What's new
What's new

Looking for tips- breaking 2 tenths/foot

lucky7

Titanium
Joined
Sep 6, 2008
Location
Canada
So using a surface plate I can reliably, altho not fast, get to one tenth (or maybe better- I’ll just say a tenth given limitations of my equipment and environment) flat parallel and square/foot on projects that are less than 30” long and 12” wide. With parts that I cannot rub directly on a surface plate I’m at 2 tenths/foot and occasional false starts. Tips? Techniques? Yes, I know about maintaining constant temp, avoiding sun etc. Just wondering how guys rescrape jig borers and similar parts that can’t get to a surface plate.

L7
 
Once you come to the realization that nothing is flat, round or straight, and is usually made of rubber, you life will improve!

What are you using for marking? how thick? have you tried alcohol?

Peter
 
Yes, I know about maintaining constant temp, avoiding sun etc. Just wondering how guys rescrape jig borers and similar parts that can’t get to a surface plate.

L7

This made me laugh. Place I went to work back in the early 2000s had a nice stout little Deckel LK jig borer; guys were getting weird results off it. Noticed that after lunch it got direct sunlight for 2-3 hours... hahaha, mentioned it to the boss and a couple days later he'd blacked out the appropriate windows.

I also remember a friend in the states who worked on Devliegs, but they had a Sip #4 or so in the shop and the Sip guys came over to tune it up. This would have been the early 90s I guess. Just recall he said they wore white shop coats and were badass scrapers, not sure if they had lasers then but they fixed a few microns of error. Coolest thing is they had mad respect for the Devliegs.

If you dig into the Moore book there's a lot about temp control. I remember Forrest talking about some number of seconds or something to take a print, then a recovery time, scrape, wait, print, etc. Hunting tenths over large areas is a fascinating subject. Wish I could add more than anecdotes and curious what Richard and others have to say.
 
I have been following this thread.... There are many factors one needs to do to achieve tenths let alone millionths. I have been waiting to see if anyone was going to write. The experts here....I believe Lucky knows the answer too, but just trying to trying to back up his theory. I have seen him scrape straight edges. We are discussing this in my Facebook forum now too. Those Sip techs scraper to 40 PPI and .00005" per foot. I met them at John Deere in Waterloo IA while I was teaching classes there. Machine tool builders who go out and repair their factory machines have an advantage over someone scraping on there machines. Design of machine makes a big difference. It's easier to scrape a machine with a lot of mass that can absorb the heat or cold slower. They sit on 3 points compared to say a Bridgeport mill that were designed to only hold .0002" per 12" accuracy. So Stan what are you trying to scrape?
 
What am I trying to scrape? Anything I can get my hands on. Only partly joking- I’m trying to up my skills to rescrape a 8x18 surface grinder and eventually a jig borer. Both are waiting in line in the back corner. B-ports, a small T&C grinder, and lathe parts (not a bed) have been successfully done, as well as the usual straight edges, angle plates surface plates etc.

My reality is that I’m working in an insulated heated garage. Not a temp controlled CMM room. Maybe I’m on a fools errand with this limitation, but am still trying to push my ability. Temp control has been careful waiting hours before printing after getting into garage as well as care avoiding heat input from hands. Two ceiling fans are being used to try to equilibrate temp.

Mostly use thinned canode in the winter (with closed windows won’t use solvent- hadn’t tried ETOH) and have been able to get good prints off the surface plate

Three point mounting on stable substrate is understood. I do have a darn good concrete floor on piles (reality of living on top of old lake bed- but gives a remarkably good floor) and have done tests on where to stand to get repeated same level readings.

Thanks to anyone for tips.
 
How good is your surface plate? A surface plate is usually produced convex. Exactly opposite what you really want for a scraping master. What do you get when you print two straightedges against each other?
 
Last edited:
This is an interesting topic.

I'm currently scraping a lathe. I'd like to think I've reached .0001"/ft, but I'm probably just fooling myself. My measurements indicated that I'm in the ballbark, but maybe I'm taking the wrong measurements.

Scraping is all about precision metrology. If you can measure an error, you can scrape it out. So then the question is, how do you measure errors in machine ways?

For example, a lathe bed:
You can do a lot with just a precision level. It can measure .0001"/ft, maybe .0002" with some confidence -- but for some measurements it can give us the total error for the whole machine, regardless of length.

Autocollimeter?

Set the whole lathe bed on a giant surface plate?
 
gbent, thanks for comment. Surface plate was A grade 30 x 48, bought new, only used by me so very light usage. I don’t own an autocolimator, or repeat o meter, so by measuring on a grid with Starrett 199 master level it is in tolerance- this is best as I cam tell- perhaps not good enough technique to claim grade A plate. Will defer to experts on this.

A surface plate needing to be convex to be in spec is new to me. I think I understand what you are saying, that a plate will bend ‘flat’ to a better tolerance with load. Problem is how much load? I thought, and correct me if wrong, plates were in spec for a max load for a given size and thickness of plate and were made ‘flat’ without weight on them.

Have tried a 48” straight edge against 60” both scraped on this plate. Good result. Will try smaller SE’s soon.

L7
 
A Starrett 199 is a 10 arc second level, or 48 microinch/inch. Your total tolerance is 336 microinches for a grade A plate your size, or a 7 inch distance at the least graduation on your level. There is no meaningful calibration a 10 second level can do on a surface plate.

Even a single electronic level is not useful to calibrate a surface plate, as your moving around the plate will affect the level. The plate must be checked with dual level heads, set up in a A minus B configuration so errors due to slight movement will cancel each other. This would be done with a 1/2 arc second level or better.

With the levels you can check the accuracy of the straight edge directly.
 
Ok, the best I can do here is my big gun, a Riken 0.02mm/meter level. If my math is correct that is 4 arc seconds with a length of the level base of just under 8” (sorry for mixed systems). From what you say, gbent, not sensitive enough to qualify my surface plate. I’m left having to assume the surface plate is still in spec. Perhaps a dangerous assumption, but I’m not doing this for a living and have nobody to answer to but myself. Comments about using the 0.02mm/M Riken level directly on a straight edge?

Could you comment about what you meant by surface plate needing to be convex to be in spec?

L7
 
@gbent, Robin Renzetti has several videos about lapping in a surface plate. He is well aware of the issues mentioned in your second paragraph above, and is using a technique using weights to compensate for the movement of the level tilting the plate. Basically, he's using a level and an identical weight that are moved symmetrically along the measurement axis.

I'd be interested in your comments on this technique. His description is in his video here:

https://youtu.be/6nY7uW1uG9s?t=1693
 
A mistake on my part- I work exclusively in imperial measurements in my shop- didn’t realize before doing the math this morning that my metric Rikon was 4 arcsec vs 10 arcsec of the Starrett 199. No where near 1/2 arcsec. Still thinking about gbents convex comment- do you mean that with use/wear that’s generally in center of plate, the plate will become more flat before going out of tolerance. Same thinking in slightly relieving center of a Starrett level?

Have also been reading up on how accurate a guy’s eye is in splitting graduations on a level. I think I can reliably split into quarters. Is this acceptable measurement practice?

L7
 
I eliminated my comment about a plate needing to be convex. I was told that by someone I believed in several decades ago, but I can't find any specification to back it up. I may have mis-understood. The speaker may have meant that a properly resurfaced plate will be convex, but it doesn't seem to be a requirement. Its easy to understand why a plate should be convex, as deflection from weight would improve the flatness. Wear would also improve the flatness, just like lathe tailstocks are left slightly high.

As for the youtube guy moving a counterweight around the plate? I really think he is picking fly specks out of pepper. I have seen systems where people moving around the plate (or machine) are a source of error if not canceled out. And I don't think there is any way someone could calibrate a plate with a spirit level without moving.
 
Ok, the best I can do here is my big gun, a Riken 0.02mm/meter level. If my math is correct that is 4 arc seconds with a length of the level base of just under 8” (sorry for mixed systems). From what you say, gbent, not sensitive enough to qualify my surface plate. I’m left having to assume the surface plate is still in spec. Perhaps a dangerous assumption, but I’m not doing this for a living and have nobody to answer to but myself. Comments about using the 0.02mm/M Riken level directly on a straight edge?

Could you comment about what you meant by surface plate needing to be convex to be in spec?

L7

I'm currently using a 200mm x 800mm granite spotting plate in conjunction with a 0.02mm/meter level to tweak a reference surface and I made the foul error of setting this column with the three point cradle directly over a crack in my slab. Yesterday I managed to nail the corner to corner twist test dead nuts to a Tenth over the full surface which the level confirmed however my slab is giving me major troubles. the surface is still dead nuts on but the slab shifted with a change from my datum in the realm of about 0.0004" overnight and since it started raining my slab appears to be moving quiet a bit. the slab at my factory is in a pretty poor state and I have serious problems with my lathe twisting if I juggle machines around so its to be expected I guess. I'd ideally like cut out sections and poor dedicated/isolated foundations for each machine when I sort out my layout

PXL_20210107_072659813.jpgPXL_20210107_072659813.jpg
 
Glen, it depends on the use of the plate. If your laying emery cloth on it like most machinist do. Most misuse all machines by putting short parts in the middle. On inspection plates next to a grinder, they set the part in the middle to measure or sand and as time passes the worn area is in the center. It wears more in the middle so we would scrape it high in the middle like a bullseye a few tenths high in the middle. If I was using it to scrape a table top I would scrape it flat.
 
gbent

I have experimented with reading the level using an inexpensive video camera from HF. That way I can read the level with the same perspective and be near the table- or away. I confirmed that my weight shifted the level on the surface plate. I also found that the side of the plate I was standing on made a significant difference. This was with a 2 x 3 ft plate with a stand I modified to put screws under the 3 plate supports, allowing me to level the plate.

A level with a bluetooth camera would be very useful for leveling machinery. You could be turning the bolt or nut and seeing what the level was doing at the same time...

lucky7

My understanding is that estimating to 1/4 is acceptable. If you have a sine bar the level will fit on you can use gauge blocks to test both the level and yourself. Unless your foundation is good- be ready to stand in the same spot until you are done, or you will be testing your patience and sanity as well.
 
Ok, I stand corrected on not being able to read the bubble without affecting the surface plate. I had never considered putting a bluetooth camera on my level. I presume this means I could use my phone as a display? This idea is very good. I could have used just such a system to assist in leveling a lathe today. The camera also allows a standoff from the level to minimize radiant heat transfer to the vial. We are discussing accuracy levels where the radiant heat is significant.

I used to have spirit level that was a 10 second vial with coincidence mirrors that allowed accurate reading to 2 seconds and a shallow angle clinometer. The clinometer was graduated to 2 seconds, but with the 10 second vial it settled much faster than a 2 second vial. I suspect the right fellow could write software to allow the camera to replace the mirror system and allow one to read a 10 second vial to 2 seconds. The clinometer feature is also very valuable as it allows you to check the straightness of a surface without having to first level it.

For someone who is unfamiliar, when you expect this level of accuracy (pun not intended) you need to move the level along a straight surface such as a straightedge to make sure you don't get a yaw error introduced into the system.
 
My modest offer on ebay for this level was accepted. They were correct in that it was in good condition- but someone ignorant had messed with the adjustments. The pictures did not insert where I wanted then, but I do not know how to correct it. The maladjustment is not too obvious in the first picture, but very obvious in the second. It does have a V in the base, so I set it on a round test bar to get the roll set, and then had to readjust the zero.

Being able to adjust the level to the slope is very nice, but keeping track of the changes was challenging at times. The last picture shows a crude but effective way to set a temporary zero point.

The dial is not centered in the plastic top surface with the zero line, and there is a gap which hinders reading the dial. At some point I may make an aluminum cover the replace the plastic one, and build into it both a fixed zero mark, and a rotating adjustable zero with a vernier. That would allow getting readings to .0001 off the dial, with centering the bubble. The way it is now I have been setting the dial to one of the .0005 marks, then adding or subtracting the .0001 reading from the bubble. Lots to keep track of.20200906_213440.jpg20200906_215218.jpg
 

Attachments

  • 20201108_235037.jpg
    20201108_235037.jpg
    82.2 KB · Views: 39








 
Back
Top