Quartz Surface Plate?

[PekkaNF]

So, I had a chance to take a closer look at my haul tonight.  Spoiler alert: Don't get too excited.

Good news is that I was able to measure flatness deviation.  Bad news is that it wasn't difficult to measure....

I know  I had the same feeling. Now, after few measurements/samples I have concluded: Generally glass (compared to a rock) seems (locally) more smooth and flat, but has greater error over a long distance + flexes like there is no tomorrow.

Have you measured diagonally and on each side how much there is error from level, like I did on Reply #28 top picture on this thread page? There is hope, if the plate is relatively true, even though it has hills and walleys....if it is inclined or twisted - ahem.

Pekka

[Manxmodder]

well a piece off glass will let you no watts hi and watt is not put engineers blue on the table then place the glass on top remove the glass the glass that has blue on is hi and the glass that has no blue on is lo if the blue is all over the glass then the top is flat

The only problem with that is if you put a piece of glass on a surface plate it might well be flat but if you put it on anything else - especially with a chunk of iron on it - it will bend and not be as flat as it was when you checked it.

Bedding the piece of float glass evenly down onto a piece of granite counter top with something like thin consistency tile adhesive or Plaster of Paris may work.

That way the glass is sufficiently supported to prevent bending/twisting distortion.....OZ.

[PekkaNF]

plaster of paris is not stable...shrinks forever

Thermal expansion of materials is different and on this scale it ads up. One material/piece if pretty sure.

I haven't found a relatively big piece of glass that is flat. We have a guy working here who used to make machines for a float glass mill and from discussion we concluded that this kind of accuracy is hard to obtain from a standard glass. You might get lucky and find a piece that is cut from the middle of glass (web?) and on the middle of the cycle. Edges are "crabbed" and restrained....you don't want from too close to edge, there is a little tilt. Not much but enough to ruin it for our purposes. When molten glass is cooled it's run over rollers - that produces waviness...you might see that when you run the TDI over the glass and see regular swing.

So, it might be enough to lap a lot of small objects and on many purposes, but I find it too hard to accept as a surface plate - unless ground and for a lightweight objects.

Pekka

[BaronJ]

I did try some off-hand "lapping" of just two pieces; a smaller one on a larger one.  I did this mostly to test the"feel" of the operation, which was interesting to say the least.  When I was using water with toothpaste (yes, you read that correctly) it was smooth in some spots and then just sort of "stopped" as though a corner dug in, or the friction or suction became strong very quickly.  It didn't slide smoothly as I expected, rather in a bit of a "jerky" motion.

While toothpaste is fine for lapping, maybe too fine for what you are trying to achieve, it needs to be kept wet otherwise as you have discovered dries out and grabs.  Fine grinding paste in oil works well but it still needs to be kept wet.

I did notice that the formerly shiny surface has begun to take on more of a matte finish.  I hope this indicates something beneficial, and not just some residual toothpaste haze...

Yes the toothpaste is starting to abrade the polished surface and take down high spots but bear in mind that the high spots on each surface will rub each other so you need a third plate to work against the other two.  Eventually you will get one surface that is matt all over.

[krv3000]

I see I will have to take sum pics at work we have a very very  large bit of glass for testing bed plates on press tools to see if they are flat the plates are normally 10ft long by 4ft wide by 4 inch thick  yes granted the glass  its 12 inches thick and gets recalibrated every year but the principal is the same  blue on the plate put the glass on and see watts watt  as for the type of glass to use at home go to the scrap yard and get ether a side window out of a panel van as they are tuff end the bit I use at  home is a  old oven doer in sum old car books to check if your cylinder head is flat they just say any peace of glass big enough  to do the job

[krv3000]

if you intend to bond the glass to a suitable top just use builders adhesive  comes in a variety of cullers and is made for the job 

[sparky961]

So, among the other things that life has required of me lately I've spent about 4 hours working on flattening one piece of counter top that I obtained.  Considering that when I started I was seeing a 0.010" swing in some places on the indicator, lots of progress has been made.  It's been going well enough that I've switched from the 0.001" indicator to the 0.0001" indicator to better discriminate the deviation.  I'm down to about 0.0015" max, with lots of large areas not moving the needle more than 0.0005".

Two words: Silicon Carbide.  That's all you need to know.   Maybe diamonds would cut faster, probably in fact.  But I don't have diamonds.  I did have an old green Silicon Carbide grinding wheel that was too small to mount but otherwise in good shape.  I thought "what the heck, I'll see what it does".  I used the side of the wheel to take down the high spots prior to lapping, not to mention leaving the abrasive grit between the two plates for faster cutting action.  I've been going back and forth between the stone and another piece of countertop.  I have three on the go now, and I'm using one of the larger pieces to span a greater distance across the largest one.  I had noticed some dishing after the first session, which I hope to rectify with the larger piece.

I'd be very pleased if I could get this locally flat to within +/- 0.0002 or so using this method.  I may start running into problems with deflection around that point though because that's about how much it deflects under it's own weight when you pick up an edge.

There's been lots of talk about bonding various things to others in this thread.  What about making a stack of granite with epoxy between?  I guess even something like builder's adhesive should be fine.  Differential expansion shouldn't be an issue because it's all granite.  The little bit of epoxy shouldn't affect that.  Rigidity will be increased significantly.

There are some pretty distinct stages one can observe:
1. Smooth - Two pieces begin rough and offer little resistance to sliding over each other
2. Grabby (very scientific, I know) - When using a liquid with grit, this happens when you're starting to get pretty flat.  There's strong suction that causes the plate to move with greater force or stop completely over irregularities.  It seems to indicate that you're gotten rid of most really high spots and are starting to get flat.
3. Slick - The big piece of rock you're sliding around begins to feel almost like it's floating.  It will just about spin on it's own.

My advice if anyone wants to try this is to have patience and perseverance.  It's not the most exciting thing to do but it does seem to be working.

[Manxmodder]

Looking good,Sparky. I'm watching this effort with interest as I do have a few decent sized pieces of this granite counter top lying around looking to be used for something......OZ.

[hanermo]

I went through this exercise, to align my scratch built VMCs linear rails, 2200 mm long.

I found a vertex ground CI  straightedge, about 150€, new in the UK. Accurate to 0.04 mm.
A better option is a granite long parallel, but they are about 600 for 1.4 m long iirc.
Still thinking about it.
Machine builders square spirit levels, == 0.02 mm/ metre, were about 70£. I got two, and they make aligning stuff quite easy.

Testing the straightedge on a real granite surface plate, 600x440 mm, proves that you can easily see errors of about 0.01 mm, with a light behind it.
Putting a cig paper as a shim on either end, shows the lift clearly.

Since you dont need perfect, testing from the center to the edges should get you close.
Its feasible to sand steel pretty flat, ime.
I used a big, heavy, large belt sander.

Marking with a spirit permanent marker pen and sanding it off, shows you can get decent control on maybe 0.02 mm depth or so.
The edges get rounded.
The rails are mounted, and very well aligned, except at one spot. Its either high, low, twisted, or wanders from the front rail in some direction.
Plan is to put a dti on the ram, and check for rail error in the 3 axes.
Then just use a screw-jack to push it straight, and then tighten the mounting bolts.

You can easily feel any error over 0.01 mm while pushing the ram back and forth. Asain, cig paper as shims proves this quite easily..
The blocks are strong, and wont come to any harm from hand-forces.
I used 35 mm linear rails and blocks.

Hope this helps.

BTW - Pics of the machines will only come once they are polished, for commercial reasons.
Machine is commercial quality, with commercial components, and costs, and I need to clean them up before pics.
Hope to get them polished inside a month.. but this is not a reliable estimate.

[sparky961]

I've had the chance to spend a few more hours rubbing two stones together and I'm surprised to say that I wasn't really bored throughout the process.  It's rather relaxing and very rhythmic; albeit very messy as well.  It's to the point where I have one fairly large slab pretty much floating around on top of the other with little resistance.  Actually, a few times it just about got away on me and slipped off the other side onto the floor!  These pieces aren't light!  It's a bit strange to spin one atop the other with almost no resistance.  Hours of fun, in fact.

Now, the two issues I'm struggling with are as follows:
1. Flatness doesn't seem to be improving beyond around 0.001" max deviation over about 6".  Lots of places sit closer to a few tenths plus or minus but not as many as I'd like to see.
2. How to translate indicator readings into "map" of surface, most importantly the high spots.  The way I see it, a + or - reading could mean a small surface variation at the indicator tip, a large variation at the tip, or that the base is sitting on/in a variation that's causing the tip to erroneously read high or low. 

How do you sort all of this out? 
Is it possible without the fancy gadgets and software that professionals use? 
Would 3-point contact of the height gauge base be of more benefit than the flat base that can rock a bit on imperfections?
Is this where it becomes easier to use a level? 

For #1, I'm thinking it could be just that I haven't done enough work on it yet.  But what I don't want is to keep working and not see any improvement.  Hence #2, I want to be able to ensure its getting better, not worse or just shifting the variations around.

[jcs0001]

Sparky:

Not sure if you've seen this.  Evan describes using a cheap laser level (capable of generating a horizontal line) and a tripod to mount it on to get an estimate of the flatness of a surface:

http://bbs.homeshopmachinist.net/threads/18917-Qualitative-flatness-testing-cheap-%28pics%29

John.

[PekkaNF]

There is pretty good way, if you have time and good level. I have some proof that there is method for one spirit level and for two....Any solid information?

Procedure is somewhat like this:
http://www.mate.tue.nl/mate/pdfs/1698.pdf

There must be somewhere a punch list instruction, but this should give a good foundation on theory.

Pekka

[BaronJ]

Hi Sparky

Would 3-point contact of the height gauge base be of more benefit than the flat base that can rock a bit on imperfections?

Use three ball bearings to stand your indicator base on.  If necessary a touch of super glue will hold them in place.
 

[sparky961]

Thanks for the new info and tips. I'll give an update once I've made some progress.