How to model/predict airflow? (and some GS1000 specifics)
Posted: Wed Apr 01, 2009 12:45 pm
Howdy folks,
I've done enough lurking and searching to know that the answer to pretty much any question about airflow is going to be "it depends"
So, hopefully a slightly more intelligent question - beyond gut instinct and a degree of common sense how do you model airflow to check that what you think will happen really does? Thus far, beyond trying different configurations and trying to draw inferences from temperature variations, I've had a modicum of success with.... cigar smoke Whilst this undoubtedly does nothing favourable for internal components, in the grand scheme of things I should probably be more concerned with what it's doing to my own internal components.
Are there any more scientific or accurate methods which I've missed and which are within reach of Joe Public? (ie. I don't have access to a wind tunnel/aerodynamics test lab)
Ok, case specifics - Zalman GS1000, bottom mounted PSU, largely similar to the much favoured Antec P182, does a couple of bits better and a couple (to be fair, probably a bit more than a couple) of bits worse.
3 exhaust fans, 2 top, 1 top-rear (all Noctura 800RPM 120mm)
1 input fan at bottom middle, in front of PSU and behind HDD bays (of which four are in use), again a Noctura 800RPM 120mm.
Military surplus jet turbine top down CPU blower (Zalman CNPS-8000 - I know, I know, it came out of a HTPC case, throttled down to 5v it approaches more bearable Cessna noise levels)
Small volcano of a graphics card (HD4870 - but they all seem to run hot and the stock cooler is suprisingly restrained).
Temperatures are fine, CPU cores run at 30-45, mobo at 35-40, GPU at 55-80. All the HD4870s seem to run hot but the stock cooler seems to do a good job at ensuring heat doesn't leak into the rest of the case. I managed to get idle temps on the GPU down from 70 to 55 by fixing the cack handed bios (there was no dynamic underclocking in 2D/low 3D usage) - this had seemingly no effect at all on mobo temps so I infer from that it keeps it's heat well contained. HDD temps I can't tell right now, they're off a raid card so I can't get easy SMART readings - however they pass the touch test with flying colours and the cigar test confims a decent pull of air from the front of the case through the HDD bays. I can only assume that the rising column of air at their back from the bottom mounted intake fan creates a low pressure area to either side of the column which acts to pull air through the front HDD bays.
Anyway, question time - and I appreciate this has come up before - PSU orientation? The instruction book which comes with the case specifically notes that dual fan PSUs be installed upside down with the 120mm intake facing up into the case. My gut instinct tells me this must be wrong - I figure it's running the risk of short circuiting the bottom intake fan and potentially fighting with the GPU fan for air, but perhaps it does work to extract a bit more heat out the case (although how much heat radiates down?) - I figure the PSU is more than up to sucking a bit more heat out given that typically it'd be at the top of the case and dealing with a lot more. So which is best? Gut instinct says let the PSU suck nothing but cold air and exhaust only the heat it itself produces but Zalman (who probably have more testing facilities and qualified engineers at their disposal than I) say the reverse.
I've tried both. Makes no difference whatsoever Temp differences across all components register no changes large enough to be significant. Noise wise my non-calibrated night-club going and motorbike riding ears fail to tell any difference; I suspect that less ravaged ears or sensitive testing equipment might possibly find favour in the Zalman approach as it places a fan inside the case rather than facing out (although a decent temperature sensing PSU, which this is, barely spins that fan anyway).
So all in all I'm not entirely sure why I should be bothered either way But for some reason I am, if I can get a 1% efficiency bonus for free then I guess I'm the kind of person to whom that sort of thing oddly appeals
I've done enough lurking and searching to know that the answer to pretty much any question about airflow is going to be "it depends"
So, hopefully a slightly more intelligent question - beyond gut instinct and a degree of common sense how do you model airflow to check that what you think will happen really does? Thus far, beyond trying different configurations and trying to draw inferences from temperature variations, I've had a modicum of success with.... cigar smoke
Whilst this undoubtedly does nothing favourable for internal components, in the grand scheme of things I should probably be more concerned with what it's doing to my own internal components.Are there any more scientific or accurate methods which I've missed and which are within reach of Joe Public? (ie. I don't have access to a wind tunnel/aerodynamics test lab)
Ok, case specifics - Zalman GS1000, bottom mounted PSU, largely similar to the much favoured Antec P182, does a couple of bits better and a couple (to be fair, probably a bit more than a couple) of bits worse.
3 exhaust fans, 2 top, 1 top-rear (all Noctura 800RPM 120mm)
1 input fan at bottom middle, in front of PSU and behind HDD bays (of which four are in use), again a Noctura 800RPM 120mm.
Military surplus jet turbine top down CPU blower (Zalman CNPS-8000 - I know, I know, it came out of a HTPC case, throttled down to 5v it approaches more bearable Cessna noise levels)
Small volcano of a graphics card (HD4870 - but they all seem to run hot and the stock cooler is suprisingly restrained).
Temperatures are fine, CPU cores run at 30-45, mobo at 35-40, GPU at 55-80. All the HD4870s seem to run hot but the stock cooler seems to do a good job at ensuring heat doesn't leak into the rest of the case. I managed to get idle temps on the GPU down from 70 to 55 by fixing the cack handed bios (there was no dynamic underclocking in 2D/low 3D usage) - this had seemingly no effect at all on mobo temps so I infer from that it keeps it's heat well contained. HDD temps I can't tell right now, they're off a raid card so I can't get easy SMART readings - however they pass the touch test with flying colours and the cigar test confims a decent pull of air from the front of the case through the HDD bays. I can only assume that the rising column of air at their back from the bottom mounted intake fan creates a low pressure area to either side of the column which acts to pull air through the front HDD bays.
Anyway, question time - and I appreciate this has come up before - PSU orientation? The instruction book which comes with the case specifically notes that dual fan PSUs be installed upside down with the 120mm intake facing up into the case. My gut instinct tells me this must be wrong - I figure it's running the risk of short circuiting the bottom intake fan and potentially fighting with the GPU fan for air, but perhaps it does work to extract a bit more heat out the case (although how much heat radiates down?) - I figure the PSU is more than up to sucking a bit more heat out given that typically it'd be at the top of the case and dealing with a lot more. So which is best? Gut instinct says let the PSU suck nothing but cold air and exhaust only the heat it itself produces but Zalman (who probably have more testing facilities and qualified engineers at their disposal than I) say the reverse.
I've tried both. Makes no difference whatsoever
Temp differences across all components register no changes large enough to be significant. Noise wise my non-calibrated night-club going and motorbike riding ears fail to tell any difference; I suspect that less ravaged ears or sensitive testing equipment might possibly find favour in the Zalman approach as it places a fan inside the case rather than facing out (although a decent temperature sensing PSU, which this is, barely spins that fan anyway).So all in all I'm not entirely sure why I should be bothered either way
But for some reason I am, if I can get a 1% efficiency bonus for free then I guess I'm the kind of person to whom that sort of thing oddly appeals