Mini 12 volt ATX PSUs

[Tim B]

Good Evening Ladies and Gents,

I have recently bought two 12 volt ATX power supplies. These are now employed powering my 2 rackmount home PCs. These PCs are both 2.4GHz P4s. So far (as I am typing this) the PSU is working like a dream.

Now, for those of you who thinks that a man with a pair of custom 12Volt rackmount PCs on his desk can't possibly be trusted, I'm intending to add another two 12 Volt rackmount watercooled PCs over the next year, with a 12Volt 3U Server early in 2009. There is a point to this, though, in that the design of the new rackmount boxes is very close to the design of a completly sealed marine computer system.

Hopefully the power supplies and water-cooling will prove itself over the next 12 months, and the technology will be ready to be transferred to a yacht.

The product link is: http://www.itx-warehouse.co.uk/Product.aspx?ProductID=253

Cheers,

Tim B.

[TerryKing]

..snip.. There is a point to this, though, in that the design of the new rackmount boxes is very close to the design of a completly sealed marine computer system.

Take a look at: http://www.boatdesign.net/wiki/Onboard_Computers#Low-Power_Multi-purpose_Onboard_Computers
on our Wiki. Several of us here are interested in working on these areas.

I have a mini-iTX system board that runs with no fan.. which I intend to run in a waterproof enclosure of some kind.. My test equipment and tools are supposed to arrive Real Soon Now, and I can get back to work (Just moved to China..).

Are the power supplies you have rated for use in a vehicle 12V system? The issues of transient protection etc. are something I need to understand better.

I'd like to start a thread on 12V powered low-power general-purpose computers for shipboard use..

Let's figure some stuff out!

[Tim B]

The ones I'm using don't have automotive protection, but some others in the range do. I think it's a reasonable assumption to expect them to work ok.

The advantage of an ATX-based system is processing power. It's a trade-off between processing and electrical input power.

for more info see http://www.mini-box.com/s.nl/sc.8/category.13/.f

Cheers,

Tim B.

[Tim B]

Rackmount PC layout for anyone who's interested

[BillyDoc]

Tim and Terry,

I've been thinking about this as well, and one idea I have had is to simply flood a watertight box containing the computer mother board with silicone or mineral oil, and then re-circulate the oil to a hull mounted (or integral part of the hull) heat exchanger. I would put old-fashioned heat sinks on the high heat items like the CPUs, and make sure there was plenty of oil circulation over them. I'm talking about an aluminum hull, which of course makes this much easier. All that water on the other side of the hull should make a great heat-sink.

This isn't going to work with a DVD reader, but some company (Seagate?) recently announced that they will be bringing out a line of solid-state drives next year, and those could probably go in the oil as well. Perhaps the power supply could too after removing any fans.

It would sure be messy to work on, but I rarely have to actually open up my box for anything.

My main concerns are . . . what is that oil going to dissolve and can I get enough thermal transfer into the oil?

Any thoughts?

BillyDoc

[TerryKing]

BillyDoc, we should look more at what the Extreme Gaming guys are doing. I know some of them run processor heatsinks that are water cooled. I THINK I saw something about a system running in Freon etc.

I hope we don't need anything like that kind of compute power for typical onboard computers running NAV plus Monitoring and control, and Entertainment / DVD Movies at anchor. The newer VIA fanless systems can do all that I think.

I would like to look at using one of the 'Heat Pipe' coolers to move heat from 'inside the sealed box' to the outside if needed.. I'd LIKE to use an ABS plastic box with sealing cover so I can get at things easily. One of my kids designs stuff that goes on mountain-tops all over the world, in boxes like that. Look at:
http://www.nrgsystems.com/store/product_detail.php?cd=10&s=3147
to see the box. (Photo below, too)..
These boxes are rated like this:
- Weatherproof polycarbonate, meets the following specifications:NEMA type 4, 4X and 13, IEC: IP65

I need to look at the details of these specs, but I understand that these units pass the 'pail of salt water' test...

I have a VIA EN12000EG Mini-ITX System board that I hope to fire up next week...

Tim B, maybe we're ready to start a more general thread about onboard computers, and continue the power supply discussion you started there. What do you think??

[Tim B]

I had an interesting situation yesterday, when trying to get an adequate power distro setup. It seems that 3.5" hard drives are very sensitive to the voltage on the +12V rail. I found that the voltage drop along the supply lead was too great for the drives to spin-up. The solution to this is to use a 2.5" HDD which runs on +5V.

For casing I am a great advocate (particularly with water-cooling) of sealing the case permanently. That way, the motherboard is in a well-controlled environment. In the case of an on-board system I doubt you'd need to open the case any-way.

The DVD drive on the above PDF doesn't need to be there particularly for marine operations. A sealed SATA connection would be preferable with the DVD drive in a more accessible location.

There are a number of solid-state hard-drives available upto 64GB. That's more than enough for a Linux install + data. More storage can always be added via SATA/USB.

Tim B.

[TerryKing]

I had an interesting situation yesterday, when trying to get an adequate power distro setup. It seems that 3.5" hard drives are very sensitive to the voltage on the +12V rail. I found that the voltage drop along the supply lead was too great for the drives to spin-up. The solution to this is to use a 2.5" HDD which runs on +5V...snip..
The DVD drive on the above PDF doesn't need to be there particularly for marine operations. A sealed SATA connection would be preferable with the DVD drive in a more accessible location.

Tim, was the 12V being supplied from the ATX power supply? Maybe it's current capability was not enough.. I've wondered about operating stuff like that (and LCD display etc) directly from the +12V supply.

DVD: Totally agree.. this should be totally separate from the system case. SATA is an easier-smaller cable to waterproof, although I have seen ribbon cables like an ATA100 going thru RTV Silicone and I expect that would be fine... The DVD drive could be in a very protected area below decks, and even then is a separate simple NEMA4 box with a snap-lock cover. You only need to open it to put media in..

I'm about to work more on a list of what a "Multipurpose Onboard Computer" would have for capabilities. I'd like to start a broad discussion on that soon...
It's raining like heck in Shekou/Hong Kong for the first time in 2 weeks so I have a good reason to stay inside :P

[Tim B]

Yes, the supply (from on ATX PSU) goes to a distro box, then to the computer. although it would appear that my idea of 12 volts and that of the PSU developers differ slightly. I was expecting some voltage drop, and it is unlikely that it would be an issue for on-board use (as we're starting well above 12 volts).

Tim B.

[TerryKing]

http://www.boatdesign.net/forums/showthread.php?t=19458
now has a thread for a general discussion on onboard computers, powering them etc.

[SAE140]

My prejudice is against water-cooling - yet something else to go wrong on the boat.
Suggestion: mount the motherboard etc in an ally box with fins on both the inside and outside. That will take care of passive (convection) cooling for everthing except the CPU.

I still use Socket 7 motherboards on the boat which I underclock to get the heat down. The remainder of the heat is conducted to the case via a conduction cooler: simply a brass plate epoxied to the CPU to which has been soldered a heavy-duty copper braid (ex automotive grounding strap). A brass plate soldered to the other end is bolted to the case. Keep life simple.
Colin

[TerryKing]

(..snip..)The remainder of the heat is conducted to the case via a conduction cooler: simply a brass plate epoxied to the CPU to which has been soldered a heavy-duty copper braid (ex automotive grounding strap). A brass plate soldered to the other end is bolted to the case. Keep life simple.Colin
Colin, please check in as we work on this heat/transfer issue over in the other thread mentioned below. There is a similar idea in which a 'heat pipe' goes from Processor thru a cabinet wall to outside. A passive vapor-liquid heat pipe can move much more heat energy than an equivalent copper or silver bar.

I agree with the Keep It Simple.. no powered parts approach!

((Added a photo below.. This is for getting 100 watts out of a 3 Ghz multi-processor. Something like this should work passively for 15 or 20 watts inside a sealed box ))

See: http://en.wikipedia.org/wiki/Heat_pipe

[SAE140]

Hi Terry
thanks for the reply.

I went over to the Wiki thingy - and blow me down - I've got a CPU cooler on the shelf which looks exactly like the one in the top photo.
Two copper pipes with carefuly sealed ends. I thought it was just some kind of wacky construction technique !
'Master Cooler' is the brand name. Up for grabs if anybody want to play with it.
Cheers
Colin

Almost forgot ...

If you guys are focussing on engine management/ battery state and tankage indicators and so forth, then checkout: http://www.vermontficks.org

There's a Seatalk interface and GPS repeater and much more there.
http://www.vermontficks.org/gpsd.htm is a good place to start.

The guy uses PICs: nano-watt technology - no need for cooling <G>.

[TerryKing]

If you guys are focusing on engine management/ battery state and tankage indicators and so forth, then checkout: http://www.vermontficks.org
VERY Small World! Jon Fick is an old friend of mine from IBM (I'm from Vermont, living in China the next 2 or 3 years).. I'll talk with him about this stuff!

I have developed things with PIC microcomputers and I'm working on an NMEA0183 Engine Monitor with NMEA Multiplexor for several channels.

Small World! Thanks...

[donncha]

In response to the comment about oil cooling:

You wouldn't need any heat sinks at all as there is enough heat transfer from the oil directly to the mother board and cpu. The only thing you have to worry about is any moving parts.

Oil has been used several times else where. I saw a guy once who simply dumped all his components barring the hdd and the cd drive into a fish tank filled with oil and ran the computer form there.

I myself was considering using an ammunition box filled with motor oil(as it's non conductive and has good thermal properties) to run a fully contained/waterproff pc for on-board usage. Or you could simply just use ordinary passive cooling (i think oil is more fun).

Look at http://www.mini-itx.com/projects/ammobox/

[BillyDoc]

Thanks donncha,

I thought the oil idea might work! Cooling AND salt protection for the price of one. The only hitch I see is the oil dissolving something. Like the "rubber" inserts at the ends of some capacitors, for example. On the other hand, silicone oil should be OK. You can get the stuff by the gallon at some cosmetics supply places, I've used it for ultra high pressure transmission (75,000 psi) where petroleum oils turn into solids. I don't have any thermal data for it though. A high heat content would be nice. But then, just about any liquid should have a higher heat content than air.

I'm building an aluminum boat and I've been thinking of extending a "fat" frame at a convenient location so I can bolt a second aluminum box containing my motherboard, etc., to it with some thermal compound. Inside the computer box I could machine some stubby fins oriented vertically, and on the motherboard I could use conventional heat sinks also oriented vertically. Hopefully the back of the box where it attaches to the frame would be "cold" relative to the oil, thus causing it to convectively fall downward. The "hot" components would make the oil rise, and if I design it all so there is an easy path for this circulation to follow I think I could have a nice passive cooling system, (almost) direct to the hull. Those people with non-thermally conductive hulls (wood, glass) could do something similar by making a heat exchanger for the back of the computer box and placing the entire box and heat exchanger in the bilge below the water line. Two lines rising to a sea inlet should result in convective circulation of the water within this heat exchanger, and again you have an entirely passive system. The whole heat exchanger could be made from copper and thermally coupled to the aluminum box (or the box made from copper) via an electrically non-conductive material like is used for mounting transistors on heat sinks. Copper would have the advantage of resisting attack by all those biologicals . . . but you would sure need to make sure you weren't setting up a nasty galvanic corrosion problem!

BillyDoc

BillyDoc

[TerryKing]

For the non-extreme methods, here's a pointer to a product that apparently works well in Marine and other hostile environments like chemical plants and irrigation systems:
http://www.actonrust.com/electrical.htm

I'm thinking of trying this on some samples, and in a fanless 15 watt Mini-ITX for onboard use.

Although the ABS plastic NEMA4 type boxes are great for marine environment, I'm thinking about one of those aluminum gasketed mini-briefcase boxes with snap fasteners. The thin aluminum should do a good job of inside air to outside air or inside air to heatsink/aluminum hull heat transfer.

With only 15 to 25 watts total, this isn't a big problem. If someone wants to run a 150 watt 3 Ghz multi-core system and a big hard drive in a box, that's another story...

[SAE140]

Thanks donncha,

I thought the oil idea might work! Cooling AND salt protection for the price of one. The only hitch I see is the oil dissolving something.
BillyDoc

With flooded oil-cooling of electronics, you may also need to take into account the different dielectric constants involved; air being 1 and transformer oil typically around 2.2 Especially on high-speed FSB equipment. I guess the only way to know for sure is to test the technique ...

You'd also need to figure out a reliable means of sealing the case, especially with the need for multiple leads to pass through it. You'd also need to fabricate some kind of expansion facility (as on fluid-filled compasses), and a drain plug of course - just in case you ever need to gain access to the innards.

Many oils will harden pvc over a period of time, but this shouldn't present a problem if flexibility isn't required.

'best
Colin

[BillyDoc]

With flooded oil-cooling of electronics, you may also need to take into account the different dielectric constants involved; air being 1 and transformer oil typically around 2.2

Yikes! I didn't think of the dielectric constant at all, and you are so right. These circuit boards with multi gigahertz signals blasting all over the place look like pure magic to me under the best of circumstances. And I can't imagine designing one without the use of a very sophisticated design program . . . that would take into consideration the dielectric constant of the surround to simply be able to send those signals from one point to another without being swamped with reflections and who knows what else.

Thinking about it I can't see how signal propagation would NOT be strongly effected, unless the solder mask layer has a similar effect and this has already been factored in. This is a nasty problem. I guess I'll just have to do some experiments.

Thanks for pointing this out SAE140, this is a very important point to consider.

BillyDoc

[donncha]

also have a look at
http://www.tomshardware.com/2006/01/09/strip_out_the_fans/

They found that the cpu had to be sealed as the difference in the dielectric constant affected it.

"We have the following explanation for this phenomenon: On the motherboard in the area of the CPU base, the oil is responsible for increasing the capacitive resistance between the individual wiring. In short, the oil acts as a dielectric material. Since very high frequencies occur on the motherboard, the capacitive resistance goes down. Accordingly, this then influences (or tampers with) the digital signals, particularly in the area of the CPU base. After all, 939 pins are located there in a very tight space."

As I thought only the cpu is affected be the change in the dielectric constant.

You would still want to be careful.

I have heard of successful trials done without this step in sealing the cpu. But it was with older pc's, perhaps the less amount of pins or lack of all together solve this proplem. If you are using a mini-itx motherboard or something with a built in cpu this might not be as much of a problem.

Also watch out for the corrisive effect of oil on plastics and rubber.

[BillyDoc]

Hi Donncha,

Interesting link! One thing though, I've tried to seal against oil with RTV silicone before, and eventually the oil penetrates it. It can take a few months, but . . . Epoxy works well, though. And another type of material which might work and is somewhat flexible are some of the "casting" urethanes. I haven't tried these (or looked at the manufacturer's site) for oil resistance, though.

BillyDoc

[mydauphin]

There is an oil used in X-ray equipment that is great. It has consistancy of vegetable oil but it does not react with any materials. It has highest isolation properties. It is not cheap. Last time I bought it 15 years ago it was $200 for 5 gallons, Exxon made it. I dont remember name - but it exist and it works really good.

[mydauphin]

The whole heat exchanger could be made from copper and thermally coupled to the aluminum box (or the box made from copper) via an electrically non-conductive material like is used for mounting transistors on heat sinks. Copper would have the advantage of resisting attack by all those biologicals . . . but you would sure need to make sure you weren't setting up a nasty galvanic corrosion problem!



If you have an aluminum boat, you can not have any copper on board. Not even pennys. Eventually copper will ooze green stuff, copper sulfate and it will eat your boat alive. Aluminum is not a bad heat conductor. Also there is no need for salt water cooling. A small tank of fresh water made of metal, perhaps aluminum will dissipate heat from inside of computer box to outside that is all that is needed, unless someone wants to fit Cray in a ship.