Printable View
Contact wasn't great. Looking at low-90s/high-80s for initial temps at 22c water temps on the remount. I'll let this run awhile.
EDIT: for the record, the fan is really loud at 100%. Anybody running 120-140mm fans at comparable noise levels on a 2+ fan radiator will see extremely low water-to-air deltas already.
@ TRICHEESE - not trying to pick a fight with you, just commenting that it won't matter if the" top of the block is connected directly to the base" because air is not nearly as efficient at cooling the CPU as is water. The point about properties not changing was in your reference to Martin210's testing. Apologies if it came across as snide - that wasn't the intention.
Please continue the discussion by PM if you want - let's keep the focus here on nateman_doo's block and the testing in progress.
Ok... here is the problem that we see.Quote:
Originally Posted by TRICHEESE
1. For you to get that kind of delta you would need a great fan on that cpu block, because of the difference in mediums.
As vapor said, the noise coming from that 1 fan is better adjusted @ the rad.
If you dont understand difference of mediums after being said 4 times, then i give up.. actually i already give up, i am just now going to defend martin, vapor, and shazza.
Lemme remind these guys exactly who you are... and what i did to make you retire from milling:Quote:
Originally Posted by Martinm210
You guys have no idea who martin is do you?Quote:
Originally Posted by Martinm210
I did a lot of sponsoring for him b4 the vendors started sponsoring.
He has done a lot of tests, and he's also spent a lot of his own out of pocket money..
This is why a lot of people respect martin.
Thanks for the test... and yeah we were talking about how the noise is better placed @ the radiator.Quote:
Originally Posted by Vapor
Hehe..yeah the block making business will wear you out in a hurry if you want anything more than an experiment.. Your blocks were pretty cool, but I still kind of like my old thin line block..never was as good as a fuzion but I thought it looked cool:
http://www.xtremesystems.org/forums/...rtin+thin+line
IMHO its pretty impressive for any homeade block to perform similar to an apogee gt...blocks have come a long ways from the old days and manufacturers have been doing a lot of testing and r & d along the way. The measurement of success should come from just having a custom one off block that is truely hand crafted and works.
I may try to make my own again someday...I kind of miss that old thin line block.:D
Ok thanksQuote:
Originally Posted by shazza
EDIT: @naekuh: As with when i was conversing with shazza i said that i understood the difference in mediums.
So now your accusing me of attacking shazza which i was not. And on top of that your also accusing me of attacking martin who only supplied the data (i have only admired his work when i have seen it before, so i do know who he is) and i have not even talked to him in this thread. Where the hell you got the idea that i was attacking Vapor from i don't have a clue, i have only supported the block going to Vapor and i trust his work with blocks and base what i buy based on what information i get from him and skinnee
i think Martin is just compensating for somethingQuote:
Originally Posted by NaeKuh
okay, gang. Let's agree - no one's attacking anyone here. I think the issue is people are just getting frustrated and it's sometimes hard to understand questions/comments in the written format.
So, let's calm down and wait for Vapor and nateman_doo to comment on this latest testing round.
I am just sitting here waiting to hear. Just like everyone else. He said his first mount was a bad one, so now just still waiting.
15min left on this mount, I'll have more accurate numbers to post after I get the data uploaded.
First mount was really bad due to mediocre contact....this mount is faring better, but still not looking so good :eh:
First mount: 97.11c after being normalized for 30C water temps, contact was good but not amazing.
Compared to:
http://skinneelabs.com/assets/images...SimpleTemp.png
Mount 2 is underway.
As for 'radiator power' it's extremely complicated with the initial data.
At normal temperatures (75-80C), the CPU dumps 217-220W into the loop via the block. I know this from using block in water temps, block out water temps, the flowrate, and the specific heat of water.
As temperatures increase, CPUs consume more power. Some 20C above my normal running temperatures, the power consumption is going to be higher. Add to that a big change in the balance of primary cooling vs. secondary cooling and the figures I have from other blocks are basically incomparable to assess the radiator power. The 'fix' to that is to run this block fanless (as I originally planned to do), but considering the high temps, I don't plan to run more than 2 runs on this block--I just don't want to risk the longevity of this testbed. The second mount of this block with the fan at full speed is my final run with this block.
That all said, the heat-absorbed by this block is 225W on the first run. Which means that the extra heat absorbed from running at higher temperatures outweighs the heat dissipated by the radiator on the block (though only by a few watts). This either means the power consumption/absorption is way up or the radiator power of the block at a ~5C air to water delta is minimal, or something between the two.
Mount 2 came in at 95.4c with great contact.
Overall, puts it about 20C behind the best blocks right now on my testbed.
are there any photos of the internals?
It's soldered/brazed together.
http://rdefined.com/phpbb/images/smilies/hide.gifQuote:
Originally Posted by Vapor
That's alright, Nateman. Back to the drawing board, bro.
and thats with the fan on 100%?
You have the same radiators, in all the tests, with the block and fan at 100%?
whats the pressure drop? would be a good way to know if the design is restrictive enough to give the surface area a speed up of water flowQuote:
Originally Posted by Vapor
Yeah, your fan at 100%, everything else completely identical between tests.Quote:
Originally Posted by nateman_doo
There wasn't any noticeable 'radiator effect' either, as I detailed a few posts earlier (but there may have been a radiator effect, just negated by the increase in heatdump from the hotter CPU).
I don't have a manometer, but flowrate was 1.62GPM, roughly the same as the Heatkiller 3.0LT/Cu.Quote:
Originally Posted by Manicdan
How many radiators are being used, and what is the ambient temps in the room?
He's adjusting for 30C water temps... ambient and the amount of radiators doesn't matter. And, he's also already posted that there is no noticeable radiator effect from the block.Quote:
Originally Posted by nateman_doo
how do you "adjust" for temps? why not hook up radiators, pump water through them, and read what the software says?
Here are my simple thoughts on the subject:
1) The original tests were done using one dual radiator with weak fans. In this configuration, I assume the fan on the new waterblock helped improve cooling, since the radiator/fan combo was not sufficient to remove the heat. (Still not sure how much the design of the waterblock helped - e.g. copper/fins).
2) Vapor's test was done at the other extreme - two triple radiators with higher speed fans in push/pull and lots of pumping power (I'm assuming Vapor was using the test setup outlined here, or perhaps an updated version. He can confirm. As has been mentioned before, this setup was designed to take the radiators and fans out of the picture and get a true representation of how the CPU blocks compare.
3) This is why I didn't think the testing would resolve all of the debate - one could still make the point that the "hybrid" block has the advantage of being able to work effectively as an air cooler, should something go wrong with the water loop. Nateman_doo has said he ran his system without water using this block, so I assume it works. As to how it compares to high end air-cooling HSF, I don't think we have seen that data yet - as I don't think that was the point of the block's design.
4) There seems to be confusion around the point of whether or not the "copper-with-fins" design for the waterblock will in fact improve the cooling efficiency of the block, compared to the current CPU waterblocks. What I took from the comments by Martin and others regarding the thermal properties of water and air is that water is much more efficient at heat transfer, thus the affect of the "external" copper/fins/fan will be insignificant compared to the heat transfer that takes place inside a top end water block - thus the focus on internal design (as opposed to trying to get any heat transfer from the external part of the block). This scenario would apply to a loop that has a decent radiator/fan configuration.
5) It's still good to see someone taking the time to learn to make their own waterblocks, and I'm sure nateman_doo will keep coming up with new things to try - and I hope we can encourage him to do so.
nateman did his test w/ his i7 920 running at 2.79. Many people with i7's who venture into water cooling run their systems at 4+.
Again, shazza; wise beyond your years. (forgive me ahead of time, I am enjoying a few adult beverages-its the 4th)
You are correct that I have been using this block for weeks with no water, and I never hear the fan kicking on for daily computer usage which for me is gaming, and cpu folding. Not hearing the fan kick on means the smartfan never sees the CPU above 59°C.
This was the first prototype and I had a few design changes to implement, but I am runnig out of time. I am going away for some training and sending this block to vapor was a hasty decision on my part to quell the masses. I have plans for new internals already but I figure there is no where to go but up at this point.
I just can't see how 4lbs of copper with close to 50 fins could perform so poorly, when it works with air alone and no water?
Many others just run stock numbers as well. Not everone OC's there stuff for daily use, but still it should be able to hang. I am at a loss for words.Quote:
Originally Posted by rehpyc
And not just that, but alot of people run water to run quietly.
I can run my loop passively and still get acceptable temps. It's awesome.
yea, I will be the first to say that the fan I used for this is NOT quiet at ALL on full blast. I don't much care about noise though. @ 50% it is fairly quiet.
That's what most of us coming from air over to water think. It comes back to the difference in mediums. Water is so efficient at transfering and storing heat that the copper itself becomes an insulator of sorts in comparison(it can't move heat fast enough at great distances to keep up). This is also why copper bases are so thin. If mass and surface area was the key, water blocks would look like air cooling tower sinks.Quote:
Originally Posted by nateman_doo
Thermal properties of water is just so different all that matters for cpu/water exchange is basically the bottom few millimeters of copper that is in contact with the IHS over the cores. The rest of that 4 lbs of copper is only contributing to the water/air heat exchange. That water/air exchange is where size matters...and that's because air technically is an insulator, it sucks at heat exchange and it also sucks at storing heat especially the storage piece.
And I'm no expert, I dropped thermo when I was in college many years ago, my experience comes from my own random experiments. I suppose a lot of block design today is a result of experiments though...there are too many variables to pin down theoretically. If you could optimize using theory alone someone like bill adams or cathar would have already done it...:)
Anyhow, don't give up...its a neat idea and something different is always cool. Maybe a standard copper waterblock with heatpipes soldered to a more conventional aluminum hsf would be another idea to the same hybrid idea. Pumps do fail and not everyone wants massive radiators.
Then by what your saying the new design improvement I came up with will work much better. I MAY be able to bang out one before I go away and send it back to Vapor, but thats assuming I finish all the other projects I am working on.
These results are actually worse than I expected them to be. But I did figure they wouldn't be anywhere near what your testing showed Nate.
One of the reasons I started this thread was to get you to send the block in for testing. I was a bit concerned with all the people over at EVGA that thought this was the greatest thing since ice cream and wanted to buy one without seeing any independent testing first. Hopefully you haven't sold too many of these. Based on Vapors results you're going to have some unhappy customers if so. I trust you'll be posting the results over at EVGA.
That being said, I hope you're not too discouraged. This hobby needs people trying new things to help move it forward. Hopefully you've learned some things and your next project will be better for it.
Mass and surface area are very key.Quote:
Originally Posted by Martinm210
Ideally you want the least mass and most surface area at the base, and most mass and further surface area everywhere else on the block. If you get the base right, the rest of the block is just icing on the cake. This hybrid idea could actually work quite well if you have a base type array on the topside of the block. It would hurt flow though.
Only the bottom few millimeters of copper thickness is effective at transferring cpu heat to the water..the rest of the 4 lbs isn't helping the cpu/water heat exchange...it can only work to dissipate heat from the water.Quote:
Originally Posted by BeepBeep2
Even as a radiator, what do you see...very thin copper tubes attached to very thin fins and ginormous surface area...but again little mass/thickness in contact with the water. Waters thermal coductivity and storage is so good..large masses of copper cause more harm than good. This is also why heat pipes came along for air cooling...material properties are key and completely different that air..
Let's summarize.
Quote:
Originally Posted by nikhsub1
Quote:
Originally Posted by nateman_doo
Quote:
Originally Posted by nikhsub1
I don't need to now do I?Quote:
Originally Posted by UrbanSmooth
Smart man! :)Quote:
Originally Posted by n00b 0f l337
So I said what the outcome would be and I was proven right, is it still nonsense? Or are you nonsense?Quote:
Originally Posted by johnksss
Quote:
Originally Posted by nikhsub1
I wasn't going on about anything - I was stating facts and most of you were basically telling me to get lost.Quote:
Originally Posted by johnksss
shazza, my comments weren't deleted because they were correct... I hate to say I told you so... the last bit I disagree with however, the results pretty much resolve it all - it doesn't work... However, I do like the out of the box thinking so he should continue to tinker IMO.Quote:
Originally Posted by shazza
I thought just the opposite really.Quote:
Originally Posted by Sideroxylon
Quote:
Originally Posted by Vapor
It's ok, nikhsub, my original posts got deleted for the small bits of laughs and jabs compared to the larger bits containing pertinent information and opinion. The results of Vapor's tests confirmed my intuitions, but grandma shazza didn't like my witty jabs :[
nateman, continue to be innovative, but you really need to hone in on what market you are wanting to attract. as a water block itself, your block performs poorly. as an air block itself, your block performs poorly. with how big of a block of copper it is, it'll cost too much. at the very least, hopefully you learned a thing or two from Vapor's testing to be able to do your own effective testing for the next revision you make to your block. keep your head up and continue tinkering. at the very least, you'll find that something like this works for you and maybe a few others.
Wow, it did a lot better than I thought it would do. On his EVGA post, the heatkiller test was done at 18-19C higher ambients based on GPU, system, vreg temps (not even arguable...1 fan in its position is irrelevant for system or gpu temps, and apparently also for vreg temps, and Realtemp min temps are not relevant since it was reset), so correcting for ambients his block performed about 10C worse than heatkiller in his own test at lower overclock. And 20 C worse on vapors test at higher overclock. So really to me, testing matches up.
When I tried to cool my i7 overclocked with an old zalman with heatpipes (improves heat transfer with air) + about 8x surface area in copper fins over his block, It was still worse than 20C difference, though maybe newer zalmans would fare better.
When Vapor was testing it, I thought it would throttle the entire time at 100C....so again, I was amazed at the performance.
New revisions are already on the drawing board that will certainly perform better. The problem I am running into is time. I only have 2 hands and the clock is ticking.
I would have liked to seen these tests on stock clocks as a baseline before an overclocking test. Maybe the tests would not have reflected as poorly and might have reflected data closer to mine?
watercooling is about removing a much higher heat load than aircooling can do. stock clocks work fine and almost always near silent with even stock cooling.
BTW, We all bow to your superiority. Your crystal ball is all knowing.
I don't ever quit.Quote:
Originally Posted by nikhsub1
Why? The higher the heat load the more pronounced the performance differences are... this is what we want... why would we want tests that numb the numbers?Quote:
Originally Posted by nateman_doo
I didn't need a crystal ball to know the outcome, that is my point. I am glad you never quit, I don't think anyone here wants you to quit, least of all me :)Quote:
Originally Posted by nateman_doo
IMO 20 degrees under the "top" performing blocks is a huge accomplishment for a man with a drill press in his basement. I cant see what you will come out with with your new tools/resources. Keep up the good work natemandoo.
ReggieSanchez proud supporter of camp BlockMan
Errrr your thinking is half correct... so i give you half credit.Quote:
Originally Posted by BeepBeep2
Mass is not important in relation to efficiency of medium which is used to pick up in relation to surface area.
In water, where you have a greater absorbtion on the water side vs disappation of heat on the air side, you want as little mass on the copper as possible, and more surface area so water can pick up.
If you have more mass where water picks up or disappates, you have excess.
This excess is usless in a ambient system, because of the difference in mediums.
Now in a Chilled environment mass is important.
Why? because your not sheding heat which an ambient system does, you are pounding cold.
So you want MASS, to keep the block Cold....
Anyhow Nate i was talking to vapor, and we both said it would be interesting if you threw a TEC in between the fins and that top plate.
If you had something to pull sub ambient, and the fins cooling the hotside on the tec you might have an interesting product.
Because if i calculate correctly, you can probably throw a 60-90W tec on a 12V power rail without much problems.
Then you will have 45-65W of cooling on your CPU component, and your rad can shed the excess heat off.
You will lose effiency in your system, however you would get some interesting results... especially on a long idle period followed by quick load sessions.... because your prechilling the block each prefront load with that MASS.
You wont pull sub ambient because your rad would turn into a heater if your water went below ambient.
But you would pull work off your rad, and probably get closer to ambient.
+1Quote:
Originally Posted by nikhsub1
Going to stock clocks will just compact the data. Instead of 20C behind, it'll be maybe 10C behind, but C/W will remain the same.
And it doesn't change the data from the overclocked tests.
For the record, NaeKuh is the one who is gung-ho on TECs, I'm not as much :p:Quote:
Originally Posted by NaeKuh
But I do think there is potential to airsink a TEC to a waterloop, for simplicity's sake (dual loops aren't required).
As for mass of the block, with an extended load mass becomes irrelevant. Higher mass resists change, but it doesn't change the eventual equilibrium, only how long it takes to reach equilibrium.
im gonna disagree with the TEC too, how would you not waste the energy by cooling down the water thats just going to a radiator?
because Naekuh is just looking for someone to start playing with TEC's so he can drop his ever lasting TEC project that has been static for 3yrs on them. :p:
TEC's are an interesting proposition I agree, but I am still hesitant to go down that road...
Agreed. I will never be convinced that they are a 24/7/365 solution...Quote:
Originally Posted by skinnee
water in a closed loop works on the principle of equalibrium.Quote:
Originally Posted by Manicdan
This has been proven with math and with lab tests from almost every tester we had.
Now you have TEC... which is something that pulls heat and moves it to another side.
It doesnt work on simple properties of convection and radiation, it pulls heat @ the cost of work, and moves it to the hot side.
Now the wonderful thing about nates's design with a tec on top is that he has a safety net.
Whats that? the a protection for him incase the TEC dies, or if his TEC is too powerful.
But at the same time it will be limited to the tec size and cooling potential due to block size.
SO now he has mass + something that pulls heat regardless of where its at + water safety net.
The TEC will charge the block to hold ambient... or as close to ambient as possible.
Under long and heavy load sessions the tec will become usless, i agree, 45-65W of heat is tiny.. but when were looking at a 180W to a 200W...
That would bring that heat down to almost dualcore levels... and thats a win in my book.
What does that mean? A GPU might be possible with a MCR220 with a CPU on the same loop.
^ Major Win for SFF builds...
We are STAGNET in water.Quote:
Originally Posted by skinnee
The next big migration for us will be TEC's on PWM Controller, so we can see cake and eat it too.
Meaning we get efficiency!!!!
Cam mark my words i am calling it NOW.
i still dont see much point for a TEC, its going to be like a small chiller, except cooled right above the cpu by heatsink. if your case is already restrictive on size, then the airflow will probably be double dipping and trying to cool the radiator and the heatsink with the same air. except now theres an extra 80W TEC to worry about too.
i still think the best thing is to use as little copper as possible for reduced costs, add in a few heatpipes since they give great heat removal for their size restrictions, and they allow for more usefull shapes of heatsinks. and a fan that when used with the hybrid block can push out about 100-150W without seeing 100C (it would be to protect from overheating during idling or light use, it would be tough to ensure that 218W can be removed by a heatsink alone)
TECs have two major, major issues.
First, reliability.
Second, power consumption increase vs. performance gain is just awful. Take what your CPU consumes, multiply it by 2.5x, and you get maybe 5-10% more performance. Just an awful performance/watt ratio.
What about low power (20w) air-cooled TEC's used in series used as water chillers?Quote:
Originally Posted by Vapor
ie. Rad > TEC > CPU > Pump > Rad :shrug:
That would be a decent way to do it, but even then the watts/MHz is really inefficient. :shrug:
and the initial cost for TECs is still very high, and those things are stupidly fragile. i think a all in one phase change is easier and produces better cheaper results then TECs could for a noviceQuote:
Originally Posted by Vapor
Its a good idea.... but the design isnt the best one:
1- it is a block with a lot of mass (I disagree with NaeKuh)
2- the 3 plates (base, top, fin) mess up the convection of heat
3- when welding plates of the fin to the top it may have not enough contact with the whole surface
4. fins way too thick and too close to each other
What you can do is (chek design):
1- craft a high performance base: no more than 5°C compared with EK Supreme HF
2- fins and top just 1 plate
3- thin fins and with more space between them
4- use copper tube as shown in the desing to place the fitting inlet and outlet.
:up:
.
i appreciate hearing that from you.Quote:
Originally Posted by nikhsub1
I am not a fan of tec's as others might be. i only breifly messed around with them. i am shifting focus on the internals. The new tools i aquired after i built this block will aid in the work.
reggie was right - this was built with a drill press & a cross vise.
This was done with hand tools as well... started out as an aircooler - This block is what Bill Adams copied to make the MCW6002... anyway at the time, this was a VERY good block.Quote:
Originally Posted by nateman_doo
http://www.overclockers.com/waterblock-roundup/
I have a few things ordered in the main to begin the ACB v1.1
When you say craft a high performance base, what do you mean craft it withing 5°C? Do you mean the same thickness?Quote:
Originally Posted by JASSAF
Quote:
Originally Posted by johnksss
nate, will you quit posting here till the test are done?
this nonesense seems never ending.
So I said what the outcome would be and I was proven right, is it still nonsense? Or are you nonsense?
no, i was pointing at you and your way of bringing it across as nonsense. ain't no one your kid up in here, regardless of whether you like the idea or not.
Quote:
Originally Posted by johnksss
really? correct about what?
did no one ask for anyones approval. so im not sure what your going on about.
I wasn't going on about anything - I was stating facts and most of you were basically telling me to get lost.
refer to upper post.
i never went on about the performance of the block..only about how you went about presenting your "facts"
nothing more nothing less.
Why the fighting? :confused:
If you have a grudge, which I hope you don't, at least keep it off the threads :)
make a base plate similar to HEATKILLER or KOOLANCE 360 (a flat base with microchannels), the sides and top of the fins will make contact with the top cover. A greater area of contact: BEST heat conductivityQuote:
Originally Posted by nateman_doo
high performance?: without the fan, the waterblock performance must be competitive with the EK Supreme HF.
I recommend copper plate: 5 / 32 thick (H 57 x W 57 mm). fins: 2.5mm and microchannel of .014 "Thickness
microchannels ..VIDEO: http://www.youtube.com/watch?v=As-LZAdpPeo
.
not fighting, sir.Quote:
Originally Posted by Vapor
just answer his question.
and why am i the one with a grudge?
I get what your saying now and its along the lines of what I was thinking to do. Before I start with micro channels, I am going to try my luck with 1/16". Once I get a feel for the micro milling I will find some smaller taps.Quote:
Originally Posted by JASSAF
Now, back to thermal conductivity... If copper has a much higher thermal conductivity then water, why do larger amounts of copper do so crappy at pulling heat from a cpu base?
You still need to disperse the heat from the copper itself :)Quote:
Originally Posted by nateman_doo
Fair enough. There is lots of internal surface area in the block, just not as much directly above the CPU.Quote:
Originally Posted by wez
i think vapor said it best, more copper is not going to hurt results, it just takes longer for everything to reach the balanced point.
for best temps you want to have the water as close as possible to the source
for best removal of heat, you want to have as much surface area as possible, and as much water flow as possible. the reason small blocks do well is from the accelerated water flow.
dumb question, whats your guys opinion of heatpipe direct touch, then have the water block above that? as long as the contact is very good, i cant see this causing to much of a problem with removing heat from the water, while it can have a very good heat removal from the air
hmm TEC project sounds interesting...Quote:
Originally Posted by skinnee
I reckon it did pretty well for what it is. If you dropped a rad off the test bed the results would close up - the test bed itself is magnifying the difference by reducing the effect the active fan cooling would have in the average system.
Its never going to beat a high performance block in a system with minimal deltaT, but if you were adding a gpu to a single radiator loop it could well make a difference and even end up in favour of Nate's block depending how silly you were being with your heatdump. As you increase the load on the rads this block should start to pull back towards the others and may at some point overtake them.
Certainly not the block for my systems, or most for people on here (and raging in this thread), but don't dismiss this flat out as it certainly has its place. If you can't add to radiator cooling for whatever reason this block could certainly help you out. Well done on the first revision - I reckon with some hard work refining you could knock 10c off those temps.
Those internals could really be worked on a lot. Check the OCZ Hydroflow internals and see whether you reckon you could solder the top cleanly to the pins. Thats the best way I can think of to get both good contact to the top for cooling and acceptable heat transfer to the water. God knows how you'd direct flow though. Keep it simple to make for now as if you are really serious about doing this youll be on revision 10 before revision 2 is completed ;). The base is where you're going to get most of your improvements, but for god sake keep it original, everyone uses micro pins or micro fins these days. Besides youll pretty much junk youre heatsink if you do it that way. Ever felt the top of a copper HK3 with a 4.6ghz i7 under it? Its under 30c when the cpu is in the high 80s. That type of base wont shunt you enough heat to the heatsink (unless you intend on it being merely a failsafe).
Keep the heatsink style, the fan downdraft directly on the block top will be far more beneficial to you than switching to any tower/heatpipe type of design. Maybe look at a heatsink more like the old Alpha PAL8045 or the swiftech MCX462 instead of the skived copper style? Pins and a large base are the way to go when you are cooling a huge contact area. Fins are good at transferring heat along their length from a small central heat source.
I think you can work on it, but aiming for the top blocks is out of reach with a hybrid block design - but you can make that design useful in most average joe loops. A have a mate who can only fit 1x220 rad for his Q6600 and wants to add a GTX260 to the loop. I said no way, with a refined version of this that could be possible. It'd certainly be a cheaper and easier route than new case extra rad and fans anyway, even if not ultimately as good at cooling.
Glad you're not easily discouraged, it's gonna be really hard work if you take it on :).
Its because the copper is a solid so the heat must travel through the mass of copper, when water has the fluid speed advantage. At 1.5GPM, the water is moving a lot faster in those tiny block channels (v=q/a) than copper can move it as a solid.Quote:
Originally Posted by nateman_doo
Think about holding a 1 ft piece of copper pipe in your bare hand and heating one end with a torch. It would still take a good 10 seconds or so for the heat to make its way to your hand before you have to let go. Now think about the speed of water in that pipe...its much faster...now think about the tiny cross sectional area in the block at the nozzle inlet...its even higher...
In essence the waters thermal conductivity is increased by its flow rate.
There is more going on there, but that helps me make sence of it a little.:cool:
excellent analogy. I like it. Shooting for 1/16th channels at the moment. See if I have the steady hand.Quote:
Originally Posted by Martinm210
Nice!
I've been thinking about slitting saws. I wonder if you can make a custom arbor to hold a series of slittings saws together spaced apart?.
I had trouble with the commercial arbors before and made my own butb it was only for a single blade and took a long time. Seems like a bigger mill and large diameter arbor should be capable of running multiple saws together. Might be a good way to speed up the otherwise painfully slow prcocess. Not sure but you might think about it..
Most efficient way to mill slots like that is a horizontal milling machine with all the cutters ganged on the spindle.
Ahh...never saw one until you said that and I did a search.Quote:
Originally Posted by Waterlogged
Sure would be nice..
Until I win that lottery round, I'll try making my own arbor sometime. These slittling saws are getting pretty small for the type of cutting we're talking about. Seems like if you could find some with a large arbor, you should be able to gang a few together on a vertical mill?? Even if it was just a few at a time, sure would be faster than one cut at a time...probably more precise too because the saws would be pushing the microchannels against each other which should cut down on fin distortion:shrug:
theres alot of ways to make it once you get the right cutting tool, not sure if you can make your own very easily, would make it one quick cut
Quote:
Originally Posted by Martinm210
:wierd:. .Martinm210 .:shock: :
that's crazy .... or just for heroes :shakes:
microchannels ..VIDEO: http://www.youtube.com/watch?v=As-LZAdpPeo
.
This was Naekuh's TEC block in the works...it took forever...twice!...lol!Quote:
Originally Posted by JASSAF
http://img84.imageshack.us/img84/839/ntc4rc8.jpg
No pain..no gain..:D
Serously though, I don't see why you couldn't, especially if you made an arbor that was keyed for these types of saws and oversided in outer diameter to support the blade with the same oversized spacers.
http://i8.ebayimg.com/01/i/001/2f/d3/b5da_1_b.JPG
That video looks cool on cnc, but try it by manually milling at much lower RPMs. Cutting one block like an EK supreme will take you all day just to cut the base and that's assuming you don't screw up one of the steps and have to start all over again.
If I had plans to make more than one, I'd sure try it. It's not all that different than running a dado set on your table saw and I've done that plenty. On the mill, I consider some of the larger endmills and facing mills much scarier to run than a slitting saw. While I have had those grab/walk/ruin a work piece, I've never really had much incident with the slitting saws. My biggest problem was getting them to run true on a low cost arbor....in the end I made my own. It worked pretty well other than being extremely slow, although I'd probably do it differently now that I tried it once. I think you want the arbor to be even larger in diameter leaving just enough saw blade to make the necessary cut depth...this would go along way in preventing flex/vibration.
Just makes me wonder what some of the current mass production manufacturers are doing. I'd bet some of them have some sort of gang setup...but they never share the juicy efficiency bits with us garage builders..:)
There has got to be a better way to cut microchannels. Horizontal mill would be ideal, but what's the next best thing without a horizontal mill or CNC?
JASSAF, :eek: I cringed watching that video, speeds and feeds are totally wrong for the tooling and material. . .probably programmed by CAD/CAM. :shakes:
Martin, There is an attachment for Bridgeport type vertical mills that allows such milling, it's a bit of a pain to set up though from what I remember about the few times I was forced to use it because the real horizontal mill was already booked. For non Bridgeport's, I'm not aware of anything commercially available. Even with a regular horizontal mill it can be tricky to get consistent depth of from all the cutters, with an arbor floating freely on one end it'd be near impossible.
Why not stack some of these up to make one? Use spacers the same size as the blade with. I am curious where you find those saw blades and if you can stack a few of them.Quote:
Originally Posted by Martinm210
I have an R8 collet.
Hmm, how thick is this blade? I wonder how hard it is to get microchannels as thin as those in current top blocks using such design? (like EK HF/CPU-360 ..).
BTW, nateman: if you intend to use microchannels yourself (and imho it's way easier then trying to make thin micropins), you might experiment with depth of microchannels. Increasing it let Supreme HF show way less restriction compared to old Supreme. And that also probably reason why CPU-360 with shallower channels is more restrictive too then HF..
You can also dig arround forum. I recall seeing one DIY block here that had rather close performance to EK HF .. hadn't bookmarked/subscribed that thread though :(
Found some information on those slitting saws. Also that you can't stack them. Wouldn't it be sweet to stack a few and make one pass. There are probably too many variables when doing something like that.
Quote:
Also known as metal slitting saws, these cutters are designed for thin slitting and slotting jobs. They are thinner than staggered-tooth milling cutters and don't have side teeth. Slitting cutters are ground concave for free cutting through the workpiece. When choosing a cutter, consider the diameter and the number of teeth. A cutter with a smaller diameter provides better rigidity. To determine how many teeth your cutter should have, follow these guidelines:
• Always have at least three teeth in the cut.
• Use more teeth for thin and hard materials.
• Use fewer teeth for soft materials and larger sections.
Note: Use cutters individually. Do not gang-mount on a single arbor.
General Purpose High-Speed Steel Cutters— Use for cutoff, slitting, and slotting applications. Cutters with a 2 3/4" dia. are also known as screw slotting saws; other diameters are also known as plain metal slitting saws. All have a standard keyway. Cutter diameter tolerance is ±0.015"; thickness tolerance is ±0.001".
Smooth-Finish High-Speed Steel Cutters— Use on very thin materials (wire, tubing, and extrusions) that require light delicate cuts. Also known as jewelers' slotting saws. Cutters do not have a keyway (unless noted). Cutter diameter tolerance is ±0.015"; thickness tolerance is ±0.001".
Solid Carbide Cutters— Last at least five times longer than high-speed steel. Cutters can be used for climb (down) milling versus conventional (up) milling for better tooth engagement and part finish. Cutters do not have a keyway. Cutter diameter tolerance is ±0.015"; thickness tolerance is ±0.00025".
R8 Collet:
http://www.mcmaster.com/catalog/116/...e/2807ap3l.gif
standard 1/2"
http://www.mcmaster.com/catalog/116/...e/2807ap1l.gif
and the general purpose slitting cutter:
http://www.mcmaster.com/catalog/116/...e/3061ap1l.gif
I will pick up the last 2. The R8 collet is more expensive then the 1/2" shank arbor.
Nate, Pick a general purpose saw with a 2¾" or 3" dia and a 1" arbor hole, then use this arbor. I know it's expensive as hell but you'll get much better results than the 2 in those pics.
6th time repeating this statement in the same thread...Quote:
Originally Posted by Martinm210
guys... theres a reason this statement has been repeated 6 times.
ok heres a question, if i take a normal waterblock and replace the top with a copper one thats horribly tall and just a solid brick of copper. would that hurt anything? the flow would be identical, the base would be identical, but extra heat would be absorbed by the top. i doubt this would alter the temps, infact i think it would help for a few minutes, but after an hour would be the same as the original.Quote:
Originally Posted by NaeKuh
i think less emphasis needs to be made on how much copper is used, and more about the distance between the heat source and the water's surface area
Yup, it would help store the heat the first minutes, and then slowly releases it to the water. Like a giant heat capacitor.Quote:
Originally Posted by Manicdan
7th time repeating this:Quote:
Originally Posted by Manicdan
As martin and i have said 7 times again.Quote:
Originally Posted by Martinm210
Your difference in mediums is gonna hurt you up the butt.
Can it work? sure it can... but as i said your doing a fine balance eq of the water side in pickup vs air side at pickup on cpu block and release.
Water has such a high specific heat, that's why you dont need a thick honking piece of copper.
You in short DONT want a temp gradient forming inside the copper piece or flow of water itself.
Which would happen in tall pin setups.. and big mass plates.
Also for air water disapation, which you are looking at, not collection, your igonring the fundamental of radiators which this states:
http://www.xtremesystems.org/forums/...1&d=1144713834
His base is not like flat tube of a radiator.
Its like a circle round tube, or a box tube... he lost a ton of efficiency just by having that red circle inside his box flow.
When blocks did have something close to that.. a big middle tank which had a temp gradiant, how did we fix that?
We introduced Turbulence... to get rid of the red spot.. but i dont think just throwing in a copper top would add turbulance.
All the tweeks vapor is doing to the blocks, are actual Base warps... where he is intentionally overbowing the base.
We went though all these paths for the current blocks you are seeing today.
We went from a flat base.. to pin matrix, to dips and injectors, then back to pins and injectors, and now uniform flat bases with uniform flow.
Ouch. for $180 that thing is crazy. I totally understand in buying good quality tools, But I have to get my feet wet with the other one. (plus I already ordered the 1/2 arbor already this morning) :shrug:Quote:
Originally Posted by Waterlogged
@NaeKuh, you didnt answer my question. you gave lots of info about physics but your not separating the statement "that's why you dont need a thick honking piece of copper" with which part. can you simply say that an efficient base is needed?
Those other ones tend to "hump"* the work piece after a while. I have one of those goods ones in my personal collection at home. I don't slot much but when I do, I know I'll always get a good cut from it.Quote:
Originally Posted by nateman_doo
* this means only part of the saw is doing the cutting as it running out of round.
So did I waste my money, or will it at least function enough to get a prototype out?
It's hard to say really, I've had several arbors that were fine for a few weeks, I've also had a handful that have humped right out of the box. It's not the end of the world in your case as it's not high production. When it does start humping, all you have to do is rotate the cutter on the arbor every few cuts to keep an even wear pattern going on all the teeth. I just prefer to not mess with it at all.
nateman_doo:
if it works or not a heatsink on a waterblock?..... That is the question :confused:
how to do it fast and affordable : buy this block (http://www.sidewindercomputers.com/hecu30ba.html)
designed a new cover and makes the performance test. :up:
my tools:
http://img293.imageshack.us/img293/1903/cncfinal001.jpg
nate... take your original block.Quote:
Originally Posted by nateman_doo
Flatten the fins out and try to get the top base thinner as possible.
Then create another base which u can add a fin section on to clamp a tec in the middle.
Base - Top - Tec - Topfin
In telling you... try it with a 90W TEC in the middle.
Cool the hotside with a fan, the cool side will constantly pull heat from the hotside.
You would get interesting results... i can vouch for it.
great idea, I find TEC's incredibly interestingQuote:
Originally Posted by NaeKuh
wow thats a great idea, i almost want to buy one and get something like a zalman 9900 and find a way to get the base off to expose the heatpipes and make my own design. i so wish i had more time and better tools to mod.Quote:
Originally Posted by JASSAF
Nice setup you got going on there. You know, thats not a bad idea going with the HK plate.Quote:
Originally Posted by JASSAF
Fyi, just don't buy one of the cheap ebay arbor. I did and it did do a very good job holding the slitting saw true. My 1/16th saw wobbled enough to cut a 1/8th inch slot. That's why I made the brass one from some 1" stock I had handy.Quote:
Originally Posted by nateman_doo
It probably depends on the saw arbor hole, bigger 1" sizes as waterlogged noted should be better. I thinl my teeth on my saws were too coarse also and my rpm rate too low. I used WD40 in a spray bottle for coolant while cutting.
Keep us posted on the fabrication...always interested to see more of the work while its in progress..thanks!
Martin
well, I will know if its worth damn by tomorrow. Nice thing about buying from McMaster is that its in my mail box the next day. I bought a crapload of 1/16" mills also, though I dread milling that many without a CNC to hit the "go" button & walk away.
Yes it would end up hurting performance. Why? Because as the post I quoted below says, the 'horribly tall solid brick of copper top' would pick up heat from inside the case then end up transferring that heat INTO the water having the exact opposite effect that you would want. Naekuh is right about one thing, we seem to be repeating ourselves over and over. I already explained that the top to any WB has ZERO effect on cooling in regards to size and material...Quote:
Originally Posted by Manicdan
Quote:
Originally Posted by Alexandr0s
There are no hard and fast rules for machining... you can do whatever you like as long as it works.
A lot of what makes a good machinist is creativity in tooling and fixtures. This could include "going against the rules" and gang mounting slitting saws... what they mention is stacking saws to achieve a thickness. This is a no-no becuase the blades will be forced apart by the pressure of the chips and end up causing all sorts of trouble. If you were to gang mount .014 saws with .014 shims though, you wouldn't have any problem at all as long as you took multiple passes.
Another reason they recommend against it is that most slitting saw arbors aren't set up properly for gang mounting... but if you make your own arbor, then things change.