What's new
What's new

DIY Reference Straight Edge

If you can find a copy of Connelly's Machine Tool Repair, he goes into how to generate straight egdes and even surface plates by alternatively spotting three surfaces to one another in a set progression. I have not tried it, but it must work. If you think about it, someone had to make the first flat surface. I think you have to have confidence in your scrapping ability.
 
One has to be careful of expertise found on the internet. Anyone can post something plausible and assert it to be the final word in a particular topic. Without knowledgeable technical editing, peer review, and in the absence of critical comment, there's no way for the naife to assess the material for technical merit. If he accepts the material at face value, he can make a significant investment in time and or money pursuing flawed expertise and may come to naught for all his time and trouble. What does he do then?

I looked at the material Mr Swenson posted and from my experience (I rebuilt machine tools for a living for two years and maintained the scraped references used in the process) I submit his stuff is better than most. Its faults lie in its ommissions.

Mr Swenson casually brings in one of the masterpieces of precision technology (Moore's "Foundations of Mechanical Accuracy") tacitly suggesting his techniques in accordance with it when in fact the techniques he promotes are but roughly adapted and the results you can expect are 1/20th the accuracy routinely addressed by Mr Moore.

The images Mr Swenson uses to illustrate the working together of three edged to acheive straightness are well done as far as he went. The process Mr Swenson attempts to convey is better explained in Porter's "Engineering Reminiscences" Chapter XXI and of course the relevant chapters of Connelley's "Reconditioning Machine Tools."

Swenson states his process is best suited for linear straight edges suited for checking the tables of woodworking equipment rather than as a scraping reference for machine tools where the accuracy required is a full order of magnutude higher. I agree.

He fails to mention temperature and the effect of physically handling precision apparatus and the use of cleaning solvents (alcohol) having a high heat of evaporation.

Neither does he address how to deal with the indications the uses of transfer media (Prussian blue and similar colors in oil) brings forth.

He slips by discussing how to "remove material" in detail (filing and stoning instantly come to mind) by merely alluding to it. This is highly skilled work and a few paragraphs devoted to "material removal" in some detail would be a welcome addition to the naive reader.

His omission of the importance of cleanliness where lapping is concerned is particularly telling.

So, Bioman, if you desire to make a straighedge suited for woodworking and you're capable of dedicating considerable time, you have good hand filing skills, your native wit leads you to avoiding the pitfalls of unequalized temperatures Mr Swenson's discussion may lead you to making some very serviceable straight edges.

If you desire to make a camel back straight edge or a fitter's flat for precision scraping, then Micheal Moore's booklet and goods are a better choise than Mr Swenson's.

Remember that precision references like straightedges are dependent on the stability of their materials. Second class materials like black structural steel poses significant disadvantages not the least of which is its softness and ready deformability.

[ 08-17-2005, 02:39 PM: Message edited by: Forrest Addy ]
 
OK, time to come clean. I admit to having already started this process and put a few hours into it. I purchased three six foot by ½” by 4” lengths of hot rolled rectangular steel.
As I will be using these reference edges to setup woodworking equipment, I can be happy with a few thousands across the entire length of the edge. I need six foot references as I have larger industrial woodworking equipment to span the needed distances. As I need two edges to do most set-up I can't afford the cost of a pair of Starret's.

I have the hand technique/coordination part down from years of woodworking with hand planes, so filing is not an issue, I can handle that. What I am confused about are his descriptions of rotation as regrds the three stock pieces.

If I label the three pieces of iron as A – A’, B – B’ and C – C’ respectively (A being one end and A’ being the other end for example); do I rotate to reference as follows:

A- REFERENCE……………..B – B’…………..B’ – B
A – REFERENCE…………….C – C’…………..C’ – C
B – REFERENCE…………….A – A’…………..A’ – A
B – REFERENCE…………….C – C’…………..C’ – C
C – REFERENCE…………….A – A’………….A’ – A
C – REFERENCE…………….B – B’…………..B’ – B

Of course filing off the high points between each rotation and repeating as needed.
Will temperature be a concern if I am not looking for 10,000 accuracy?
Am I heading in the correct direction here?
 
Hi Bioman:

Lived in Waco 8/51 - 8/59.

The effort you intend to expend on such unsuitable material has to be worth considerable. Temperature changes from handling will not come much into play until you are way down the road with your file.

Seems that you would want these close before you touched them with a file, like machined. Hot roll can be amazingly crookedy.

Here is a link to how Les Watts ( a long lost relative on my mom's side) did what you are wanting to do.

http://home.alltel.net/leswatts/straightedge.html

John
 
Bioman, here's how they usta do it back in oldy times.

I'm assuming you have no machine toools available to you. Clamp all three hunks-o-arn together and dress their working edges flat and square using the beam of your best framing square as a first trial straight edge. When the reference surfaces clean up as square and straight as you can make them mark one as a temporary reference and work on the other two. Mark the ends of all three to prevent confusion during the process.

Blue in each straight edge in turn to the reference without flipping them end for end or using other averaging tricks. Do so by scraping and/or filing holding the blue pattern all the way to the edges so they are keen and square. Keep track of the marked ends as you go so you don't get an accidental flip.

When both working pieces blue in set the temp reference aside. You now have two "working" straight edges with identical error. The next step is to work their high points off gradually and alternately. This is called "scraping straight down" and is a vital skill when attempting to hold parallelism as the way bearings are re-scraped. Blue one working edge and use it to print the second. Scrape and stone ALL the blue marks. Blue the second and use it to print the first and repeat until you get a full pattern on both simultaneously. That's blue A scrape B. Clean of A and blue B, scrape A and repeat.

Here the averaging process is under deliberate control Vs a rigid rotation of parts through a scraping process. You have to clean off a lot of blue and take many prints but when you get done and if you work consistantly the two working parts are fairly straight after the first scraoing session.

When both parts are scraped alternately until both clean up, use one to rescrape the reference that has been gathering dust. Re-scrape the two working parts to the reference (now's the time to be careful of handling and temperature) and then scrape them straight down alternately - again.

By the time you repeat this process (subsequent repetitions go far faster than the first because you're rapidly converging on a true plane) three times your straightedges should be almost right on the money and it's time to check them with blue in rotation with flips from end to end.

Remember the spots in the edges are as important as the spots in the middle so scrape carefully. Your scraping spots should be a uniform speckle all over the reference surface.

Naturally if you mill or plane the reference edge at first you'll certainly shorten the cycle.
 
Be advised that because the shapes you are making the straight edges (more like straight areas) out of can only be rotated end for end upon each other exactly two different ways, a true plane is improbable, since there are two shapes I know of that can make all three surfaces match perfectly in all 6 possible arrangements (that is, each to each, and then each to each with one rotated end for end):

If all three are True Planes (that which we hunger and thirst for)

Or (heres the devil) If all three are sections of a helix.

In summary, without a third way to rotate them and mate them, the edges may have twist in them (eg: if they were plates, and were square, there would be 4 ways to mate each plate, not the two that a rectangle has)

I acknowledge that this may have little effect on your application, but is more for others who might read this post.
 
Bioman, that web page is a Dummies' Guide to Sir Joseph Whitworth's Principal of Symmetrical Distribution of Errors.

Connelly's book mentions the technique, but doesn't describe the process in great detail, and certainly doesn't advocate it.

Trust me, unless you have a lot of time on your hands, buy a cheap imported granite surface plate, and use that as a master reference.

For useless trivia, here's a photo of Whitworth's original surface plates. Where's the third one? :D

http://www.scienceandsociety.co.uk/results.asp?image=10311926&wwwflag=2&imagepos=1
 
John,

Les Watts is a frequent contributor to the Yahoo DIY CNC group, and his technique, using an angle grinder (!), seems specialized for building linear rail mounts:

If I have been doing it a bit I can get an eight foot way measuring straight and flat to a thousandth or so in one very hard days work.The final surface will still feel lumpy....you can easily see and feel a one thousandth high spot. But even with a few bearing points per square inch you have a good surface for mounting linear slides.
I know you can use a sensitive level to interpolate between sections that have been individually scraped, but there are all kinds of potential landmines (accumulation of errors, twist,...)
 
There are several articles on the internet discussing the making of a reference straight edge. The claims for accuracy are attractive, as is the significant cost savings. Has anyone tried any of these methods? Any thoughts would be appreciated. Here is a URL to one of the articles.
Sorry, we couldn't find that page

This URL is broken. Does anyone know where the information can be accessed now? Thank you!
 








 
Back
Top