Stacking Fans To Decrease Temp!!!

[BinarySuperman]

I have always been wondering, if you stack like 3 fans on top of each other will the cpu temp go down, if so is fan stacking a good method for overclocking and cooling???

[William]

Usually gives worse results if I remember correctly. It doesn't do a ton to increase airflow at all.

[BinarySuperman]

thanks

[nihili]

I've got two fans, one on each end of a duct. I see about a 10% increase in fan speed over using either one separately. I don't know if that translates into an increase in CFM.

nihili

[adamtekh]

i once expermented with this , i for some reaseon got about a 25 % increase in CFM

i noticed no dif in temps

i posted this on this fourm , ppl didnt belive me much. but it was so fricking loud , and u can for cheap get a 92 with more cfm

i said screw it

i did this with 80 mm fans

[nihili]

I think it only makes sense to do on a duct. But just to be clear, you said you got a 25% increase in CFM. Did you measure the CFM or are you extrapolating that from the RPM?

nihili

[adamtekh]

guestimation , im no math wiz... being 16

[BinarySuperman]

what does CFM mean??

[Colin]

CFM is cubic feet per minute.

While stacking fans is not supposed to work, in my early days of Athlon overclocking, two stacked 80 mm fans on a fandapter made from a 24 oz. Bud can dropped my temps 3 C. I currently use two stacked 120 mm Panaflow L1As on my bong cooler with similar results.

[kevin_bouchard]

in theory stacking fans would decrease air flow and add stress on the fans, because they would be forcing each other. Think about how the air coming from a fan goes, it moves away from the fan in a windtunnel formation, not straight, so adding another fan would only speed up the circular motion of the flow. But if you could have 2 fans,same rpms, and one spins clockwise and the other counterclockwise but both of them blow the same way and stacked them, in theory it would increase airflow alot and the power of the flow(static pressure), just like a turbine.
hope this helps

[adamtekh]

yes conter rotating fans , would work ....... i never tested i wanted to , but u try to find me a fan that spins conter clock wise

[Intraveinous]

Many of the Pabst Fans spin counter-clockwise...
Peace
John

[ButcherUK]

According to the sunon website stacking fans provides a benefit to airflow at high static pressure, but they blow more than one on it's own if there's not much static pressure.

[The Overclocker]

stacking fans will increse how straight the fans blow air, but i dont think it will inprove the cfm

[Maximus Nickus]

Of course it will increase CFM! Not much though.
Because if the air is already travellingt at a high speed when it hits the fan the fan will have to use less effort thus higher rpm - cfm... also if the fan is real rubbish and the other one ain't then it would work better, but then I would just have the good one!
It would also reduce the last fan in the sequences life.

[dugans]

I read an article on this somewhere (forget where) and they reported: an increase in rpms on 1 fan, slightly lower temp and a lot more noise.

I have a duct in my xp that works great: 2 empty 5 1/4 bays with an 80mm fan blowing room air to a cheapo hsf: cpu temp 43c average. Without the 80mm the temp was around 48c, and without the duct I couldn't oc! (The case sucks, and it is stuffed in a desk with no airflow-nowhere else to put it until I run some more cat5)

I don't think stacking fans is worth it from what I've read- minimal temp improvement, a lot more noise.

[nihili]

Stacking fans does give an increase in RPM. But that doesn't always translate to better cooling.

Basically stacked fans work to overcome static pressure. So if you're trying to push air through a tight spot (like a heat sink) then you can get some benefit from stacking. How much benefit will depend on the sink. I doubt that a thermoengine would benefit much because of it's free flow design. However one of the newer dynatron's with the closely packed microfins might benefit significantly.

Ducts also increase static pressure, particularly if theyre made from dryer ducting like mine is. Also anytime you decrease the size of a duct, as in doing an 80 to 60 conversion, you'll see an increase in static pressure. In these sorts of cases running the fans in series can show substantial improvement in the amount of airflow, and hence in cooling.

However in free flowing circumstances, as is the situation with most case fans, stacking will produce little or no improvement as static pressure is not the limiting factor. In these sorts of circumstances you're better off running fans in parallel, one per opening.

An analogy: Stacking fans is like having four wheel drive. It won't make you go any faster on the freeway, but it will let you go faster in rough terrain.

nihili

[BillA]

for the air supply used in the radiator testing I used 4 Nidec TA600DC model A32590-10 fans

nominal specs are 240 cfm and 66 dba

in parallel after 2 were running, the 3ed contributed nothing, and air would be pushed backwards through the 4th - powered or not

so then I stacked them in 2 pairs and with 2 running the non-powered ones would windmill, kick off a 3ed and the output went up about 30%, light off the 4th and it was a blowin' jenny

the highest output I actually used (and therefor recorded) was about 3 and 1/2 fans (they are all on rheostats, at 2A ea. - big pots) where the output was 202 cfm at a static pressure of 0.25in.H2O

a lot of air, and a lot of noise

if more air at a higner pressure is needed (which is normally the case), stacking fans can be effective

all of the above relates to ducted output as I was pushing that air through a 6 in. duct

nihili's comments are quite correct

be cool

[Maximus Nickus]

I would never bother stacking, I'm trying to make my PC silent!
Anyone know of a good hsf? Theres a really heavy 80mm one for £60 from overclockingstore but I think thats OTT! lol:D

[The Overclocker]

the air from the first fan comes off the blades at 45degrees or so depending of the fans blades tilt, this mans the second fan blades cannot 'catch' the air and the second one is useless, however it does improve as the fans depth increses as it is at a higher pich the second fan can catch the air

[RangerJoe34]

no, stacking two fans doesnt do anything good.....i did this with a delta black label and a ys tech 38cfm (both 60mm fans) on my HSF, it was HELLA louder and the temps went up. even if the top fan has a higher cfm, it will be stressing the bottom one because of the higher cfm's. and if you put the one with lower cfms on the top, it will slow down the bottom one, so it is a lose/lose situation.

[jbell]

Originally posted by kevin_bouchard
in theory stacking fans would decrease air flow and add stress on the fans, because they would be forcing each other. Think about how the air coming from a fan goes, it moves away from the fan in a windtunnel formation, not straight, so adding another fan would only speed up the circular motion of the flow. But if you could have 2 fans,same rpms, and one spins clockwise and the other counterclockwise but both of them blow the same way and stacked them, in theory it would increase airflow alot and the power of the flow(static pressure), just like a turbine.
hope this helps

there is an airplane either in use or in production to be used... it had that very design of two baldes spinning in opposite directions of each other... it reduced the amount of fuel to run on a normal flight and also increased its power...

[Maximus Nickus]

There you go, two fans blowing in opposite directions and you ca n oc to 5ghz.... Another vodka please!

[nihili]

Originally posted by jbell


there is an airplane either in use or in production to be used... it had that very design of two baldes spinning in opposite directions of each other... it reduced the amount of fuel to run on a normal flight and also increased its power...

Volvo/Penta makes an outboard with two inline props spinning opposite directions

nihili

[dugans]

Originally posted by RangerJoe34
no, stacking two fans doesnt do anything good.....i did this with a delta black label and a ys tech 38cfm (both 60mm fans) on my HSF, it was HELLA louder and the temps went up. even if the top fan has a higher cfm, it will be stressing the bottom one because of the higher cfm's. and if you put the one with lower cfms on the top, it will slow down the bottom one, so it is a lose/lose situation.

One of the things I remember from the article was that they used 2 of the same fans: rpms were not an exact match (when running alone), but close.

[nihili]

I get an RPM bonus using very disimilar fans at each end of a duct. One is about a 19cfm 60mm, the other is an 80mm that I'd guess at about 30 cfm. Both of the fans run faster than they did alone in the same duct. The duct is 6-8 inches long with 90 degree bend, so I doubt that the increase in RPM is due to the rotation of the air coming off the fan.

nihili

[BillA]

since my (measured) experience with the thick-bodied Nidecs was different than what youall (Texas here) were reporting,
I decided to test

using 120mm Panaflo FBH-12G12L fans (68cfm nominal rating)
blowing into a 6in. duct:

1 fan pushed 71cfm
2 fans stacked pushed between 27 and 62cfm (unstable fluctuating output)

2 fans stacked with a 2in. fitted square 'spacer' had the same output as the close-coupled pair

scratching my head, saying wtf
I then tested a Danger Den Cube

1 fan mounted on the intake side (13/16in. standoff) pushed 34cfm
2 fans (the "puller") close-coupled to the fins) moved 8 to 11cfm, fluctuating

I was/am surprised
the radiator 2 fan recommendation (if the fans are the same) is all wrong

amazing what a test can reveal

be cool

[nihili]

How did you measure CFM?

nihili

[BillA]

Kurz model 441S thermal anemometer (calibrated)
sensor in center of 6in. duct, 10 - 12 diameters from fan (or radiator)
fps reading multiplied by 0.177 to convert to cfm (avg. velocity, cross-sectional area, etc.)

not adjusted for RH or barometric pressure

but the numbers are indeed real (and perhaps the ONLY real ones you will encounter, I've seen no others in OCing land)

be cool

[jbell]

Originally posted by nihili


Volvo/Penta makes an outboard with two inline props spinning opposite directions

nihili

Penta is awesome! I think I have seen something about the prop once before...

[ymNA]

to measure the cfm...

materials needed:
- 1 plastic box with a volume of 1 cubic feet - must be made of clear plastic
- fans need to be measure
- a stop watch
- one assistant
direction:
- Cut a hole on one side of the plastic box (need to fit with the fan you want to measure the cfm)
- give the stop watch to your assistant
- turn on the fan
- put the fan on the hole you just cut on the box - your assistant need to start the watch at the same time to be accurate

and see how many box of air you fill in one minute...

[ButcherUK]

Originally posted by ymNA
and see how many box of air you fill in one minute...
How do you do that, it's not like you can see the air filling the box...

[ymNA]

Originally posted by ButcherUK

How do you do that, it's not like you can see the air filling the box...
oh, sorry i didn't know, ignore that then

[ButcherUK]

Originally posted by ymNA

oh, sorry i didn't know, ignore that then

er, WHAT? you didn't know air is not visible to human eyes?! okaaaay...

[dugans]

Originally posted by ymNA
to measure the cfm...

materials needed:
- 1 plastic box with a volume of 1 cubic feet - must be made of clear plastic
- fans need to be measure
- a stop watch
- one assistant
direction:
- Cut a hole on one side of the plastic box (need to fit with the fan you want to measure the cfm)
- give the stop watch to your assistant
- turn on the fan
- put the fan on the hole you just cut on the box - your assistant need to start the watch at the same time to be accurate

and see how many box of air you fill in one minute...

rotfl!



And I hate posts that just say rotfl, but........!!!!!!!!!!!:D

[adamtekh]

stackin fans is not worth the noise , just buy biger fans

[adamtekh]

or go with water cooling

or even refridgeration

unless of course u want to test out submersive cooling

[nihili]

Originally posted by BillA
Kurz model 441S thermal anemometer (calibrated)
sensor in center of 6in. duct, 10 - 12 diameters from fan (or radiator)
fps reading multiplied by 0.177 to convert to cfm (avg. velocity, cross-sectional area, etc.)

not adjusted for RH or barometric pressure

but the numbers are indeed real (and perhaps the ONLY real ones you will encounter, I've seen no others in OCing land)

be cool

Marvelous, I love real numbers.

So as I understand it.

When you stacked Nidecs pushing through a radiator, you were able to get a 30% increase or better by stacking fans.

But when you stacked two Panaflo fans in a variety of ways, you got a decrease in CFM of between 15 and 70%.

Add in to the mix that the data from Sunon's website indicates that by stacking one *should* see an increase in CFM in all but free air situations where there would be no difference. I am, btw, inclined to take Sunon's numbers as real also. If you differ on this point, could you let me know why.

I'm disturbed by the Panaflo results for a variety of reasons.

First, they are unlike the other data. That by itself doesn't show there's a problem, but it's at least cause for curiousity.

Second, although you stacked the panaflos in three different settings, you got similar results in each. Normally that would be a good thing, but in this case it's worrisome. The best explanation we have of why you would see a decrease has to do with turbulence. But turbulence should have decrease through your three setups, so performance should have improved. However performance remained the same. So turbulence can't have been responsible for the decrease in performance. But if turbulence wasn't the culprit, what was?

Third, in all three setups with the Panaflos you got wildly fluctuating results. I'm assuming you ran the fans long enough to allow for stabilization if it were going to happen, so something in the setup was causing the CFM to fluctuate. Again the best candidate is the chaotic nature of turbulence. That sort of explanation works well for the tightly stacked fans, less well for the spaced fans, and not very well at all in the radiator setup. So once again we are left without a clear explanation of the results.

So something caused a fluctuating decrease in performance, but it probably wasn't turbluence. What else is there? The only things I can think of are poor power supply or a defective fan. I'm not saying these are the only possible answers, theyre just the ones I can come up with off the top of my head. Whatever the cause, it has to be consistent across the three setups. Did you try changing the order of the fans, and did you test each fan individually? Were there any potential anomalies in the power setup?

Please understand that I'm not doubting the numbers you provided, nor am I calling your competence into question. I'm merely trying to understand *why* those numbers exist.

Any ideas?

nihili

[BillA]

good post nihili
I've been thinking about this non-stop

the power supply is an Acopian A12H1300, set to 12.00VDC; and after 5 min or so, that's where it stays

the fans are the alu bodied variant, 13 yrs old, purchased a year ago as NOS - but who knows
in any case, they are both the same so probably not "defective"
they are however the low speed model, which also means low torque
the specs are 0.2A, 1700 RPM @ 32 dba and 67.14 CFM @ 3.6 mmH2O (this info from Panasonic tech support)

Sunon, and all major mfgrs have real numbers - and valid recommendations

re "free air" results: that is in fact what the single, stacked, and spaced tests are
the discharge duct has less then 1Pa of pressure (measured)
- and probably why the single fan output was over spec

some comments on the Nidecs: they too are old but are thick-bodied and draw 1.9A, high torque for sure and quite suitable for stacking
and they do work quite well in this configuration (but what a roaring !)

in my now rather extensive air flow testing I have come to appreciate just how compressible air is
the radiator flow setup was over 20 feet long and one could observe the pressure "waves" occuring and gradually decreasing
- unstable flow regimes are not difficult to create

my best guess (up to now) is that the greatly decreased output is due to the pressure pulses from the leading (5 bladed) fan precluding the second fan from getting a clean "bite" so it's windmilling and blocking the output of the first
and this is exacerbated bu the very low torque fo these fans (given their size)

I don't see the 3 1/2in. thickness of the very coarse-finned Cube as doing much to smooth these pulsations

ideas ??

be cool

(edit due to hitting the wrong button)

[nihili]

I hadn't considered pressure waves as a culprit. I agree that the radiator wouldn't do much to smooth them.

I'm thinking and have some ideas, but would like a bit more information if you've got it.

1. Were the fluctiations random or rythmic?

2. In either case were the fluctuations slower or faster when the fans were spaced farther apart? In any case, can you give some idea of how fast the fluctiations were?

3. I'm assuming that in a single fan setup there are no appreciable pressure waves apart from the intial one. Is this correct?

4. I'm also guessing that you turned both fans on simultaneously, correct?

nihili

[BillA]

1) random in terms of their period, repeating in terms of amplitude
(but they would not be random in the mathamatical sense)
repeating groups of varying waves

2) no discerenable difference, est. 3/4 to 3 sec. or so

3) not really, there is always some fluctuation (given the sensitivity of the instrumentation)

for example; the 400fps number (71cfm) was actually 390 to 410, and every 5 sec or so twice that range or more
- while the inst mfgr states 5% accuracy, engn practice gives these measurements a 15% tolerance

but for sure the variation with the single fan was far, far less
variation will normally be a % of the reading

4) tried every possible combination, front first, back first, and both, and "rolling" starts (with the Nidecs it can make a difference), but always the same

and ??

be cool

[Colin]

Originally posted by BillA
since my (measured) experience with the thick-bodied Nidecs was different than what youall (Texas here) were reporting,
I decided to test

using 120mm Panaflo FBH-12G12L fans (68cfm nominal rating)
blowing into a 6in. duct:

1 fan pushed 71cfm
2 fans stacked pushed between 27 and 62cfm (unstable fluctuating output)

2 fans stacked with a 2in. fitted square 'spacer' had the same output as the close-coupled pair

scratching my head, saying wtf
I then tested a Danger Den Cube

1 fan mounted on the intake side (13/16in. standoff) pushed 34cfm
2 fans (the "puller") close-coupled to the fins) moved 8 to 11cfm, fluctuating

I was/am surprised
the radiator 2 fan recommendation (if the fans are the same) is all wrong

amazing what a test can reveal

be cool

Bill - If I am reading this correctly, one fan pushing and another pulling on a Cooling Cube don't work according to the numbers (CFM)? How do we explain the 3C temp drop at full load I got when using this configuration? The majority of posts I have read by folks who have tried this dropped full load temps 2C to 3C.

[BillA]

Hi Colin,

I'm thinking the "problem" is related to my using low speed (low torque) fans

you have any ideas ?

be cool

let me add to this:

fan mfgrs make different blade designs for pushing or pulling,
and normally the appropriate model would be used
this would produce the best results
(but I suspect that most OCers just slap on 2 of the same)

how did you do it ?

2ed edit (since no one has yet commented):

with the Panaflos there is no evidence of a beat frequency
with the parallel Nidecs this is not so; one can hear them, depending on the setup, moving in and out of phase - and this will affect the total output
once the Nidecs become locked in-sync it precludes flow rates immediatly above and requires the application of full voltage to break up the feedback loop

unfortunatly I do not have an optical tach

[nihili]

I'm sure you're right about most OC'ers just slapping two of the same fans on. In fact, I've never seen fans advertised in a way that distinguishes which end of a duct they would go on.

My current thinking is that it has to do with the effect of various pressure waves. When you start a fan, you get a positive pressure wave forward, and a negative wave backwards. Given the low power of the fans, these waves may have a substantial effect on fan speed as they pass through the fans. This would in turn create other waves. The key here is that the fans are more affected by pressure waves than they are by their own motors.

I'm still trying to think my way through a couple cylces of waves. But it gets complex after the first waves pass through the fans and I haven't had a quiet spot to work through what happens then.

I'll say more after my 2 year old goes down for his nap. ;)

nihili

[BillA]

The key here is that the fans are more affected by pressure waves than they are by their own motors.

for sure, the difference is audible - they will not rev up
when they try, that's the "high" flow
- but then their mutual interference drives them back almost to the stall point

be cool

[nihili]

Ok, another question, this time about blade shape. I'm guessing that the blades on the Panaflows are broadest at the tip. I'm also guessing that the side of the blade that faces against the flow is convex while the side of the blade facing with the flow is concave or flat. I'll even be so bold as to guess that the back of the blade is more convex at the tip than at the hub.

Are any of these guesses right?

(Thinking about Bernoulli)

nihili

[BillA]

pretty much
without taking sections through the blade, it seems to be a constant chord
- simply getting larger away from the hub (about 2X)

bear in mind that this fan is now 3 generations past

be cool

I would add that the angle of attack decreases towards the tip
(related the the velocity)

[nihili]

With respect to pressure waves, there are four cases to consider, positive and negative pressure waves moving backwards and forward through the fan. In my descriptions I consider only the leading edge of the wave. The trailing edge may be treated as the opposite sort of wave travelling in the same direction.

Positive Forward - As this sort of wave reaches the fan, the fan should begin to speed up because of the reduction in net static pressure. But as the fan begins to bite into the wave, there are some factors that would mitigate and might potentially overcome the increase. First, the air in the wave is denser and hence offers more resistance to the blades. An underpowered fan might have difficulty with this. Second, if the wave is moving faster than the fan is pushing air, the wave will push air through the fan. THe described shape of the blades (having a longer back face than front face) could act as an airfoil. The resulting Bernoulli effect would actually produce a rotational force acting against the motor. If the fan is not able to handle the increased airflow, some of the wave will be reflected back, resulting in a positive wave travelling backward through the flow.

Positive Backward - As this sort of wave reaches the face of the fan, the fan should begin to slow down because of the increase in net static pressure. In this case considerations of density will exacerbate rather than mitigate the change. In the given setup we don't really have to worry about air being pushed backwards through the fan. In this case the reduction in fan speed would likely prevent reflection of the wave (unless the wave were very large). In fact it seems more likely that the drop in fan speed would result in a forward travelling negative wave.

Negative Forward - As this sort of wave reaches the back of the fan, the fan should begin to slow down because of the increase in net static pressure. However, this slowing will be mitigated as the wave moves through the fan due to the decreased density. By the time the leading edge of the wave has passed through the fan, the fan should be increasing in speed. In this case as the fan initially decreases speed we should see a decrase in CFM through the fan. This decrease would lead would lead to a positive pressure wave being sent backwards.

Negative Backward - As this sort of wave reaches the face of the fan, the fan should begin to increase in speed because of the decrease in net static pressure. This increase should be increased even further as the wave passes through the fan. The increase in fan speed may send a positive pressure wave forward.

The important thing here is that as the initial pressure wave from the rear fan reaches the front fan, the front fan is slowed down and reflects a wave with a positive edge back toward the rear fan. When this edge reaches the rear fan it slows that fan and reflects a negative wave forward. When the negative wave reaches the front fan, the front fan is slowed and sends a positive wave backwards.

THe key is that the front fan responds to a forward moving positive wave by slowing down rather than speeding up. Most fans will speed up, but given the description of the blades, I'm guessing that this fan actually slows down. This sets up a pressure oscillation in the area between the two fans with the resultant snafu in airflow.

Their are two blade features that are relevant and known. One is the curve back edge which allows a Bernoulli effect on the front fan. The second is that the blades widen toward the tip. They've compensated for increased tip speed, but haven't attempted to equalize force along the blade. This means that pressure changes at the tips might exert more force than the motor could handle. Of course with an good sized motor, neither of these factors would matter. But with the light motor they might.

Does this sound vaguely right?

nihili

[adamtekh]

u know to much realy

[BillA]

thinking out loud here, . . .

it's clear that it's the presence of the second fan that's preventing the first from running at speed
and that the output of the first, even straightened by the radiator fins, can almost stall the second

these fans will run in a stalled condition (against a closed duct)
so I would conclude that it is the pressure interaction between them, rather than any imparted flow resistance, that is causing the difficulty

a comparison of the 3300 RPM thick-bodied Nidec to the low speed Panaflo is interesting, but I'm not sure how instructive
- the camber of the Nidec is much higher and the leaading and trailing edges are parallel

your wave analysis implies a degree of sinchronicity that I doubt exists, the erratic hi-lo cycling indicates that there are several flow interaction regimes, all of which are negative

it seems as if a strong low pressure is being created between the two
I did not think to connect a manometer between them when they had the 2 in. gap

for sure I agree that the low speed/power motor is a factor

I'm trying to scrounge up some loaner fans to do some testing, hopefully JoeC will come through
would be nice to run the same setup with more powerful fans

be cool

adamtekh, one can never know too much

[nihili]

I didn't mean to imply a high degree of synchornicity with the wave analysis. A forward positive wave and backward negative wave will be started when you turn the fans on. But once secondary waves start generating all heck is going to break loose.

Yes the fans will run in stalled condition. Will they run at speed?

I don't see anyway for the area between the fans to be at a lower pressure with both fans runing that it would be with only the forward fan running. But I'm presuming that in such a setup we would see better results than we did in your test. So there has to be something more going on than simple negative pressure. The only other things i've been able to come up with are density and Bernoulli effect. Any other ideas?

When we're done with this puzzle, I'd be interested in finding out what effect running fans at 7 volts has on their CFM.

I have to admit I'm pretty close to stumped on this one.

nihili

[BillA]

I did not (mean to) say that there was a negative pressure
(I don't know how such would be possible)
only that the fans seemed to be behaving as if that were so

when stalled, the pitch changes; but not having a good tach . . . .

some fan mfgrs have (provide) 7V P-Q curves
the output is not hard to measure, the static pressure part is a PITA

here is an Intel example (http://www.ecom-answers.com/PQcurve.pdf )

be cool

[nihili]

Thanks for the link. I've been debating wether to use Deltas at 7v or some High speed Sunons. It looks like the free air performance is roughly proportional to the percantage of full voltage, but that the percentages get worse as back pressure increases. Given that, the 7volt trick doesn't look like a good way to design a system.

Let me know if you come up with any ideas on the Panaflo situation. The only thing I've come up with that might give the effect of negative pressure is Bernoulli effect. But frankly it's hard to imagine how you could get a strong enough effect from that under the circumstances.

nihili

[nihili]

BillA,

About loaner fans. I've been meaning to buy a matched set of 80mm fans. I've pretty much decide on the Sunon 50CFM 40dBa fans. I'd be willing to have em shipped to you, let you test em for a week or 10 days, and then forward them on to me.

Interested?

nihili

[BillA]

nihili,

super, another brick in the wall
I'll pm you my address

got a link to the P-Q curve for that model ?
I would like to relate fan/radiator thru put to the P-Q curve

and THANKS, Bill

be cool

[ButcherUK]

Originally posted by nihili
Thanks for the link. I've been debating wether to use Deltas at 7v or some High speed Sunons. It looks like the free air performance is roughly proportional to the percantage of full voltage, but that the percentages get worse as back pressure increases. Given that, the 7volt trick doesn't look like a good way to design a system.

Given that it would be interesting to see if a PWM fan does better than a voltage lowered fan at low speeds, care to try that Bill?

[BillA]

ButcherUK

is your thinking that a PWM controlled fan vs a voltage controlled fan will have different power (or torque ??) characteristics ?

just asking - I have never even thought about it

I have a PWM "kit" (un-assembled) that I bought for use with a TEC
I suppose it would work

what is the advantage of PWM for fans ?

I did see a bunch of circuits on Low-Noise.de for fan control

I use 150W rheostats ($6 ea.) for the 2A Nidecs as I often wish to run them at full voltage

be cool

[ButcherUK]

Yes the theory behind using PWM at all is that you get better torque at low RPM. Because PWM works by pulsing the power on and off, the on state is driving the motor at full power so you get higher torque than if you have lower power to the motor. Whether this actually works well for fans or not is another question. Also PWM can run at full voltage, if you lengthen the pulses enough they merge and you get full power, most PWM controllers have this is their upper bound.
Edit: Just thought to add the main drawback of PWM, it tends to make the motor get a lot hotter than just dropping the voltage, this shouldn't be an issue except in badly designed fans.

[BillA]

makes sense, never thought to use the PWM kit for that

for the Panaflos which nihili and I have been discussing it would not apply, their poor performance is at 12V

be cool

[ButcherUK]

Yes I was more thinking of the higher power fans, in particular I'm curious as to whether my 2 ys-tech 120s will be better in parallel or series, be interesting when you have some more data.

[RPM_Computing]

Well I am not a mathematition but here is what I did.

1. Stacked 2 fans one on top of the other and did not notice any improved temperature results.

2. Took the two fans and created a "V" shaped design with both of the fans pointing at a 45 degree angle on to the heatsink so that there would be doubble the air flow to my heatsink. Then I made a duct to a 4" nidec betav fan blowing air on top of the two fans and have noticed a much greater reduction in temps.

I do not have any monitoring devices that show RPM's or temp's, otherwise I would have reported. Any way this was a very cost effictive solution to get my temps down as I made the ducts from a cerial box and used black electricial tape to seal the duct. No power tools were required to build this, just a pair of scissors.

A dividing wall was also use between the "V" fans mounted on to the heatsink but I did not notice any difference either with or withought. The diagram below is obviously not to scale but if anyone chooses to build their own this should give you a good idea of what I am talking about.

[Owenator]

THis is a very stimulating discussion, it re-kindles the fluid dynamics part of my brain!

As I recall at the flowrates (speeds) we are talking about air is still pretty much behaving as a fluid so I doubt compressibility is too great an issue, but pressure waves are still very possible.

In my fluids studies I noticed that turbulent flow to the inlet of a fan can cause a lot of variation in performance. Have you tried to add a flow staightening device between two fans? I have seen square grids of sheet metal placed between fans to staighten the flow so that the second fan gets essentially laminar (non-turbulent) flow at its inlet. In this way you can construct an essitially multi stage air pump. I will try this this week end as I am getting four days off! Wahoo!

Right now I have a 120mm Panaflo H blowing onto my heater core rad and an 'identical' 120mm Panaflo H. Both are running at 7V. My case is supplied by two 120mm Panaflo M's blowing in. I know the flow is not balanced as the outlet fan is rated at ~60 CFM at 7V and the two inlets are rated at ~50 CFM at 7V. Even with the modest power supply fan at maybe 20CFM there is a substantial flow imbalance. I am thinking of going back to my four 80mm Panaflo L series as outlet fans after the rad but I am now considering a flow straightener after my radiator to give them clean smooth inlet flow.

Any thoughts?

BTW

nihili,

this may be amusing after the persons comment that you 'know too much'. I once heard a joke that a professor came into his class after grading an exam and said "If ignorance is bliss, why aren't more of you smiling?" I know I keep smiling ! :)

O

[BillA]

Owenator,
I'm certainally NOT qualified to comment on fluid dynamics theory,
but experimentally I can "see" the "waves" even in the range of 50 fpm

one of the trials described above was through a radiator with continous flat fins and transverse tubes
this was done to look at (also) straightening the flow
with no difference

your situation would seem to be different in that you are using different fans for the supply and exhaust

but I suspect the most relavant difference is the use of "H" fans instead of the "L"s I've got

very interested in any info you can share

be cool

nice professor joke <g>

[Owenator]

BillA,

I am using a heater core but I have no air flow measurement tools. The heater core has flat passages from the top to the botom and "corragated" (for lack of a better term) aluminum sheets zig zagging between the passages. The air flow resistance is pretty low. I can hold it up to my mouth and breathe through it easily. I notice better air flow with fans sucking air through it rather than pushing at it. My two H fans are on either side of the rad with adapters from the 120mm fan to the 7" square rad. The fan rad setup is inside my case which is 'pressurized" by the two H fans blowing in. I realize my fans are too close to the rad to overcome turbulent effects but they push air through very well. My idea is to go back to a four 80mm fans connected edgewise so I get an effective 160mm x 160mm fan area. With the fans set up like this I can relocate the fans and rad to get about 6" between teh rad and the fan intakes. I originally did this but went to the 120mm as a 'tweak'. I even tried four temp controlled 92mm fans once but it was too noisy for my tatstes (I really want silent).

O