BUILD Fanless bookshelf NAS

[Dice]

As a rough figure and using the formula from the Wikipedia article on the stack effect, it should be possible to get 2 - 4 liters / second airflow through the case via natural convection already at a dT of 20 Kelvin.

dT of 20K is huuuge... Unless the server is placed in a winter-environment & outdoors.
What components in the presumed build would generate a dT of 20K in relation to ambient temperature and still work within recommended temperature range? ....not your laptop hdd's for sure. They'll be on the "sauna side of life".
The only thing going up through that chimney is the life expectancy of those laptop hdd's...
SSD's tho... ;)

I might sound crazy, but I'm drawn to try it using cardboard and light bulbs. Stay tuned ;-)

The main problem in doing this experiment and trying to transfer conclusions to the case of useless laptop HDD's is that light bulbs are designed to work without additional cooling. Hard drives are not. They tend to get excessively overheated just sitting on a desk while doing intensive io, such as the regular scrubs.


I must admit I'm a bit too inspired by this discussion for my own best :P
SSD's, planning for all out passive setup, a highly efficient PSU, a chimney setup, a backup fan that ideally would be programmed to only run during intensive io.. It would've been such a sweet build.
You'd probably get away with a pico psu if you choose the right components. Then your powersupply is both cheap and weak, but efficient, small and without moving parts. (I use a pico psu for a i3-6100 + 1x SSD, +1x 120mm fan for router)

 

[lorchi]

Not at all, just my attempt at humor.

Oh. Err... sorry. Well now, that's also funny ;) ;)

 

[lorchi]

dT of 20K is huuuge... Unless the server is placed in a winter-environment & outdoors.
What components in the presumed build would generate a dT of 20K in relation to ambient temperature and still work within recommended temperature range? ....not your laptop hdd's for sure. They'll be on the "sauna side of life".
The only thing going up through that chimney is the life expectancy of those laptop hdd's...
SSD's tho... ;)

You're right with your concern, but you didn't get my idea of separating the mobo and the drives.
I want to try the following:

Creating a chimney chamber for the mobo & CPU & PSU only, about 30 cm tall. The heat source (ie electronics and heatsink) have to be placed at the bottom, because only the column of heated thinner air drives the flow in the chimney. Using the formula from Wikipedia about induced flow in a chimney, a 5 x 10 cm2 of unobstructed area gains a flow of 2 liters already at a hypothetical dT of 20 K. This is supposed to mean that creating a chimney effect should indeed be possible. The mobo & CPU should ideally be placed flat at the bottom.

Next iteration is to make the intake slightly wider and the chimney slightly thinner at the top, following the acceleration of the air. This way there will be lower pressure at the top. This is what I first have to try using cardboard and watching the flow using some smoke source.

Then its time to think about the chamber for the drives. Both chambers have to be thermally insulated from each other.

The drives should as well be placed in the lower half of the case. The drive chamber has its own air intake, and also follows a chimney design. But: its exhaust enters the exhaust chimney of mobo & CPU & PSU, before the final chimney top. Even with much lower dT, the air from the drives is supposed to get sucked into the main exhaust (see the Bernoulli principle).

This way the drives, themselves equipped with a small heatsink with fins in the drive chamber's airflow, should stay with a dT of less than 5 Kelvin, even at load.

Edit:

must admit I'm a bit too inspired by this discussion for my own best :p

So I strongly invite you to letting us make it happen :):D:)

 

[Bidule0hm]

should stay with a dT of less than 5 Kelvin

That's pretty much impossible without a fan (or a lot of room for a (very (very (very))) big heatsink on each drive).

 

[lorchi]

That's pretty much impossible without a fan [...]

Not at all, if you control the parameters. Example:

Try this heatsink: 0.9K/W, 100 x 160 x 40mm
It is big enough to bolt 2 notebook harddrives onto it. You just need the skills and tools to drill 2x 4 screw holes at the right positions into it.
Take 2 of Western Digital Red WD10JFCX. They have 1.4 Watts under load (which is admittedly very little in comparison to others).
With a slight breeze or convection along the heatsink, you get:
dT = 0.9 K/W * 2 * 1.4 W = 2.52 K

Just know that a slight airflow must be there. And use seriously designed parts. This brick of metal for instance is a joke as a heatsink, pure marketing crap. Additionally people cram this thing into a bay without airflow. Haha.

Ps: This heatsink is slimmer and cheaper, with half of the effectivity (1.8 K/W). In the above example you get a dT of about 5 Kelvin.

 

[Bidule0hm]

You forgot some things :

The Rth of dissipators are usually measured at 80 °C and it'll be higher at 30 °C.

Then you have the problem of the drive/heatsink interface: even with thermal paste there's a Rth here too.

Then if you put the heatsink on the cover side of the drive the heat will have to go through the thin walls of the cover to get to the heatsink so one more Rth. If you put it on the PCB side then you'll short the PCB so actually that's not an option at all. And usually the cover isn't flat and there's the label(s) to remove so you'll void the warranty.

And I did say "or a lot of room for a (very (very (very))) big heatsink on each drive" so yeah, if you have room for a heatsink that takes the space of 15-20 drives just to cool 2 drives and you're certain the air will never go higher than 20-25 °C in the room then sure, you can use this solution ;)

 

[lorchi]

Meanwhile I wanted to know it. Here are some pictures of my first chimney test. :D

I built a 20 x 20 x 31 cm size cardboard box with the inner air tunnel you can see on the pictures.
The first two pictures show the setup. The NAS-lamp ;-) On the left there is the rear of the case, on the right is the front. It is closed with plastic wrap (not on the first 2 pictures).
It has two intakes, one 5 x 18 cm in the front, and one 3 x 18 cm in the back (above the ATX panel). The exhaust also has a clear path of 5 x 18 cm (the opening is somewhat 6 x 18 cm).
The exhaust is in the front, so that the case can go into a bookshelf. It is diagonal, so that the warm air accelerates up to the top (a horizontal outlet would just act as an obstruction).

I tried it first with a 25 W bulb and aluminum foil as heat spreader. The chimney effect was there, but to see it, I had to stop breathing in its proximity. Later I tried 40 W, and it was more stable. The rear intake showed only a fraction of the effect of the front one. A 4-sided intake down to the mobo level would be best for the motherboard when it is placed flat as I have done it.

As a conclusion I would say that it works of course better with higher loads. This means, for low loads, like an Avoton, it is not necessary. A good passive heatsink is enough, but it hardly creates airflow over the mobo. For higher wattages I would actually use it. I consider it for my next desktop PC, because it convinced me. But spending it more height would definitely increase the effect in all cases. The airflow is faint - I could not feel it with my hand, but it's there.
With this experience I would build a desktop (e.g. i7) completely vertical with more height, just one long square tube, at whose intake the motherboard would be mounted vertically to benefit better from the flow.
My own experiment worked well, but a huge passive heatsink like the NoFan CR-95C conflicts with my air path design. A higher case would solve this problem.

This means that for a small NAS with an Avoton and 4 notebook disks or SSDs I would make a DIY tiny case, with the disks at the bottom and the mobo above, and enforce airflow with a fan, just in order to make it small.
The chimney adds a lot of volume. The concept works. But it's a little effort to build it.
I thought about creating a separate chamber for the drives as I had explained above. But for minimizing the effort, I might simply place them at the walls of the air intake (giving the case 5 cm extra in depth or so).

 

[maglin]

Ironically this cooling solution not only costs more to run than a fan or two but for it to work the components have to get hot for it to work. This defeats the purpose of cooling. But I do like the ingenuity of it.


Sent from my iPhone using Tapatalk

 

[lorchi]

Ironically this cooling solution not only costs more to run than a fan or two but for it to work the components have to get hot for it to work. This defeats the purpose of cooling.

Yes and no. I will try to rephrase and summarize, also for myself:

• The chimney is designed to support a heat sink at load. In other words:
• It is designed to keep a CPU within a certain range of delta T, say 20 - 30 Degrees.

Characteristics:

• Smaller heat sink and narrower chimney: delta T goes up, air moves faster.
• Larger heat sink and wider chimney: delta T goes down, air moves slower.
• Taller / higher chimney: more airflow is created with increasing delta T -> delta T rises slowly
• Low chimney: less pressure difference, delta T rises steeper

So the comparison with a fan needs to look at the two sides of the curve:

• At low loads, a chimney cannot bring delta T towards zero. Only a fan can do that. Without a fan, delta T would stay where it is.
• At high loads, a chimney replaces a fan.

Suitability:

• It's not suited to cool a system with many 3.5" HDDs, especially in the tropics.
  
• It's well suited to cool a powerful server with an SSD, especially in temperate climate.

Costs:
The cost is me sitting there and cutting cardboard ;-) No but seriously.
Compared with people buying tower cases, there is no extra cost both in terms of volume and money. In practice, there is a cost of development because it's not available on the market. The problem: The market wants to sell high-tech. So there's a need to cut cardboard or other suitable material.

Conclusion:

• Chimney = better with tower case size, better with powerful device (server / desktop)
  
• Enforced airflow = allows for smaller sizes (bookshelf), allows for low delta T (cooling of HDDs)
• A bookshelf chimney works for small, medium-load devices without many HDDs, so it's not flexible enough for a conventional NAS.

 

[lorchi]

You forgot some things :

The Rth of dissipators are usually measured at 80 °C and it'll be higher at 30 °C.

Thanks for your feedback.
I actually didn't know that it is measured at 80°C, that's good to know. Rth depends on the induced convection I suppose. I assumed some convection to be there, as described above. So Rth might be lower than what you might mean, but it might be higher than what I counted with.

Then if you put the heatsink on the cover side of the drive the heat will have to go through the thin walls of the cover to get to the heatsink so one more Rth. If you put it on the PCB side then you'll short the PCB so actually that's not an option at all.

Yes I meant putting it on the PCB side - to screw both metal pieces together. The PCB can be protected with insulating foil. Or copper washers can be used to provide some clearance while still creating a thermal connection.

So the total thermal resistance is apparent, but should be manageable ;-)

 

[lorchi]

Anyway, as for my bookshelf thingy I have ruled out the fanless design. But it was fun though ;-)

 

[Bidule0hm]

If you count on 4 washers and 4 screws to transmit the heat then you'll have a problem because the Rth will skyrocket.

Yep, in the end the concept is smart and theoretically possible but in reality a fan, even a very slow and quiet one, will be far better than natural convection ;)

 

[Dice]

......or just get SSD's ;)

 

[lorchi]

If you count on 4 washers and 4 screws to transmit the heat then you'll have a problem because the Rth will skyrocket.

There will be a simple way to mount an HDD onto a heat sink. Take any.

in reality a fan, even a very slow and quiet one, will be far better than natural convection ;)

Yes I have to agree (mostly). At least for a small case that appears without doubt. :(

 

[lorchi]

......or just get SSD's ;)

I would love to. You have some left over? ;)

 

[Mirfster]

I built a 20 x 20 x 31 cm size cardboard box with the inner air tunnel you can see on the pictures.

<--- Cardboard Rocks... ;)

 

[joeschmuck]

This was a pretty interesting thread to read but I find the idea of making a truly fanless NAS very possible. You could do it if you could make the entire case a heatsink and mount every component to it so the heat properly dissipates. This would require some metal fabricating but it could be done and I'd do it in sections that could be screws together for ease of use. The motherboard heatsink would be completely custom of course (you could not use any add-on cards and would be stuck with just enough room to mate/demate SATA cables) and the most difficult piece to manufacture but if you have the skills you could do it. A separate heatsink plate would be for the hard drives mount or better yet, just SSDs and nothing special needs to be built but I'd still make it a heatsink just to draw more heat out of the case. You could anodize the case a cool color (blue feels cool) and of course the outside of the case would look like a lot of cooling fins but it can look nice and you could put a design into it.

And I'd use one of those Pico power supplies already mentioned.

This could be a cheap build if you have a CNC machine and a huge block of aluminum laying around. For accurate measurements of the motherboard you would need one of those devices with an articulating arm so you can create a map for the CNC pattern, or you could use a micrometer and a good ruler, and make a sample out of foam or plastic.

Now I want to do this project, JUST... NEED... MONEY...

But getting back to reality, while it may be possible to create a truly fanless design, it is much easier to go with a fan design. Shoot for the larger 120mm+ fan sizes and you will never hear one operating at 7VDC but it will throw a lot of air. You are likely to hear only the hard drives clicking away, especially if you opt for 2.5" laptop hard drives because they tend to park the heads very often. I don't know if there is a way to override that feature but I think it will be what you hear and irritate you. It would irritate me. My WD Reds are very quiet, even after 3 years on continuous operation.

As for a case design, if you wanted to make your own, I like the flat box idea and mounting the hard drives in the base but I'd like to see the air flow be pulled across the hard drives first, then across the top of the motherboard so you would have to factor in some wind vanes in order to direct the air flow properly before it exits out the top. It would be easier to push the air in the top and then it would hit all the motherboard components, and then flow over the hard drives but that will only warm up the drives, but likely not so much.

 

[maglin]

This could be a cheap build if you have a CNC machine and a huge block of aluminum laying around.

LoL. Thank you for making me laugh. I do happen to have a CNC mill laying around. But that block of aluminum will cost around $80. If you want to machine say 1" fins into it then more like $180. Countless hours (maybe 10) spent modeling the motherboard in CAD. And another few hours doing CAM. It would be much cheaper to cut up old heatsinks to fit all the various heat producing parts on the MB.

 

[joeschmuck]

It would be much cheaper to cut up old heatsinks to fit all the various heat producing parts on the MB.

But where is the fun in that, and then it's not much of a heatsink "case".

 

[maglin]

Ahhh. You are saying to make an entire case with a CNC. :)