Lapping waterblocks?

[Top Nurse]

Well I have been using a piece of glass with W-D sandpaper wet to lap some blocks lately. Been thinking of getting a 8" x 10" granite surface plate that is flat within .00005" to do this as I have been wondering just how flat a piece of glass really is. Anyone know?

 

[Homer]

probably flat enough to not notice a difference between that and your granite thing.

 

[BillR]

Well I have been using a piece of glass with W-D sandpaper wet to lap some blocks lately. Been thinking of getting a 8" x 10" granite surface plate that is flat within .00005" to do this as I have been wondering just how flat a piece of glass really is. Anyone know?

The surface plate is a good idea, but the granite idea bothers me. If you slip off your lapping material the granite is too soft and can become easily damaged.

You can get a cast iron surface plate where they sell machine tools, a bit less susceptible to surface damage, no pun intended.

I use glass as well, but I use a mirror. Using a mirror you can see any obvious optical distortions that would be caused by a non-flat surface.

I also test my glass plate by placing a drop of water on my cast iron plate and laying the glass on top of the cast iron plate. The water squeezes out in all directions equally and creates surface suction thus verifying it’s pretty flat.

To remove the glass you are testing “slide it off” don’t lift it, I have had one break that way.

Luck

 

[eumskickin]

If you buy a finished cast iorn lapping plate it will be $
if you want something a little stronger look for the pink granite plates (you can find them on ebay from time to time)
you could use pre-hardened ground flatstock if you want to do it on the cheap ( i have some left over when i leveled the tail stock on my lathe. ) look for the mom/pop places that make there own or order some made by timkin of starrett

go with 1/4 thick 3" wide and 18" long stip . place it on something flat ( like a steel set up table or granite block) and lap ontop of it and it will get it flatter than you need.

 

[Erasmus354]

The glass should be fine for lapping the WB.

 

[Homer]

The glass should be fine for lapping the WB.

I conquer!

 

[BillR]

If you buy a finished cast iorn lapping plate it will be $
if you want something a little stronger look for the pink granite plates (you can find them on ebay from time to time)
you could use pre-hardened ground flatstock if you want to do it on the cheap ( i have some left over when i leveled the tail stock on my lathe. ) look for the mom/pop places that make there own or order some made by timkin of starrett

go with 1/4 thick 3" wide and 18" long stip . place it on something flat ( like a steel set up table or granite block) and lap ontop of it and it will get it flatter than you need.

If you actually hand scraped your tail stock level I am impressed. I did an apprenticeship in a machine shop and was given two cast iron plates, a scraper and blue die. The end result was two finished surface plates, one of which I was allowed to keep.

Not many people do it that way anymore

 

[eumskickin]

If you actually hand scraped your tail stock level I am impressed. I did an apprenticeship in a machine shop and was given two cast iron plates, a scraper and blue die. The end result was two finished surface plates, one of which I was allowed to keep.

Not many people do it that way anymore

hand scraping is amazing, give a better finish then grinding and is becoming a lost art.

after i finished the tail stock i fould out problem #2, the arbor on the jacobs chuck was bent and i have to replace the chuck and arbor no. i think some one droped it when i was not there. now if its not bolted down i keep it locked up.

 

[DFI Daishi]

i use mirror glass as well, for the above noted flatness verification.

it's not perfect, and for a waterblock (coldplate, in my case) that is macined to better than 4 decimals of flatness, i doubt that lapping is doing anything to increase the block's degree of flatness.

i only lap when i tear things down and put them back together, to remove oxide deposits.

 

[Top Nurse]

Well the cast iron plate is totally out of the picture. About $100 for a granite surface plate as opposed to close to $250 for the cast iron I think I will do the mirror instead I got a block that definitely is concave.

 

[Homer]

whoa 100 bucks for granite for lapping? dont do it man, you would maybe get a 1* difference in temps, if that. not worth it imho

 

[Top Nurse]

whoa 100 bucks for granite for lapping? dont do it man, you would maybe get a 1* difference in temps, if that. not worth it imho

With a concave block?

 

[Homer]

maybe, theres no way to be sure unless you spend the 100 bucks on it and try. you can always flatten it to a degree with the glass, probably as close as will be noticable (sp)

 

[DFI Daishi]

Well the cast iron plate is totally out of the picture. About $100 for a granite surface plate as opposed to close to $250 for the cast iron I think I will do the mirror instead I got a block that definitely is concave.

now, i'll admit that this is going out on a bit of a limb, but if one of those german blocks, that are being sold at premium prices, turns out not to be all that flat........would you not expect some truely good customer service to get that block returned and have a replacement sent free of charge?

seriously, the cold plate that i had fabbed up at one of the small machine shops here in town is good to within the 5th decimal place in terms of absolute flatness. i only paid $60 CDN for that one-off part. given what A-C charges for mass-produced blocks, i think that you have some real reason for complaint.

 

[eumskickin]

did you get a concave Ac block as well ? was it as concave like the pictures i posted ? mine was of by about .002"

if only i had a mill with a 90' adjustable rotery table i could make some sweeeeeeeet blocks.

 

[Top Nurse]

The first XT I got was not flat either and slightly concave, but I figured it was the first production run and they hadn't got their setups done right so I just lapped it. I just got a silver XT block in and after your post I checked mine with a semi flat surface (side of a gage block) and it was concave too So I looked around on the AC site to see if there was some kind of quoted spec on flatness and not finding it I asked in their forum at this thread. As you can see they weren't to forthcoming about any info so I assume that either they don't want to talk about it or this is a normal block. When they won't say much I just figure it's up to me to make it right for myself. Besides I am not sending a block back to Germany to have it returned in the same condition.

Someone else also gave me this link to a similar thread about 6 months ago. My silver block appears to be in the same condition as this one. Nice finish, but not flat. When I started lapping it I noticed that only the outside of the block was touching. The center portion isn't even touching the sandpaper at 1200 grit. I'm not interested in pretty, just flat. So it is off to the glas store and auto shop to dig up some more supplies.

 

[Top Nurse]

So tomorrow I will get a nice thick mirror from the glass shop. My silver block is about .002" concave. So what would be the best grit to start out with? I tried at 1000 grit, but it is going to take forever at that grit. Also I should point out that this silver is considerably softer than a copper block, so please take that into consideration as well. From the bottom of the block to the screws it is about .010" so I would like to finish up at about .004" max material taken off. I also value a flat surface over a mirror look so a brushed look is okay.

 

[condac]

Thank god I'm home, and that home is hardforum.com. This is realy [H]core stuff. I never thought about if the block concave. My "friends" laugh when my other friend removed the text printed on the old AMD core just to get that extra flatness. I thought it was a good idea. If the surfaces isn't flat it will depend more on the thermal grease. Well I'm of to check my blocks flatness.

 

[wwparrish]

You can start at any grit you like, I would suggest around 200 grit for th initial work and then progress to 400, 600,800,1000. Look for an assortment pack of wet+dry.

Go as high in grit as you like, have time for. A final light touch up with jewlers rouge makes it like a mirror but as I am sure you know, its more important to get it flat than have a mirror shine.

 

[Bio-Hazard]

200 hundred is a bit on the rough side don't you think................ When I'm bored I might lap a block only if it's scratched up from a lot of removing and installing and sitting in a desk drawer................... But I start with 800 and finish with 2000.................

 

[zer0signal667]

I think it's prudent, in this forum at least, to mention that glass is a superCOOLED liquid
A sheet of glass that has been sitting in a window pane for the past 20 years is sure to have slouched a bit and will not be very flat. The mirror glass is a good idea - mirrors are likely to suffer the same fate after time, but at least you will be able to tell somewhat more easily.

 

[DFI Daishi]

well, i usually start with a 320-400 grit wet, and progress from there to 800, 1000, 1500, 2000 wet grits. once things are flatened out with the 320, which can take quite a while, then the subsequent passes with finer grits go pretty quickly in my experience. since silver is soft, maybe start with a 500-600 grit, rather than a 320-400, but it sould turn out as a faintly hazy mirror once you have it up to 2000 grit.

i never really worry about any special markings when i am lapping, since the difference between the smoothed areas and the unsmoothed areas is pretty visable to my eye.

as for the glass really being a liquid.......this one has been kicked around for years, and i know that there are textbooks preaching both sides of the debate. i personally think that glass is an amorphous cristaline solid, and that the evidence presented that old windows are thicker at the bottom results in older glass manufacturing techniques being unable to produce panes of uniform thickness, and installers having the presence of mind to put panes in with the thick side on the bottom, where it does not present a liability to the overall strength of the window.

 

[cnick79]

I don't see why somebody would spend the time and money on something for only a marginal increase. I can see giving your block a quick lap to clean it up after it's been used and/or scraped up. If your block looks flat then chances are you'll have good contact. If your block isn't coming in contact with your chip while using thermal paste then you have some serious lapping problems.

You're better off looking towards parts that will be more effective at cooling instead of trying to make a block perfectly flat.

 

[DFI Daishi]

I don't see why somebody would spend the time and money on something for only a marginal increase. I can see giving your block a quick lap to clean it up after it's been used and/or scraped up. If your block looks flat then chances are you'll have good contact. If your block isn't coming in contact with your chip while using thermal paste then you have some serious lapping problems.

You're better off looking towards parts that will be more effective at cooling instead of trying to make a block perfectly flat.

well, i can understand this point of view to a degree. the way in which you worded this makes me kind of suspect that you have not read the whole thread, since the OP has pretty clearly indicated that she DOES have some serious problems with her block not being flat.

howerver, i also understand the point of view for she who doesn't want to return/replace a really expensive waterblock from overseas. a little time and some elbow grease can fixt the problem, along with a few dollar's worth of sandpaper.

the thing that baffels me is that she actually continues to purchase from the company, given that they sent her a defective block once already.

 

[eumskickin]

if i buy any other blocks from them i am going to sharka's and bringing some tools to check the block ... i won't buy any thing else from them unseen/untested.

 

[cnick79]

well, i can understand this point of view to a degree. the way in which you worded this makes me kind of suspect that you have not read the whole thread, since the OP has pretty clearly indicated that she DOES have some serious problems with her block not being flat.

howerver, i also understand the point of view for she who doesn't want to return/replace a really expensive waterblock from overseas. a little time and some elbow grease can fixt the problem, along with a few dollar's worth of sandpaper.

the thing that baffels me is that she actually continues to purchase from the company, given that they sent her a defective block once already.

You're right I didn't read the entire thread. But

a little time and some elbow grease can fixt the problem, along with a few dollar's worth of sandpaper.

No need to spend any more money then you should.

 

[Vertigo Acid]

I think it's prudent, in this forum at least, to mention that glass is a superCOOLED liquid
A sheet of glass that has been sitting in a window pane for the past 20 years is sure to have slouched a bit and will not be very flat. The mirror glass is a good idea - mirrors are likely to suffer the same fate after time, but at least you will be able to tell somewhat more easily.

When with the urban legend die?!?!?!
http://dwb.unl.edu/Teacher/NSF/C01/C01Links/www.ualberta.ca/~bderksen/florin.html
http://math.ucr.edu/home/baez/physics/General/Glass/glass.html
http://www.spectrumglass.com/Library/ScoreArticles/BitAboutGlass.html
etc.
it is *not* a super-cooled liquid, it is an amorphous solid

 

[Top Nurse]

Well I suppose I can forgive any company that makes a mistake. Now if I ever buy another block with this problem from Aqua Computer then it will be time for everyone to put on hard radiation suits over in Germany. But being the cynical kind of gal I am you do have to wonder about something. Did they know about this problem and continued to ship them until the word leaked out by some crazy woman that posted eumskickin's pic on their site or did they not know about it until I posted the pic and my findings?

After I brought up the issue of the clogged jets and pin arrays you think they would have been going over everything with a fine tooth comb before shipment. The next shipment of XT blocks I saw all seemed to be pretty clean, though not stellar. In fact I was heartened by seeing the next shipment come in with little QC stickers on some bags of parts and such. Never saw any QC stickers on the machined parts boxes though Makes me wonder as to whether they even have a QC department? The Silver XT I reported on was just machined a couple of weeks ago and it was after I reported the first crop of problems with the clogged jets and pin arrays....

I mean all you really need is a decent size granite surface plate, a height gage, a dial test indicator, some ID and OD mikes, a gage block set, and various plug and thread gages. If you went with high buck stuff it might set you back only a couple of thousand Euro's. Every production run should have the first part totally QC before starting production and they should be randomly checking per QC protocols.

So will I continue to buy Aqua Computer gear? Of course I will as I like the design and some stuff they make just isn't available anywhere else. However, I will continue to beat their little buns whenever I think they are not paying attention to the things I think they should be.

 

[Top Nurse]

When with the urban legend die? it is *not* a super-cooled liquid, it is an amorphous solid

Well I can see your point, but I live in a building that was built in the late twenties. Much of the surviving glass from that era is definitely manufactured the old way and I think they call it float glass. I have heard that they manufactured it by floating the molten gass over something till it cooled. Thus with old glass you get varying thicknesses over the same pane. I kind of like it as it gives a slight disturbance to the view outside that I found pleasing. However, if you are going to use glass for lapping DO NOT use float glass

 

[zer0signal667]

When with the urban legend die?!?!?!
http://dwb.unl.edu/Teacher/NSF/C01/C01Links/www.ualberta.ca/~bderksen/florin.html
http://math.ucr.edu/home/baez/physics/General/Glass/glass.html
http://www.spectrumglass.com/Library/ScoreArticles/BitAboutGlass.html
etc.
it is *not* a super-cooled liquid, it is an amorphous solid

I guess it's all in the semantics then. Sure, it's solid for all practical purposes. However there is no phase transition that dictates a transformation from liquid to solid upon cooling or vice versa. There is a "glass transition temperature", but that is measured almost qualitatively. Are you aware of a phenomena called creep? Not the guy outside your window, but the slow deformation, or flow, of a solid material as a function of time, temperature and stress? It's fact that a crystalline solid can flow, so I don't think it's unreasonable that an amorphous solid could as well.

And Daishi - an "amorphous cristaline (sic) solid" is sort of an oxymoron, isn't it..?

Sorry, didn't mean to throw the thread off topic, I was just warning people of old non-smooth glass.

 

[Vertigo Acid]

Are you aware of a phenomena called creep? Not the guy outside your window, but the slow deformation, or flow, of a solid material as a function of time, temperature and stress? It's fact that a crystalline solid can flow, so I don't think it's unreasonable that an amorphous solid could as well.

Can't argue with that, because I know nothing of the topic. But statements made about glass panes being thicker on the bottom as a result of them "flowing" are wrong. The thickness difference is a result of the process used to make them, as Dashi stated.

 

[Top Nurse]

I can't seem to get any decent info off the AC Forum and I don't have a depth mike or a 0-1 throat mike to gage the thickness of the bottom plate. So do you think it will be close enough for government work to just use my dial caliper to check as it only is graduated in .001"? Anyone have an idea how thick these nozzle coolers are on the base as I have heard they are quite thin? Do you think I can reasonably take off .010" without causing problems?

As an aside I think I can guess how this block got concave. If I was making this I would mill the block bottom first and use that surface as a reference point in the rest of the jig. So they flip the part and hog out the inside, slot the pin array, put the o-ring groove in, drill, bore and tap as needed. Voila its done! So assuming they started out with a flat block the only logical conclusion is that they warped the block while machining it. Either that or the spindle bearings are shot to hell, which I doubt unless they bought a used CNC mill. And since the TwinPlex's are all flat that just leaves the first supposition.

 

[wwparrish]

While it might provide little comfort, since I have followed water cooling and computers in general I have never heard of anyone sanding off by hand so much material that the block would leak or deform under normal clamping procedures. However I do applaud you carefull approach to the issue.

LOL the glass thing is a HOOT, a nice BRAND NEW 8x8 pane of float glass is like 99 CENTS at Home Despot, Lowes, corner hardware store, etc. The main problem is finding a guy to cut it for you, if they dont have one already made. scheez. And its damn flat enough.

I got a jewlers faceting machine sitting here Model 8826 by MDR manufacturing Los Angles CA . It is designed to, and does, put the finish on diamonds.and provides as flat and smooth surface as one could want. Its too much trouble to set it up for a large heatsink, I use a plate of glass, Of course I could be an idiot, it wouldnt be the first time someone pointed that out. /shrug.

 

[eumskickin]

i mesured the thickness of my block after laping and i got 0.0341" with my 0.0000"-0.5000" no. 232 mic from starrett (tested it with some a grade gauge blocks and it was on the money) or about 0.866 mm thick

my block was 51.14mm (2.0133") x 51.02mm (2.0086") not square with a starrett 50mm to 75mm no 436

all my mesuring tools are starrett. i have a kenedy chest full of them + the big ones in there wooden box's ( i have some brown & sharp but there days are numbered)

 

[DFI Daishi]

I guess it's all in the semantics then. Sure, it's solid for all practical purposes. However there is no phase transition that dictates a transformation from liquid to solid upon cooling or vice versa. There is a "glass transition temperature", but that is measured almost qualitatively. Are you aware of a phenomena called creep? Not the guy outside your window, but the slow deformation, or flow, of a solid material as a function of time, temperature and stress? It's fact that a crystalline solid can flow, so I don't think it's unreasonable that an amorphous solid could as well.

we talk about creep a fair bit in geology classes. these days i just love driving by an overpass, and seeing a case study in action.......so yeah, i'll ammend myself that even the most solid of solids experiences some degree of creep under the force of gravity and with the changes in temperature that it undergoes naturally during its lifetime. i was under the quite istinct impression that the strength and ridigity of window glass was such that creep of any measurable magnitude would take thousands of years. then again, those are the points of view presented by texts that say that glass is not a liquid of any sort.

And Daishi - an "amorphous cristaline (sic) solid" is sort of an oxymoron, isn't it..?

quite right you are. things are cristaline or amorphous, and to mix the terms in that way is incorrent. i have, however, heard that terminology tossed around previously in this context.

so, to say it properly, glass is: a non-repeating, non-regular arrangement of regular, repeating crystals, as well as other impurities, in a solid state. properly phrased this time?

 

[DFI Daishi]

So will I continue to buy Aqua Computer gear? Of course I will as I like the design and some stuff they make just isn't available anywhere else. However, I will continue to beat their little buns whenever I think they are not paying attention to the things I think they should be.

all right then. i guess that i can afford to be a little harsher in my stance with DFI product, since there are plenty of other motherboard manufacturers producing boards out there with everything that i want for my next build.

 

[DFI Daishi]

But No need to spend any more money then you should.

qite right you are. also notice that i said that i don't understand why she would buy a second block of this type, in silver this time, as opposed to copper with the first, after the first block had a concave base. i don't see the point, and i would not want to support a company selling flawed product.

she has pointed out her reasoning, however that doesn't mean that i agree with her choice. it appears that you do not agree either. i think that is enough said.

 

[Top Nurse]

i mesured the thickness of my block after laping and i got 0.0341" with my 0.0000"-0.5000" no. 232 mic from starrett (tested it with some a grade gauge blocks and it was on the money) or about 0.866 mm thick

my block was 51.14mm (2.0133") x 51.02mm (2.0086") not square

Well I could care less that it wasn't square as commercial stuff like this is usually +/- .050 on non-critical dimensions. However, the fact that the bottom thickness is only 0.0341 tells me that in all probability they warped the base while cutting it.

 

[Top Nurse]

qite right you are. also notice that i said that i don't understand why she would buy a second block of this type, in silver this time, as opposed to copper with the first, after the first block had a concave base. i don't see the point, and i would not want to support a company selling flawed product.

she has pointed out her reasoning, however that doesn't mean that i agree with her choice. it appears that you do not agree either. i think that is enough said.

Well that is because I ordered the Silver XT block before I took delivery on the first copper block. I also didn't check to see how flat it was until I decided to sell the first block as it was a bit tarnished so I started lapping it and that was when I said it's a tad off, but it didn't seem like .002 and I didn't bother checking. After someone else reported a concave Cuplex Pro I then got serious about checking it.

I had made an issue of some other QC related issues by trying through proper channels, but that didn't seem to resolve the problem. So after the second time I pointed them to the the posts here on the [H] and it seems like they got a bit more serious about there QC. However, as I pointed out before I didn't see any QC stickers on machined components so this time I brought it up on the Aqua Computer manufacturer's forum where the whole world could see. Now they have said that all new products are going to be delayed while they fix their CNC mill. So while I feel somewhat tweaked I also took the steps to resolve the problem, which was all I ever wanted to do in the first place.

The measure of a company is not usually what they do or don't do, but with how they resolve lifes little issues. They are taking the proper steps by halting their production until they figure out what is going on. I also would be really suprised if anything else gets out of their factory again that isn't right. So if push comes to shove I could do with Innovatek for my reservoirs, pumps, or electronics and Alphacool for my blocks. But as I have pointed out I don't expect to have to go there.

 

[Cathar]

Well I could care less that it wasn't square as commercial stuff like this is usually +/- .050 on non-critical dimensions. However, the fact that the bottom thickness is only 0.0341 tells me that in all probability they warped the base while cutting it.

More likely it warped for two or three reasons:

1) Heat during cutting. When the metal is this thin, especially silver, it will warp from the heat of the cutting. One of the last base-plate preparation stages of the silver Storm/G5 base-plate is a grinding wheel prior to time on the lapping machine, and even here the grinding wheel does not always give a flat finish due to heat - despite the base being bathed in coolant the whole time.

2) O-ring compression during assembly. Again, with base-plates this thin, the compression of the O-ring seal will slightly flex the piece, and this almost always creates a concave flex. Evidence of this is achieved by measure the base-plate flatness before and after assembly. If non-flatness is increased after assembly, then it's an O-ring compression issue. I've seen this issue affect 5mm thick base-plates, and can only imagine how difficult it is to control with a 0.85mm thick base-plate.

3) Handling abuse during assembly. Silver is a soft metal as far as metals go, and in fact so is high purity copper (although copper is a lot stronger/stiffer than silver). Bending 0.85mm thick silver is pretty easy to do. As we can see with the XT's design, it has quite large areas where there is essentially no structural support due to the cross-cut pins. This effectively creates a ~30x30mm area of silver that is 0.85mm thick, and such would only require minimal effort to bend/flex. 1kg (2lbs) of pressure would likely be enough to flex the silver, and probably 5kg for copper.. Unless the middle plate that is bracing the tops of the pins is perfectly flat (see points 1 and 2 above), and the tops of the pins themselves are perfectly flat relative to the top edge of the baseplate, then the baseplate will warp under minimal handling pressure, and especially when mounted against a CPU.

The ultimate question though is just how much effect does this non-flatness have on performance. Thermal pastes are designed to work happily with 50 micron gaps (0.002"), and up to 100microns in a pinch (0.004"). Also remember that it's the non-flatness that's directly above the CPU die that's important. While edge-to-edge non-flatness might be 50 microns, the part that's touching the CPU may only be bowed by 10-20 microns edge to edge, which is something that a thermal paste can easily handle.

In my experience, slight non-flatness doesn't affect temperatures as much as it affects overclocks. While the overall CPU temperature may not vary greatly (0.5-1.0C), it's the CPU's overclock that tends to suffer the most, especially moreso with bare-die CPU's.