PDA

View Full Version : About Those Carpenters' Levels



John Garner
04-09-2008, 10:11 PM
Good Evening, All --

An earlier thread dealing with level instruments compares the sensitivity of a premium carpenter's level, which the maker claims is "0.5 mm PER 1 METER", to the 0.005 inch per foot sensitivity of a standard millright / machinist level such as the Starrett Model 98 in its 6-inch-or-longer versions. The difference between them seems small -- 1 part in 2,000 in the first case, 1 part in 2,400 in the latter -- but the numbers alone are misleading.

The typical carpenter's level maker cites "attainable accuracy" assuming that the residual decentering of the bubble within the vial is not discernable, whereas the millwright level maker cites the tilt needed to displace the bubble 1 increment of vial graduation . . . typically 2 millimeters for European, Asian, and newer British vials, 1/10 inch for U.S. and older British vials.

So how well can the bubble actually be centered?

Well, one of the premier makers of carpenters' levels, Stabila, says on their website "The eye readily detects differences of one ten thousandth of an inch, so the reading of the difference in proximity to a reading ring from one end of the bubble to another is the least of anyone’s problem." (I'll editorialize: I'm skeptical.)

In contrast, Princeton University's Dr. Philip Kissam concluded in the early 1950s that the uncertainty in "reading" the position of the bubble in a vial graudated in 1/10 inch increments is, under good conditions, 1/5 of a division. This agrees well with a similar study performed by Kern Instruments of Aarau, Switzerland in the early 1980s that determined a skilled operator could reliably read the position of the bubble in a vial graduated in 2 millimeter increments to 1/4 division. (1/5 of a 1/10 inch graduation is 0.020 inch, as is 1/4 of a 2 millimeter graduation after rounding to the nearest 0.001 inch.)

What is probably the best way to state the sensitivity of level vials is to declare the vial's radius of curvature, which is unfortunately uncommon. Radius of curvature can be calculated from the second-best data format, tilt over bubble displacement -- a simple example being the 0.005 inch per foot per 1/10 inch graduation of the Starrett Model 98-12 vial . . . 0.1 inch divided by 0.005 inch per foot = 0.1 inch times 200 feet per inch = 20 feet.

I didn't find a radius-of-curvature value on Stabila's website, but the 1970 Interim Amendment to the U.S. Federal Specification GGG-L-211C requires a carpenter's level vial " . . . a radius of 7.645 inches, minimum, equivalent to a vial sensitivity of 45 minutes maximum . . . ". One major U.S manufacturer of carpenters' levels, Empire Level, hints at the sensitivity of their carpenters' levels by offering unmounted vials for sale that range in sensitivity from 38 arcminutes per 1/10 inch to 45 arcminutes per 1/10 inch of bubble displacement.

If a generally-recognized-as-superior Stabila carpenter's level is twice as sensitive as the competitive Empire level, the Stabila vial's radius of curvature would be in the 18 inch range. A full order-of-magnitude less than the radius of curvature of a millwright level vial.

Bottom line: If a high-grade carpenter's level is the best you have, use it. But don't try to fool yourself into thinking it's almost as good as a millwright's level.

John

J Tiers
04-09-2008, 10:42 PM
I'm the one who directly compared them.....

And I agree it is all in the definitions. I assumed, perhaps incorrectly, that the 0.5mm/m was sufficient to deflect the bubble over a "significant measurable distance". That would be something like a line division, etc.

The typical 0.005 per 10" type level is readable to better than that.

I have not used a carpenter's level of the type describes, and it may be impossible to read one as accurately as is implied by the "spec".

1/10,000? easily visible? NOT!

Forrest Addy
04-09-2008, 11:32 PM
According to the Moyers Company their better acrylic level vials are calibrated for 30 arc minutes per 2mm along the vial; that's a 1/114 slope.

http://www.wamoyer.com/tubular_vials.html

You can find level vials for practically any application. Carpenter (actually "framing") levels are made to several vial resolutions but the most common is the two line vial representing 1/8" per foot (center bubble to bubble tanget with line) so that it can be used for plumbing drains but that is not cast in concrete. The level can be eyeballed much closer than that; 1/64" per foot is readilly attainable if you use reversal technique.

A carpenter will get inensely annoyed if forced to use a machinist level for structural carpentry. It's too damn accurate. He'll be chasing the bubble all over the place. He wants a level for carpentry not machine tool leveling.

1/8" per foot (1/96 slope) equals 36 minutes about. 1/64" (0.015") per foot (or 1/750 slope roughtly) equals about 4 1/2 minutes. That's a long way from 0.005" (1/2000 slope) per 10 inches for the Starrett #98 style machinist's level and ten times a long way for a 10 arc second master precision level (1/20,000 slope).

A good framing level is all but useless for the final leveling and alignment verifications for machine tool accuracy but handy for lots of other stuff around the shop. Better have one in the roolroom near your shop grade machinist levels and your boxed and babied master precision levels.

wierdscience
04-10-2008, 12:50 AM
So,why would a person in a HS setting need a precison level?


A carpenter's level is for a carpenter and a machinist's level is for a machinist.

That said what are you using it for?If you just want to set your shortbed lathe level so the coolant doesn't run all to one end then a good carpenter's level is fine.
If you want to do the same with your bridgeport mill,then a carpenter's level is fine.

A level is practically useless for setup unless ballpark is all you want,so forget that.


The cheapest carpenter's levels aren't even good for carpentry so they are out.Most have curved vials set willy nilly into a plastic or screendoor aluminum extrusion and are no better than a sardine can filled to a line with water.The better levels,glass filled resin,aluminum or bound wood are better,but the older cast iron milled edge levels are tops.A good CI level when you can find one will run $80-100,about what a good 98 will cost,but it will be more useful.

I have at least two 98's,I seldom ever use them.I have an old Stanley 16" CI level with a 3" vial that I use for most things and it works fine since the 98's can't find a visable error.It's as close as a level is capable of getting so what more would one expect?

Sure you can get master levels,I have one at work for checking over the lathe beds every so often,but how many HSMer's can split with $500+?

J Tiers
04-10-2008, 08:33 AM
I use the 98 every so often to check (now that it was set once) the lathe for twists. Usually all is well, sometimes not

I saw a significant improvement when I used it the first time to set the bed up w/o twisting. Decreased chatter, decreased the "shift" when reversing the carriage, generally better.

Anyone is free to do whatever. Apparently the P.O. of mine did nothing at all, and didn't even know how to set up the back gear lever, OR the countershaft...... And it didn't make his nose fall off.

oldtiffie
04-10-2008, 09:56 AM
There are two messages here - different but not necessarily inter-dependent:
- using a level; and
- getting something/anything, a lathe, a mill - level.

Put all the references, theory, quotes/cites etc. aside and try the damned level for your-self. Set it level - on an adjusted level surface. Now progressively pack one end with shims until the bubble just moves to one graticule/"mark" and record the shimming. Now leave the shims under that first end and start shimming the other end until the vial moves one graticule/"mark" "back". Work out the difference between the height of the two packs of shim and you have the sensitivity per graduation of the level. The sensitivity per unit length of the level is easy from that point

If you have access to a a "dressed and true" magnetic chuck from a surface grinder - a 5" x 10" or a 6" x 12" will do (the bottom and top faces must be flat and parallel to a very high order of accuracy - say less than 1/2 a thou between highest an lowest (or better if possible). Put the "fences" up around the periphery of the chuck. Set the chuck on parallels or "same size" packer/s as required. Put several new say 1/2" steel balls on the magnetic chuck - switch magnets "off" - of course. Now see where the balls run to. It might just surprise you how accurate this is.

The sensitivity of both methods can be considerable enhanced by starting the lathe or mill motor and drive but with the spindle in "neutral" (ie "between gears") - put a bowl of water on the machine (lathe/mill) and see that the surface of the water "shimmering" slightly. Now try the "level" and the "ball" methods and you should find that their accuracy and resolution are considerably improved as the "stiction" between the vial fluid and the vial in the case of the "level" and the balls and the chuck in the other is reduced or eliminated.

I know, I know, I know its not in accordance with "the old sacred texts as done since time immemorial etc. etc. ........." but those texts are just references and guides and they are not the "Dead Sea Scrolls" or holy-writ. All that they describe is a method - not a ritual. Same applies to "Machinery's Hand-book" and the like.

Sure, what I've said is a bit "agricultural" but as with a lot of "agricultural stuff" it works quite often - not always but often enough to be worth trying.

dsergison
04-10-2008, 11:03 AM
hold on there...

they dont mean you can see a 1/10,000 difference in bubble position!

they mean that a 1/10,000 difference in level will result in a bubble position move you can see.

which for a vial gradutaed in 1 part in 2,000 this means "seeing" 1/5 of a division.

oldtiffie
04-10-2008, 11:28 AM
hold on there...

they dont mean you can see a 1/10,000 difference in bubble position!

they mean that a 1/10,000 difference in level will result in a bubble position move you can see.

which for a vial gradutaed in 1 part in 2,000 this means "seeing" 1/5 of a division.

Maybe its different "down here" in up-side-down OZ, but my 0.02mm per 1 meter (ie 1 in 50,000) machine level is the calibration of the vial graduations ie 1 graduation movement = 0.02mm per meter. My "Precision Machinists Level" is the same.

http://i200.photobucket.com/albums/aa294/oldtiffie/Frame-level1.jpg

http://i200.photobucket.com/albums/aa294/oldtiffie/Frame-level2.jpg

http://www.machineryhouse.com.au/Products?stockCode=Q208

http://www.machineryhouse.com.au/Products?stockCode=Q205

As J Tiers has rightly said on a couple of occasions recently, the base of the level must be "flat" else the level and its vial/s are of no practical use for the intended purpose.

The machine square is at least to Grade 1 standard so far as I am aware but even Grade 2 2 is quite adequate for my purposes as regards being flat and square in addition to the 0.02mm per metre accuracy and calibration.

lazlo
04-10-2008, 11:35 AM
I know, I know, I know its not in accordance with "the old sacred texts as done since time immemorial etc. etc. ........." but those texts are just references and guides and they are not the "Dead Sea Scrolls" or holy-writ. All that they describe is a method - not a ritual.

Thanks Tiffie. You're probably right -- the level/two collar method described in all the machinery texts is just theoretical pedantry. Just fire up the lathe and Get 'R Done.

oldtiffie
04-10-2008, 12:06 PM
I know, I know, I know its not in accordance with "the old sacred texts as done since time immemorial etc. etc. ........." but those texts are just references and guides and they are not the "Dead Sea Scrolls" or holy-writ. All that they describe is a method - not a ritual.



Thanks Tiffie. You're probably right -- the level/two collar method described in all the machinery texts is just theoretical pedantry. Just fire up the lathe and Get 'R Done.

Thanks lazlo.

I didn't say that at all.

What I did say is that whilst these "texts" are valid there are or may be other methods that will - as you say "Get 'R Done" - as well.

Too much is made of the text and not enough of the reason or substance.

The way that some would have it, they are cast in stone and start off with "Thou shalt/shalt not ......................... " and inevitably the self appointed custodians, High Priests and "Holy Rollers" want to "take charge" to the extent that no doubt nor question/ing is allowed under pain of eternal damnation etc.

Some of the performances are not much short of an Evangelical "voices and messages" effort.

I'm probably well beyond damnation and salvation - nothing new there - but I'm used to it and I'm staying put.

Perhaps it all started in Haiti - and that Voodoo doll with all the big needles in it is me - no risk!!

Oops - I can hear François Duvalier ("Papa Doc") and his "Tonton Macoutes".

I'd best be off - quickly!!

pcarpenter
04-10-2008, 12:07 PM
Just fire up the lathe and Get 'R Done.

Yeah...round is a relative term and highly overrated in my book:D Sounds like Tiffie may find a wood lathe more to his liking:D

Seriously, however, I have used a really long (24"?) number 98 (.005/foot) as well as their master precision level (.0005/foot). I now own one of the import .0005/foot levels and it seems to work well too but it arrived slightly off calibration and that's an annoying process...I digress.

Working with the .005/foot #98 is handier when starting the levelling process. The master precision levels are quite sensitive and take a while to allow to settle as each change is made. Handling has to be kept to a minimum as warm hands will change the reading. Its this extra sensitivity that is actually useful when finalizing adjustment...even moreso than its precision.

In the discussion of all of this its important to separate the precision of the instrument from the desired degree of precison of the levelling job. If you use a measuring instrument with precision exceeding the desired precision of the operation then you can emphatically know what you ended up with--even if you are satisfied with some twist. If you use one less precise, then you can only guess at the real end result as the instrument is not capable of showing you what you actually settled for.

Paul

oldtiffie
04-10-2008, 12:39 PM
Yeah...round is a relative term and highly overrated in my book:D Sounds like Tiffie may find a wood lathe more to his liking:D

Seriously, however, I have used a really long (24"?) number 98 (.005/foot) as well as their master precision level (.0005/foot). I now own one of the import .0005/foot levels and it seems to work well too but it arrived slightly off calibration and that's an annoying process...I digress.

Working with the .005/foot #98 is handier when starting the levelling process. The master precision levels are quite sensitive and take a while to allow to settle as each change is made. Handling has to be kept to a minimum as warm hands will change the reading. Its this extra sensitivity that is actually useful when finalizing adjustment...even moreso than its precision.

In the discussion of all of this its important to separate the precision of the instrument from the desired degree of precison of the levelling job. If you use a measuring instrument with precision exceeding the desired precision of the operation then you can emphatically know what you ended up with--even if you are satisfied with some twist. If you use one less precise, then you can only guess at the real end result as the instrument is not capable of showing you what you actually settled for.

Paul

Thanks Paul.

Its not the Carpenters Levels that I'm having problems with - or the Carpenters' (yours??) rules either that I'm having trouble with. (I agree with your post).


Sounds like Tiffie may find a wood lathe more to his liking

Finding a wood lathe is not a problem for me either, as "going against the grain" is second nature to me - though I suspect that finding a lathe that "would" is a problem for some others. Perhaps they need to "wood(en)" it.

S_J_H
04-10-2008, 06:31 PM
I have one of the Empire carpenter levels-
http://www.empirelevel.com/true_blue/default.asp
they say it's accurate to .0005" per inch. So does this put it at .005" per 10" ?
Anyhow it seems to be a decent level for 30$.

Steve

lazlo
04-10-2008, 06:36 PM
I have one of the Empire carpenter levels-

they say it's accurate to .0005" per inch. So does this put it at .005" per 10" ?

It's got an extruded "Aircraft Aluminum :D" body -- how flat could that possibly be?

JCHannum
04-10-2008, 06:51 PM
That's 6061 Heavy Duty Aircraft Aluminum. It's got to be good. Just don't leave it in the sun.

S_J_H
04-10-2008, 07:09 PM
The surfaces on this level have what looks like a coarse flycut finish. I just tested it on my lathes. A .005" shim on one side of the ways is easily spotted on the bubble. You can get these levels at the big orange store.
There is an article in an older hsm or maybe it was projects in metal using some sort of beam and plumb bob dohicky. I just glanced over the article bit it looks pretty slick at amplifying any bed twist distortion.

oldtiffie
04-10-2008, 09:04 PM
The surfaces on this level have what looks like a coarse flycut finish. I just tested it on my lathes. A .005" shim on one side of the ways is easily spotted on the bubble. You can get these levels at the big orange store.
There is an article in an older hsm or maybe it was projects in metal using some sort of beam and plumb bob dohicky. I just glanced over the article bit it looks pretty slick at amplifying any bed twist distortion.


Thanks Steve - muchly.

You've made my day!!

A "Carpenter's" level - oh dear me, and made in the USA too - and not another mention of it anywhere!!!

That's far too good for them pagan barbarian "Wood Butchers". They must be "put in their place" (and KEPT in it).

Jeez - the Welders and Sheet-metal workers etc. might be next if we don't get this under control!!!.

What-ever next?

That just "will not do" - at all.

The cheek of them!!!

Must send the "thought police" and "cavalry" from here to sort that sort of heresy out - once and for bl**dy all!!

kendall
04-10-2008, 10:57 PM
The surfaces on this level have what looks like a coarse flycut finish. I just tested it on my lathes. A .005" shim on one side of the ways is easily spotted on the bubble. You can get these levels at the big orange store.
There is an article in an older hsm or maybe it was projects in metal using some sort of beam and plumb bob dohicky. I just glanced over the article bit it looks pretty slick at amplifying any bed twist distortion.

One of my carpenter levels is also rated at .005, won't say it is but I know it's 'slower' to settle down then everyone elses levels. And like you say, if I set it up level, then place a shim at one end I can detect it.

The weight and string level is something I've used quite a bit. Had a tall triangulular frame with a pendulem hanging from the apex, string crossed a bar that I'd put a mark on, then I could true to that mark and get great accuracy.
It's awkward to use, but can be easily 'slammed' together at home with no special tools.

ken

John Garner
04-11-2008, 09:40 PM
I sent E-mails to both Stabila and Empire Level, asking about the sensitivity of the main vials on their best-grade levels, expressed either in terms of 1) tilt necessary to move the bubble 2 millimeters from center or 2) as the radius of curvature.

No response from Empire yet, but a quick response from Stabila provided a stock answer: "All of our vials are the same no matter what angle they are set at. Vial facts are: .029 degrees = .5mm/m = .0196/48" 1.7 arc min accuracy 14 arc min sensitivity."

Ignoring the obvious error in the inch-unit citation, I wrote back: "I'm not sure I fully understand your statement '1.7 arc min accuracy 14 arc min sensitivity'."

"Does that mean that visually-perceptible bubble decentering occurs when the vial is tilted 1.7 arcminutes, and that the bubble center will be 2 millimeters from the vial center when the vial is tilted 14 arcminutes, which would be equivalent to a 490 millimeter radius of curvature?"

"I do appreciate your time and consideration is answering my questions. I'm trying to determine if Stabila levels are sensitive enough to substitute for the more-sensitive-but-MUCH-more-delicate millwright's level (such as the Starrett Model 98) when roughing-in precision machinery."

To which Stabila replied: "Mr. Garner, what "arc minute" and arc sensitivity" implies is the curve of the vial itself. The more flat a vial is the more accurate it is, also it will take longer to show level. In the construction industry we serve, having a level that takes a long time to achieve level is not very
practical. Stabila tries to have both speed and accuracy in our vials. My
opinion would be Stabila levels are not accurate enough for someone like
yourself. I would suggest you look at "Hahn & kolb" they are another German
company that makes high quality accurate machinist levels."

At lunch time I drew a 1/4 meter Stabila-type torpedo level from the toolroom and set it on a level-within-5-arcsecond surface table. As expected, the bubble centering error was imperceptible. I then slipped a 0.004 inch leaf from a thickness gage under one end of the level stock and observed the bubble. The bubble still appeared to be centered.

I was able to detect a slight change in bubble position because of the 0.004 inch thick shim only if I watched the bubble while removing or replacing the shim.

So, if I assume that my interpretation of Stabila's original message is correct, that moving the bubble 2 millimeters within the vial takes a tilt of 14 arcminutes, and that the least-perceptible bubble decentering takes a tilt of 1.7 arcminutes, Stabila believes that the threshold of apparent bubble movement is essentially 2 millimeters / (14 arcminutes / 1.7 arcminutes), which is essentially 1/4 millimeter.

Which hints that their website is supposed to say decentering of "ten one-thousanths of an inch" is visible, not "one ten-thousanths of an inch".

It'll be interesting to see what, if anything, Empire Level will have to say.

John

lazlo
04-11-2008, 10:23 PM
Great post John!


"Mr. Garner, what "arc minute" and arc sensitivity" implies is the curve of the vial itself. The more flat a vial is the more accurate it is, also it will take longer to show level. In the construction industry we serve, having a level that takes a long time to achieve level is not very
practical. Stabila tries to have both speed and accuracy in our vials. My
opinion would be Stabila levels are not accurate enough for someone like
yourself.

W. A. Moyer, who makes the vials for the Starrett 199, the Kingway alignment jig, also makes vials for the construction industry. Don (the head guy there) explained the same thing to me: the precision vials are ground on something like a centerless grinder, and then sorted by accuracy. The very straightest and roundest vials have 2-3 second accuracy, and are sold at a premium. Those are used in the Kingway alignment jig, by the way...

The Starrett 199 Master Precision Levels get the 10 second vials.

The Starrett 98 Machinist Levels are 80 - 90 seconds accuracy, or about 8 - 9 times less sensitive than the Master Precision Levels. They're made on a whole different process/factory operation than the ultra-precision vials.

I'd imagine the construction levels are at least another 10x less sensitive than the Machinsit Levels. I think most of the construction levels use plastic vials, so that probably limits how straight and round they can be "ground."

darryl
04-12-2008, 01:44 AM
Well, this is all interesting. I wanted to make a vial once, and I would have except I didn't have any glass tubing on hand. I wondered how much of a bend the tubing would take before it broke, and whether there would be enough curve to actually make the bubble behave properly. I also wondered if the bubble would exhibit any 'stiction' to the glass walls. At the time I didn't know what the fluid was, and today I still don't. Someone told me it was alcohol, someone else said light oil, or mineral oil.

Aligning it was not going to be a problem unless the bubble had stickage. I didn't consider a calibration at the time, nor did I consider whether ambient temperature would have affected the alignment.

dp
04-12-2008, 02:02 AM
The bubble is a lack of oil or what ever liquid in the vial so I think it doesn't have stiction. But I'd start with a really clean vial and boiled oil to kill any fungus potential. Think I'd toss in some inert gas, too. Something that has a low coefficient of expansion and no affinity to mix with what ever liquid was in there. I'd suspect nitrogen and mineral oil would work as a hunch.

Forrest Addy
04-12-2008, 02:50 AM
If you are talking of making a level vial, the radius of the bend (or for precision vials, the barrel shape) in the tube and the liquid in the vial has to suit the sensitivity you are striving for. Oils and organic solvents are the usual fluid but I understand they use ethyl ether and alcohol in the most sensitive levels.

You can make a level vial, that's no big trick. Making one having specific sensitivity that responds reliably to minimum tilts may well be a trick.

Dawai
04-12-2008, 09:42 AM
I'm buying one of these today.. the last day of the sale...
http://s7.sears.com/is/image/Sears/00948295000?qlt=90,0&resMode=sharp&op_usm=0.9,0.5,0,0
$19.00 Craftsman digital level sears item 00948295000
Mostly cause it is digital, I used one in Illinois recently and I liked it..

The online ordering crapped out here, possibly linux?? or ??

torker
04-12-2008, 09:53 AM
A question??? David shows the digital level. Taking all things into consideration... how would any level compare to that? I'm thinking of the human sight variables.
IE... my old (now unused) high quality vernier caliper. I haven't been able to see the grads on it for a long time now and even if I could.. how would that stack up against even the cheapest digital caliper/mic/level?
Thanks!
Russ

J Tiers
04-12-2008, 10:13 AM
It would stack up VERY well..... The digital level have VERY coarse divisions in general....... Possibly 1/10 degree....... to the nearest 0.1 degree plus or minus one count........

In other words, probably an effective accuracy of around 1 quarter degree.

lazlo
04-12-2008, 10:56 AM
In other words, probably an effective accuracy of around 1 quarter degree.

The Mitutoyo and Fowler digital levels for sale in all the MSC/J&L/Penn Tool flyers claim accuracy to +/- .1 degree, so a quarter of a degree for the Sears model sounds about right.

lazlo
04-12-2008, 11:03 AM
You can make a level vial, that's no big trick. Making one having specific sensitivity that responds reliably to minimum tilts may well be a trick.

I know there's a HSM article about making your own level vial (I may even have it somewhere) -- is the vial ground, or did the guy just heat-seal the ends of a test tube or pipette?

ligito
04-12-2008, 11:13 AM
Jeez, if you wan to test for level, just pour some Mercury on it and adjust til it stops moving in all directions.:)

Dawai
04-12-2008, 11:22 AM
A friend was needing money, he had a great laser transit-level by Bushnell.. a two foot level that bolts onto the transit..

I pulled a $100 bill out and offered it.. he refused it but took $50.. they are all over ebay for less than that..

I would have gave him the $100 without the level if he had needed it.. he was just too proud to accept it. I'm too broke these days to be that generous.

slugger
04-12-2008, 09:32 PM
For most guys using the average machine, A very good carpenters level, like my Stabila, will get you so close that you will probably not have any issues with your machine. There are many other factors in other than a clean room environ that can affect machine accuracy (temp fluctuations come to mind) so if a guy doesn't have a master precision level, he can still get pretty dang close with a good carpenter's level and some careful methodology.

oldtiffie
04-12-2008, 10:32 PM
Jeez, if you wan to test for level, just pour some Mercury on it and adjust til it stops moving in all directions.:)

Good move ligito.

A dozen or so ball-bearings will do as well.

Watch out that the OHS/"heavy-metal" police don't getcha thou' with all that mercury around.

SHADOW
04-12-2008, 11:00 PM
I have dealt with a number of vials over the years needing replacement or resetting. They were filled with alcohol and ether, the colored ones just have added dyes. Bubbles can indeed stick. I was told the mixture coagulates over time and it forms small clear blobs that naturally find their way to the top and when the bubble comes along it hits these and stops. These then become a constant problem. They have usually been stored in extreme conditions and are somewhat older.

I aquired an engineer's survey level, a Deitzgen (cira '50's) with a 6" vial, having a 2 1/2" bubble. It was formerly a government item and in it's box was a spare unmounted vial (unexpected bonus). It basically needs fastening to an appropriate base.
http://img507.imageshack.us/img507/6097/levelvialph1.jpg

oldtiffie
04-12-2008, 11:38 PM
For most guys using the average machine, A very good carpenters level, like my Stabila, will get you so close that you will probably not have any issues with your machine. There are many other factors in other than a clean room environ that can affect machine accuracy (temp fluctuations come to mind) so if a guy doesn't have a master precision level, he can still get pretty dang close with a good carpenter's level and some careful methodology.

Agreed Slugger - absolutely.

Your method will do most of the time at best result for minimal cost and effort.

Using a "super level" on its own is a problem as there is too much effort (and frustration) in trying to get the bubble into the vial at all let alone into the centre only to have to have it upset as you go onto leveling and adjusting elsewhere.

It is far better to use a "staged" approach - using "coarser/less-accurate" levels as you go on a progressive basis up to the standard that you are satisfied with and that meets your needs.

There may well be compromises that have to be made as well as things that will "have to be lived with" and/or "worked around".



There are many other factors in other than a clean room environ that can affect machine accuracy (temp fluctuations come to mind) so if a guy doesn't have a master precision level, he can still get pretty dang close

That is a very valid practical point that is often over-looked or ignored.

The "level set-up" will change with ambient temperature changes. It may even change between "cold" (before starting work) and hot-warm (after the machine has been running both under and free of load).

It would be unrealistic to expect otherwise or to attempt to re-level with every temperature change.

Perhaps its best to do the leveling at the average temperature in your shop when you use your lathe/mill or what-ever.

Evan
04-12-2008, 11:41 PM
In the thread that Fasttrack started about levels mounted on ball bearings I described a way to level a lathe using a cheap line laser level and two 90 degree prisms from a cheap pair of binoculars. Back when I did the thread on qualitative flatness testing using a laser level I went on to develop the method that I described in Fasttrack's thread. Last fall I was extremely fortunate to buy not just a horizontal mill at a very good price but also a box full of the highest quality Swiss and German instruments, including a good frame level.

Now, you would think that as soon as I got my hands on that level I would be running around checking things like my lathe. To tell the truth, I haven't used it to measure anything until tonight. I already knew that lathe was straight both from it's performance and the fact that I had optically aligned it a year or two ago.

Because of this thread and the previous one I decided to check and see just how effective my dirt cheap laser system is. I put the master level on the cross slide with a shim to bring the bubble on center and then ran it to each end waiting several minutes for it to settle.

You can see for yourself what the difference is from one end to the other.

http://vts.bc.ca/pics3/level1.jpg

http://vts.bc.ca/pics3/level2.jpg

lazlo
04-13-2008, 05:03 AM
You can see for yourself what the difference is from one end to the other.

That Swiss frame level is very nice Evan -- it's 0.02 mm per meter, or 5 second accuracy. That's twice as sensitive as the Starrett 199 Master Precision Level.

It looks like your lathe is leaning backward (away from the operator) at about .135 mm per meter at the headstock, and .11 mm per meter at the tailstock.

oldtiffie
04-13-2008, 06:48 AM
That Swiss frame level is very nice Evan -- it's 0.02 mm per meter, or 5 second accuracy. That's twice as sensitive as the Starrett 199 Master Precision Level.

It looks like your lathe is leaning backward (away from the operator) at about .135 mm per meter at the headstock, and .11 mm per meter at the tailstock.

Sorry laslo.

Nope - don't think so.

If you have a close look you will find that the level is centred/zero-ed on the long lines/graticules with the black "spots" at each end.

It is only 3 of the divisions/graticules "off" = 3 x 0.02mm/m = 0.06mm/m = 0.06/1,000 = 0.00006 = atan 0.0034 arc degree = 0.2063 arc minute = 12.3276 arc seconds.

It would not matter if the lathe were out 10 degree at that setting as the ends are within a "whisker" of being parallel and so are co-planar. If Evan's lathe bed is straight and he gets a similar reading in the centre of the bed then that lathe is as good as can be asked for and will meet any spec of any modern lathe of any size from any country and from any manufacturer.

Evan's lathe and set-up just has to be the bench-mark.

I will be interested to see if any can better it.

Further, Evan's Machine Square is German and at least as accurate as any machinist's level - "Starrett" or otherwise.

My "Chinese" "Measuremax" machine square is of the same specification as Evan's and I have every reason to be confident that it is as accurate as Evan's.

The machinist square has capabilities beyond the machinists precision level - but with equal accuracy - as it will show vertical and square to very high orders of accuracy as well. It is my precision level of choice as my machinists level is only a "one trick pony".

I thought Evan's use of the cross-slide (or saddle) - as suggested seperately by J Tiers - was excellent thinking as the "shift" of the saddle as opposed to just measuring the deviation/wear in the bed was brilliant. In other words the "error" was determined where it counted - at the tool position.

Evan's pics are repeated here.
http://vts.bc.ca/pics3/level1.jpg

http://vts.bc.ca/pics3/level1.jpg

Pics of my "MeasureMax" machine square (as posted previously) - for comparison - follow:

http://i200.photobucket.com/albums/aa294/oldtiffie/Frame-level1.jpg

http://i200.photobucket.com/albums/aa294/oldtiffie/Frame-level2.jpg

Evan
04-13-2008, 06:51 AM
It looks like your lathe is leaning backward (away from the operator) at about .135 mm per meter at the headstock, and .11 mm per meter at the tailstock.

(edit)I make it as no detectable difference. Both ends lean the same. But that is dependent on the shim I placed.
Which of course doesn't matter since it all leans together. One thing I discovered when running my seisomgraph is that the foundation of the house is quite flexible. The siesmograph I built is essentially a very sensitive tiltmeter and can detect the effect of somebody walking near the house. I had problems maintaining the alignment of the instrument when it snowed or rained because of the effect of the multi tons of water deposited around the house had on the bending of the earth under and around it. Of course it wasn't even, especially when it rained because the rain that landed on the house was directed off the roof and onto the surrounding ground.

I can't quantifiy the actual amount of deviation from level this causes because the seisomometer is uncalibrated in that respect. That isn't relevant to it detecting earthquakes. It would be nice to know but I can't think of an accurate way to determine it because the instrument is so sensitive.

[added]
I should say that when I said I brought the bubble on center I didn't mean exactly on center, just to where it was readable as you see it in the pics. I also made similar measurements directly on the lathe bed with the same results. It doesn't vary appreciably anywhere along the length of the bed.

John Stevenson
04-13-2008, 07:11 AM
I have a 18" builders level here that I use daily in the shop.
It's the only tool that comes to hand where I can reach over and scratch my back with it.

Next question? How will I know it's level as it's behind my back and will changing over to a machinists level improve the quality of the itch ?

Seriously I do use it for that but in the course of buying up job lots of gear at sales, and measuring equipment here fetches zilch for some reason I have loads of levels of all sorts, some are supposedly very expensive and accurate.
Simms Sons and Cooke spring to mind here as a manufacturer but not certain.

Scrap man gave me a frame level [ although I didn't know they were called that ] about a year ago, still wrapped, have to dig that out and see just what it is.

Edit got the wrong name

http://www.ast.cam.ac.uk/~ipswich/History/Troughton_and_Simms.htm

lazlo
04-13-2008, 11:11 AM
It looks like your lathe is leaning backward (away from the operator) at about .135 mm per meter at the headstock, and .11 mm per meter at the tailstock.

If you have a close look you will find that the level is centred/zero-ed on the long lines/graticules with the black "spots" at each end.

When the legend says ".02 mm per division", I've been interpreting that to mean each tick on the level, where the major ticks are there to help you count how many divisions off you are. Can someone (John Garner?) confirm?

So at the headstock end of Evan's lathe, he's off 6.75 divisions x 0.02 mm per division = .135 mm per M.

At the tailstock end of Evan's lathe, he's off by 5 divisions, or .11 mm per M.


It is only 3 divisions "off" = 3 x 0.02mm/m


Evan's Machine Square is German and at least as accurate as any machinist's level - "Starrett" or otherwise.

It's Swiss, not German. And as I posted above, it has 5 second accuracy, which is twice the sensitivity of the Starrett 199 Master Precision Level, and half the sensitivity of the 2.5 second level in the Kingway Alignment system.


I thought Evan's use of the cross-slide (or saddle) was excellent thinking as the "shift" of the saddle as opposed to just measuring the deviation/wear in the bed was brilliant.

Yes, we all know you're in love with Evan -- do you want to be alone now? :)
Not only did Evan invent the idea of putting the level on the cross-slide, but he invented the level too. :rolleyes:

Evan
04-13-2008, 11:13 AM
Robert,

The bubble is in the same position at both ends. Where are you seeing a difference?

lazlo
04-13-2008, 11:37 AM
From your pictures in post 35, it looks like the reading are different by 1 - 1 1/2 divisions, but maybe that's the angle you took the picture. In any case, I was just reading off the vial scale in the middle of the night -- my 5 month old was up sick last night... In either case, your lathe is extremely well leveled.

By the way, you say that the lathe was leveled with a laser setup -- did you do the laser autocollimator thing with the binocular prism? Do you have any pictures?

Evan
04-13-2008, 01:08 PM
I don't have pictures of the process as I wasn't intending to post about it at the time. Perhaps I will replicate the setup and do a thread about it. I know you may find this hard to believe but I don't post about all the shop related things I do, even if they work.

lazlo
04-13-2008, 01:29 PM
I know you may find this hard to believe but I don't post about all the shop related things I do, even if they work.

I thought the HSM Forum was your personal Blog? :p

Seriously, I have several commerical laser line generators and little first surface mirrors (from when I built the ubiquitous laser alignment jig for my Mill/Drill). It sounds like I need a beam splitter, and I'm not sure how the combining/interference setup looks?

Evan
04-13-2008, 01:46 PM
No beam splitter of interference is used. It is simple geometry and one of the reasons I didn't post about it previously is that you need sufficient distance in front of the lathe to obtain sufficient resolution. I will put up a drawing of the setup later but right now I have some work to do outside.

To put you on the track just do the math on the possible visible angular displacement of a beam over 30 feet. That's the length of the arm I have with the laser 15 feet from the lathe. I can even go to about 30 feet x 2 by going to the end of the hall in the basement.

John Stevenson
04-13-2008, 05:51 PM
Went digging tonight to find those levels.

The frame level is unfortunately unmarked.

http://www.stevenson-engineers.co.uk/files/level1.jpg

Other than the company's name and department. Malc the scrappy used to handle all their scrap before he retired.

The other level isn't a Troughton / Cooke as I thought although I'm sure I have one of those as well somewhere but a Hilger & Watts.

Nice bit of kit.

http://www.stevenson-engineers.co.uk/files/level2.jpg

All adjustable and God knows what else, I think it also gets cable TV.

Thrown out by the local college when they closed.

http://www.stevenson-engineers.co.uk/files/level3.jpg

Never used it and they probably didn't. It's too sensitive, took ages to get the bubble somewhere near just to take the picture.

About all it's useful for is dick slapping rights, mine is more accurate than yours ner ner.

That plastic builders level will do me 'cause it's impossible to scratch your back with this monstrosity.

.

J Tiers
04-13-2008, 07:43 PM
Never used it and they probably didn't. It's too sensitive, took ages to get the bubble somewhere near just to take the picture.

That plastic builders level will do me 'cause it's impossible to scratch your back with this monstrosity.

.

That's (almost*) why I use a Starrett 98...... it can be got on level, or if a bit off absolute level, it still reads.... And you can always read it to a fraction of a line, so all I need is to get it to read at this end just where it did at that end.

Another nice thing is that the 98 is sneered at, and so maybe you can get one from someone who has acquired a 'fine accurate level" and feels they don't need that mere 98 any more. Mine cost me $10, but I also got some toolmaker's buttons and a couple other things that I don't recall, in with it. I've used them all, so I guess I got my money's worth.

(* it's too short to scratch your back with, that is what doorposts are for)

aboard_epsilon
04-13-2008, 07:59 PM
Even the crudest levels can be accurate in the right hands

just turn them end for end ...and if the bubble ends up in the same place ...then all is good.

so the bubble does not actually have to be in the middle to be right ....as in a damaged or wrongly assembled one ...
just so long as its in the same place when you turn it end for end.

same goes for the precision ones .check them out by turning end for end .

all the best..markj

J Tiers
04-13-2008, 11:08 PM
Well, the "hands" can't make the level indicate a finer angle than it is capable of.......

So the 98 is a good compromise between sensitivity and usability.

A regular 1/8" carpenter's level is very usable, but isn't sensitive enough to make it worth anything for lathe leveling.

A "master precision" level is plenty sensitive enough, but almost unusable unless you start out with a much less sensitive level (like a 98) first. (assuming it isn't a crappy defective chinese level. A good chinese level would have only the usual problems)

oldtiffie
04-13-2008, 11:10 PM
Even the crudest levels can be accurate in the right hands

just turn them end for end ...and if the bubble ends up in the same place ...then all is good.

so the bubble does not actually have to be in the middle to be right ....as in a damaged or wrongly assembled one ...
just so long as its in the same place when you turn it end for end.

same goes for the precision ones .check them out by turning end for end .

all the best..markj

Thanks Mark.

Right on.

In most cases the level (bubble) is not adjustable but the factory does a pretty good job of it most times. Reducing number of "returns" and "lost sales" is a good incentive to get it right.

Also the main body of the level needs to be firm, stable and straight.

I find that 1/2 to 1mm per metre is pretty good as it can be used from start to finish of leveling without jumping around between levels of varying accuracy as I "get closer".

My 0.02mm/m level and machine square stay in their box most times (just about always actually).

One of the handiest levels I have is an Electricians "Torpedo" level as its "pointy" ends gets into some awkward places and the magnetised strip on the base can be a God-send both in use and for keeping it on the table and to stop it being brushed off onto the floor.

These levels are available in most good hardware stores and are really good value.

The machine square is very versatile indeed as a square for setting up and checking.

I use my "Carpenter's" levels to set the "knife edges" (well - they are about 3mm wide) on my balancing rig for setting my surface grinder wheels. If done accurately, the balance is very sensitive indeed. I have no problem detecting "out of balance" and "balanced". If I just lightly push the balanced wheel (and hub - on the balancing spindle provided) the wheel will travel to the end stops without changing speed. That is about as much as I can ask for. The results on the grinder after balancing and dressing speak for themselves.

Incidentally, just that some can get a "handle" on this "old fashioned" "non-digital" "Carpenter's Level" bit, I ran a few figures on the 1mm/mm level.

1mm/m = 1/1,000 = 0.001 = arctan 0.0573 arc degree = 3.4377 arc minutes.

A 0.5mm/m level - readily available - will be twice as accurate - ie better than 2 arc minutes.

The standard machinists digital level/inclinometer has an accuracy of 0.1 are degree = 6 arc minutes.

A vernier caliper and a standard rotary table are doing well to better 0.1 arc degree either.

My "Measumax" ("Chinese") digital protractor and vernier protractor has an accuracy of 5 arc minutes.
http://www.measumax.com/ViewPage.php?ID=88

So, really, its just a case of "horses for courses" as the HSM-er requires and can achieve with what he has got or can get for the tasks that he has.

Pretty much the same for most things - inside the HSM shop and out of it.

lazlo
04-13-2008, 11:41 PM
In most cases the level (bubble) is not adjustable but the factory does a pretty good job of it most times.

All master precision levels, including my Starrett 199, the Mitutoyo 960, the Viz, and my Russian Wyler clone all have set screws to calibrate the vial. Most will have sealing wax on the set screw to correspond with the calibration date.

My Viz came with the vial putty all messed up, and it was a long and drawn-out affair to re-set, and re-calibrate the vial. It takes so long for the vial to settle, you have to be very patient. A bottle of scotch helps a lot...

J Tiers
04-13-2008, 11:47 PM
The 98 has adjustment also, a screw at each end.

Some carpenter levels can be adjusted, many cannot. You unscrew the bezels to loosen a bit, and turn them until they read right. Not all can be turned, and not all have enough movement capability. And, not all need it, either.

John Garner
04-16-2008, 12:10 AM
Well, still no response from Empire Level about their True Blue vials.

As far as Evan's lathe leveling job . . . in looking at the photos, it strikes to me like the bubble moved a bit less than 1/2 of a division with the move of the carriage. It's entirely possible that the apparent bubble movement is parallactic, but if we assume for a moment that it is A) real and B) due solely to uniform twisting in the bed of the lathe, it represents a twist of under 0.01 millimeter per meter -- one part in 100,000, about 2 arcseconds -- over the length of the bed.

To say that, though, I have to make an assumption. The label on the level stock clearly states the vial sensitivity is 0.02 millimeter per meter, but doesn't complete the thought by telling us the corresponding displacement of the bubble in the vial. Inferring a supplemental "per 2 millimeter graduation increment" is reasonable given that the instrument is European.

After a bit of arithmetic, we can conclude that the vial radius of curvature is 2 millimeters divided by (0.02 millimeter per meter) = 100 meters.

Let's compare that with the vial in the Starrett 199, which has a sensitivity of 0.0005 inch per foot per 1/10 inch graduation. The arithmetic here is that the radius of curvature is 1/10 inch divided by (0.0005 inch per foot) = 2400 inches = 200 feet, which is just under 61 meters.

Bottom line, Evan's Stiefelmayer instrument is (100 divided by 61) = 1.65 times as sensitive as a Starrett 199, NOT twice as sensitive.

In much the same way, John Stevenson's Hilger & Watts level is probably old enough to be graduated in 1/10 inch increments. Its vial radius of curvature is therefore 1/10 inch divided by (0.001 inch per 10 inches) = 1000 inches.

So how about a "millwright or machinist level" such as a 6-inch or larger Starrett 98? The Starrett 98 vial is graduated in 1/10 inch increments, and each increment represents a tilt of 0.005 inch per foot. The vial radius of curvature is 1/10 inch divided by (0.005 inch per foot) = 240 inches.

If, as I speculated from my series of E-mails with Stabila's representative, the radius of curvature of a very high grade carpenter's level vial is about 18 inches, a millwright's or mechanic's level is more than ten times as sensitive as the high-grade carpenter's level . . . and a master level is more than a hundred times as sensitive as the high-grade carpenter's level.

lazlo
04-16-2008, 12:34 AM
That Swiss frame level is very nice Evan -- it's 0.02 mm per meter, or 5 second accuracy. That's twice as sensitive as the Starrett 199 Master Precision Level.


Bottom line, Evan's Stiefelmayer instrument is (100 divided by 61) = 1.65 times as sensitive as a Starrett 199, NOT twice as sensitive.

John, you lost me there. Starrett lists the 199 as accurate to 0.04 mm per meter. Evan's Stiefelmayer is 0.02 mm per meter. But you're saying it's not twice as sensitive, because the graduations are 2 mm == .0787" apart, where the Starrett 199 has graduations .1" apart?

http://catalog.starrett.com/catalog/catalog/groupf.asp?groupid=88

No. 199 Series Master Precision Level

* Ground and graduated main vial of 10-second accuracy; one division equals 1/2 thousandth (0.0005) of an inch per foot, or 0.04mm per meter

John Garner
04-16-2008, 12:48 AM
lazlo --

That's exactly what I'm saying.

Look at it this way: the bubble in a Starrett 199 moves 0.1 inch when tilted 0.04 millimeter per meter, the bubble in Evan's Steifelmayer level moves 2 x 2 millimeter = 4 millimeter, roughly 0.16 inch, when tilted the same amount.

John

Evan
04-16-2008, 12:49 AM
To say that, though, I have to make an assumption. The label on the level stock clearly states the vial sensitivity is 0.02 millimeter per meter, but doesn't complete the thought by telling us the corresponding displacement of the bubble in the vial.

Actually it does, but in German. It says "1 Stricht Ausschlag = .02 mm pro m". That translates to "1 stripe = .02 mm per meter".

darryl
04-16-2008, 01:31 AM
Well then. I think a glass tube can be bent that much without coming too close to snapping. One of these days I'm going to make a vial and play with it. Here's my plan- obtain a piece of plate glass about 1 inch wide by 6 inches or so long. I'll do the best I can to make the long edges perfectly flat. Then I prepare the tubing by heating and drawing out both ends to a smaller diameter so they can be reheated to seal them easily. Then clean the tube and stand it upright inside of another tube with the fluid of choice in it. When the fluid has infused the tubing nearly to the upper end, I heat that end and seal it. Turn the tube over and seal the end with a finger, then level it off to check the size of the bubble. Adjust as required, then seal that end.

Now I stand the plate glass on a long edge, and lay the tube atop that. With a shim at the center, I press down on the tube at both ends until they touch the plate glass. I glue it this way. The tube will have a small arc and this assembly should be quite stable. The assembly now is fitted into whatever I make for a base, with an adjustment to align it.

A piece of lexan could be etched with lines, which get filled with ink, and this side placed towards the glass tube and fitted as a protective cover.

For the math guys, what thickness of shim would give a curvature similar to what's being talked about? Say, radius of curvature of 100 meters, for example.

I could easily do a test to see what the breakage point is for bending the glass tubing. I'd use the same length of glass tube, a piece of flat tile or similar, and shims- basically creating the same setup as I've described. When the shim thickness is at the point where the glass tubing snaps before both ends can touch, I maybe cut that thickness in half and assume that to be a safe bend radius for the tubing. I think the curvature for a sensitive level would be less than this anyway.

I tried to keep this as short as I could, so I did not elaborate on any of the other fine details that would no doubt require attention.

I know I'm nuts, so no point in asking if anyone thinks I am for coming up with this idea. Can anyone think of any obvious drawbacks, or other issues? I know the accuracy of the tubing would be one.

oldtiffie
04-16-2008, 01:34 AM
My level and machine square are 0.02mm per meter which is 0.02:1,000mm = 1:50,000

The arc tan of this "gradient" is 0.02/1,000 = 0.00002
gradient angle = arc tan 0.00002 = 0.0011 arc degree = 0.0688 arc minute = 4.125 arc second

Just to put that into perspective, that is more than 60 times as accurate as a 5 arc minute rotary table, vernier of digital protractor, dividing head etc. etc. I have 20 second grarticules on my theodolite tribrach (base on which the theodolite rotates) and that is not too bad. Setting it to 10 seconds (by interpolation) is bordering on impratical and getting it to 5 is a waste of time as the conditions and accuracy required don't justify it. So 20 second it is most times.

The graticules/lines are 2mm apart. For instance, a less accurate but more sensitive level may be the more practical. Having a very accurate and very sensitive level can be and more often than not is a total PITA. John Stevenson alluded to this very thing in saying that his 0.01mm/m - that in NOT a "typo" - level was so hard to set that it was bordering on being useless for his purposes.

The "finer"levels have a very bad habit of "wandering off" on a random basis for no evident reasons. They are very sensitive to light and temperature changes.

As mentioned previously, the gradient or accuracy is one thing, but sensitivity may well be another.

Also as mentioned previously, using levels of this order of accuracy other than when you are VERY close is an exercise in futility. "Work up" to better accuracy with increasingly accurate levels.

The "Starrett" even if less accurate may well be more sensitive and is a damn side easier to use.

Evan's method of putting the level on the carriage - also suggested by J Tiers - and agreed by many others - is a very real approximation of how the lathe tool will be affected. It further tends to negate or cancel out the effects of any wear on the lathe bed and under-neath the apron in those pats that slide on the lathe bed ways.

Combine that with turning test-pieces (the ultimate proof) and even "Rollie's Dad's Method" as dictated by circumstances and considerable accuracy can be achieved.

As said else-where, my 0.02mm/m levels are a PITA to use and they stay in their boxes. I would much rather use the nominally less accurate but possibly more sensitive "Starrett" levels to get a better faster result.

(Jeez - I've just re-read that!! I can see some saying that "Tiffie has got religion - he likes "Starrett". I can see me being beatified yet!!! "Saint Tiffie" - err well, I don't think so - but I DO like "Starrett" (the old original USA-made ones that is).

But most times my very good machined "box" 1.0mm/m or 0.5mm/m carpenters levels will do most of my work.

Its "horses for courses".

Forrest Addy
04-16-2008, 02:18 AM
I suppose many know this but the better level vials are barrel shaped not bent. The early level vials were glass and bent by glass blowers over forms to get the desired radius then cut, closed on one end by hot working them, filled, and plugged. You can plainly seel this design in older levels.

The modern barrel shaped vials are injection molded with a blind end, filled then plugged and ultrasonically welded. High precision glass vials are lapped barrel-shaped, tested on a sine bar, closed blind on one end and to a pin hole on the other, then filled with ether/alcohol mixture, stopped with an elastomeric compouind (I think; it looks like Aluminastic,) and finally graduated.

The plastic vials regardless of angular resolution are very consistant and can be knocked out by the thousands using auromated processes. The glass high precision vials are still produced by hand crafted cottage industry methods and thus cost $35 to $300 each depending on quantity and features.

Geier and Bluhm sells high presision vials in brass capsules furnished with the adjustable posts. They sell in small quantities to individuals and in larger quantities to Starrett (How many #199 levels does Starett sell per year?) and other bulk customers. Geier and Bluhm vials aren't cheap but I submit they are the go-to place for any one seeking precision vials for home brew master precision levels. They have them down to 4 arc seconds and if you want to pay extra they will make them finer yet. Remember that higher resolution leve vials reach a point of diminishing returns at about 10 arc seconds. Their settling time goes up with finer angular resolution. If you want to obtain finer reading it would be better to secure longer vials having a larger spacing (5 MM for example) between 10 arc second graduations than fewer arc seconds per 2 mm graduations. Vials this fancy are custom made and if the maker has the graduation masks there's no great increase in costs to make a few specials Vs standard models.

Anyone desireing finer resolution than 4 arc seconds should look into coincident levels (some resolve 1/10 arc second) and electronic levels available in multiple switched resolutions. For example the Federal electronic level (now Mahr) which in its more sophisticated form can have two leveling heads connected to a sophisticated differential gaging head. There's fancier yet but I've run on to long.

There are 1,296,000 arc seconds in a circle. Working the math the radius of the wall of a 10 arc second vil havng in 2 mm between graduations = 425 ft radius. How much swell is that in the barrel shape of a level vial intended for a 24 mm long bubble?

This thread started with a question about a humble carpenter level and look how far we drifted pursuing the topic.

The end result of all this fancy leveling is more alignment and establishment of datum planes than brute perpendicularity to the force of local gravity. So don't get lost in leve per se. Instead understand that leveling is a means of obtaining the alignments and planarity originally established at the factory. Thus machine tool leveling is a means to an end not an end in itself.

John Stevenson
04-16-2008, 04:27 AM
Has anyone took the curvature of the earth into account in doing these calculations ?

.

Peter Neill
04-16-2008, 04:45 AM
Has anyone took the curvature of the earth into account in doing these calculations ?

.

The curvature being approximately 0.0015" every 12".

Peter

John Stevenson
04-16-2008, 05:04 AM
The curvature being approximately 0.0015" every 12".

Peter
So on that big TOS lathe I actually need to have an error from one end to the other ?

Thank Christ for that I can sleep easier at night.

I have already worked out that when leveling the shop door needs to be shut because of wind.
This is apparent on a very windy day when it can be 0.00000002" off over 80"
God knows how people with Myfords and South Bends get on ?

.

Peter Neill
04-16-2008, 05:18 AM
God knows how people with Myfords and South Bends get on ?

.

We don't have that problem of course, and can happily hold 0.0001" over 20" :D :D

Peter

oldtiffie
04-16-2008, 06:52 AM
So on that big TOS lathe I actually need to have an error from one end to the other ?

.................................................. ..
.................................................. .
.

Great John.

I suspect that there are a lot of graduates of "Tos" school.

At our place it isn't the shop door that needs closing because of "wind".

J Tiers
04-16-2008, 08:33 AM
I have already worked out that when leveling the shop door needs to be shut because of wind.
This is apparent on a very windy day when it can be 0.00000002" off over 80"
God knows how people with Myfords and South Bends get on ?

.

Drifting into the bozone here...........

There seems to be a "bubba" contingent..... you slam it down on the bumpy concrete floor and "git 'er done".

The next possible step seems to be ridiculed as accuracy better than 0.00000002" off over 80"........ Quite a silly suggestion, of course.

In point of fact, it is perfectly possible to be a mere 25 thou or so off over the length of the bed, and have problems as a result.... With 5 to 10 thou per foot error, there can be noticeable rocking and wear problems on a carriage which may be 10" or more long.

I can see no practical utility in getting much past what a "98" can do. That is getting on for a couple thou max over the length, which should be quite adequate to avoid problems..

If you have no other level than a carpenter's level, go ahead and try to use it. it probably won't hurt. But don't make the mistake of considering it the best precision tool for the purpose.

Lew Hartswick
04-16-2008, 08:47 AM
The curvature being approximately 0.0015" every 12".

Peter
That sure sounds like a lot but I know for a fact that when building a
long water fall (as in a shopping mall) it has to be taken into account.
...lew...

Evan
04-16-2008, 09:06 AM
The curvature of the Earth is much easier to see than most people think. There is a beach on Murtle Lake here that is perfect for observing the curvature. When sitting on the sand you cannot see the white sand beaches on the opposite shore about 5 kilometers distant. Stand up and they are clearly visible.

More important in John's case is the effect of what are called Masscons, or Mass Concentrations. They use flying gravitometers to find such things as large quantities of buried iron.

John Stevenson
04-16-2008, 09:16 AM
More important in John's case is the effect of what are called Masscons, or Mass Concentrations. They use flying gravitometers to find such things as large quantities of buried iron.

We know about this as the planes flying in to East Midlands International have altered their flight path because the workshop was affecting the compass's.

.

Capt Turk
04-16-2008, 09:52 AM
As long as a fresh bourbon and coke don't spill, or tump over, when you set'er on the ways, I say "Gitter Done" :D :D

Evan
04-16-2008, 09:56 AM
This reminds me of an argument I use to illustrate the fallacy of astrology. People think that the gravitational influence of Jupiter is important to how their life will turn out when in fact the influence of the milk delivery truck that was parked in the driveway was much greater.

lazlo
04-16-2008, 10:37 AM
The curvature being approximately 0.0015" every 12".

I posted a review here of Mike Morgan's scraping book (which is excellent, by the way), and someone here was fussing when I mentioned that Mike had a section on the curvature of the earth.

Mike is a professional machine tool rebuilder, and he was pointing out that when you align and re-scrape a large bed mill, grinder, or lathe, that the ways are easily long enough that the .0015" per foot is enough to induce a pretty significant error.

Evan
04-16-2008, 01:45 PM
Hmmm. Just how, I wonder? If the level is used to detect the planarity of the ways in the same manner as I did then a change in the gravity normal caused by the curvature will not affect the reading in measurable proportion to the change. Only if the level is turned 90 degrees to measure along the length of the bed will a change be seen, and that assumes the lathe bed is tangent to the gravity normal. Measurements perpendicular to the ways will read the same even if the lathe were long enough to encircle an ideal smooth Earth.

The only way that a deviation due to the curvature of the Earth would be noted in the perpendicular orientation would be if the bedways deviated from straight in the horizontal plane. The deviation would then be very hard to detect because it would be equal to the sine of the angle that the ways deviate from straight times the total deviation expected from the curvature. This would be many orders of magnitude below the detection ability of the best level.

Deviations from true planarity in the longitudinal direction are of far less importance since they will only affect the tool height in respect of the work. Even that assumes that the lathe extends at a tangent to the gravity normal and even then the affect on a level in the perpendicular orientation is minute.

In other words, the change in gravity normal with length will not affect the proper use of a level in determining the planarity of the lathe bedways perpendicular to the turning axis no matter how large the lathe. The effect over distance on planarity in the longitudinal direction will be so small to be of no consequence in nearly all cases. The 50 foot lathes that I routinely observed in operation at Cariboo Pulp and Paper would have a maximum tool height error of only about .002" at the tailstock, an insignificant amount when turning a 6 foot diameter roller 40 feet long. That assumes the roll stays perfectly straight in respect of the lathe.

Optics Curmudgeon
04-16-2008, 02:11 PM
.0015 inches per foot? Better check that math, that implies that the Earth has a radius of 1000 feet. Using an average radius of 3955 statute miles, the real value is more like 8.62E-7 inches per foot, or 0.862 microinches.

Joe

Forrest Addy
04-16-2008, 02:29 PM
I guess its time to recount an experience I had.

In 1996 I moved a big floor mill across a customer's shop to a foundation which had been prepared for it. It was quite an evolution but in the end the 80 ft long runway was leveled as close as my helper and I could get it using a couple of Taft Pierce 6 arc second levels using reveral technique. The bedways were set on grouted wedge levelers a system far more rigid than leveling screws.

Installation and alighment proceeded from there. In the end the customer contracted another outfit with a laser system to verify what i had done. No problem from me. I was confident in my work and interested in a comparison of the two systems. The laser showed my work to be very slightly high in the center. The tech and I were interested in the reason for the difference. Both laser and level showed consistant and repeatable readings. There were no significant heat sources to interfer with our readings. Our readings showed consistancy not enough to actually quantify the chordal height. We checked and re-checked. It was the laser guy who suggested the difference was the Earth's chordal height in 80 ft: we had verified the Earth really was round - or local gravitation had a lump in it.

Makes sense. Levels refer to gravity and define a spherical surface perpendicular to the local force of gravity. Lasers produce a beam of coherent light which defines a line having practically zero curvature in the Earth's gravitational field. If both apparatus are sufficiently sensitive, 80 feet will be enough reckoning distance for the difference between the two systems to become apparent. An arc second along the Earth's surface by the way is about 101 feet - 1/60 of a nautical mile. So what is the chordal height of a 4000 mile radius in 80 feet? Nah! The height in that distance is vanishingly small but why did we get the consistant readings? Another machine shop mystery.

lazlo
04-16-2008, 02:42 PM
In 1996 I moved a big floor mill across a customer's shop to a foundation which had been prepared for it. It was quite an evolution but in the end the 80 ft long runway was leveled as close as my helper and I could get it using a couple of Taft Pierce 6 arc second levels using reveral technique.

The laser showed my work to be slightly high in the center. The tech and I were interested in the reason for the difference but cause both laser and level showed consistant and repeatable readings and there were no significant heat sources to interfer with our readings. We checked and re-checked. It was the laser guy who suggested we discovered the difference was the Earth's chordal height in 80 ft

That's basically Mike Morgan's explanation. I'm at work, but I can quote the salient paragraph when I get home. I think I've posted it here before...

Optics Curmudgeon
04-16-2008, 04:17 PM
Fearing the worst, I've gone back and redone my original numbers. Originally I was using a spreadsheet I set up for a spherometer, which uses a "less rigorous" calculation, expecting the ratio of R to r to be relatively large compared to what we are discussing. The result is an increasing error as we get to very small angles. A purely trigonometric analysis gives a value of 3.129E-7 inches drop over 1 foot, and for Forrest, a value of .000935 high in the center of an 80 foot span. That's really good, using levels, regardless of their rated accuracy. Of course, I could still be fooling myself somehow, and invite independent confirmation.

Joe

Evan
04-16-2008, 04:49 PM
6378000 meters radius

40074155.2056 meters circumference

111317.09779 meters per degree

80ft= ~24.384 meters

.000219 degrees per 24.834 meters


cos (.000219)x radius = 6377999.9999534094878915819570174

Difference is .00004659051210841804298257045375234 meters

The rule of small angles say we can call it a triangle so divide by two to get .00002395256 meters or .02395256 millimeters

.02395256 millimeters = 0.000943014173 inches

Hey, wadda ya know? That's close enough to be called agreement.

BTW, metric is sooo much easier for this sort of calculation.

Forrest Addy
04-16-2008, 05:03 PM
"BTW, metric is sooo much easier for this sort of calculation." Maybe so but how do you convert back to cubits and quintals and other units what I was fetched up to unnerstand.?

lazlo
04-16-2008, 05:40 PM
A purely trigonometric analysis gives a value of 3.129E-7 inches drop over 1 foot

That's what Mike Morgan has too:

"Because of the curvature of the Earth, a level displays an error of approximately .00003" per foot. With this in mind, a surface that is 10' long when scraped or adjusted to indicate dead level along its entire length is actually .00015" low on each end. This is a minor problem with a surface this small but when setting up a machine 60' long, this curve begins to be an appreciable problem. With a machine of this length, the error on each end would be .0009" or nearly one thousandth of an inch. If the machine were required to hold part tolerances of less than .0005" this would be a problem."

By the way, that quote is out of the chapter on Machine Alignment, which is outstanding.

oldtiffie
04-16-2008, 06:40 PM
What we are talking about here is geodesy, which is often confused with but is related to geodetics and used in surveying and GPS - amongst others.
http://en.wikipedia.org/wiki/Surveying
http://en.wikipedia.org/wiki/Geodesy
http://en.wikipedia.org/wiki/Geodetic_system

Most Land survey work is in relatively small "parcels" in which the difference between the distance between 2 points measured over the chord (straight line) and curve/arc (of the earth surface) is so small that any differences distance and in elevation (height) - both real and relative - are so small as to be of no consequence.

Thus most land survey work is "Plane" survey which assumes a true horizontal plane.

When surveying over longer distances and/or larger areas the differences do become significant and so the survey becomes a geodetic work where the survey is conducted on and between points on a sphere. Spherical geometry is used instead of plane geometry. Gravity becomes an issue as well.

The curvature of the earth is most noticeable at the horizon.

Geodetic survey work requires and uses instruments and measurements of such precision that we can only contemplate them.

Astronomy is the same - sort of - mostly. Evan and others have demonstrated that as well over time in several previous threads.

So Forrest was quite correct and perhaps inadvertently demonstrated the relevance of geodetics.

This is what Forrest ran into on his machine bed. To do it once was extraordinary - but to repeat it was fabulous!!

That was damn fine leveling (surveying??) Forrest!!!

John Garner
04-19-2008, 12:33 AM
Evan and Company --

I've been thinking about Arc-Chord Separation question, and have come to believe that the 0.00094+ inch separation over an 80 foot length overstates the true value by a factor of two.

The "formula" is Arc-Chord Separation = Radius x [1 - Cosine (Subtended Angle / 2)], which would be 6378 kilometers x [1 - Cosine(0.000219 degree / 2)] = 6378 kilometers x [1 - Cosine(0.00011 deg)].

An alternative method of calculation would be Separation = (Chord Length / 2) x Tangent (Subtended Angle / 4) = 40 feet x Tangent (0.000219 degree / 4).

The first calculation is somewhat treacherous on to perform on a pocket calculator, because of limited accuracy in the computation of Cosines of very small angles. Done in Excel, though, both algorithms provide a calculated result of a wee bit less than 0.0005 inches, as does a very inexpensive Casio calculator only if using the second calculation.

Why? The Sines and Tangents of Small Angles are very nearly linear functions of the Small Angle magnitude, but not so with the Cosines of Small Angles. The Small Angle Approximation for Cosines is [1 - Cosine(Small Angle)] is approximately equal to 4 x [1 - Cosine(Small Angle / 2)].

John