Okay, it is true that hot air rises. However, I'm now of the firm opinion that for the purposes of quiet computing the effect is simply too weak to matter much.

I finally got around to slapping together a test rig for my solid state airflow generator--simply a 60watt light bulb at the base of an aluminum chimney tube. This tube is about 3 feet tall, which is pretty bulky to stack on top of a computer already. Incandescent light bulbs are notoriously ineffecient, so I'd guess that it's giving 40+ watts of heat right off the bat. The 4" diameter tube allows for completely straight upward airflow, with minimal air resistance (there's plenty of room for air to move around the light bulb).

The amount of airflow generated is almost undetectably low. A piece of paper needs to be carefully balanced for the tiny airflow to affect it at all. In contrast, an essentially silent undervolted fan has no problem pushing the same piece of paper around.

Based on these results, I'm abandoning my experiments into using a light bulb chimney to generate solid state airflow. Unquestionably, undervolted fans are the way to go. Of course, you all already knew that.

I'd think the calm air around your typical lamp would have shown the same thing. Always good to experiment though, you never know what whacky idea will actually work out well until you try it!

What happens with a 100W bulb? Does the size of the intake opening for the chimney change anything?

Most heat loss/transmission from a light buld is via radiation.

You give up too easily. I think your heat pipe is too short. Look at a furnace chimney stack....it extends above the roof. Partly works by pressure differential. I'd abandon the light bulb for heat and use heat tape (keeps gutters free of ice) wrapped around the upper part of the pipe. That would not block airflow.

Just cut a hole in your roof or rig up some other exit point.....run up a tall stack at least a few feet above the roof. You can get rain caps for these things at hardware stores. Bet that would give you measureable airflow.

"I'd think the calm air around your typical lamp would have shown the same thing."

Not really--a typical lamp doesn't have much of a vertical offset to create a stack effect, nor are they largely enclosed to maximize heat retention. (Doing so would be rather counterproductive toward the main goal of providing illumination.)

"Most heat loss/transmission from a light buld is via radiation."

No kidding. In this case, most of the radiation hits the aluminum tube nearest to the bulb. This aluminum gets hot quickly, at which point it can warm air flowing upward through the tube.

"What happens with a 100W bulb? Does the size of the intake opening for the chimney change anything?"

I haven't tried a 100W bulb because the results at 60W are so vanishingly minimal I don't expect it to be worthwhile. The size of the intake as it is is as big as will make any difference (the full 4" diameter, minus the small diameter of the light bulb fixture). The airflow bottleneck would be the cross section where the light bulb has maximum diameter. However, with the puny almost non-existent airspeeds involved, air resistance can be ignored.

"You give up too easily. I think your heat pipe is too short."

Well, I know the principle can work if scaled up sufficiently. The real question is whether or not there's any point in doing it at all if the scale needs to be so large. Even at 3 feet by 4 inches, the bulk is comparable to a fully isolated silent fan with an insane amount of muffling. (Line the tube with 1" thick foam, and put the fan at the midpoint. This leaves a 50mm diameter airflow path.)

Hmm...THAT, I might try next. I'll see what effect it has on a loud fan first...

How would you have used this chimney thing, if it worked?

It's kinda trivial that putting a bulb inside a computer makes it hotter, since there's no way in hell that it's going to move more heat out of the case than how much it produces.

However, there might be some point in designing the airflow path so that it goes upwards. Remember for example the small cubical mac that had no fans?

I suppose it would also make a difference how the air was getting into the case...it would be important that the only inlets were below the heat source, and that they were completely unrestricted (filters would be out of the question).

I have a feeling though that for significant convection to take place, the inside/outside temperature delta needed would normally make it too hot for the components anyway...

Remember how they commonly overheated and died?

what if you spray painted the inside of the tube black so it absorbed more heat?

Hmmm. Someone else had done a chimney and had it work before. Aha. Found it:
http://forums.silentpcreview.com/viewtopic.php?t=5089

Granted, this is a water-cooled setup, but the concept is probably similar enough. I believe an article of the same name is in the archives, too.

My plan was to design a negative pressure case around the light bulb chimney.

Air is exhausted upward from the case through a CPU heatsink exhaust duct. This passes through the PSU, placed in a vertical orientation over the CPU to keep airflow in the upward direction. Above the PSU is the light bulb chimney, along with some strategically placed aluminum foil to block light/IR radiation from directly hitting/heating the PSU.

Needless to say, the air leaving the chimney would have been hot because of the lightbulb. The light bulb would have heated up anything downstream of it--but there would have been nothing downstream of it except chimney tube.

I am surprised there hasn't been a fluid dynamics pedant around to set the thread straight. IIRC the fallacy is that "hot air rises". I am not qualified to take the job, but IIRC the reality is that hot air is lighter (or less dense or both?) than cold(er) air, and it is the effect of gravity on the cold air (causing it to fall) which displaces the warmer air up.

If all my drivel is vaguely correct, then perhaps the limit of your chimney was not allowing enough length (well height, as per Mike and others) or else to alter the design to benefit from falling cold air?

Increasing height would take advantage of a different effect, pressure differences at different heights. Higher up, there is less pressure. I believe that some of the cliff dwellings at Pueblo were cooled by this effect, relatively tall buildings with openings at the top and bottom.

But still one should be careful about coming to conclusions based on this experiment, probably more than half of the heat is being lost to air outside of the tube IsaacKuo. First, as you noted most of the heat is being radiated to the pipe which acts as a heatsink. Second, the air inside is confined by the pipe. Third, the surface area is roughly equal inside and outside of the pipe, and surface area is directly related to heat transfer to air, which would indicate roughly equal amounts of heat transfer to air inside and outside of the pipe. So the more heat is probably being transferred to the air outside of the pipe because the air outside can move more freely.

In effect, you're testing a <30 Watt heat source.

That said, I wouldn't expect to feel that much air motion...if you keep your hand over a 100 C steam house heating radiator there isn't that much motion of air.

Now, this honestly IS a myth. It's true that pressure decreases with height. This is due to gravity. Gravity is not magically shielded by a chimney tube. Thus, the same effect applies both within and outside a tube--you get no free lunch.

No, the true effect which matters is "the stack effect". This is a real effect, important in analyzing chimneys and tall buildings.



This shouldn't be the case, because the air outside is relatively stagnant and has no stack effect to generate airflow. However, with the incredibly low airflows I was getting, the difference between heat loss within and without would be minimal.

Tonight I'm repeating the experiment; this time I'm wrapping the light bulb within layers of aluminum foil. This should contain almost all of the heat, to be lost to air rising upward through the foil layers.

I still don't expect it to produce a useful amount of airflow, but there's no harm in giving it a shot...

As lm has pointed out, doesn't it seem kinda counter intuitive to introduce more heat to a system in order to try to cool it? Adding a, say, 50W heating element inside a computer case just to make sure that there is some airflow just doesn't make much sense to me.

Yes, I know about Peltiers. You introduce another X watts through the Peltier to get it to remove Y watts of heat. But X > Y, and unless you have an easy way of venting this heat, it's going to heat up the surrounding air.

If using a chimney works for a PC (I don't know either way) wouldn't it make sense just to have a PC case built like a tall chimney with open intakes at the bottom?

Imagine : a standard ATX case with openings at the bottom of the case, on stands elevating it about 2 inches. The top (or most of it) is removed and a funnel is attached to it. Connect a long tube to the funnel, extending, say, 5 ft.

I don't think this will work unless you have huge sinks with heat pipes on the graphics card and CPU. A desktop configuration might work better than a tower configuration.

And if you have huge sinks and heatpipes, you don't need the chimney. You'd have Zalman's behemoth case.

Do you have a soldering iron or some other heating element instead? Somehow the thought of wrapping a light bulb in aluminum foil really worries me. If you do go ahead, please make sure none of the foil touches the base of the bulb (insulate the base with electrical tape first is my suggestion).

Silly question; but you have insulated the outside of your pipe?

This is just what I was thinking. As was pointed out by another poster if the aluminum tube is not insulated the air just outside of the tube will get almost half of the heat. You need for the temp of the exiting air to be as high as possible to maximize the air flow. How tall the tube is also important. Three feet is probably too short. Also the light bulb is probably gives off about 85% of its power rating as heat perhaps more.

wont be long now

For those who keep asking, he has the lightbulb ABOVE the computer. the bulb isn't inside the case.

I think Andy was right about the tube acting as a heatsink. I think you should try to insulate the tube. I think you should also paint the inside of the tube black so that the infrared light coming from the bulb doesn't just reflect all the way down the tube and is instead almost completely turned into heat.

Also, don't wrap the bulb itself in foil, it'll pop from the heat. It's the same idea as the prank where you spraypaint a bulb black to make it pop.

I think with insulation and the inside of the tube painted black, you should get some noticeable airflow. the heat won't have anywhere to go except up through the tube

I repeated the experiment, with aluminum foil loosely wrapped around the light bulb to capture most of the heat (leaving airspace to let air flow upward within it).

The temperature of the surrounding aluminum tube went down dramatically, while the felt temperature of the air at the top of the tube went up.

The strength of the airflow also increased quite a bit in relative terms, but it was still stupendously pathetic. I still hard to carefully balance the piece of paper in order to detect an airflow effect on it. It simply isn't in the same ballpark as a silent undervolted fan.

have you played around with http://chuck-wright.com/calculators/stack_effect.html ?

your 4" pipe has very little area and 3' is fairly short. i think all the suggestions of insulating the pipe are a good idea, but i think your pipe is just sized wrong.

also a light bulb in a 4" pipe must take up almost all the area of the pipe... this restricts air flow. there is a rule of thumb for how much area you can block in a pipe without disrupting flow too much (people who make in-line industrial flow meters worry about this a lot). i forget the exact number, but i imagine google would know... i'm guessing its around 10%. so a larger diameter pipe would help in several ways.

just my .02. hope you get it to work for you, this is an interesting project.

"My plan was to design a negative pressure case around the light bulb chimney. "

Well that is a mistake, if the PC case is at a negative pressure it will be trying to suck air down the tube. You need a Positive pressure case.

Remember that in the hypothetical application of this stack, you'd have all the heat generated by the computer contributing to the air movement, assuming flow through the box and into the chimney isn't impeded at all. That'd be a fair bit more than 40 watts, I should think.

Unless I missed it.

Can you tell us what case u have, and the innards.

I would also suggest that you dont have the lightbulb at the top of the tube, you have it in a box roughly in the middle of the tube.

2 reasons, More airflow around the bulb due to the box. The heat from the bulb will go up the tube above it therefore drawing air from below.

Negative pressure case, I dont understand how thats going to work, surely you want to have all case fans blowing air INTO the case, this should also force air out of the chimney.

Andy

I believe he wants to have no fans at all, just this chimney thing sucking air out of the top of the case.

This reminds me of something.

My girlfriend's family has a cottage on the cape. The chimney is solid stone. If you try and make a fire in the dead of winter, you can plan on sucking ash, because all of the cold air initially begins to rush down the chimney to feed the fire and pushes into the house. After about ten minutes, however, the chiminey gets enough heat to create sufficient updraft, and it starts working properly.

In terms of thermal output, it's probably quite a bit more than a 100 watts...but it still seems applicable. That being said, I still think undervolted fans are clearly the way to go.

No fans at all, not one.

I wish Isaac the very best of luck,

Still, luck aside, air needs to get in, so air can get out.

I am no scientist, but I can see a major flaw in the plan, negative pressure, airflow is a must.

I have thought about an interesting case design, I have a few ideas floating around my head, they all involve fans, but I am power hungry more than I am silence hungry. Powerful components produce heat, heat is the enemy, the enemy must be destroyed, the weapon is fans, the fans become the enemy.

I have a balance that I like, fast performance, quiet, but not silent, if silence was the ultimate goal then VIA have the answer, utterly fanless systems, laptop drives are very easy to silence, but they are way to slow for me.

I cant use a system that goes like a snail carrying a large weight uphill through treacle.

Anyway Isaac, what is on the inside of that case of yours.

Andy

What he meant by a "negative pressure case" is that the chimney will provide the negative pressure instead of a fan. It's the same as having one exhaust fan: it provides negative pressure in the case so that air is drawn in elsewhere in the case.