Building a Lapping Machine

[Neomoses]

Update: I got my surface plate today, and I have access to an optical flat/monochromatic light source. This weekend I'll try to do some hand lapping on the new base and see how flat I can get. Then, sometime in the next few weeks, I'll try motorizing it and see how well it works. I'll keep this thread updated with my progress.

[Michael]

I always used a fly cutter... works great for me.

[RoboTech]

I have not tried this myself (yet), but I wonder what kind of a surface you could get by putting the finished baseplate up on a surface grinder? I realize you would have to use a small machine vice on top of the mag chuck and optimally should switch to a different wheel (?) for copper... But since most waterblocks are small and you just want to do a clean-up on the surface - maybe it would produce a better surface than finish cuts on the mill?

It seems like a well dressed wheel would give a nice FLAT and relatively smooth finish. Any thoughts? Not as good, equal to, or better than hand lapping on glass?

I bet some of you guys have already tried this???

[RoboTech]

What, no surface grinders in the crowd?

Well then, what about using a single-point diamond flycutter? I realize this would require a "tight" mill (spindle bearings and gibs/ways) for optimum results but should be a relatively easy final step for anyone who has access to a good vertical mill. I would think this process could produce a flat, even surface as good or better than average hand lapping?

Any diamonds in the house?

[Gulp35]

Neo Where and what is your flat surface?

[Neomoses]

Well, here's an update, and I guess a picture's worth 1000 words.


What you see is a hand lapped block sitting on the 9"x12"x3" granite surface plate. I'm trying to get an idea of how flat of a finish I can get by lapping by hand, so I'm gonna lap a few blocks and then go test them with an optical flat. We'll then have some before and after numbers to compare.

[BillA]

a picture may be worth a thousand words, but not that one

how did your lapping lessons come out ?
a picture of the fringes will suffice

but wait, let me guess
-> it had a pretty fair hump in the center (or thereabouts)
I know this because I too am expert at making humps - as is everyone else doing it by hand

sooo, I have started to build a DIY lapping machine
off eBay


it has a 5" diamond lap, but no problemo -> 8" here we come

note that while this approach may work out ok for me in a lab setting,
for production it would be about worthless

Swiftech has a custom manufactured 42 inch diameter Lapmaster
(Neo - you really should stop your twaddle about "first rate" this or that, you do not have a clue about what is being done by the 'big boys')

[bigben2k]

Neo,


for lack of an optical rig, you might consider taking a picture of the block's reflection of a perfect grid, then analyse the grid for linearity.

It won't be anywhere near as accurate, but it might be a good start. At worst, it would be a good theoretical exercise.:shrug:

[BillA]

you have outdone yourself Ben

please describe the specific procedure by which YOU could quantify anything at all by such methods

what 'tools' ?
how to 'set up' ?
how to 'calibrate' ?
how to measure ?
how to calculate the 'non-flatness' ?

a suggestion for the future Ben:

every time you get set to propose something,
ask yourself for the answer

you would have fewer posts, for sure

[N8]

I can tell you a truly subjective, non-critical way to measure the flatness. IF you do have a granite flat with the correct paper work to authenticate it's degree of flatness (and your granite flat is in near perfect shape still), you can get a rough estimate of how polished-flat your surface is by how hard it is to pull it back off your granite flat.

Notice I said 'polished-flat'. You can have a really flat surface that is not polished, but it will not suction-adhere to the polished surface of the granite flat. You need two very flat, polished surfaces. I have had polished-flat surfaces stick together that required me to pry a razor blade under them to get them off my granite flat. Remember, this is only an estimate, and gives you no real data. It is purely subjective, but does give you an idea of how polished-flat your surface is. This is just air (lack of air) suction between the surfaces.

You could try this with glass, but glass is not as flat as a good granite flat, plus glass breaks. Glass is really not very flat (in the scientific sense.) One side of a piece of glass is pretty wavy, the other is flatter. Has to do with the way they make glass.

[bigben2k]

Originally posted by unregistered
you have outdone yourself Ben

please describe the specific procedure by which YOU could quantify anything at all by such methods

what 'tools' ?
how to 'set up' ?
how to 'calibrate' ?
how to measure ?
how to calculate the 'non-flatness' ?

a suggestion for the future Ben:

every time you get set to propose something,
ask yourself for the answer

you would have fewer posts, for sure

Fair enough, I'll add details to this (ludicrous?) suggestion.

Needed: quad lined paper.

Rest the block at an angle, on top of the quad paper: the quad paper should be resting on top of the granite slab (available?), or a pane of glass.

The block's surface should reflect the grid.

Take a picture of the block's reflection. Have the film processed, and scan the picture (BMP preferable).

Open up an editor (heck, MS Paint would do!), and redraw the line over the reflection, in another color (i.e. red).

Look at the fringe on both sides of the line, to identify deviations.


Of course the quad paper is usually a 4 or 5 square per inch, so you'll have to repeat the procedure in diagonal increments of that reflection.


Error sources: The quad paper should be fairly straight. There may be an error from the camera's lens, as it may distort the picture, so I'd avoid using a macro feature. There may be additional error/distortion from the photo processing lab equip, similar to above (but should be minimal). There may also be an error from the scanner.

There will be a loss of resolution at many steps: transfer from film to paper/print, and from the print to the scan.

It would be a long process, and now that I type it out, I can see how innacurate it would be: at best, it might point out deflections that should be visually observable, but the quad paper idea would still be an excellent source to point this out.


Alternatively, there is metric quad paper out there, with a 1mm grid. It was easy to find, in Canada...

[N8]

Here is another subjective method that I use - BUT you must have good vision. You can hold the polished surface up and look at it at a very low angle, almost like you are looking at the edge. Have a low light source behind it (I like to look at reflections from outside light through a window). You don't want bright light. I can see the imperfections (scratches, flat/unflatness) pretty well, but I do have really good eyes. Not for everyone, but something you could try.

[bigben2k]

What about a mechanical inspection?

Attached a ball to a rod, mount the rod to a pole, move the block back and forth under the ball, and mesure the deviation at the other end of the rod.

As long as the ball isn't too heavy, it should minimize the damage to the block's surface.

[BillA]

12" Lapmaster
http://cgi.ebay.com/ws/eBayISAPI.dll...category=25279

$11.50 for the moment

[Alchemy]

Originally posted by N8
I can measure flatness mechanically with a digital micrometer down to 1 micron of deviation, but with optics you can measure down to about 1/100 of a micron.

That's 10 nanometers, and is really quite impressive.

How is this optical measuring done?

Alchemy

[myv65]

Well, I'm way late to this game, but I have a perspective that hasn't yet been mentioned in this thread. It's one that has been talked about elsewhere. Given its nature, I'm surprised no one has mentioned it.

The method of mounting will have an impact on flatness. You can measure a block on the bench til your eyes are all squinty and possibly know nothing of how it will look as-mounted.

Many blocks are mounted with screws to holes in the motherboard. This means you've got the die pushing up in the middle and four constraints at the corners. As a rough ballpark, I tossed some numbers into one of Rourk's equations. It isn't perfect, but close enough for roadwork.

Say you have a block ~70mm on a side with circular die (yeah, I said circular, the equation is shorter) of radius = 5 mm. If the (copper with E~15E6 psi) base plate is 5mm thick and responsible for carrying the load, you can calculate deflection vs load. At 20 lbf, this comes out to ~8E-5 inches. For the metrically challenged, that's about 2E-6 meters.

The situation is obviously different if you have a mount that presses down directly above the die.

If nothing else, at least consider "flatness" below this level to be of extremely limited benefit.

I ran this through in a hurry, so no guarantees to accuracy. . .

[BillA]

to which must be added the deformation of the bp due to the heat being applied only in the central area
- and of course this appraisal should account for the stiffening effect (if any) from the furniture on the backside of the bp

and consider the effect of the mechanical loading plus thermal deformation on the CPU (or CPU + IHS) as well
- then their interaction

aiiiiii

where is that ANSYS fellow ?

myv65
I trust you have noted that I am extremely careful to NOT characterize the effect upon cooling of the percieved flatness

[myv65]

Originally posted by unregistered
myv65
I trust you have noted that I am extremely careful to NOT characterize the effect upon cooling of the percieved flatness

Duly noted!

[chambochae]

what about vibro lap machines used for grinding rocks. They use loose grit.

http://manchesterminerals.co.uk

(site doesnt like linking, go shop, stone polishing, vibro lap).

comparatively cheap, dont know how well they work though.

[MadDogMe]

Link does'nt work for me...