Cylindrical squares or blocks

[lazlo]

consider a stepped cylinder with half of its length of one diameter and the other half a second, coaxial, diameter. Imagine then that the base of the surface gage contacts one step and the dial indicator tip contacts the other step. Would the dial indicator reading change as the stepped cylinder is rotated around its axis?

How about if, instead of having a stepped cylinder, we have a cone. Will the indicator reading change if the cone is rotated around its axis?

John, you're re-describing why taking a measurement at the base, as with Forrest/Harvey's method, doesn't detect taper or lean.

I've explained that several times since page 2, so I'm in complete agreement

[lazlo]

If tiptop made his measurements using a surface gage and indicator, using the technique explained by Forrest Addy

I got the impression from Jay's previous posts that he was doing those measurements with a DTI in the mill or lathe spindle, and that those measurements were the differences in the top and bottom DTI reading on opposite sides of the cylinder.

If that's the case, Jay's measuring 3 1/2 thou of taper, and the second two measurements, taken at the starting point after you've rotated the cylinder a 1/2 turn, indicate that there's virtually no lean. That's what you'd expect on cylinder turned on a lathe.

If those measurements are at the base with a surface gage on a surface plate, then that's 3 1/2 thou of eccentricity at the point the DTI contacts the cylinder (i.e., it's off the perpendicular plane with the base by 3 1/2 thou). To calculate the lean, you still need two measurements at top and bottom of the cylinder...

The Lucas Method data reduction algorithm isn't appropriate if the Forrest Addy / Machine Shop Trade Secrets Method of testing is used.

Agreed, but I think we're talking about two different tests. The surface gage/surface plate method can only check out of roundness at one point: it's basically checking whether that point where the DTI touches the cylinder is perpendicular with the surface plate. So as we've both mentioned, it can't detect taper on the cylinder.

By the way, the trick to Moltrecht's method which separates the taper error from the lean error is that after you've made the first two measurements at two points 180° apart, you return the mill/lathe spindle (and DTI) back to the first point, and then rotate the cylinder half a turn. Then take the third measurement.

If you don't do that 1/2 turn of the cylinder (and the third measurement), you can't distinguish between the taper of the cylinder and the lean of the cylinder.

[John Garner]

If tiptop made his measurements using a surface gage and indicator, using the technique explained by Forrest Addy and shown in the illustration old tiffie excerpted from Machine Shop Trade Secrets, the test results indicate that 1) tiptop's "cylinder" leans somewhere around 3 1/2 mils in one direction only, and 2) that there is some differential out-of-round between the top and bottom gage contact areas . . . but his measurements can't rule out taper or stepping.

As a mental exercise, consider a stepped cylinder with half of its length of one diameter and the other half a second, coaxial, diameter. Imagine then that the base of the surface gage contacts one step and the dial indicator tip contacts the other step. Would the dial indicator reading change as the stepped cylinder is rotated around its axis?

No.

How about if, instead of having a stepped cylinder, we have a cone. Will the indicator reading change if the cone is rotated around its axis?

No again.

Finding the stepping or coning requires either measuring the diameter of the pseudo-cylinder at the two gage-contact points or inverting the pseudo-cylinder relative to the surface gage and dial indicator and again rotating the pseudo-cylinder around its axis.

The Lucas Method data reduction algorithm isn't appropriate if the Forrest Addy / Machine Shop Trade Secrets Method of testing is used.

John

[lazlo]

Hi Jay,

Two opposite sides measure, "0" and ".007 respectively. Then the two opposite sides that are 90 degrees off measure, ".004 and ".004 respectively.

If you refer back to that formula I posted on the 2nd page, that's saying your cylinder has a taper of 3 1/2 thou, and virtually no lean.
Have you leveled and adjusted the lathe with the two collars method?

Keep at it!

Robert

Say the first measurement was D1, and the second measurement 180° opposite was D2:

Taper = (D1 + D2) / 2

Now the third reading, with the cylinder rotated and the DTI back at the start is D3:

Lean = [(D2 - Taper) - (D3 - Taper)] / 2

[lazlo]

Testing perpendicularity in the manner Forrest Addy describes is a common practice in metrology, but it's important to realize that the technique is totally blind to a difference in the diameters at the of the would-be-cylinder being tested at its "hard" and "indicator" contact points.
...
as with the conventional cylinder square, the surface-gage-and-indicator won't find taper in the "square cylinder" object.

Thanks John -- I've made that point several times in this thread:

Nope - wrong again. In both the case/s of Forrest and Harvey they both used known/proven squares, set their bases (in Forrest's case, the ball of the mast of his surface guage) to the base of the reference square and zeroed the dial indicator.

Jay has a wrist pin that he's trying to check for taper and lean. You can not prove the wrist pin by just checking the base. All that will do is show you the error at the base. Checking the base will not give you enough information to calculate the taper and lean of the cylinder.

You must run a DTI along the length of the cylinder to determine the tilt and lean.

On the last page of Machine Shop Trade Secrets that Tiffie posted, Jim Harvey is showing how to check the perpendicularity of a square block, so taper is not an issue. As Jay is finding, it is an issue with a cylindrical square:

[John Garner]

Alternative to a cylinder square

Testing perpendicularity in the manner Forrest Addy describes is a common practice in metrology, but it's important to realize that the technique is totally blind to a difference in the diameters at the of the would-be-cylinder being tested at its "hard" and "indicator" contact points.

Consider as an alternative to a conventional cylindrical square the variant commercialized by Hermann Schmidt, pictured on their website at https://www.hermannschmidt.com/produ...&idproduct=229

Hermann Schmidt's picture makes the squareness-checking process self evident, but as with the conventional cylinder square, the surface-gage-and-indicator won't find taper in the "square cylinder" object.

A roll-your-own "square cylinder square" should be fairly straight-forward, and the meaningful errors are constrained to a single plane . . . unlike a conventional cylindrical square that really needs to be measured and evaluated in both the "North / South" and "East / West" perpendicular planes.

Of course, if you wanted to made a four-sided "square cylinder square" . . .

John

[dp]

Sorry folks, I guess my mind was not working last night, I got my measurements wrong. This is what I came up with this morning after thinking about what I had posted. Two opposite sides measure, "0" and ".007 respectively. Then the two opposite sides that are 90 degrees off measure, ".004 and ".004 respectively. I am not sure why I was not able to get a cut on the end that was square or closer to square than that.

Sounds like the cylinder is tapered, no?

[tiptop]

Sorry folks, I guess my mind was not working last night, I got my measurements wrong. This is what I came up with this morning after thinking about what I had posted. Two opposite sides measure, "0" and ".007 respectively. Then the two opposite sides that are 90 degrees off measure, ".004 and ".004 respectively. I am not sure why I was not able to get a cut on the end that was square or closer to square than that. I am going to have to ponder this for a while. The reason for my exercise with these pins is to hopefully come up with something I know is square to use as a check or reference or clamping block for my other set up peices, such as what oldtiffie did in post # 17.

[oldtiffie]

"Get 'er done"

.......................................
......................................
First I would like to say that I am not a seasoned machinists or even apprenceship grade.

.........................................
........................................
Then I set it up in my old SB in a four jaw and steady rest and faced off the end. Then I skipped .200 on the next cut to leave a ridge on the outside. Then off to the surface plate. As best as I could tell one side had a lean of .00425, the opposite side was "0" and the side 90 degrees out was "0". So I guess my machining techniques leave a lot to be desired today. Maybe after class I will try something different if you all are not to bored with this subject? Jay

Thanks Jay.

First I would like to say that I am not a seasoned machinists or even apprenceship grade.

You have nothing to explain or apologise for - at all. You are doing OK - and I'd guess as good as many who claim to be Machinists. Many do things on a "monkey see - monkey do" (ie "rote learning") basis and get "thrown" when something new or the the same thing in a different way comes up. You have shown that you can learn and have an inquiring mind as well as tenacity and the ability to "roll with the punches".

So, you owe an explanation or excuse or sub-ordination to nobody.

That was good work - very good work. I was going to post pretty well that in my next post - and you've pretty well done it for me - many thanks.

All of this presupposes and requires the cylinders to be as round and straight as you can get them. If there are errors, they can be identified and "worked around".

In either case, you will have quite serviceable "squares" that should suit most work in most shops with reasonable care.

I thought later after making mine, that instead of buying new material (mine was $100 per meter), I could have used the column from a smaller pedestal drill. If the roundness and straightness are anywhere as good as the cylindrical or centre-less ground finish looks, they should do the job very well at minimal cost - none if you get lucky!!

Now all you need do is mark that "zero" point and your are "in" - a good job well done.

I guess you will appreciate why I wanted mine as big a diameter (3") and as "stumpy"/short (9") maximum as that way any errors in the base will (only) be multiplied by three at the top of the tallest (9") one/s. Just use the "zero error" part for use on your machine - the rest probably doesn't matter. Get a size that will fit in your lather steady-rest - makes it a lot easier to set up and face the ends off!!!

Another "quick and nasty" (but surprisingly effective) way of checking the "run-out" at any point on the circumference of the cylinder.

Get two metal washers - any size - even "punched" will do, that are say 1/8" to 1/4" thick. "Roundness" or "flatness" is not an issue. Just use your pedestal grinder and put a bevel (any angle) all the way round one side (to ensure contact above any bevel on the end of the cylinder/s). Leave a say 1/16" to 1/8" "land".

Fix the washers to the mill table (bevelled side down, "land" side up) - nuts, screws and "T"-nuts will do - so that the "lands" contact the cylinder (on one end) about 60 degrees (not at all fussy) apart.

Stand the cylinder on its end between and touching both washers. Put a dial indicator on the top of the cylinder surface - hold with a magnetic base or similar. Pre-load the indicator and set to zero. Rotate the cylinder against the washers while reading the indicator. Note where "zero" (two - at mid-defection points) is and mark them. That is your "zero" marks. Just set them to your work each time and any errors anywhere else won't matter. No errors at all is ideal, but the least possible in the circumstances is good enough most times if used with care.

"Get 'er done".

[tiptop]

First I would like to say that I am not a seasoned machinists or even apprenceship grade. I have only been at this for two years and I heard that after you serve your apprenticeship and then work for at least 10 shops for at least 2 years each you may be able to open your own shop. Having said that and knowing I will not live long enough to get the doors open this is where I stand. I took the time to play with one of these today. I checked it for being round as best I could with my caliper set up on different height gage blocks and found they were round. Then I set it up in my old SB in a four jaw and steady rest and faced off the end. Then I skipped .200 on the next cut to leave a ridge on the outside. Then off to the surface plate. As best as I could tell one side had a lean of .00425, the opposite side was "0" and the side 90 degrees out was "0". So I guess my machining techniques leave a lot to be desired today. Maybe after class I will try something different if you all are not to bored with this subject? Jay

[oldtiffie]

Fits

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

Me, I'm still new enough at this that I impress myself when I manage to cut, bore or turn +/- .0005, and I wouldn't be surprised to find that becomes my ultimate limit as a home shop guy, both for ability and need.

Frank. "1/2 a thou" is pretty good going in many shops - and all (and often more/better) than is required on many jobs.

An example.

"Tolerances" are one area that is often over done as regards level of accuracy that is sought - particularly if one part already exists and you have to make a part to fit to get the fit required - on that job.

Let's say you have a 2" shaft that has to be a light running (or any) "fit" with a part you have to bore out.

If the tolerance for a light running fit was, say:
2.0010"/2.0005" for the hole; and
2.0000"/1.9995" for the shaft;
the tolerance zone for the required fit when the parts are to be put together are:
highest: (largest hole - smallest shaft) = (2.0010" - 1.9995") = 0.0015" (+ve denotes "clearance" - ie not "interference"); and
smallest/least: smallest hole - largest shaft = (2.0005" - 2.0000") = 0.0005" (+ve denotes "clearance" - ie not "interference")

So the "clearance range " is (0.0015" - 0.0005") = 0.0010".

Now if say the shaft exists, and is say 1.9998" the hole can be bored to 1.9998" +0.0015/+0.0005 = 2.0013/2.0003" which gives the correct clearance fit tolerance range of 0.001" ( a full "thou") for the shaft and hole.

My point is that although the individual tolerances may be "tight", when taken together, that are surprisingly less and "better".

A similar approach is used for "interference" or "transition" fits - but watch out for the "signage" of the figures/values.

(It helps to sketch it out and it becomes more obvious and less confusing).

My point is that where "tolerance" and "fits" are concerned, if you "work the figures" a "difficult" "half a thou" may well be achieved using an "easy" "full thou".

Many parts in many shops are "made to suit the other part" and stay that way - so take advantage of "stuff".

Same applies to "errors" in say "cylinder/master squares".

If there is an error or "lean" of say 0.001" in a square, just rotate it 90 degrees and the "error" will "disappear" for use - but do check - just in case.

The engineer's/machinist squares at LittleMachineShop.com are excellent for most shop work as they are not more than 0.0006" "out" along the blade and parallel(ism) is very good too.

So, a bit of care and logic can get very good results by just using and taking advantage of what is available so that otherwise "difficult" jobs become less or or even "easy" (easier??).

Forrest's advice on the "gaps"was excellent and reasoned - as it always is.

I hope this helps.

[Frank Ford]

Frank's proposal is the equivalent of trying to determine the flatness of one of the original three plates by matching against the other plates, before all the scraping happens.

If you get lucky, and they're all close, then you can probably get a reasonable measurement out of it.

But if the wrist pins, or 3 plates, are way off, you won't be able to get a useful measurement.

Exactly. I wasn't even thinking of making any kind of adjustment. I was just wondering how accurate a visual check would be, presuming the pins were exactly square on the ends, or really close. Can one reliably see a tenth, presuming all things are super clean?

Me, I'm still new enough at this that I impress myself when I manage to cut, bore or turn +/- .0005, and I wouldn't be surprised to find that becomes my ultimate limit as a home shop guy, both for ability and need.

[oldtiffie]

Engineers/Machinists Square

I had been looking = without any result that I could use - for the details re. accuracy of Engineers/Machinist squares iaw/to DIN 875 for Class 1 (inspection/reference) and 2 (general machine shop use).

I could not find anything on British Standards (BS series) or the US series either.

The best I could do was at Littlemachineshop.com (LMS) at:

http://littlemachineshop.com/products/product_view.php?ProductID=2742

I expect that these are to DIN 875/2 or US grade B or 2.

Never the less the results for what-ever grade they are are not bad at all.

(Extracted from LMS link/page)

Includes 2", 3", 4" and 6" squares
Made from high quality tool steel
True right angles inside and outside
Beams and blades are precision ground
Blade is hardened
Machinist squares made in the US by Products Engineering Corporation (PEC).

Length of blade is measured from inner edge of beam to end of blade.

These squares are square within 0.0006" over the length of the blade. The two sides of the blades and beams are parallel within 0.0005" over their length.

I would appreciate getting the links to/for the applicable standards.

TIA

[oldtiffie]

Squaring up

If you have multiple pins, how close could you get to proving the accuracy of square, straightness, and taper by working clean, standing them on end and shoving them together to see what kind of crack of light you get between two in different rotated positions?

Thanks Frank.

I think I can answer that question.

The tubes were set up in the lathe with each end being set to a maximum TIR between highest and lowest being 0.02mm (0.0008") with maximum indicated differences between opposing jaws on my 4-jawed chuck of 0.01mm (0.0004").

I first tested them as pairs by putting them on my mill table (to show those without a surface plate that it is not always needed for even very good HSM shop accuracy). I put one against the other and rotated one each 90 degree and checked again. There was no discernible variation in the width of the narrow line of light. I inverted the same one and repeated the process - same result. I repeated the entire process for the other of the pair - same result. I then repeated the process for remaining 6 (3 pairs) of tubes - same result again. I also carried out a check with the Class 1 square - "spot on".

So I now knew that the ends were square to the axis of the cylinders.

It is a direct quote from my post #1 at:

http://bbs.homeshopmachinist.net/showpost.php?p=355980&postcount=1

I used those "squares" with great success to "true-up" my angle-plates, details of which are in the same post/thread.

I tested my angle plates with my Class 1 (DIN 875/1) try-square (aka Engineers/Machinists Square) and all was good.

To answer your question in regarding "light" in the "gaps". It is at least as good as you will get with a precision try-square.

To further check things out on the squares and to get the "multiplier effect" I put the stock of the try-square on the "barrels" of the cylindrical squares with the blade across the ends. I did this at least four places on each end - good result. I reversed the try-square and put the stock on the ends and the blade on the "barrels" - same good result.

So, all in all, I got what I think is a very good result in "fairly ordinary" shop with similar quality tools.

Perhaps I was just "lucky" - perhaps - or perhaps I didn't know what I was looking for or at - I hope not - but it's possible I suppose.

I "went" for the widest practical diameter so that any of my "squares" were more "stubby" than "tall/thin", both for accuracy and especially to get a much and as big a surface as I could get on the mill/machine table due to "location"/"square", area on the table, ability to carry a load (eg squaring angle-plates etc.), and resistance to distortion due to clamping etc.

The tube I used was precision cold-drawn steel with an OD of 3" and nominal 1/4" wall-thickness and nominal 2 1/2" i.d. (Yes that's right - we still make, use and can buy "inch" stuff in a "metricated" country (OZ)).

There is a "self-checking" method using 4 discs/"rounds" that is new to me but very obvious!!:

http://en.wikipedia.org/wiki/Machinist_square

[lazlo]

There's merit to Franks proposal. It strongly resembles the three surface solution to proving flatness

The three plate solution solves this problem neatly in that the blue indications can be taken up on vellum and preserved.

There's a big difference Forrest -- in the three plate method, the three plates start off way out of flatness, and through sucessive series of scraping and spotting, you gradually improve the flatness of all three.

Frank's proposal is the equivalent of trying to determine the flatness of one of the original three plates by matching against the other plates, before all the scraping happens.

If you get lucky, and they're all close, then you can probably get a reasonable measurement out of it.

But if the wrist pins, or 3 plates, are way off, you won't be able to get a useful measurement.

In any event, any further participation in this thread and I'm going to be criticized, so I'm signing off. I think Jay knows how to check his wrist pins.