Forum Home
Press F1
Thread ID: 66196	2006-02-14 20:40:00	Adding cooling	The_End_Of_Reality (334)	Press F1

Post ID	Timestamp	Content	User
430383	2006-02-21 04:53:00	Ok tweak'e, you have given me good excuses to play a bit more :) I can measure rpm easily enough, one of the things I did was to convert a Xenon car timing light to trigger off an audio oscillator. I observed that if I put my hand in fron of the fan, about 4 - 6 inches away, the fan noticeably slowed, but I didn't record anything, just playing. But it does show that what happens downstream can affect things upstream, a bit like fluidic logic devices if anyone can remember those. I didn't have to calculate fan area for the air speed tests as I was just comparing air speeds, but I have done so to get an estimate of fan flow rate. The outer diameter is 76mm, and the hub diameter is 44mm. This works out at an annular flow area of 0.003m². So the flow rate is 0.003x3.2 = 0.0096m³/sec = 20cfm This is somewhat lower than the flow rate for the later model Toyo fan of 31.4cfm, though the later fan is 25mm deep as against 22 for the one I used, and it is also rated at a slightly higher current. The flow discrepancy is also one reason I am dubious about the (absolute) validity of the measurements, but just comparing flow speed with and without a perforated plate, with the anemometer in the same positions each time, should be reasonable. The other thing I noticed when the cardboard tube and spacers were attached to the fan, was that the 'dead zone' of air due to the hub extended at least the tube+spacer length of 105mm. That is, the anemometer doesnt spin when placed on the centre line. However the anemometer is big compared to the flow area and will affect the flow. What is really needed is a tiny hot-wire anemometer probe and meter, but these cost $$$$$$. ;) Edit: I'll see if I can get some longer cardboard tube, about the depth of a tower case, or preferably use an airtight cardboard box, then the effects of the hub dead zone could be eliminated.	Terry Porritt (14)
430384	2006-02-21 05:13:00	Marconi, HP, Marconi... I love your workbench, Terry. The test rig I considered was a box with holes in opposite ends, and a manometer to show the internal pressure. That could be used to show the effect of different size fans, holes (with and without fans) , obstructing grilles, etc. Axial flow fans aren't really ideal for modern computers, and they're probably used because "they've always been used". That 's a fairly pathetic pressure, far worse than I imagined. A good centrifugal feeding ducting to the places where cooling is needed would be far more effective. Some people might complain about the noise, but why have they bought 7.1 speaker systems? :D Some of the video cards look to have centrifugal blowers. Though that's just my impression. Some might even be just stirrers. That seems to be what a lot of the axial flow fans in PC boxes do. If there's any back pressure at all, or input constriction, they don't move much air.	Graham L (2)
430385	2006-02-22 22:58:00	I've combined some of tweakes ideas with Grahams, and mounted the fan on top of a cardboard box, with a hole cut in the front at right angles to the fan: . imagef1 . net . nz/files/boxflow . jpg" target="_blank">www . imagef1 . net . nz The idea was to try to get a more uniform flow through the hole without a central dead spot . Then I measured air speed through the hole with and without a perforated plate, and with the fan blowing into the box, and then sucking out of the box . In the picture at the front left is the anemometer, then the timing light to 'strobe' the fan, a digital multimeter to check fan voltage 12v +/- 0 . 05v, and a digital frequency counter to get fan rpm . The Marconi oscillator at the back on the right triggered the stroboscopic timing light . The white blob on the fan is for strobing . Results: Now with the fan blowing into the box, the flow out of the open hole registered 1 . 9metres/sec, and when the hole was covered with the perforated plate, 0 . 9metres/sec . That is a speed ratio of 47%, which is comparable again with the 40% open area of the perforations . Interestingly the fan speed increased from 2730rpm to 2840 rpm when the plate was put in front of the hole, presumably because with a lower flow it is doing less work . With the fan turned over and sucking air into the hole, the anemometer readings were more variable, and were more sensitive to how close to the hole the meter was held . Possibly, because now air had to be sucked through the anemometer . Anyway with the anemometer placed right against the hole, the air speed going in was 1 . 7 to 2 . 1 metres/sec with an open hole, and 0 . 7 to 1 . 1metres/sec with the perforated plate . This time, with the fan sucking, the fan speed didn't noticeably change when the perforated plate was put over the hole . I can't explain that one . So the overall conclusion is unchanged . The use of perforated grills will reduce flow in about the same proportion as the percentage open area of the grill, that is, assuming no other holes in the case . (The use of a Griffins Wheaten carton did not seem to confer any special properties to the air flow)	Terry Porritt (14)
430386	2006-02-23 02:07:00	what speed was the fan doing when sucking into the hole? as exspected the fan rpm went up on the blow test, the more pressure the fan is under the less grip it gets on the air. i would have exspected tha same on the suck test, mayby it allready was going fast? i wonder if a lot of the varition in the readings is due to haveing a flexable box? it could be just the fan design, its not good with inlet restriction.	tweak'e (69)
430387	2006-02-23 02:40:00	I suspect the fan is more sensitive to back pressure than to input constriction . To overcome back pressure, it's performing work . If the motor power is limited (by its impedance -- these 12V DC fans are actually three phase AC motors) it must slow down . If the input is less it just moves less air, so will maintain its speed . tweak'e: "blowing up" the carton won't affect the air flow . Once it reaces equilibrium, it's a constrant volume . I suppose it's easier to stir air than to move it . Centrifugals have two styles of rotor . One has forward facing, the other back facing vanes . The forward facing vanes cause the power needed to increase with back pressure, or input flow restriction . It tries to maintain the flow . The reverse facing vanes cause the power to limit at some point . A total shut off of either input or outlet does no harm . With a "forward" rotor, you can burn out the motor .	Graham L (2)
430388	2006-02-23 03:04:00	Sorry, I didn't quote the fan speed in sucking, It was 2710 rpm without the grill and 2700 with the grill. However don't read too much into the last 2 figures, as the digital frequency counter at that speed reads around 0.0452 Khz = 45.2Hz with a gate or counting time of 10 seconds to get that last decimal place. Then it is multiplied by 60 to get rpm, so speed is about +/-10 rpm or so.	Terry Porritt (14)
430389	2006-02-23 03:44:00	Wow Terry, you have put a lot of time and effort into this, thanks for sharing it with us, it was very interesting :thumbs:	The_End_Of_Reality (334)
430390	2006-02-24 19:33:00	Just been doing a little tinkering with the CPU fan and ITE System Guardian. I was able to get the CPU temp down to 24°c idle in a 21°c room :D and a full load temp of 38°c. I am fine with that, I will still drill out the rear holes, but not sure when, might not have time this weekend but... Now all I need is to tinker with the GFX card fan, what programs can do that? He He, I got the ° sign :D	The_End_Of_Reality (334)
430391	2006-02-24 20:53:00	Putting a damper on your enthusiasm, I'd be a bit suspicious of those low temperatures, they sound too good to be true :) But since presumably the same sensors are being used as originally when you reported 47ºC, then there is an enormous improvement, Good oh !	Terry Porritt (14)
430392	2006-02-24 21:43:00	I realise that it is unlikelyto be the CPU core temp, but the heatsink temp, as you said seeing as it was the same sensor that previously reported 47°c that is a huge drop, now the CPU fan is on full all the time :D though I can't hear it because of the other fans so it doesn't matter. I might even get lower when I drill out the rear fans, but I am not too hopeful :p	The_End_Of_Reality (334)
1 2 3 4 5 6 7 8 9