Schematics for Delta PSU

[Jockelill]

Ah, cool, I’ve only seen it in 5V version, should work just fine then . I’ll post some picture of the PSU with the Noctua fan later, but in terms on noise it made a huge difference!!

 

[Phipli]

Careful folks, Noctua quote unloaded speeds. Its like torque and power with cars. A sports car might do 150mph, but it won't pull a plough as well as a tractor with the same power.

Check the airflow and compare temperatures. I would be very cautious replacing a fan. To maximise life, and minimise power usage and noise, you're meant to aim for a certain location on the fan curve (airflow vs pressure) where small changes in pressure don't cause significant changes in speed.

Out of curiosity, how do the power ratings compare between the stock and replacement fans?

 

[Phipli]

So in the attached (just a generic plot to show a typical curve), you're aiming to stay to the right of the hump in the static pressure curve. If you get too near the flat bit the speed is unstable and changes wildly - it isn't good for the fan.

This is on top of making sure you have enough airflow to cool the computer / psu. Noting many consumer fans are designed to cool large open cases, not tight, convoluted PSUs and cases with a single fan.

 

[cheesestraws]

So in the attached (just a generic plot to show a typical curve), you're aiming to stay to the right of the hump in the static pressure curve. If you get too near the flat bit the speed is unstable and changes wildly - it isn't good for the fan.

I would love it, if you have time, a quick 'fans for computer people' guide. You have unusual expertise in this area, and selfishly I'd really like to learn more about it.

 

[Phipli]

I would love it, if you have time, a quick 'fans for computer people' guide. You have unusual expertise in this area, and selfishly I'd really like to learn more about it.

I'll put something together I'm trying to work out what the simplest setup is to test things. Measuring pressure with a manometer is a good start because they're basically free... then I'm pondering if I can calibrate a cheap aftermarket mass airflow meter for a car to work at our low flowrates by running the heated wire at reduced voltage... or possibly get away with no airspeed at all and assumptions.

I hate how there is so little data on computer fans available publicly. I'm used to manufacturers giving me detailed data in the hope I buy their fans.

 

[Phipli]

I would love it, if you have time, a quick 'fans for computer people' guide. You have unusual expertise in this area, and selfishly I'd really like to learn more about it.

This is a nice read. I've only skimmed it, but it covers common pitfalls wrt assumptions people make.

AnandTech Forums: Technology, Hardware, Software, and Deals

Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

www.anandtech.com

What I'd add is old macs have a higher resistance to airflow than modern computers due to the way they were generally designed to use a single psu fan.

 

[Jockelill]

Careful folks, Noctua quote unloaded speeds. Its like torque and power with cars. A sports car might do 150mph, but it won't pull a plough as well as a tractor with the same power.

Check the airflow and compare temperatures. I would be very cautious replacing a fan. To maximise life, and minimise power usage and noise, you're meant to aim for a certain location on the fan curve (airflow vs pressure) where small changes in pressure don't cause significant changes in speed.

Out of curiosity, how do the power ratings compare between the stock and replacement fans?

Stock fan had 0.12A at 12V and Noctua is 0.11A, but yes, you’re quite right. I did compare a few fans with similar specs and they where all in the same flow range, but I can’t say now if they where all unloaded or not. Anyway, when I get time I will do a thorough test with K-thermocoupler and one of my Multichannel loggers. It’s difficult to measure these relatively low pressures, and also to find a good measurement point, but a temperature measurement should be a good start at least.

 

[Phipli]

Stock fan had 0.12A at 12V and Noctua is 0.11A, but yes, you’re quite right. I did compare a few fans with similar specs and they where all in the same flow range, but I can’t say now if they where all unloaded or not. Anyway, when I get time I will do a thorough test with K-thermocoupler and one of my Multichannel loggers. It’s difficult to measure these relatively low pressures, and also to find a good measurement point, but a temperature measurement should be a good start at least.

Yeah, for my tests I was considering a larger external fan on a known quantify duct connected to the inlet, to increase the pressure differential to characterise the case, then compare the characteristic with manufacturer fan curves. Would make measurements easier.

Sounds like you're on it. Most of these fans don't have a hugely pronounced knee so hopefully picking something with a similar or better pressure and air flow should see you through.

Interesting that the noctua is similar power in this instance. Previously I've seen people replace a fan with a noctua one third of the power. My thought is that it is unlikely to be three times as efficient, so in its design case (which may not be the same) it is likely doing 1/3 of the work.