[bobJandal]

Hello again.

Thanks to everyone who contributed to answering an earlier question of mine.

I am very curious to know how to build machines from first principles. That is the root / basis of my questions.

One of my burning questions is:

Given that a reference flat must be originally produced from a set of three plates, and some machines have bed lengths longer than a surface plate can be made (unless you are a pharoah, and have armies of slaves to do the scraping for you), how can a bed that is longer than any reference plate you possess be scraped true ?

I ask this question because I believe the plate has to be shifted along, and once shifted along, I do not understand how any guaruntees can be given that it is in a coincident plane with where it was the first time.

Can someone please enlighten ?

[HuFlungDung]

If the machine bed is twice as long as the reference, and each half of the bed is scraped until it is flat, and then the reference is placed in the middle of the bed, and still touches with so and so many points of contact all over, then I don't have a problem with believing that the bed is as flat as the reference.

[Mcgyver]

i can see how you'd be chasing your tail on this, haven't had to deal with it myself. after doing some head scratching and comparing so to minimize the amount of scraping, maybe scrape flat the low part, then by 50% overlaps bring the rest into the same plane? wonder what Connelly has to say on it.

[Geof]

If the machine bed is twice as long as the reference, and each half of the bed is scraped until it is flat, and then the reference is placed in the middle of the bed, and still touches with so and so many points of contact all over, then I don't have a problem with believing that the bed is as flat as the reference.

Hu; surely if your reference is true to +/-0.00001 over its length the bed which is twice as long will be +/-0.00002, etc, etc for increasing multiples.

But thinking about it more maybe not; I suppose it depends whether you introduce any systematic errors. With length multiples greater than two if you always used your reference in the same orientation you may finish up with a bed that hogs or sags.

[Rustybolt]

First level the bed as level as you can get it. If the area where the headstock was, is used you have a pristine datum.On most lathes the very end of the tailstock area has little wear as well. Having leveled the bed you can now measure where the low spots are.
If your gage is 1/2 the length of the bed, scrape in thirds. From each end towards the middle or lowest spot.
There is a method of using a simple laser level to determine the high and low spots as you move along. It has something to do with the interference patterns. I'm not up to speed on it, but people who have done it say it is very accurate. I think I first heard of it on 'The Practicle Machinist' site.

[bobJandal]

Geoff -

That was exactly my issue - is there any way to reproduce the machine bed to the same tolerance as the reference without any further tools, like Rustybolts laser level ?

Or is a departure from the reference an unavoidable consequence ?

But on the topic of further tools, like levels and collimators and the like, would scraping them not be serveral multiples of difficulty harder, since the feedback loop of check against the reference, scrape, check etc now requires the level check too, which does not allow a marking pattern that IS what you want, but rather tells a person how much different the marking is from ideal ?

How can one scape to something they can't see ?

Is not the eternal tradesmans curse looming over the head and wallet of anyone who attempts such things - "darn @#$% - cut too much off" ?

[Geof]

bobJandal: "That was exactly my issue - is there any way to reproduce the machine bed to the same tolerance as the reference without any further tools, like Rustybolts laser level ?

Or is a departure from the reference an unavoidable consequence ?"

I think the definite answers to these two questions are No and Yes. When you use a tool or machine to make another tool or machine there is always a degradation in accuracy and precision. In a different thread someone made this dogmatic statement which led to the counter statement that this implied at the beginning of time a celestial being must have created a master machine and everything humankind has made since then has been a degraded copy (or something to this effect, I can't remember which thread it was).

But I think you have to take the sentence; "When you use a tool or machine to make another tool or machine there is always a degradation in accuracy and precision." completely literally and exclude any human intervention affecting the end result. As a simple example I once worked on a manual mill with the feedscrews very badly worn by having blocks bolted in position so I could check the table movement from a reference point by measuring between the blocks with an inside micrometer and dial gauge. My intervention using instruments with a much higher level of precision meant I could make jigs with holes placed to within +/-0.0005" on a machine that had up to 0.025" wear on parts of the screws.

"But on the topic of further tools, like levels and collimators and the like, would scraping them not be serveral multiples of difficulty harder, since the feedback loop of check against the reference, scrape, check etc now requires the level check too, which does not allow a marking pattern that IS what you want, but rather tells a person how much different the marking is from ideal ?

How can one scape to something they can't see ?

Is not the eternal tradesmans curse looming over the head and wallet of anyone who attempts such things - "darn @#$% - cut too much off" ?"

So I think you do not have it quite correct phrasing it in terms of 'scraping to the tool' I think it is more a case of; make segments of an intended final structure using a reference that has been generated from your three flats. Then assemble these segments and tweak the assembly by referring it to some other absolute reference such as the fact that light travels in a straight line for your laser or gravity always points straight down for a level.

But you are correct about the eternal curse hanging over the wallet of anyone attempting such things. In my example above it took something like 16 hours to drill and bore 8 holes in some large cast flanges that were part of a big drill jig and I killed a few flanges with misplaced or oversize holes. But it saved the expense of buying a Jig Boring machine which had the required higher level of accuracy and installing it in a climate stabilized room.

[nikolatesla20]

Hi,

I would like to mention that I do not think that a machine, built by another machine, will always be slightly inferior. The main thing pointing to this, is that someone had to create the first machine to start with - so how did they do that? For extreme flatness you can simply use optics to see how it looks. One can tell by looking at reflections, quite easily that something is not perflectly flat, and the reason is because light is such a short wavelength (into the nanometers). So if you use light as your guide you can approach fine precision on a flat piece of work. You can use optical flats and a laser, for example.

Also, flatness would be easier over large areas, for example the first lead screw invented was about 5 feet long, and had a 1 foot long nut, to compensate for pitch differences in the screw threads (the nut spans such a large distance that it basically nulls out any pitch differences)

-niko

[NC Cams]

A plumb bob will make a VERY accurate level if you make the string long enough and the bob pointed enough. The Egyptians in biblical times were the first I knew who used the tool.

Once you declare one end "level" you then move the bob from one end to the other. If the table slopes right/left or fore/aft, it will show up as not being plumb to the declared level "reference" end. You can scrape/hone to return the bob to plumb as you traverse it down the bed.

Keep in mind that twist/bow of the bed can/should also be minimized by shimming before you start to scrape/lap. This helps minimize the amount of material removal. Again, use the bob to do any rough pre-straightening.

BTW, a simple machinists level will also suffice a leveling a bed way. At some point, you do have to use a known reference to quantify straightness. The plumb bob is a universally recognized "first pass" method of leveling stuff as gravity is pretty good at consistantly sucking the bob to the lowest point, especially if it is supported by a thin and long enough string.... You should be able to get to within tiny fractions of a degree of plumb using this method.

[nikolatesla20]

Wouldn't you also be able to declare something "flat" by comparing it to a very taught string stretched between two posts?

Also, if you simply take an object, and spin it while you carve it (like a simple wood lathe) or even a small metal cylinder, then by default a round object is equal distance all around, and you can use that in a rolling technique to guarantee constant distance from something...some basic ideas anyway....

-niko

[NC Cams]

Re: taught string between posts = not really. Gravity will tend to cause it to droop no matter how tight you try to get it. Even the most rigid beam can't overcome the force of gravity and it droops between the supported ends.

The turning idea works ONLY at the fixed length. Bend or twist in the bed or warpage in the bed will show up as taper or an irregular OD. If you move the tail stock or if the cutting tool deflects, you've lost your straightness and you'll get an incorrect sizing that may NOT be related to a bed straightness/flatness issue. Besides, just because something is round doesn't mean that it is straight - ask anybody who has turned a tapered pin.

When checking straightness, you always have to use a reference and/or method that you KNOW is straight. This is why lasers work - they are using KNOWN straight, non-deflected light rays.

The plumb bob takes the known constant of gravity and FORCES the string that holds it to be straight. If you make the string long enough, you can use it (along with a bit of trig) to measure straightness of a bed to within a tiny fraction of a degree.

[Geof]

Re: taught string between posts = not really. Gravity will tend to cause it to droop no matter how tight you try to get it. Even the most rigid beam can't overcome the force of gravity and it droops between the supported ends.....

The string, probably a thin very high tensile wire, deflects in the vertical plane but not the horizontal so you can determine that something is straight in one plane.

Even in the vertical plane it should be possible to use a tight wire because the sag is a catenary; it is predictable and can be described mathematically, so a nomogram could be contructed for points along the wire.

[nikolatesla20]

Well thankfully lasers are quite affordable nowadays and can be used to measure quite a few things

So if you used a plumb bob, would you :

1. make a small square block and mount the plumb bob on the side hanging downward

2. then put the plumb bob on one end of the table, and mark the position of the string on the cube

3. move the plumb bob to the middle and check the string pos, and then move to the other end of the table and check the string,

4. deflection in the strings would show how off flat the table is?



-niko

[handlewanker]

Hi all, I once had to recut all sliding ways on a lathe bed by hand, due to there being .013" of wear halfway down the bed to the chuck end. This was a lathe bed, four feet long and the only straight edge I had was a 2 foot ruler from a Starret combination set.
This had been checked on a surface table and was found to be within .002" of parallel over it's 2 foot length. On this job I used the unworn parts between the raised vees and the sides of the ways as referance, to slide a hand planer table mounted with ball bearings and so was able to cut the unwanted metal away to allow a rescrape.
The two foot length of the rule, wiped with mechanics blue, soon revealed any high spots from the hand planer and a precision level placed across the slides checked for parallel and twist. However if there had been no referance points to work from, where do you start?
The same way they did it in the old days, that is if you are going back to those methods and not sending the job out to someone with a planer or Mill with the capacity to cut the full length in one pass, and grind it true.
To do it the old way from a raw casting required chipping the surface of the casting with cold chisels using templates and straight edges to ensure longitudinal accuracy.
Straight edges were made using three to produce the one required. Length being at least one third of the bed length for those long planer beds and gun barrel lathe beds and so it progresively got better and better untill scraped referance plates were used to get surface truth, finally the carriage was prepared and then blued and mated to it's sliding surface.
If you stop for a moment and think, given a pair of flat slide ways of approx 2 feet in length and 1 foot apart, and a carriage slide of 1 foot length, and then place them together with blue on the long slides. If you have as little as .002 variation in any direction either humped or hollowed or twisted it will soon be revealed.
This is what made a machine tool fitter worth his weight in gold. Anyone who has done any serious scraping will know that if you produce a hollow with at least .001" depth then you have to scrape the whole of the surrounding area down to match it.
Nowadays the time taken to produce a machine in this manner and the cost woudl be more than anyone would pay. So it is machines to make machines and the better they get the better they will get.
I've still got the hand planer if anyone wants to hand do a lathe bed, and wants the design.
Ian.