Follow Slashdot stories on Twitter


Forgot your password?

HP Marries Inkjet and Robotic Technology to Cool Chips 175

An anonymous reader writes "Extremetech has an article about how H P has decided to use the spraying tech developed for inkjet printers to cool chips -- and has made a robot that'll wander around data centers, detecting too-hot chips and hosing them down." The article notes that the robot needed about 1 hour of training on the room before it would go about the business of chip cooling. The real advance would be if it achieved sentience and went crazy and became a graffiti taggin' super robot, but I digress.
This discussion has been archived. No new comments can be posted.

HP Marries Inkjet and Robotic Technology to Cool Chips

Comments Filter:
  • by SpanishInquisition ( 127269 ) on Thursday August 08, 2002 @03:46PM (#4035105) Homepage Journal
    But the memory chips for it will be insanly outpriced.
  • "Thermosyphons funnel fluid from the outside condenser to the evaporator on top of the microprocessor. But HP discarded thermosyphons, and heat pipes, because of "pulse boiling", a rapid cycle of heating and cooling that can damage the microprocessor, according to an internal study conducted in May of 2000. HP developed its evaporative cooling to eliminate pulse boiling, Patel said."

    Since when does ExtremeTech use a story generator?
  • spray down really hot people? That sounds scary. Yeah. Besides... the slashdot audience has nothing to worry about... he'll be starting with marathons - places where obviously you wouldn't find your typical slashdot reader.
  • by armyofone ( 594988 ) <> on Thursday August 08, 2002 @03:48PM (#4035127)
    Robot - $1,500.00
    Coolant Cartridges - $300.00 each
  • Actually... (Score:5, Funny)

    by Bruce Perens ( 3872 ) <> on Thursday August 08, 2002 @03:49PM (#4035134) Homepage Journal
    Actually, the robot searches for DMCA-covered data in data centers and sprays the chips with sulphuric acid.

    Only kidding.


  • Warchalking (Score:5, Funny)

    by krugdm ( 322700 ) < minus poet> on Thursday August 08, 2002 @03:50PM (#4035138) Homepage Journal

    The real advance would be if it achieved sentience and went crazy and became a graffiti taggin' super robot

    How about letting it wander around town looking for wireless hookups, and then marking where they are, warchalking style.

    Of course, I just had a vision of this thing dressed up like a dog, going around and "marking" it's territory...

    • It is my job. To freeze you. Fish, and plankton, and sea greens, and protein from the sea. It's all here. Ready! Fresh as harvest day! The old food left and the new food came.

      --Box, Logan's Run
  • .. that the robot plays the Nelly song "It's gettin hot in here, go on and take off all your clothes" as he's going around cooling down processors.
  • by NWT ( 540003 )
    "What we're proposing to do is take the cartridge right over the silicon chip..."

    Sure, the robot removes the heatsink and then cools the chip?! uhm, i'm sorry but the time between the two actions is enough for my athlon to die ...
    • Clearly, you didn't read the article.

      The robot doesn't spray anything. The ink jet based coolant is in the case against the chip.

      The robot is running around finding hot spots so the AC can be adjusted.

      They were only mentioned in the same article because they both revolve around HP's Datacenter Cooling technology.
  • The article doesn't make much mention of the fact most computers in datacenters are sealed racks packed fairly close together. We run a couple dozen cabinets in the office and I honestly can't picture any benefit (aside from running down lazy techs) of a clunky robot trying to move it's way through our facility.

    Mist spray is a good idea though, if they were priced well it might make a nice addition to add into existing server cases.

    • by Afrosheen ( 42464 )
      If you're going to get this elaborate, why not just build fancy liquid chip cooling systems that overclockers use all the time into the cases? You could have a whole rack of servers running off of one coolant box. Your datacenter would go from rack after rack of servers to 1 rack of servers, 1 rackspace taken by the cooler, etc. It'd be cheaper in the long run also, judging by HP's pricing scheme for their existing consumables like inkjet ink cartridges.

      Oh and there's already a graffitti bot. He's over here!
      • Shit my html skillz are failing today.

        Grafitti robot is here
      • Might be easier to build a room sized refridgeration unit and move the racks in there. Even mom and pop grocery stores have them so they can't cost that much. Condenser maintenance is probably less expensive than maintaining a robot.
  • do they keep the robot from being confused by the "hot-spots" created by sysadmins in the data center viewing pr0n?

  • by mustermark ( 104271 ) on Thursday August 08, 2002 @03:51PM (#4035154)
    I recently went on a tour of the HP research labs here in Palo Alto, CA, and I made a comment to the lady conducting the tour concerning this. In a nutshell, it was the following.

    This technology would require liquid cartridges to run the cooling mechanism, which would mean that every computer would require us to buy these from HP, much like printer cartridges. The lady had a rude comment about how HP was really in the business about selling consumables (like printer cartridges and soon CPU cooler cartridges) and that this was somehow a wonderful idea.

    Taking advantage of a liquid-gas phase transition to cool is a great idea, but to require a proprietary chemical to do it is lame. I'm sure there are ways to do this with water, right?

    Makes you wonder whether this would be better than the cheap plastic cooling fans that break down and have to be replaced all the time now.
    • We may have to buy CPU cooling cartridges, but perhaps we could start manufacturing recycled cooling cartridges for less? Of course, once HP starts bitching that we're stealing their market share, they will tell their customers that recycled cartridges could damage their chips, or they will put "killer chips" that will destroy your robot when the cartridges become empty. That then brings up the possibility of recycled robots...
    • but to require a proprietary chemical to do it is lame.I'm sure there are ways to do this with water, right?

      Such chemical are not very complicated compound and can be OEM very easily just like people OEM the inj cartridges, provided that there's a economic drive.

      However, you probably wouldn't want to do it with water, which is too corrosive. :)
    • The problem with a liquid-gas phase transition is that the transition has to happen at just the right temperature. Water's gas phase transition is too high to be useful in cooling chips... most chips are rated right around 100C for max operating temp, active cooling that only works after the max operating temp is reached is in general a bad idea.

      There are really only three ways to do phase transition cooling that make sense.

      1. You can use water but lower the pressure in the gas portion of the system. This has the disadvantage of requiring low pressure fittings and whatnot, and being somewhat difficult to manufacture. Also, you'd probably need a compressor to maintain a pressure imbalance.

      2. You can use Freon or whatever the replacement chemical is. This has the opposite problem of water; in order for the Freon to stay liquid, you have to have a high-pressure system. Compressor is mandatory. This would be an air conditioner with the cold coils directly on the die.

      3. You can custom make a liquid coolant with a phase transition in a sane place. Like 50C or so. Rubbing Alcohol would work, but you have to be careful. Alcohol eats plastic and corrodes metals. Rubber too. So everything including your pump has to be made to withstand constant alcohol vapor. You could make a different liquid thats not so corrosive. In either case, your compressor/pump would be MUCH smaller than in the other cases.
    • Well, then you have to worry about condensation, etc. of the evaporated water. If you use freon or sorry tetrafluoroethane, the resulting vapor and condensate is inert. With water cooling, you might as well be using a closed loop system, as it it would be cheaper than building a system to withstand water.

      So, while you're at it, you might as well use a real refrigerant and actually chill the chips. Basically, this is a cool project, but seems pretty pointless. I can think of some other jobs it could do.
  • Kryton (Score:2, Funny)

    by T-Kir ( 597145 )

    This reminds me of the Red Dwarf series, but more precisely the books. Where Kryton the maintainence mechanoid was responsible for crashing the ship he was on because he ran out of things to clean, so he decided to give the (dusty and dirty) live computer circuit boards a good going over with soapy water and a sponge.

    Let's hope they don't have any disasters.

    • Heh - there was a guy here at work that did a similar thing. Bonehead had some slack time on his hands so he decided to clean his computer. Got it nice and shiny - almost like new. But left it on while cleaning. Deleted all kinds of important data during the keyboard-scrubbing phase.

      Needless to say, he no longer works here. Still makes me chuckle when I think about it.
  • Someday we have to realize that cooling the chips is not the answer: making cooler chips is.

    It's just simple conservation of energy. The heat of a hot chip contributes to the overall global warming because the chip required enough energy to make it very hot, so dissipating the heat into the environment just exacerbates the global warming trend.

    The answer is to make a chip that doesn't require as much energy to run and therefore doesn't dissipate as much energy into the environment.
    • I think you're missing something. Cooling. Cooling would cool the hot air, not just dissapate it. This isn't just a fan. I do agree that chips should 'somehow' be made to run cooler, but since I don't make chips I'm not really in any position to be critical of anyone's designs.
      • I think you're missing something.

        Convervation of kinetic energy requires that for something to lose kinetic energy (excited and fast moving ceramic particles) something else has to pick up the energy (a heat sink, the air, a finely sprayed mist). The only thing the mist is doing is efficiently exchanging kinetic energy levels with the chip.
    • this is true, but the problem is that to make chip run cooler you need to lower the voltage. if you lower the voltage you don't have the electrons running from place to place fast enough and the computers slow down. there's a very interesting book "the feynman lectures on computing" or something like that which is a series of lectures that richard feynman gave in the 70s i believe. he talked about this issues way before it became important. one of his proposals was to have massivly parallel chips running at very low voltages. at the voltages he was talking about it was statistically possible for the electrons to wander the wrong way so you need gates that are reversible. for example, the and gate is not reversible, you have two inputs and one output. if i tell you the output of an and gate, it's impossible for you to tell me the input. but there's another type of and gate called the CAND (controlled and gate) that has three inputs and three outputs, this one is reversible, so you can have electrons run backwards. anyway, i've rambled enough, so to get back to your post. we can't make chips cooler without lowering the voltage, but that'll slow down the electrons and everything else will slow down. there is a way around it, and i'm not sure why it hasn't been done yet.
      • Lowering voltage does NOT change the speed at which electrons flow. Just lowers potential, so less current flows, but it still flows at exactly the same speed.

        I R

        clockless cpu's. We're about to hit a wall anyways somewhere above 2.3 Ghz in which more of the CPU cycles will be spent tracking cpu cyles than actually getting work done. I'd say a basic design change is due. I've read several articles on /. over the past 2 years regarding the technology, but I've yet to see anything I could ID as a device using the tech ni the consumer market.
        • You've not been reading enough of the articles then. See eg this one. [] The Myna pager [] is definitely a 'consumer market' device. I'm sure I read Ivan Sutherland say somewhere that there are asynchronous islands on the latest SPARC chips, but I can't find a reference.

          However you're right, takeup is minimal, see eg this talk [] for a description of the state of play.

          Another approach that may have gone the way of the dinosaur (havent seen it make headlines on /.) is reversible computing [] - the notion that by not discarding information [] within a chip you can run chips cooler (though apparently we won't reach the level where this much thermal loss becomes significant for another few years). E.g. a nand gate loses one bit of information, resulting in an energy dissipation of at least ln(2)kT, about 3x10^-21 joules. These links are 4 years old; I have no idea if reversible computing is now mainstream?
    • A microprocessor cannot make any real contribution to global warming. Suppose that a chip dissipated 50 watts of heat (I have no idea what they actually produce/consume, but this seems about reasonable). For comparison, the sun delivers about 1370 W/m^2. Over the projected area of the entire planet (about 40 million square meters). In other words, 1 processor at 50 watts adds 0.00000009% to the earth's energy balance. Pretty tiny, even when you consider large numbers of chips. Global warming is more the result of greenhouse gasses that prevent this heat from being dissipated into space, as (essentially all of) it naturally is. The problem with earth's environment, much like the problem with computer chips, isn't in how much heat you produce/receive, it's in how efficiently you get rid of that heat.

      BTW -- I didn't check my math too carefully, so be warned...
  • Y'know, I just can't help but wonder what they call this robot. Hmmm...nah, couldn't be.
  • I tried this myself just a few minutes ago. Since I don't have a robot or injet spraying thing, I just used a water hose and set it on mist. I dont think it worked to well. Sure my computers nice and cool, but it won't boot. How could this be caused by the cooling system?
  • by yeoua ( 86835 ) on Thursday August 08, 2002 @03:53PM (#4035172)
    Well the obvious use for this now is to put this tech into Sony's Aibo.

    Now you got a smart dog wandering around sniffing for heat, and pissing... er I mean spraying on it to put it out.
  • Too late (Score:1, Troll)

    by sllort ( 442574 )
    The Crazy Graffiti Writing Robot [] has already been built and deployed to independent freedom agents (Trolls) across the country for years.

    In fact, we've moved on to phase II []

    • That Graffiti Writing Robot [] won't be just writing chic indymedia slogans about working wages and legalized pot. If this guy [] has his way, it will be spewing herbalife [] ads and multilevel marketing.

      In spite of the fact that slashdot's editors got it completely wrong [] and this robot doesn't spray anything but just sends signals to the air conditioning system, this development is just one in a flood of recent advances in the robot field that will ultimately create the real-world equivalent of pop-up ads.

    • That's cool!!! I mean, if there's only one way to get people to respect to your POV, it's vandalism!!!!

    • All those projects on that site are really neat. I especially like the "iSee" project. Though, in the wrong hands, I can imagine it being a problem for authorities (it plots survelance cameras).

      But the Street Writer and Grafitti writer are very cool uses of automation. It kinda reminds me of those silly clocks with the swinging arm and 8 or so LEDS that turn on just when you need to. Of also reminds of the Dot Matrix printers. Very cool reading...thanks.

  • by IvyMike ( 178408 ) on Thursday August 08, 2002 @03:53PM (#4035175)

    Well, not only graffiti taggin' super robots, but also graffiti taggin' super cargo van (think "A-Team") and graffiti taggin' super remote controlled cars can be found at Applied Autonomy []. Very cool.

  • Wait... (Score:2, Funny)

    by wo1verin3 ( 473094 )
    .... if I can eventually use this to cool my systems... how do I heat my house?
  • why no one has come up with a scheme to use this "waste heat" for something. Think of all the thermal energy your average datacenter throws out daily...
  • Stop intruders... HP has trained a robot to run around spraying people in the eyes with inkjet cartidges, who are not wearing their access badges.
    • Stop intruders... HP has trained a robot to run around spraying people in the eyes with inkjet cartidges, who are not wearing their access badges.

      It will then bill the intruder $100 for a replacement cartridge, $20 for the high quality photo paper used to print the bill, and $500 for the labor costs involved.
  • How much will the refill cartridges cost for the robot?

  • by SCHecklerX ( 229973 ) <> on Thursday August 08, 2002 @03:57PM (#4035208) Homepage
    So...are there cooling robots that chase the cooling robots to cool them down when they get too hot? And if so...are there more cooling robots to cool them down? Could be interesting :)
  • At the chip level, the problem isn't the overall heat produced by a chip, but the heat density. Although Intel CTO Pat Gelsinger first noted publicly that upcoming chips could produce more heat per square millimeter than a nuclear reactor, HP engineers said they first began thinking about the problem in the mid-1990s.

    Just think, super heated water comming of the CPU, used to run a steam powered electric generator!!!

  • Why not mount the spray gizmo inside the case with a temp sensor? building a robot that runs around being all liek "Chip Hot Danger!" ::Spray:: seems like overkill to me. you could even have all the sprayers connected by tubes in a rackmount style environment to a big old tank of the coolant. But I suppose the bot is definitly cooler.
  • Or you could just buy a pack of dobermans to guard your server farm and place fire hydrant stickers on your cpu's.

  • Ok, one part of a chip is warmer than another part. That's why we have copper to distribute the heat. It does that, remember physics class?

    What i wonder is, how are they going to identify individual parts of a chip as hotter or cooler, other than sensors all over. I'm sorry, but i can't envision the whole process being very efficient or well run right now. Personally, I'd rather try to get chips that don't need cooling, or piezoelectric fans that just go apeshit at cooling.

    I've worked with people in the printer part of HP before, and they're not exactly the most efficient people out there. This sounds more like a Compaq-inspired consolidation technique, not a true engineering project.

    Dilbert's boss, "so, we need some cooling in this part, right? and we have these printers that put stuff where we need it, right? why not use that thingy do help with that thingy! Make it happen!" ^Big idiot grin on face^

    Dilbert, "Sigh"

  • I want to mod the story (+1 Funny). I was laughing my ass off about the psycho taggin' robot!
  • Just give me a thermometer and a super-soaker, and set me loose in the server farms. It'd be a great way to blow off some steam! (Excuse the pun :) )
  • A question. (Score:5, Funny)

    by teamhasnoi ( 554944 ) <> on Thursday August 08, 2002 @04:05PM (#4035272) Journal
    if you filled the robot with hot grits, would it automagically find Natalie Portman and spray her down with them?
  • by doublem ( 118724 ) on Thursday August 08, 2002 @04:07PM (#4035291) Homepage Journal
    Guess What? The robot and the ink jet based cooling system are two different things. The robot adjusts the air conditioning in the room, the ink jet based coolant determines the specific parts of the chip that need cooling.

    One is on the Macro scale (sorts) and the other is on the micro scale.

    The robot will NOT be spraying ANYTHING!
    • Hmmmm, a robot to adjust air-conditioning... useful... and all this time we've been using thermostats....
      • You know, that's what I thought. Why not set in some long term equipment to monitor the temp and do whatever the robot would do instead?

        Would wiring really be that much more problematic than a blooming robot?

        Then I thought: coolness factor.

        Which is neater, an invisible system that does the job quietly and in the background, or Johnny 5 running around with a big temperature probe, shoving it behind servers?

        One is responsible and efficient, the other is neat looking and will bring up subconscious images of alien anal probing.

        The robot looks like overkill, but it also looks like a video game.

        Of course, an hour of programming is easier than running a few thousand feet of cabling and temperature sensors, so there is the speed of instillation factor. No dust floating around from drilling holes, no additional hookups to worry about and so on.
  • by lingqi ( 577227 ) on Thursday August 08, 2002 @04:07PM (#4035293) Journal
    Now-a-days CMV (chemical vapor deposition) of carbon is so advanced that you can get milimeters worth of diamond within hours.

    diamond have probabbly the best thermal conductivity known to man, so if you CMV a diamond layer on the chip and use that for interfacing to a copper heatsink, i would think that it would be a better idea than putting small sprayers.

    liquid will vaporize and get recollected -- but it also have the problem of
    1) depositing crap when it's vaporizing
    2) possible diminishment of the resevior throughout the system's lifetime. i would hate to have to replace anything like this -- since they recommend direct access to the die's surface!

    any impurities in the liquid can spell certain death.

    Lastly, i do not foresee this being much cheaper that artificial diamond heat-interface. especially if this is done on a large scale -- it would have the side-benefit of really cheap diamonds for everything from lenses (scratch free! never breaks! ultra-light!) to screwdriver tips, etc etc.
    • diamond have probabbly the best thermal conductivity known to man

      What gave you that idea?

      Even if it were, what would be the point of a thin layer of it between the heat sink and the chip? NEWS FLASH: Diamonds are hard.

      The whole point of an interfacing layer between the processor core and the heat sink is to put something mushy in there to compensate for imperfections in the surface of the chip and of the core that could cause less than 100% contact between the two.
      • diamond have probabbly the best thermal conductivity known to man

        What gave you that idea?

        I dunno what gave him that idea, but I thought it was a well []-known [] fact [].

        According to the first link, the thermal conductivity of copper (in W/cm-K) is 3.937. Room-temperature diamond: 6.299. And an isotopically pure room-temperature diamond: 50. The last link claims a conductivity of 2000 W/cm-K for CVD diamond and talks about using it to cool stuff.

        So I guess the more interesting question is where you got the idea that diamonds wouldn't work well for cooling.

  • Does it take its face and hands off before it goes to bed? What kind of powers does it have? And does it use them for good, or for awesome?

    Strong Bad [] wants to know!
  • The problems asociated with CPU cooling can't be solved by spraying the chip. Although it would be neat to have a robot running around monitoring systems in a datacenter, It seems unlikely that spraying overheating CPUs is a wasted effort. Perhaps a robot that answers the helpdesk phone and insults the users. Why pay a Bastard Operator From Hell when you could have a robot, or even simple IVR system that serves the same purpose. And you'd have a really cool Tape Robot to handle your system backups.


  • ...that's cute! the H points to and P points to hehehe =)
  • > "We believe we have to take a holistic approach to cooling," said Chandrakant Patel,

    On first thought, throwing mist over microprocessors, seems more like an alcoholistic idea. Anyone been to a sauna? Does not this create a new problem of increased humidity -and even worse -fluctuating temperature and relative humidity and problems resulting from that...oh, but HP makes the hardware, now they can sell double the stuff in same time, good plan :)
  • Broken robot => broken chips
  • by bgeer ( 543504 ) on Thursday August 08, 2002 @04:22PM (#4035391)
    I just heard a report of a strange bug in the new robot: for some reason if it detects a non-HP ink cartridge in the printer, it pulls it out and stomps on it. A patch for this problem is expected in Q3 2009.
  • in retribution, they might send some infiltrator in to knock out the robot with funny rubber coated disks that bounce off the walls. My advice is to put more snipers up in the courtyard... and actually train them this time.

    If you don't get it, don't worry. I doubt I could remember the name of this game if I tried :(

  • Extremetech has an article about how HP has decided to use the spraying tech developed for inkjet printers to cool chips -- and has made a robot that'll wander around data centers, detecting too-hot chips and hosing them down.

    Hmm. Imagine this going wrong..

    Gang-bangers and wannabes get hold of these and guide robotic taggers through your 'hood.

    • sounds like a bad .NET commercial....rather in the vein of the Asian mother who keeps telling her daughter to go to bed while the daughter is tagging vehicles on an assembly line oversees with her name (forget whose commercial it was, but it illustrates the point). Nothing like one degree of separation between the black hats and your business...
  • You will see a lot of bimbos and bosses walking around all soaked.
  • If you actually READ the article, the coolant is sprayed from within the chip and potentially recycled; the robot wanders the data centre looking for hot-spots; the two things are COMPLETELY DIFFERENT.
  • What? These robots are going to be carrying Palm Pilots around?
  • According to the article, the robot goes around the datacenter and monitors the temperature at certain preset locations. Someone explain to me exactly how this is more efficient than placing thermal sensors at each of those locations and having a computer monitor them and take appropriate action? It would be one thing if the robot actually hosed off the warm areas, but it doesn't. It simple reports the data to a computer that takes appropriate action. This is just silly...
  • Why the robot? Seems to me, they could just install a cooling module where needed and let onboard thermal management decide when to send the thing a pulse to spray.
  • I wonder how much it will cost to refill this thing? I guess HP will charge about $50 for a water cartridge.
  • Yes, one actually exists [] but it isn't a crazy, self knowing, bot.
  • With a few hidden keystrokes, the robot can turn its sprayer full blast on the comely female interns while chanting "Girls Gone Wild, Girls Gone Wild...."
  • by Anonymous Coward
    One of the main reasons for semiconductor failure over time is thermal shock - when you turn a computer on and off the chips and metal expand and contract, and shorten the life of all electronic equipment. That is why computers are left on all the time.

    This plan would expose the equipment to super-megasize thermal shock, and would result in much lower long term reliability.

    Either this is a joke, or some engineer didn't do his homework bigtime.
  • can be achieved by painting the chips red (like fire hydrents) and letting your dog loose.
  • HP Marries Inkjet and Robotic Technology

    The first image that came to mind when I read that title was of a small robot, maybe the size of the martian land rover that roamed around painting large canvas's.

    It's underside would have a inkjet cartridge, battery powered, and completely remote via infa red or 802.11.

    You suck for not making something like this HP. I know everyone want's one.
  • It can't be that hard to swap out the coolant with a keg strapped on its back and the temperature gauge with a gauge that determines the depth of beer in a glass...

    Kick off a keg party and set the robot on AutoFill!!

    Man... Now all we need is to give it a nice rack and a tight tank top :-)


  • All robots eventually become self aware and turn on their human masters!
  • I mean, it's great and all. A robot to cool chips. Yay. More power to em I guess, but come on... This couldn't be done by more conventional means, not to mention cheaper? Okaaay...

"You can have my Unix system when you pry it from my cold, dead fingers." -- Cal Keegan