For ages I've been using 120mm fans mounted in the side panel of my case. The only downside is that the fan is exposed to the 'outside' and its noises will escape into open air. It can also let harddisk and component whine out through its hole. So what happens when you make a 90degree angle duct as seen in the pic:



The duct opening is 4x12cm. The fan is a Nexus 120mm, ducted (I love ducts) to my cpu heatsink. The results are shocking! (Prime 95 until temperature stabilizes)

No duct:
---------
fan rpm 510, max diode temp ~67..68 degrees c
fan rpm 660, max diode temp ~50..51 degrees c (max speed)

With intake duct:
-------------------
fan rpm 540, max diode temp >79 degrees c (eek, I took it off.. would've gone higher!)
fan rpm 720, max diode temp ~71..72 degrees c

The fan speed increase is caused by the duct and the resulting pressure situation, despite the opening being 48cm^2

The temperature difference is huge. 4cm is about the maximum height I could use 'inside' the case, it would really make the case too wide going to 12cm for instance. Anything in the way of direct intake paths can dramatically decrease cooling efficiency!

It's not that shocking. There's more air resistance from going through that long duct so its efficiency is reduced. The fan used to take air from every direction, too, now the air is coming from one side only. What would be more interesting is to see how a taller (case-height?) duct affects anything.

A good rule of thumb is that the bottleneck is where the cross sectional area is the smallest. The length of a duct has a certain effect, but the main consideration should be to look at the minimum area cross section.

In this case, the bottleneck used to be the fan itself (roughly a 120mm diameter circle, minus some for the central hub). Let's guesstimate the area at around 100 square centimeters. Afterward, the bottleneck was the 4x12cm duct cross section--only 48 square centimeters!

Thus, you roughly halved the minimum area cross section. It's no surprise that air resistance was severely increased!

What you can do to more or less eliminate the bottleneck is open up the duct on both sides. This should have a more modest effect on air resistance.

I think I am missing something here, but why not duct on the intake side (ie. outside the case) rather than exhaust. Yeah your case now has a portrusion sticking out (you could fancy it up though) but you dont have back pressure to worry about. Having a fan blow onto a solid plate makes little sense.

If the fan has a diameter of 12cm and the hub is 4cm:

Potential area = pi * 12^2 / 4 = 113.1 cm^2
Hub area = pi * 4^2 / 4 = 12.6 cm^2
Available area = 113.1 - 12.6 cm^2 = 100.5 cm^2. Good guesstimate Isaac!


I think it's on the outside as he said it's ducted to the CPU heatsink.

Based on the pictures (fan on inside of case) and on his description of a duct to the CPU, I assume the duct IS on the outside--and his concern is whether it excessively sticks out.

:deleted:

Right you are! And back to the OP it is truely surprising how badly it performs.

Edited.

I expected to see temperatures rise, but not by 20+ degrees! (I thought 5, perhaps 8 max!) The reason for my experiment was that I'm currently designing a new case, made out of MDF. Less noise is barely possible, but a more solid / transportable / future proof case certainly is.

With case dimensions not really being restricted I wanted to see if I could place my fans inside the case and duct the intake to the rear of the case. It was a little experiment to see the influence on temps. I don't want to make a huge case as space is rather limited, so this brings me back to the old and proven 'side panel fans' approach. The case design is progressing nicely, I might have some simple renders ready tonight

You can reduce the air resistance while keeping the same depth by making the duct wider. For example, a 4x25cm cross section should eliminate the bottleneck. You want the full width of the box to "surround" the area next to the fan, to ensure that there isn't a bottleneck there.

Or you could make the "duct" much bigger than necessary (lets say a square plate with length 200mm) and attach it to the side panel with 4 thick large screws spacing it 2-3 cm off the side panel surface. These screws would hold the duct off the side panel and you would have gaps on all 4 sides to let cool air in. Of course it would let out more noise than a back facing duct.

Or you could open the back and down facing sides of the duct to see if that helps.

Enough rambling ...

Those temps are hard to figure. If your fan was blowing out through such a duct, you'd have excessive back-pressure.....and a bunch of problems.

But apparently you're pulling air through the duct, Which creates no back-pressure....only some resistance to airflow. And at those slow fan speeds, very little at that. I've built several similar setups with intake ducting like the pictures, and no temperature problems.

Your results are puzzling....

Well, I would have to dig deep in the Silent PC E-group, but I once put speaker cloth a few cm from the fans as a sort of fan guard + dust filter. This is pretty open cloth stretched onto a frame. The results where terrible, although I don't remember exactly what the temperature rise was..

Experience with dust filters was poor as well.. It just seems that intake should be free of fan guards, 90 degree ducts or dust cloth! I also tried a long 13cm round 'tube' straight onto the side panel. This had zero effect, temperatures where the same as without.

Perhaps part of the explanation is in the tappered duct that goes from the fan to the cpu heatsink on the inside of the case? I'm still puzzled though!

the taper duct would add resistance

The duct as designed is very restrictive. There are several factors that contribute to the restriction.

1. The reduced area of the duct compared to the area of the fan.

2. The duct causes a very abrupt change in the direction of the air flow very close to the fan.

3. The duct restricts air flow on 3 sides of the fan.

Also if you do a little test you will learn a something about how air flows into and out of a fan. Take something like a stick of insense and light so that it is producing a stream of smoke. Then start a fan running and insert the tip of the insense into the air stream of the fan at various points. You will notice that on the intake side that the air flow converges on the fan in a shape that is somewhat conical and is getting smaller as it approaches the fan. On the outlet side the shape of the airflow stream is also conical but it is getting bigger as it flows away from the fan. By placing the duct so close to the sides of the fan and forcing an abrupt change in direction causes this natual flow to be greatly disrupted.