groove" and your "tinting". This is
well down the list. If the original paste was good,
just wipe it clean with wiping action.

Paul
 
F

Flasherly

They will be at least an inch over my Corsair
Vengence RAM, but I don't want pre-mature failure (due to the heat
drying out the tubes?)...

2. The Corsair cooler has thermal paste on it, of course. Will cleaning
the CPU with 99% alcohol be proper "prepping" before I attach the cooler?

The tubes are about "wicking" action - a portion of evaporative fluid
within that, heated, expands wicks upwards to the fins and
subsequently generates a closed circular cooling effect.

You're not going to 'wear out its pipes,' not at least in your
lifetime;- Perhaps by the year 2500, in case you want to leave a note
to any descendant prodigy.

Yeah, clean it with alcohol, that's what I also use. Unless I'm lazy
and dry wipe it with a paper towel. Both the CPU and the cooler. And
don't use those thermal pastes/tapes. I never ever do. They're
poorly regarded by reasonably available products from NewEgg or Amazon
in highend thermal pastes. Arctic Silver, for instance. Got a
"baggie" full of them, tubes accumulated over the years.

Might be OK for a pinch, most operators wouldn't know the difference.
Dropping money into a for real heatsink, for an assembler/builder, and
actively monitoring CPU temps, though, the heat mounting
paste/compound included with a lot of heatsinks, they're a joke.

A better CPU cooler, these days anyway, they're totally overkill and
awesome about adequate cooling. Shame, really, not to seat one
affixed most righteously. Think of it with impunity as the crowning
touch of your masterful competence.
 
B

Bill

Flasherly said:
The tubes are about "wicking" action - a portion of evaporative fluid
within that, heated, expands wicks upwards to the fins and
subsequently generates a closed circular cooling effect.

You're not going to 'wear out its pipes,' not at least in your
lifetime;- Perhaps by the year 2500, in case you want to leave a note
to any descendant prodigy.

Yeah, clean it with alcohol, that's what I also use. Unless I'm lazy
and dry wipe it with a paper towel. Both the CPU and the cooler. And
don't use those thermal pastes/tapes. I never ever do. They're
poorly regarded by reasonably available products from NewEgg or Amazon
in highend thermal pastes. Arctic Silver, for instance. Got a
"baggie" full of them, tubes accumulated over the years.

Might be OK for a pinch, most operators wouldn't know the difference.
Dropping money into a for real heatsink, for an assembler/builder, and
actively monitoring CPU temps, though, the heat mounting
paste/compound included with a lot of heatsinks, they're a joke.

A better CPU cooler, these days anyway, they're totally overkill and
awesome about adequate cooling. Shame, really, not to seat one
affixed most righteously. Think of it with impunity as the crowning
touch of your masterful competence.


You and Paul give great answers! I'm glad to learn that the tubes are so
stable!

I couldn't wait. I read somewhere that Corsair wouldn't put bad thermal
paste on their coolers.
I installed the CPU, wiped it with one end of a Q-tip with 99% alcohol,
dried it with the other end, and put the H-105 on.
I like the way the cooler went on compared to the Intel coolers. The
latter have generally aggravated me in the past, trying to get
all 4 of the plastic tips to come through the other side the way they
are supposed to. I didn't even think of taking a good look
at bottom of the new cooler (duh), wish I had. I had previously
noticed, through the plastic bad, that it had compound on it.

I don't claim "masterful competence" (ha, ha, ha!). I spent 15 minutes
figuring out the right way to attach the 3-pin pump connector to the
4-pin CPU_FAN header.
I lined it up with a plastic tab behind 3 of the pins (duh!)--I didn't
do that the first time. It goes on just fine in two different ways! ; )
I hope everything works when I turn it on in a day or two! : )
At least now I'm not "afraid" of installing the cooler. If I have to
redo it, now I'm not skeered!
Alot of worrying always seems to go into the 45 seconds it takes to
secure the CPU into its slot!
--Funny, when you're actually doing it, you don't worry too much. It
is well-rehearsed. I treat the CPU "delicately".

I have a Corsair 650D case. Tomorrow, I'll pull all the wires through
to the back of the case and push them through where they go.

I'll report back after "post time".

Thank you, and cheers!
Bill
 
F

Flasherly

You and Paul give great answers! I'm glad to learn that the tubes are so
stable!

I couldn't wait. I read somewhere that Corsair wouldn't put bad thermal
paste on their coolers.
I installed the CPU, wiped it with one end of a Q-tip with 99% alcohol,
dried it with the other end, and put the H-105 on.
I like the way the cooler went on compared to the Intel coolers. The
latter have generally aggravated me in the past, trying to get
all 4 of the plastic tips to come through the other side the way they
are supposed to. I didn't even think of taking a good look
at bottom of the new cooler (duh), wish I had. I had previously
noticed, through the plastic bad, that it had compound on it.

I don't claim "masterful competence" (ha, ha, ha!). I spent 15 minutes
figuring out the right way to attach the 3-pin pump connector to the
4-pin CPU_FAN header.
I lined it up with a plastic tab behind 3 of the pins (duh!)--I didn't
do that the first time. It goes on just fine in two different ways! ; )
I hope everything works when I turn it on in a day or two! : )
At least now I'm not "afraid" of installing the cooler. If I have to
redo it, now I'm not skeered!
Alot of worrying always seems to go into the 45 seconds it takes to
secure the CPU into its slot!
--Funny, when you're actually doing it, you don't worry too much. It
is well-rehearsed. I treat the CPU "delicately".

I have a Corsair 650D case. Tomorrow, I'll pull all the wires through
to the back of the case and push them through where they go.

I'll report back after "post time".

Thank you, and cheers!

Right-o. The way or thing about if you're using their mounting paste,
and I could be wrong in that the included compound is 'the good
stuff,' is if in need of a remount at some future point. You won't
obviously have it and you'll have the possibility of some better
aftermarket compounds - thick heavy-metal saturated paste compounds.
Run a torture-type benchmark, or, from your regular programs the most
demanding on the processor. Actively monitor the CPU's temps
(SPEEDFAN or similar/suitable to pull that info from the CPUs
output/thermal diode). Then try find a reported poll/forum of users
and what they're reporting for cooling with your CPU. Doesn't matter
as much what they're using, just skim their temps for comparison to
your conditions. (Wide latitudes for 35/45/65/95/125watt processors
all using heat variously.) If you're not satisfied, though, feel you
should have better - drop a few bob on a heavier known brand, one
well-received. Might be yours is a light silicon based heatsink
compound - or not, might be the benefits of another compound would be
effectively insignificant.

The residual of it all, say, between a CPU running at 57-65C and one
that runs ambient or close - to my thinking - does matter in the long
run. Rated/spec'd much high, of course, might not be a dead CPU
although I believe anomalies can occur, especially, with overclocking
that contribute to a mess, it is, when an OS reaches general
instability (figuring and knowing if its hard/software induced).
 
B

Bill

Flasherly said:
<snip>
The residual of it all, say, between a CPU running at 57-65C and one
that runs ambient or close - to my thinking - does matter in the long
run. Rated/spec'd much high, of course, might not be a dead CPU
although I believe anomalies can occur, especially, with overclocking
that contribute to a mess, it is, when an OS reaches general
instability (figuring and knowing if its hard/software induced).

Thank you! I printed out your (entire) post to remind me of the details.

Bill
 
F

Flasherly

Thank you! I printed out your (entire) post to remind me of the details.

Seat of the pants. Built a lot of 'em, overclocked some, others would
run with a wooden matchbox-sized heatsinks;- guess I like just like my
CPUs on the cooler side these days, even with today's superior case
and CPU heatsink designs. (10 years life expectancy out of a computer
isn't exactly unheard of, least not for me.) Hope your new build
serves you well!
 
B

Bill

Flasherly said:
Think of it with impunity as the crowning touch of your masterful
competence.

I didn't think of it that way, but anyone who can get all the wires and
components in the case is at least a little competent, I guess (pat,
pat, pat, lol). I think I'd be finished if I didn't have a case with
a "back" to thread wires through. I learned I liked my 650d case a
little more when I realized I was able to remove half of the 3.5" drive
holders, to improve airflow from the front. There can now be high
winds from the front air-intake through the case out the back and top
(through the radiator), except I'll probably be tempted to try to keep
the fan speed low (I can adjust the fan speed from a 4-fan controller on
the top of the case--new feature to me). I still have the H-105 cooler
fans on CPU_OPT though.

I put the SDD closer to the bottom of the case, in front of the 200 mm
front intake fan, figuring it's cooler the lower you go (since heat
rises). I put the CD/DVD drive about in the middle, just above the level
of the fan, so it would get a draft from all the air coming into the case.

I think you and Paul taught me pretty well here! I posted a little over
6 months ago, and didn't even like the idea of water cpu cooling (after
I realized it was a "closed system" I liked it better). You got me
thinking about temperatures more than usual. My last build was 5 years
ago. Pretty soon the old system can keep working for another family
member (I don't skimp on things like power supplies).

I think I have one more session before the first post. You wouldn't
want to pay me by the hour...

Cheers,
Bill
 
B

Bill

Bill said:
I think I have one more session before the first post. You wouldn't
want to pay me by the hour...

Paul and Flasherly,

The first post was *completely* uneventful!!! Very quiet... not even an
error code. Then after taking a short break, I re-inserted the 24 pin
connector, even flexed the MB a little, and heard a click, and after
that things went much better.

I selected the "optimized defaults". Windows7 loaded in 15 or 20
minutes. I'm running the 4790K CPU at 4.0 GHz (no XMP features yet), and
the cores temps are 29-31 C at idle.
On one of the first BIOS screens, there's a place where they list the
CPU temp as 22C (I'm not sure where that number comes from).

I don't have much software installed yet so, besides the install, I
haven't given the system a good test yet. I haven't even got it on the
Internet. The "Windows Experience" numbers are mostly 7.9. 7.8 for
Graphics, since I have a very modest, but quiet, GPU (GX750TI, Strix).
Far superior to the GPU (HD-4400?) in the CPU however. The first time I
ran Windows Experience, I realized my video card wasn't being used.
Another story, sort of funny--when I was finished "assembling my
computer" and was admiring my handiwork, I realized I had forgotten to
put the video card in. Fortunately I had left room.

So far, the new system is hooked up to an older monitor and keyboard and
I'm toting my one USB mouse from one computer to another. I have to do
all of the stuff you do when you change computers, then I'll be good to
go! By the way, I usually use the SANDRA program for performance
comparisons, but I don't get too caught up in making comparisons. I just
like to verify I'm getting the performance I'm supposed to be getting.

Cheers, and thank you for your support!
Bill
 
P

Paul

Bill said:
Paul and Flasherly,

The first post was *completely* uneventful!!! Very quiet... not even an
error code. Then after taking a short break, I re-inserted the 24 pin
connector, even flexed the MB a little, and heard a click, and after
that things went much better.

I selected the "optimized defaults". Windows7 loaded in 15 or 20
minutes. I'm running the 4790K CPU at 4.0 GHz (no XMP features yet), and
the cores temps are 29-31 C at idle.
On one of the first BIOS screens, there's a place where they list the
CPU temp as 22C (I'm not sure where that number comes from).

I don't have much software installed yet so, besides the install, I
haven't given the system a good test yet. I haven't even got it on the
Internet. The "Windows Experience" numbers are mostly 7.9. 7.8 for
Graphics, since I have a very modest, but quiet, GPU (GX750TI, Strix).
Far superior to the GPU (HD-4400?) in the CPU however. The first time I
ran Windows Experience, I realized my video card wasn't being used.
Another story, sort of funny--when I was finished "assembling my
computer" and was admiring my handiwork, I realized I had forgotten to
put the video card in. Fortunately I had left room.

So far, the new system is hooked up to an older monitor and keyboard and
I'm toting my one USB mouse from one computer to another. I have to do
all of the stuff you do when you change computers, then I'll be good to
go! By the way, I usually use the SANDRA program for performance
comparisons, but I don't get too caught up in making comparisons. I just
like to verify I'm getting the performance I'm supposed to be getting.

Cheers, and thank you for your support!
Bill

Let's hope that click was the latch on the
24 pin connector, mating and closing with the
motherboard portion.

With the different colors in this example, you
can see the latch on the black piece, hanging
down where the matching section on the beige
motherboard connector should be. To release it,
you need to press the top of the lever, so the
bottom portion moves away from the surface of
the other connector body. Then you can pull up
on the connector and harness. They use latches,
because without a latch, thermal expansion would
gradually cause the connector to be ejected from
its mate.

http://www.playtool.com/pages/psuconnectors/main24pin.jpg

It's especially difficult some times, to get the
ATX 2x2 or 2x4 to release, as there is no finger
room in that area.

It's almost as difficult as actuating the heel release
on the video card, so the video card can come out of
its slot. On some motherboards, that heel thing is a
pure bitch (examine it carefully while the motherboard
is out of the computer case, because it's pretty hard
to observe it later). It can be either a spring or a
slide piece. The slide piece type, isn't nearly as bad
to work with.

You can give your cooling a workout with something
like Prime95 torture test option (mersenne.org/freesoft).
For single threaded benches, some people like SuperPI.
Selecting enough digits so the whole calculation does not
fit in the processor cache. The 32 million digit calc
is used on some of the high end processors, because
they have such a large cache to work with. In the past
when I did builds, I'd take ratios of SuperPI times
in seconds, to see what my money bought me. But this
time I didn't bother, because clock rates just aren't
climbing all that much any more. What I got on my
new build, was a lot more cores. So that programs
like 7ZIP (parallel execution) go faster. I've seen
as much as a 7X speedup on 7ZIP (which, unlike some
compressors, doesn't "speed up" that much when it
hits easy stuff).

Paul
 
B

Bill

Paul said:
Let's hope that click was the latch on the
24 pin connector, mating and closing with the
motherboard portion.

It was. I was listening for the click the first time, but didn't press
the issue.
I'm trying to un-train myself from my (bad) habit of over-tightening
things.
I will try the program you suggested for testing.

By the way, I did notice that my GPU has a hook on it. I guess that means
you slide it towards the screws a bit before you try to pull it out?

Cheers,
Bill
 
P

Paul

Bill said:
It was. I was listening for the click the first time, but didn't press
the issue.
I'm trying to un-train myself from my (bad) habit of over-tightening
things.
I will try the program you suggested for testing.

By the way, I did notice that my GPU has a hook on it. I guess that means
you slide it towards the screws a bit before you try to pull it out?

Cheers,
Bill

My advice on the video card retainer, is figure it
out while you can still see it :)

There's nothing worse than not being able to see in there,
and barely being able to get a finger on it, then trying to
figure out what you're supposed to do. Even after you've examined
the mechanism in the light, you might still have trouble. The
slide ones should be less of a challenge.

You can see the heel in a picture of a video card here.

http://images17.newegg.com/is/image/newegg/14-131-338-Z11?$S640$

It's that thing on the lower right. On the spring loaded
ones, a round "button" fits over top of the heel, and
prevents the heel end from lifting. You pull back on the spring,
moving the button out of the way, so the heel can rise up.

In the slider type retainer, sliding the slide mechanism towards
the faceplate, is the "locked" position. Sliding the slider mechanism
away from the faceplate, removes the slide from over top
of the heel, so the heel can rise up.

HTH,
Paul
 
B

Bill

Paul said:
In the slider type retainer, sliding the slide mechanism towards
the faceplate, is the "locked" position. Sliding the slider mechanism
away from the faceplate, removes the slide from over top
of the heel, so the heel can rise up.

HTH,
Paul

I think I had a computer that had one of those. A "tiny white lever"
was really tough to see I'm lucky I didn't break anything! ; )
My new one has a hook, just like in your picture. I will strive to
keep it in mind if I ever need to remove it!

Tonight I'm "screwing around" transferring files (some mp3's I purchased
from Amazon.com just don't want to go), and
transferring my email files (I can make it work...it's just not working
as nice as it should--yet).


Thanks,
Bill
 
B

Bill

I recollect (I think) someone mentioning a program called SPEEDFAN for
measuring temperatures.
Where is a good place to download there (I'm "dreadfully afraid" of
getting something I don't want as part of a download)?

Thank you,
Bill
 
B

Bill

Paul said:
Thank you very much.
One more question please.

The fans on the H105 are "PWM". Do you think it is normal for 1 fan of
the 2 radiator fans on the H-105 to run while the other sits and maybe
just "jiggles" a little (at low speeds). That is what drew my attention
to it in the first place, is that I could hear some very, very light
sounds (of the 2nd fan sort of trying).
They both run together at slightly higher speeds/heat. The two 3-pin
fans are co-joined by a simple wire "coupler" and are attached to the
4-pin CPU_OPT header. I can rationalize why it does what it does, and
I can rationalize why it shouldn't be doing what it's doing! Please
tell me the right answer, if you know!

Cheers,
Bill

By the way, gigabyte included some fan and temperature software with
their MB. Its doesn't work at advertised (where do the hours go?), but
it can be made to work.
 
P

Paul

Bill said:
Thank you very much.
One more question please.

The fans on the H105 are "PWM". Do you think it is normal for 1 fan of
the 2 radiator fans on the H-105 to run while the other sits and maybe
just "jiggles" a little (at low speeds). That is what drew my attention
to it in the first place, is that I could hear some very, very light
sounds (of the 2nd fan sort of trying).
They both run together at slightly higher speeds/heat. The two 3-pin
fans are co-joined by a simple wire "coupler" and are attached to the
4-pin CPU_OPT header. I can rationalize why it does what it does, and
I can rationalize why it shouldn't be doing what it's doing! Please
tell me the right answer, if you know!

Cheers,
Bill

By the way, gigabyte included some fan and temperature software with
their MB. Its doesn't work at advertised (where do the hours go?), but
it can be made to work.

That's not what is shown here.

I see two fans with separate four pin headers in this Newegg photo
for the H105. I can't find a picture of the pump, but it would make
sense for it to use a three pin.

http://images17.newegg.com/is/image/newegg/35-181-060-Z04?$S640$

( http://www.newegg.com/Product/Product.aspx?Item=N82E16835181060 )

If a person wants to run fans in parallel, something must be
done about the RPM signal. The RPM signals cannot be joined
on the two fans. If I were to make an adapter cable, I'd either have
to disconnect RPM on one fan, or run the RPM signal of one fan, to an
"RPM only" connector to be run to some header that can measure
the RPMs for me.

FAN1 FAN2
+12V --------------X---------------X
GND --------------X---------------X
RPM --------------X
PWM --------------X---------------X

If I make an adapter cable like that, the 25KHz 5Vpp signal with
variable pulse width (PWM), controls both fans at the same time.
If the fans are a decent design, they should be within a 100 RPM
or so of one another. The match won't be perfect (due to maybe
friction differences or mechanical differences), but they should
be in the same performance ballpark.

In that diagram, only FAN1 has RPMs monitored, so you can see
the RPM rate in Speedfan.

I could make a two connector adapter, and route the RPM signal
of the second fan, to a three pin connector. Just for monitoring
and not for power. (My local electronics store (not RadioShack)
has a small, poor selection of connectors and pins so I can make
my own cables.)

FAN1 FAN2
+12V --------------X---------------X
GND --------------X---------------X (Yellow)
RPM --------------X X----------------+
PWM --------------X---------------X |
|
+12V |
GND |
RPM -----------------------------------------------+

Now, speed control is on the four pin, via PWM. Power is
drawn from that header too (so make sure the total fan
load in milliamps, does not exceed the milliamp rating
of the header. Both fans are commanded to do the same thing.

The second connector allows the RPMs to be measured.

You'll have to give me some idea what this "adapter"
is doing, for me to figure out what is going on. My
guess is, the second fan has radically wrong characteristics.

PWM fans are supposed to have something like a two
line response. When the control stays at 0 logic level
permanently (0% PWM), the fan should still be running
at a low speed. That's so your "low RPM" alarm doesn't
go off unnecessarily. You would need a more complicated
monitoring method (PWM, resulting RPM), to monitor
the fans otherwise.

RPM | /
| /
| /
| /
| -------
|
+--------------- PWM %
0 100

As to what the Corsair fans actually do, you'll need to
dig into some customer comments.

I wouldn't leave it hiccuping like that, if it was my rig.
I'd want to know "why".

*******

I found the quick install guide.

http://www.corsair.com/~/media/Corsair/download-files/manuals/cooling/H105_QSG_.pdf

On the one hand, the fan Y-cable, looks as expected. It
matches my first diagram, with RPM on the second fan
disconnected. That means, you cannot monitor RPM on
the second fan that way.

http://i58.tinypic.com/31788p3.gif

You would need to adjust the response curve, so the PWM
level doesn't drop so low, and allow the second fan to
get in trouble. In my PWM fan behavior curve, what is supposed
to happen, is the fan can't go below a regularly
spinning level. There shouldn't be a "fits and spurts"
level in the response. If the fan response curve was
a single line at 45 degree angle, extending to 0% PWM
input, then it's up to you, with your fan speed adjustment
software and method, to prevent the PWM signal from
dropping all the way to 0%. Maybe if the PWM signal
stays at 30% say, the fans will be in a regularly spinning
range.

If it was my rig:

1) Pumps runs at constant speed. The three pin header for it
likely doesn't have voltage adjustment in hardware anyway,
so this should be easy to meet. Verify, using Speedfan,
that the pump is in the right RPM range. If too high, there
could be air in the liquid loop. If too low, it could be that
the header isn't delivering +12V.

2) If I had the headers, I'd run the two fans separately.
But that would complicate fan control. It isn't that
common to have control software that couples two headers
in terms of settings and makes them do the same thing.
I expect that's why Corsair provided that Y cable. You
can use the Y cable if you want, but just make sure the
fan runs properly.

What would really be nice, is fan control software that
uses closed loop control. Where the RPM level was monitored,
the PWM value adjusted, for a "level of cooling". And that
"level of cooling" target was set by the temperature. That
would eliminate all the concerns. Only if the software
"hit the rails", and could not apply enough PWM to meet
the target, then it would alarm and put a dialog on the
screen saying something is wrong.

The same sort of thing happens with your car. There is
an acceptable control range, which varies as the car
ages. But if some control loop goes out of bounds,
maybe your car throws a code and "claims the oxygen
sensor is bad". So cars are an example of well applied
closed loop control, such that most of the time you
don't have to worry about engine management.

HTH,
Paul
 
F

Flasherly

I wouldn't leave it hiccuping like that, if it was my rig.
I'd want to know "why".

Or cut off, from a mylar connector on the PS unit, and solder leads to
a mini-MB fan connector. Forgoing the control factors for speed or
RPM monitoring.

As it is, I've decent-sized to larger fans connected to the MB's fan
output voltage provisions. Didn't used to, though. Used to splice
and run them to the PS to make sure I couldn't possibly be stressing
the MB. Just didn't seem right, at least to me, a masterworks of
modern timing, to be driving some $5 Hong Kong fan on a special sale.

Still don't care for variable speeds, especially (the lower the
better);- temps, however, I am careful about.

(Don't even care to go near what mechanics may say when addressing
factory vehicle codes, those whom recommend the StealerShip as best
situated to analyze and repair a modern vehicle. Not even with your
10'-pole barge pole. ...Defeats the whole purpose of a viable
economic democracy, defacto, as one practicing competitive marketing,
including aftermarkets.)
 
P

Paul

Flasherly said:
Or cut off, from a mylar connector on the PS unit, and solder leads to
a mini-MB fan connector. Forgoing the control factors for speed or
RPM monitoring.

You don't want to run the Corsair fans at 100%, because they're
"vacuum cleaner fans". They're like the old Dell fans, designed
to work into a high resistance load, and when spooled up,
would be obnoxious. (Things would be different, if Corsair
selected a larger fin spacing on the cooler, but that
in turn reduces cooler performance. The vacuum cleaner
design idea, is for high performance systems.)

The fans really should be controlled by a PWM system, to prevent
damage to the fan (from using voltage-mode control - you're
not supposed to run PWM fans off voltage-mode control).

You run a PWM fan off a full 12V, with the purpose being so the
control transistor has enough voltage for proper saturated
operation. If you supply the 12V PWM fan from a 7V or 5V source,
the control transistor might heat up on it. So if you want to
adjust the speed of a PWM fan, run it off 12V and use the PWM
control as intended. (I think I got these recommendations
from some official document. Like the Intel PWM spec.)

Just make sure you have enough four pin PWM headers (with working
PWM), to pull it off. I think that's why the Corsair kit
comes with a Y cable, to provide PWM control to two fans
at the same time. To conserve headers.

*******

This is from a Dynatron PWM fan spec. (Cannot figure out
where I got the spec.)

Fan operating voltage shall be 12V +/- 1.2V

So they don't want reduced voltage operation.

Duty cycle 0-20% 1000RPM +/- 200RPM
Duty cycle 50% 1800RPM
Duty cycle 100% 2500RPM

The first line basically says the fan remains
spinning when the PWM control is at 0% setting.
So the fan does not stop. And Dynatron references
the Intel fan spec (but doesn't name the document).

The Dynatron behavior, matches figure 3 on page 14.
Formfactors is an Intel site for defacto standards.

http://formfactors.org/developer/specs/4_Wire_PWM_Spec.pdf

Paul
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top