Slashdot Mirror
Water Cooling Flow Indicators
A reader writes:"There's an interesting HOW-TO about how to construct a a flow indicator for your water-cooling rig for your overclocked machine." I dunno what I'd do with the extra cycles from overclocking and watercooling - maybe run more rc5 packets though the box - crazy idea though.
I got this months MaximumPC magazine in the mail today (and, damn it's a good thing I'm not having to pay for the subscription)...it's really sad to see what an appliance PCs have become.
15-20 years ago, you'd go to the magazine rack, and there'd be plenty of computer magazines...and most of them contained _CODE_ -- people back then WANTED to learn about their computers and how they worked. Heck, at least a couple of magazines had circuit schematics. There was a viable market for hardware (as in circuitry) experimentation.
Nowadays, "hardware hacking" consists of bolting ever more ridiculous fans and cooling devices to your processor -- and we gauge this progress by how many extra frames/sec we get under Quake. It's rare (unless you go to a _really_ well stocked bookstore) to find a magazine which covers any programming -- and what you will find usually isnt geared towards beginners or intermediate skill levels.
Every time I open MaxPC, I get the sad feeling that all the joy of computing has largely been sucked out of it (or stolen away by the ease-of-use-is-EVERYTHING crowd). Byte, Creative Computing, Softalk, Kilobaud, 80micro, (most of the Wayne Green titles) were so much better rags it isnt even funny.
And this month's MaxPC issue has an article on upgrading old PCs that incessantly pounds the drumbeat of the upgrade treadmill...they say you must discard that P-II 266, because it cant run squat!
I mean who really has time to do this stuff?
Unemployed ex-dot-commers.
Ooooh. I'd never thought of it terms of getting bigger, slower fans. Very cool.
Thanks! Your post has been the *ONE* truly useful post I've read in about a year!
--
--
Don't like it? Respond with words, not karma.
Er, no. Hacking at its finest was back when one had to weild a soldering iron. And not just replacing the quartz oscillator, either.
My greatest hack was doubling my computer's memory by soldering new memory directly over the old memory, leaving one leg bent up to be wired to a page-flip lead on the CPU. Goes low: original RAM is R/W; goes high, new RAM is R/W.
Second-greatest was hacking the serial port. Ran wires from the legs of a serial I/O chip (I forget the part number) directly to the motherboard; and other wires from the chip to a port connector. Stuffed the whole thing under the integrated keyboard, melted a hole in the case to dangle the port outta...
*That* was hacking. Have gun, will solder. Whoo-hoo!
--
--
Don't like it? Respond with words, not karma.
I don't really care if it's water cooling, magic pink smoke, immersing in an oil bath, or whatever it takes to accomplish the task...
:-(
...what I want, more than overclocking, is for the freakin' roar of three fans to be silenced! My god, this box is noisy.
While I love the speed of my new computer, I kinda long for the days of my AMD K2-200, which didn't have any fans at all. Not on the CPU, not for the case, and not in the power supply. Just hung the latter outside the box for convection cooling and, oh!, was it a quiet machine...
This machine would start glowing if the fans were stopped...
--
--
Don't like it? Respond with words, not karma.
Actually it makes more sense as time goes on. The cooling system is a one time purchase, you can keep reusing it so it gets more cost effective with each new system.
The guy could have just put a PSI gauge inline. doh!
Keep the Classic Slashdot.
Overclocking is an end in itself. There's not much to be gained that can't be gained through a few bucks well spent, particularly if you factor in the overhead that is involved when setting up that water cooling rig. Extra parts, setup time, hardware failure when screwups happen(not to mention shortened hardware lifespan and bye bye warranty). The same can be said for monitoring the cooling system designed for this purpose. Trollish as it may sound, overclocking doesn't get you anywhere and is a waste of time.
Well obviously for a lot of people who don't go to quite the lengths (i.e. building water cooling and such) overclocking can be a great way to increase get performance for nothing. Two years ago (or thereabouts) I got a Celeron 300 (with the 128KB L2 cache) and by just covering a couple of pins (boosting the voltage to 2.2v) and pushing the bus speed to 100Mhz, instantly I had a 450Mhz processor (when at the time Pentium II 450s were about 4x+ more expensive) : This processor is still going strong in my firewall today. No cost to me but for a couple of minutes of tape cutting, and I derived 50% greater performance.
However for the people who go all out with case mods and cooling systems, I think the actual performance is on the bottom of the motivational reasons list. Just simply doing the project and having fun while discussing it with friends and people who share a common interest is the motivator.
Most of these people say that it would be cheaper to buy a new processor, but in actuallity you can build a good water cooling system for less than $100USD, factor in a new processor and fan, and thats quite cheap. Hell, why not buy a new processor and add a water cooling system.
---
--
Insert Witty Sig Here
I would love nothing more than to get a 1ghz+ machine that I will actually allow me to sleep when I leave it on, never mind that annoying incessant buzz. Does anyone have any links to some known high quality OEMs or whatever that assemble this stuff for you at reasonable cost? Like I would be willing to pay a significant premium, but despite my technical abilities, I simply don't have the time to spend hours running around looking for parts and/or assembling a zillion different things.
On a slightly seperate note, I think someone needs to make high-end PCs, sorta like how Mercedes or BMW is to cars, combination between a case and/or a full PC (as in assembly). Not necessarily always the fastest or most advanced technologically, but rather a framework which those things can be adapted to in a pleasant manner. i.e., nice looking, really easy to open/access, as quiet as possible, flexible, etc.
Do you know of any manufacturers that make cases and fans to accomodate this? Have any links?
What about using a waterproof motor. Then use your PC to measure the backward voltage generated by the waterflow. If the motor was very small, this wouldn't effect the flow too much.
Use it to heat up leftovers while cooling down the CPU...
---------------------------
"I'm not gonna say anything inspirational, I'm just gonna fucking swear a lot"
---------------------------
Most of these people say that it would be cheaper to buy a new processor, but in actuallity...
Not only is it cheaper to do a water-cooled rig as you suggest, but in addition the water cooling rig can be reused for all of your CPU purchases. A friend of mine has been using the same rig since the original Pentium days. Whenever he gets a new CPU, he just has to change the CPU cooling block (pretty cheap) and presto, he's back in business. So it's great, because he always gets more speed out of every CPU he buys. That's more than I can say for my CPU fan purchases - seems like every time I upgrade, I shell out $40 for the top-of-the-line platinum-coated super-groovy Orb, but it goes out the door with the old CPU every time, and it's not reusable.
Plus, if you build a shell for the hard drive or other components, those are usually completely reusable. He's been using the same 3.5" water cool shell since he started, and the same video card chip cooler. These things have lasted beyond several upgrades. Pretty slick.
What's your damage, Heather?
--Ben
300 fps Quake 3...
kids these days! don't you know that q3 caps the max fps at 250?
Overclocking is an end in itself. There's not much to be gained that can't be gained through a few bucks well spent, particularly if you factor in the overhead that is involved when setting up that water cooling rig. Extra parts, setup time, hardware failure when screwups happen(not to mention shortened hardware lifespan and bye bye warranty). The same can be said for monitoring the cooling system designed for this purpose. Trollish as it may sound, overclocking doesn't get you anywhere and is a waste of time.
...who once said that 640k of memory should be enough for anybody. But really, what the hell do we need these extra 10-20% speed boosts for at this point besides Q3 benchmarks to impress your l337 friends? I don't know what amazes me more: that the Powerbook I carry around is 10 times more powerful than the Cray supercomputers of 20 years ago, or that I still watch it struggle to resize a window under OSX.
Hrm. All right, maybe we do need to overclock our machines still. But something is terribly wrong.
TomatoMan
-- http://frobnosticate.com
On second thought, maybe I'll just stick with old-fashioned air.
While the cost/performance boost ratio for watercooling is questionable, I've got a Duron750@1000 running here with a $30 HSF. The processor cost me $50. Voided my warranty, sure, but it's _$50_ :D
The main reason behind setting up watercooling systems is for the "cool!" factor; if your nerd friends come over and see the tubes and radiators snaking their way around your machine they'll say "whoa." These are the same kinds of people who spraypaint their cases and run hardware websites obsessing about components that come on blue toned PCB.
It's sort of like how I've got a million gamepads kicking around here, but I just had to solder up a set each of SNESpads, PSX pads, and NES pads.
--
I think it seems to be missing some sort of indicator to my computer so that my computer knows what the flow rate it and can shut itself off (or adjust the clock speed) after it's gone for a certain amount of time at dangerously low flows. Otherwise it's just a visual indicator to me, but if I'm not by my computer it doesn't do me much good.
Things you think are in the Constitution, but are not.
I dunno what I'd do with the extra cycles from overclocking and watercooling - maybe run more rc5 packets though the box - crazy idea though.
AFAK, the first faulty part in a CPU when it's o/c'ed is the FPU, the most use part on distributed computing.
Even if the o/c'ed system seems rock solid, it might induce false calculating results. the Seti@home project team pointed this on sci.astro.seti a few weeks ago.
Don't get me wrong, I agree on the concept of o/c'ing, for games and general performance; but why don't physics labs use clusters of o/ced Athlons instead of 32CPU Alpha Global Servers ?
Almost two years ago I spent around $450 on building a dual celeron 333 box and overclocked it to 500mhz. Running winbench its equal in speed to a 1ghz Athlon. Tell me that wasn't worth the price. Given how fast technology is growing, a 5 year warranty on the cpu is worthless. In 5 years your box probably won't even be around.
Only the State obtains its revenue by coercion. - Murray Rothbard
paying more than the price of the processor you'll actually reach with the price of the hardware required for the cooler. I mean processors have becomed damn cheap, if you have that much time to kill, why not modding the case with some cool neons or stuff like that?...
:)
Back in the celeron I days it was cool, but even then, I could overclock my 366->550 with air-cooled heat sinks, why the heck would I want to add water/pelletiers/carbon ice to reach 600mhz? 50mhz more... hello?
--- Metamoderating abusive downgraders since my 300th post.
Remember the water-powered computer (see the Hacker's Dictionary)?
since most people don't want water dripping around, cooling systems should keep the chip above the dew point. usually less than 50-60 degrees F or so. and also keep the chip cooler than, say (ouch) 150 degrees F or so.
all this, and the processor makes about, as an insane maximum number, 100 watts of heat.
now what happens if you take a 100 watt lightbulb and put it in a box about 20cm on a side. thermodynamics says that the box will be about 130 degrees on its surface, assuming an 80 degree room. so that means if you can transfer all your processors heat to the surface of a box about 8 inches on a side, the result will be an equilibrium of about 130 degrees F. (sorry for mixing metric and standard, just trying to use _some_ recognizable numbers).
so now just use a chimney. attach a small copper block to your cpu and drill a hole or somehow attach a hose passing through the block _vertically_. give about 6 inches or so above the block for good thermal convection. run both of those tubes into a 8 inch box filled with water (uninsulated as possible).
the water will naturally circulate due to heating from the chip and recycle to the reservoir.
upon further calculation, if you have a tube about 1 cm in diameter and a convection style flow rate of, say, 1 cm (or ml) per second, the exit temperature of the water (and hence the cpu heat) is about 50 degrees F hotter than the input. that's bad!!! almost boiling, and certainly disappointing.
so you must have pump to provide at least 10 ml/sec of flow to keep the heat to only +5 deg F. i would think that any fish tank style pump could do that cheap.
anyways, i was hoping to show that pumps were unnecessary, but instead all i can say that a _chiller_ is pretty much a waste of money. (ps. doubling the surface area of the reservoir reduces the delta T by a factor of 2, etc)
i hope that someone has found this little bit of physics interesting.
muerte
What next? Will punched cards come back in vogue?
Come on, there are always people with enough time to do these sorts of things.
.com shakeout effects everyone, and is constantly adding to the ranks of people who have time to pursue projects like this. The latest group of geeks who will have time to pursue this are those from cyberRebate.com
I wish I had enough time to do these sorts of things. That wouild mean (hopefully) that I had enough money to free up my time. The
Best of luck to them.
--CTH
--
--Got Lists? | Top 95 Star Wars Line
how about a "boner" indicator to gauge how excited the user in front of the machine is, expect it to increase massively when someone mentions "VI", "kernel hacking" or "CowboyNeal" and decrease when someone (your girlfriend?) interrupts you from your coding.
or maybe a something to tell how open your eyes are (as a percentage of fully open) and give a wanring beep when you fall asleep at the terminal (maybe this could be inverted whilst at work and warn you if your eyes reach more than 10% open except when you're refreshing /.)
..Barny
Some is a good website to buy big fans:
http://www.2cooltek.com/
http://www.2cooltek.com/ (for those safe surfers)
their Panaflo 120mm fans are really nice. they move a LOT of air, and are quiet and slower than most 80mm fans that push less air. Don't even bother with anything smaller than a 60mm fan on your CPU, i suggest 80mm or even 120mm fan on your CPU, some can get by without any fan on the CPU at all if you have a good enough heatsink, and the PSU pulls off enough air from it. (most computer builders do this such as Dell, Gateway, and Compaq for their workstations/desktop models)
As far as cases are concerned, I do not really know where you can buy them with 120mm Fan slots. they all come with slots for 80mm fans. however, some cases on the net do come with a 120mm slot in the front, and/or a 120mm slot in the back, however they are usually much more expensive than standar high quality cases and are of lower quality metals/design. I think somoene could make a lot of money designing cases, and nobody does. Its a shame really if you ask me
Two infinite things: your stupidity and mine. But I'm not sure about the latter. If my sig offends you, I'm sorry.
You need to use larger fans. If you use small cpu fans, small case fans, and spin them as fast as a fucking hard drive, you will get a loud as fuck case.
I have a single 120mm fan running in the front of my case, an 80mm CPU fan, and a 120mm fan blowing out of my case. This all required a case mod, but it was worth it. I get 2x more airflow through my case than the same case with louder [read: faster] 80mm case fans. To top it off, my ambient case temperature is 10c cooler than it was with high speed 80mm fans, and my cpu temperature is 15c cooler with the big heatsink, and big, slower 80mm fan (big compared to standard 50 and 60mm cpu fans).
As a nice side affect, bigger fans can spin slower and push the same or more air, this means less power. with the same case, when i switched to 120mm case and 80mm cpu fans, it use 5% less power[than old cooling setup] when the machine is at full load. This is a very good setup. I have the quietest computer on the block, and my fans costed me like $2 more than standard fans.
some other tips on keeping your case quiet:
Screw down your hard drives using ALL the screws, and do it firmly. I use rubber washers in between my HDD and case. thin cardboard cutouts also do this job quite well.(this decreases the heat that can come off the hdd, so be carefull if you have high speed hdd's such as 10k and 15k RPM models) This decreases the amount of vibration your hard drive sends to your case, and can REALLY quiet things up.
Make sure your CDROM drive is screwed down firmly! this drastically reduces noise when a 50x cdrom is spinning off its bearings!
Replace your fans when they get noisy. Ball bearing fans last longer, and only cost 1 or 2 dollars more than sleave bearing fans.
I really wish ATX cases came standard with 120mm fan spaces. there is PLENTY of space on the case to cut a bigger hole, but they do not and expect people to settle with 80mm. You all wonder why the Mac G4 desktop is so quiet? Simple, insted of using multiple fans (60mm cpu fan, 80mm PSU fan, 80mm case fan like dell, gateway) they use a SINGLE 120mm heat sensitive fan mounted on the case directly over their oversized heatsink. the fan exausts air out of the case, and pulls it from the hot cpu. there is no need for 50 fans in a case if you use proper air direction, and hardware placement.
Two infinite things: your stupidity and mine. But I'm not sure about the latter. If my sig offends you, I'm sorry.
keep the fan attached to the motor assembly. this could help in one of two ways, the second being considerably more useful. first, if the fan is still attached to the motor perhaps it could be turned on to help boost the water flow under high load/heat situations. the more useful reason is for speed monitoring. all but the cheapest fnas now have a 3 wire connector with the yellow wire being used to monitor RPM. with a little fiddling (perhaps cutting one of the other wires) the RPM monitoring ability could be retained allowing you speed junkies to use it with standard fan speed monitoring software. i may sacrifice one of my fans tonight to test this. well wait no. unless youre using a nonconductive fluid. netiher of these will work with a standard fan. oops. hopefully somone will expand my ideas with a waterproofed motor.
Although I'm happy it solved the problem, it's sort of scary that in the future with everfaster CPUs, Video Cards, Front/Side Busses, this problem will only worsen. I wonder how many PC problems people encounter are not bugs or conflicts, but just heat.
Although I'm happy it solved the problem, it's sort of scary that in the future with everfaster CPUs, Video Cards, Front/Side Busses, etc., the problem will only worsen. I wonder how many PC problems people encounter are not bugs or conflicts, but just heat.
I do and I agree. After all, its not only the CPU that gets hot, but memory and videocard and other PCI cards and the motherboard. So I just use a window fan in front of my open box. It puts a larger volume of cooling air over the whole box. It cools not only the CPU but the whole package. (I think by about 20 or 30 degrees, but I'm not sure.)
--
on the ledge of someone with way to much time on their hands. I mean who really has time to do this stuff?
If you ignore the other uses of a tool, does that make the tool less useful, or you less useful?
Computers can be o/c'ed, because there's a safety margin in time between clocks and the time it takes for one 'round' through the flip-flops, etc... obviously no? Vendors also use it to buy them time and sell computers more (aside the safety purposes)... and that's the real reason why you should o/c. Note : Because schematics become different in a newer faster CPU (for faster 'rounds'), computational speed may not have increased by the same factor as the clock ( while o/c'ing does tend to do that, obviously no? ). And : It's always very relaxing and good for the stress-full o/c workaholic to have a warmwater-aquarium, next to the computer... mjummy gummy ...
Don't mind this dude, half of the time he goes half naked, and that's not the same as full time quarter naked, is it ??