3 Points Use also a Lost Art

I lecture this when I speak at SME Events, teach in my class on how many people will lay a mill table, grinder table and scrape them with a straight-edge and King-Way type gage and when the machine is finished it cuts off or when they are scraping the mating surface they can't understand why that surface is not coming in. When you store parts, test parts and scrape parts you need to put them on 3 points to eliminate twists. Everyone knows you can bend a lathe bed when it is cutting twists and can be aligned by adjusting leveling screws. So you know cast iron bends and if you take apart a machine and store a table on wood, on a skid, leaned against a wall it will bend over time.

If you set the parts on horses, a wood or steel table you are assuming they are perfectly the same height, the concrete floor floor is level, etc. To eliminate all those issues you set the parts on 3 - points. If you have a bent table and lay it on a granite table thinking it will be OK, scrape it and then lift it off the flat granite the table will be bent. To eliminate the issue, you put it on 3 points and it fine.

Below is one of the things I show in my classes. We take a Bridgeport table and set it on saw horse, then I have students check the twist with a King-Way, record the measurements. Then have them put the table on 3 wood blocks and re-check it. It is always better and by a lot. So if a new scraper starts to scrape the table when it is already twisted it will be screwed up when it is done.

This was tested at the recent Bourn and Koch plant in Rockford IL where we had a rebuilding class.

Check worn Bridgeport on 3 points, check Monarch EE saddle on 3 points. Small shim under bar on left. Using a 60year old HK-100 mini King-Way to check dovetail parallelism.

You can see I am an experienced rebuilder and teach rebuilding and scraping. In my opinion you have to teach classes this way.
More pic's.
Pics L to R
My students in Taiwan testing a new VMC column on and off 3 points. Also on this column it is a Triangle shape and it is heavier on the bottom, so we did not use the 30% from the ends. We set the column on a pipe and found the balance point and divided that end in 1/2 and that's where we put 2 blocks and then slit the smaller endin 1/2 and put a block in the middle. The next one is where I am scraping the bottom of a Blanchard grinder saddle with the HK-200 King-Way with a longer cross-bar. Then the home made aluminum bridge using 2 Starrett Levels we made to align the bed of a bed type VMC I helped align and scrape the turcite a few years ago in La Crosse, WI for PMR a Practical Machinist Member. I did a thread on it back then too. Note the bubble is from one of the King-Ways Sheer Tumico made under license from my Dad.

Richard,
You have made this point in past threads, and I get the idea. Similar to the airy points under my surface plate. The thing I have not seen any one ask nor do I recall it ever being brought up is what happens to the Bridgeport table when you mount it on four points of the saddle(providing the center of the saddle slide way is relieved you get four points), get everything adjusted and place that machine back in service. Does the table still remain straight as scraped on three points?

If so can you explain why

Yes as the 3 points are self leveling. When you scrape a table you make it parallel top to bottom flats, same thickness, dovetails parallel and then you match fit it to the saddle. Many grind shops place the table on 3 major points and use perimeter adjusters so the outside ends don't sage when the stone hits the work. You have to have the gib fit right too so each end of the tapered gib has the same clearance.

Some extra info too:

Warren, I wanted to write about it again in it's own thread and so new member can hear about it. I used to think that the unsupported ends of the table bent the table and I still believe it happens, but not to the to the major extent. I had lunch several years ago a couple of times with Archie Cheda who explained to me how the iron gets peened or stretched when tightening the T-bolts. I doubted that but this old dog learned a new trick. I will attach a page out of the George Sleschinger's book where he discusses this too. I have seen many flimsy tables bend when the move out over the end and bends I see this because I set a precision lever on the able and the level changes. When I scraped those flimsy tables, I scraped the base ways concave.

When we grind the Bridgeport table, I like to see a spec of .0005" to .001".

When we grind the worn tables we set them on the 3 points as described above or on 4 parallels at 25 to 30% from the ends and move them until the 4 corners indicate zero. Then we put jack screws finger tight on the outside 4 corners and one in the middle on both sides. Then block around the table and mag those blocks down to trap the table. Then grind the table top flat. I also hit the top with a soft blow hammer before the last pass. Then flipit over and grind the flats and then set it on the edges and dress your wheel at 50 degs and side grind the rest of the flats and dove My friend Rick at A&D Machine Rebuilding mills the bottom on his planer mill.

Richard-recently I have been doing some investigation to modify a frame for a 72" x 144" granite table I picked up.

I even referred back to the note book from our class and everybody had a couple different ways to measure this.

Upon much further investigation I have found that for granite tables all documentation points to 20% in from the ends. So a bit less than what we had discussion on.

Richard King said:

This was tested at the recent Bourn and Koch plant in Rockford IL where we had a rebuilding class.

Check worn Bridgeport on 3 points, check Monarch EE saddle on 3 points. Small shim under bar on left. Using a 60year old HK-100 mini King-Way to check dovetail parallelism.

View attachment 175502 View attachment 175503

Click to expand...

I recognize those Monarch parts! They look horrible! And I need to finish them. Haha!

Cash, I have been using 30% for my entire career, my Dad used 30% for his career and it has always worked for us. To me I find that illogical and can't see how the center 60% doesn't sag. I am going to ask the professors from MIT to explain that to me.

cash --

The old US Federal Specification for granite surface plates -- GGG-P-463c, dated September 12, 1973 -- calls for the table supports for a rectangular plate to be " . . . located no less than one-fifth or more than one-forth of the length and width in from the ends and sides, respectively, with the exception that the single pad at one end shall be located in the center . . . "

It also requires, for circular granite surface plates, that the supports " . . . be located at three equally spaced positions on a circle with a radius of approximately 0.7 [times the] radius of the plate measured from the center of the base surface . . ."


Everyone --

There are times when a simple three-point support isn't adequate, and in those cases pulling "wiffletrees", which are also termed "balance beams", out of the bag of tricks can save the day.

At its simplest, a wiffletree is a beam with two supports that fit against the workpiece on the top side, and a single support on the bottom side that is midway between the top-side points. Wiffletrees are especially effective when circumstances demand that a rectangular workpiece be supported near its corners.

There is also a equilateral-triangle variant of the wiffletree having thee support points on its topside, and a single support on the bottom that is equidistant from the topside supports. The triangular variant works particular well on circular workpieces. The lower supports of the triangular wiffletree are spaced approximately equally, on a circle having a radius of 0.7 times the workpiece radius, with one corner of the wiffletrees toward the center of the circle. The other two corners of each wiffletree are, of course, going be at the vertices of a hexagon that supports the workpiece nearer its periphery than the 0.7 x radius.

Each of the triangular wiffletrees pivots to fit the workpiece, while the triangle of their lower support points fits against the supporting floor or table.

For a real eye-opener into the sensitivity of a workpiece's shape to the geometry of its supporting triangle, set up a rectangular workpiece on supports at three of its four corners, and measure the surface. Then move one of the three supports so that it is beneath the previous-unsupported corner of the workpiece, leaving the forth workpiece corner unsupported, before again measure the surface. Unless the workpiece is, speaking proverbially, as solid as Gibraltar, the difference in surface contours will most likely surprise.

And finally, a slick way to stabilize, without significantly bending, a workpiece supported at its Airy or Bessel points is to partially fill small Zip-Lock-type bags with freshly-mixed plaster of Paris the consistence of pancake batter, squeeze the air from the bag, seal the bag, and then slide the sealed bag half-way underneath the workpiece. Then, while the plaster is still quite fluid, put a light weight on the portion of the bag extending outside of the workpiece. The idea is to slightly pressurize the nearly-fluid plaster so that it flows into the portion of the bag beneath the workpiece, filling the gap between the workpiece and support, before setting to form a rigid chock. Plaster of Paris typically sets in about the time it takes to enjoy a cup of coffee.

John

cash said:

Richard-recently I have been doing some investigation to modify a frame for a 72" x 144" granite table I picked up.

Click to expand...

Did Starret ever get back to you about this? ( I am curious what they had to say )

John, a similar idea to the whiffletree are supports used for telescope mirrors....i wonder if anything like that has been used to support large but not so thick plates. My smallish 18 x 24 is 6" thick...so how thick does a 10' plate have to be to similarly hold its shape with three far apart supports?

Mcgyver said:

John, a similar idea to the whiffletree are supports used for telescope mirrors....i wonder if anything like that has been used to support large but not so thick plates. My smallish 18 x 24 is 6" thick...so how thick does a 10' plate have to be to similarly hold its shape with three far apart supports?

Click to expand...

It doesn't really matter IMO. The granite plate is lapped flat while supported by 3 points. As long as you support the plate on the same 3 points, it doesn't really matter where they are.

Now, if you toss a 5 ton casting on top of it, all bets are off.

20% is well established as airy points. It's actually 21% I think, but close enough.

Here is the math from an engineering perspective. The goal is to get the slope to be 0 (perfectly flat) directly over the supports. That would minimize the "droop".

https://static1.squarespace.com/static/510573ece4b0c3c7083749ef/t/51074d68e4b0176d1d470e19/1359433064483/Airy+Points.pdf

When I ground the table on my milling machine I set it up on three points on the grinder table. A magnetic chuck at one end and two points the other.

I would not have known to do it like that without the valuable advice written on this forum.

When we designed big weldments, we'd put in three support points for the machinists to set it on. They'd 3-point it, then clamp it down at a bunch of points. At each point, they'd use an indicator to make sure that the clamping didn't tweak the part. This gives a similar effect to the plaster of paris described above.

Spherical washers on the three points are essential.

And a clamp /unclamp cycle to help a part settle .

I have several sets that are of a known height and are homemade out of ball bearings.

25% on the spread places 0 stress on the centerline of the part (why it is the theoretical perfect spot)

Richard King said:

Cash, I have been using 30% for my entire career, my Dad used 30% for his career and it has always worked for us. To me I find that illogical and can't see how the center 60% doesn't sag. I am going to ask the professors from MIT to explain that to me.

Click to expand...

Richard-this was my FIRST thought with the plate. Unfortunately I can NOT find any documentation to support this.

John Garner mentions the US Federal Spec-GGG-P-463c, I found this posted somewhere on PM a while back and printed it out. It states to use 1/4th to 1/5th.

Right now Starret Trustone is using 1/5, or 20%.

I am struggling with he fact that at 1/5th there is so much open in the middle and subject to sag if you put some weight on it. This is where my thoughts kind of get lost and I really need to defer to an engineer who understands this sag.

My theory on this-if you use 1/5th what I am thinking is you will allow the plate to sag a bit. Once it is verified and the sage is corrected then maybe it will not sag much more??? How much can granite sag before it would break?? This granite is 16" thick. Trustone states the granite will not sag more than half of its tolerance when supporting a load of 50 lbs per each square foot of surface are of the entire plate when loaded in the center of the plate. This is right from the federal spec.

So if I read that correctly-does this mean since my plate is 72 sq feet, could I theoretically put a 864 lb 1' x 1' x "x" tall in the center of the table and the plate will only deflect half its tolerance?? If so this seems to be pretty rigid to me.

I bought this plate from a local rebuilder who is now out of business. They had the plate supported on 15 points!!!! They did this as they were concerned about sag as they would build and align 10,000 lb rotary tables on it. One of the employees now works by me and he said once they adjusted the levelers they would not touch it but then certify it.

I can agree with this thought process but I can find no data to support it. I want to have the option to move the plate and table if needed and if it is only on 3 points I don't have to re calibrate it.

Here is a picture of the plate when I picked it up and you can see how every foot on the base that goes to the floor then as well has a screw leveler under the table.

Instagram

ewlsey said:

It doesn't really matter IMO. The granite plate is lapped flat while supported by 3 points.

Click to expand...

agreed, if a different support was ever done it would be calibrated/lapped while being supported from the same device. For my size plate, the starret A 18x24 is 6 inches thick and lesser grades are 4 - why? Either that 50% increase in thickness was necessary to hold the tolerance or the marketing department is wagging the dog.

Remember I am discussing scraping or machining a mill or grinder table and not supporting a granite surface plate. I want the reader to understand how to set cast iron castings up to scrape them. On 3 points and not any old which way on horses or work benches.

I had a friend who is now deceased DR. Richard Visinor who was Engineer at Honeywell and he was part of their think tank. He told me that if you have a square part say 6" x 6" x X long you use the calculation .707 x length divided by 2 to find the support points. I was in his shop after he retired from Honeywell scraping a Jones & Shipman Cylindrical Grinder and I had his table setting on 3 points at 30%. He came over and said technically if I was inside a laboratory he would have to calculate the thickness, width, length, where the mass or heaviest areas of the part were, material to find the support points. He said for a guy like me working with rectangular shapes like a mill table or grinder table out in a shop 30% would work just fine.

If you look at the base of a Sip Jig Bore which is shaped like a T the 2 points are under the heavy end under the 2 upright columns and the single point is under the table end about 30% back from the outside edge.

Heald ID grinders set on 3 points on the extreme out-side edges and not at 20 or 30%, but the machine was built from the factory like that and as one of you pointed out if the plate or part is designed, built and rebuilt or relapped at that position then who cares where the points are. Meaning pads on a plate on a stand. Now if they were placed in several spots, lapped and then you decided to remove it from the stand and placed it on a new stand on new points I can see a real problem.

There are some exceptions to the rules as I pointed out above when the casting has heavier sections, like a rack mounted to one side and it changes the mass or weight of the part. Many times when I am looking for the spot to put the single point I move it from place to place finding a break even point sliding my King-Way. Move it say one inch sideways and the bubble moves out of read and then slowly move the point until it reads better and moving it until the bubble moves the opposite side. I am looking for the best spot for the single point. (hope that makes sense).

I instruct my students to always level/align a machine by following the leveling instructions of the factory so when a customer buys that machine and moves it into his shop he will pull out the manual and level/align it that way again.

I am a rebuilder working in a shop and not a scientist inside a laboratory and as I said before I have used 30% successfully for 50 years and as I said so was my Dad. I used this method on not only Bridgeports, I use them on Studers, Myfords, Drake, Moore Jig Bores, super precision machine tools, etc. I do add extra supports where it may droop or sag finger tight jack screws.

I am not debating with the engineers, just sharing my successful knowledge. Rich

PS: Cash I see you use Instantgram...what is that anyway?

WAYYY above my pay grade...but could the 30/20% thing both be correct but for different reasons?

As in lapped in at 20% , but set up for work at 30% as a compensatory measure to account for sag in use when you actually have parts/weight on it?

hard to wrap my head around it but I can see some logic there.

Is there a practical limit as to when to use 3 point mounting during scraping, and when not to? For instance, if I have a small lathe compound, say 8" in diameter, should this be supported on 3 points? Or as another example, a 36" length camelback with a large length to width ratio?

I think it's dependent on the shape. If it is a rectangle, as Rich was mentioning above, the 30% rule applies. If you were to have a rectangle with large outriggers at one end, this would change the way the deflection occurs as opposed to a simple rectangle. It depends on how the mass is distributed on the part. But we know how theory versus pratice works.

It sounds to me from other's experience that 30% is a close enough approximation for the real world.

Or as my Dad would say, "Good enough for the girls we go out with"