Flatness Measurement

I've never hear of "fringes"... and I've been measure flatness to "light bands".. That is if the surface is not flat, then the light reflected off the surface will distort the light.

I checked my Machinery Handbook, Wikipedia, and my convert progam.. non show any reference to a "Fringe" Flatness is a distance measurement. you compairing the high of the highs and low of the lows. µm can be converted to a measure of distance as well as nm's, so please explain to me what a "Fringe" is and maybe I can help you.

if you take two flat surfaces that are transparent and place them face to face you can see fringes in scallopped patterns through the glass that waver as you press on them, showing that the glass has flexed. If you have a flat surface and a very slightly curved surface that touched in the center you will get concentric fringes. To use these as a metric you must use monochromatic light and you get reinforcement with successive half waves.

A read here will help.

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

and this too for drilling down

http://www.google.com/search?hl=en&q=%22interference+fringes%22+%2Bmeasurements&btnG=Google+Search

Hi !!!

For Checking Flatness with a Opitcal Flat & Monochromatic Light Source

It is necessary to have surface to be checked to be reflective. Usually after lapping the parts on Lapping Machine or even hand lapping dull Matt Finish is is created which needs to be made reflective using a polishing stand witha fine Polishing paper which is stretched tight over the flat surface of the stand .Part is rubbed on the paper until a reflective finish is obtained.

Part with Reflected surface is to be kept under the light source ,polished side up.and place the optical flat or top on the part to be tested for flatnness. Sliding is to be avoided in any circumstances

Light from the monochromatic source passes through the optical flatand is reflected from the polish surface. A pattern of Interferance fringes are produced,commonly called light bands.,check these bands for straightness. If thbands are straight and parallel the surface is flat. If the bands are curve,the surface is not flat but may flat enough for your requirement.

Your question is "How to tead Light bands and measure flatness" ?

Answer : Light bands or interferance fringes are craeted by a thin wedge of air between the component and optical flat. The bands or fringes,occur at right angles to the slope of the wedge and represent intervals of approximately 0.00029 mm ( 0.0000116") These bands and/or fringes are reflected from the polished surface of the component back through the clear optical flat. The dark bands seen benith the optical flat are not light waves but show where interference is produced by the reflectections of the two surfaces.It is these dark bands that are used to measure the surface flatness of the component.

The number and width of the bands seen have no relation to flatness.The wedge of air between the two surfaces control this factor. Press the center of the optical flat to achieve even spacing of the light bands for easy checking.

To ensure the accurate reading or the light bands, the angle at which the light bands are viewed is important.The optical flat should be viewed 80 and 90 degrees to the plano surface of of the optical flat,to ensure that light bands do not increase the value.

Counting Light Bands

To assess the number of light bands out of flat, count the number of bands that intersect an imaginary straight line superimposed through the center of the optical flat. If three bands intersect the imaginary straight line, the component is either covex or concave by three light bands depending on the point of pressure.

Trust above expaination satisfies your querry.

There ae several other method of inspection of flatness on manufactured machined parts : Such as use of Auto Collimator, Laser Interferrometer, Electronic Level , Use of Straight Edge and Filler / Gauge blocks , and flatness checking gauges for small parts using digital electronic probe mounted from bottom on a flat surface and moving the part on Flat surface over the probe stylus of which is spring loaded which touches and scans the peaks and valley on the object under inspection for flatness.

Udayan

udayan121049@yahoo.co.in

What a great summary and description. I've got a few more questions & am hoping you can shed some light (pun intended).

We are lapping parts and need to get a flatness of 5 light bands or better. I've got 1/5 and 1/10 optical flats, and a light box.

The parts being lapped are round, with holes in the center- aka donut shaped.

How do I get accurate readings of degree of flatness? Straight line through the center doesn't work well as the center is mostly missing, so there isn't much to count.

My real life doesn't match the nice evenly spaced consistent lines in examples I've seen. Typically we have curved lines, sometimes even full circles around a center point. I understand that this means the part is convex/concave. But, how do I tell HOW convex/concave it is or if it is within 3-5 light bands? Usually lines vary across the part also, they don't all follow the same pattern. Or do I need them flat enough that the convex/concave totally goes away?

I understand that the width/spacing of the light bands indicates the air gap between sample & optical flat, but this does add/subtract the number of lines which are there to be crossed. What is the happy medium or good measurement place?

Are there good books or websites to help with this evaluation?

Thank you in advance for your response.

-Lynnette Marks

Does this help?

yes, and no. These are the same ideals that I've found in other literature. We're not aiming for "perfectly flat", but within 3-5 light bands according to old company documnets I've been able to dig up.

I've got the plant working to bring the lapping tables back into flat themselves. Tables are at .003" across the surface, and parts being lapped still look concave/convex. I need to come up with understandable measurement/eval system for mfg to flag good vs bad parts. If part is slightly convex/concave, how to we measure the amount of flatness? I saw "ring counting" in one article, but that doesn't see right since the ring spacing depends more on the air gap than the flatness (I think), and it would vary based on how large the part is.

-Lynnette

The answer is to count fringes. Once the part has an optical flat placed upon the part, one can push on the flat to change the number of tilt fringes. When on rings occur, count the number of rings. If one counts 5 rings, then the part deviates from being flat by 5 light bands. If you want to be more precise, count dark bands as 1 and light bands as 1/2. For example, if there is a dark circle at the center, count the number of whole dark rings, say this is 5. Then if there is a white ring after the last dark ring, the part has 5and 1/2 bands.

Optionally, you could buy an interferometer from someone like Zygo. Then the interferometer will give a precise reading. But this is probably over kill based upon your discussion.

Udayan Patel & LabGuy have given a comprehensive description of how to measure flatness in fringes but it still does not get me closer to expressing this as a flatness in µm.

We measure fringe flatness under momochromatic sodium light which has a wavelength of 590nm. If I can see 8 fringes under this light would it be reasonable to say that the flatness is equal to 8x590nm = 4.7µm?

The brief answer is S = W*F/2

where S = surface

W = wavelength of light being used

F = fringes

If a HeNe laser was used in an interferometer, W = 633 nm and 4 fringes would equal 1266 nm.

Two beams of light interfere with each other producing fringes. These can beconsidered surface height differences (black fringes occur when the two beams are out of phase by 180 degrees). One of the beams is a high quality reference beam, the other beam is the surface under test. Since light reflects off of the surface being tested, the optical path is 2 times the height surface feature.

Thanks GWJ, thats the answer I was looking for. As usual I ended up learning more than I expected from the original question, thanks everyone.

At risk of stating the obvious, you control flatness in lapping by conditioning the lapping table or w.h.y. Are you using a lapping 'turntable' (sorry, abysmal memory, forgotten the name for the thing. You are correct, if against the reference flat you have curved fringes, the surface is concave/convex which is the principle "out of flat" mode of such a process. Bias the objects being lapped on the table to combat this, i..e wear the lap down where it's high (edge or centre if a turntable). The trick is to work out which way to go! Easy, except a) knowing how long it will take and b) combining this with achieving a surface finish!

This depends on the type of lapping machine. We use a machine which has a steel plate to which is attached a thin pad, the material of the pad depends on the material being lapped, we generally use felt or silk. The lapping medium we use is 2um diamond in oil which we use to flatten stainless & nickel steel parts.

OK, its years since I worked in that field, but where we needed to control form accurately we usually used bitumen lap facings. It usually produced a better surface finish as well.

Yes, our optic polishers use bitumen but it's not an option on the lapping machine we have.

If you are using a flat round plate, or an annular shaped turntable, I believe there are problems due to the speed and distance travelled by a point on the surface according to distance from the centre. This makes such plates ideal for making sperical lenses (convex or concave shapes) but less than ideal for achieving flatness. The wear rate of the lap naturally varies accross its diameter. The operator then has to try to compensate for this by using one portion more than another. It's an intrinsically unstable process with the operator applying corrections.

True but our machines naturally compensate for this. The components are placed into 3 cylindrical holders but are free to move around within the holder. We use loose fitting jigs to space the components but the group of parts within each holder tend to rotate, this moves them randomly from the centre to the outside of the lap.

Yes, I remember such an arrangement, but my point was that randomness isn't going to achieve flatness. One way or another the removal rates across the lap will vary; that is removal rate of the lap as well as the workpieces, so your lap tends to a curved state unless you're constantly trying to correct it. That's how it seemed, sometimes weeks were spent trying to find a balance.

I guess that's why the machines we use have replaceable lapping pads. The steel plate will never wear because the surface is not exposed to the lapping medium. I would also assume that the use of the lapping pads means that we cannot lap as flat as machines using a steel lap but that is not a problem in our application.

No, the steel plate will never wear but you can condition it to a certain shape before the felt or W.H.Y goes on (carborundum for instance). I'm slightly confused, it sounded as though you were after better flatness than you were getting currently. That suggested that you needed a bit more science in your laps.

The flatness we are getting is ok. For optics etc. we have the polishing done outside, the parts we lap in-house are in preparation for brazing to get vacuum tight braze joints. These parts have to be vacuum stoved to drive off contaminants & can distort slightly during the bake. We then lap the braze surfaces which need to be quite flat but not as flat as you might specify for optics. This gives us the best chance of achieving the vacuum tight joint.

I think I see but forgive me if I'm talking cods. Prior to the baking, lapping and brazing, I wonder how these parts have come into being and what they're made of. Some of the distortion might be avoidable if you can, somewhere in the upstream process, normalise the components and irradicate any 'locked-in' stresses prior to final machining.

Isn't the brazing process also a source of distortion itself?

I know what you are saying, some parts are turned & maybe milled, some are etched from sheet and those should be stress free. The vacuum bake will anneal all of the parts but has to be hotter than a normal anneal to achieve the cleanliness that we need. The brazing could introduce distortion so we have to carefully jig the parts during the process.

So isn't it necessary to repeat the cleansing after brazing?

The brazing operation is clean because it takes place in a nitrogen/hydrogen atmosphere. The ceramic/copper/steel assemblies are washed in de-mineralised water then it all goes into a clean room for final assembly. The last clean is by baking out in a vacuum chamber but because this can only be at about 350°C the bake out has to last several days.

Crikey, don't suppose it goes out of the door wrapped in a bit of cling film then!

They used to, when I started here products that customers were paying £8-10 000 for were bubble wrapped & packed in 2nd hand boxes. Now we either have custom packaging or use off the shelf foam filled flight cases. I figure it's worth spending £50 for the protection and for the customer to feel that he's getting a quality product.

Agreed,

I'm beginning to feel a strange contricting feeling,

as though we're about to fall

off

the

page!

I know this is an old post but I just stumbled across the answer to my original question in Machinery's handbook (p696, 26th edition) under surface measurement. Just goes to show that the answers are out there if you look carefully enough.

Sorry about the quality of the image but it's the best I could manage without ending up with a ridiculous file size.