measuring wall thickness 2

[idanbrk]

hi,

i need to measure the wall thickness of a certain product , the product is made of ceramics and has an egg like shape which makes it very difficult to measure, in addition the wall thickness is not constant and varies along the profile . the average wall thickness is about 8.5 mm.
i tried using standard mechanical measuring tools (all sorts of big calibers and similar methods) , an ultrasonic gauge and i even found a kind of magnet based gauge but they all have the same problem- not accurate enough, i need an accuracy of at least +- 0.01 mm , if anyone has an idea on how to do it id love to hear it.

thanks

 

[MintJulep]

http://www.starrett.com/metrology/m...urrentPage=1&displayMode=grid&sortBy=none/asc

 

[Nereth1]

You have to be so damned careful using stuff like that at 0.01, just because you double the chance for operator error to ruin the reading.

There are external calipers that look similar to what MintJulep posted, that have digital readouts and read thicknesses directly. We have some that read to 0.005mm (and are therefore probably good to about 0.01 perhaps). Get one with the right tips, or even get one and grind the tips to the right shape for your application, calibrate, and go?

 

[rb1957]

CMM ? but that'll probably only give you the outer shape.

average thickness is 8.5mm and you need an accuracy of 0.01mm ? boy you must have defined a lot of (three?) deciminal places in your dimensions.

how did the manufacturer inspect it ? (to verify that it was to drawing)

is it a complete "egg" shape, or open ?

are you prepared to sacrifice a sample ?

another day in paradise, or is paradise one day closer ?

 

[btrueblood]

A good CMM will get the inner profile as well, if you can provide access for it (presumably you have a hole somewhere in order to have tried using calipers). You may need a whole set of different probe tools to reach around corners, etc.

Another trick would be to cast (using eg. replicast or a low expansion wax) both i.d. and o.d. surfaces, with a tooling insert cast into both surfaces so that the distances from a datum can be measured and the two profiles reconstructed. The problem is that you need to be able to extract both castings, which means no re-entrant surfaces, so you are limited to only parts of the egg shape in any given casting.

 

[GregLocock]

Archimedes

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[racookpe1978]

That would only give the average wall thickness over the whole egg. Any local place or many local places could be thicker or thinner than the average.

 

[rb1957]

ultrasound ? will measure skin thickness, but the accuracy demanded will be challenging for any instrument ... precisely measuring the thickness will depend on determining the normal ... i guess you could sweep through a bunch of angles and get the minimum.

displacement (archimedes) if you can get the external volume (easy) and the internal volume (accessable ?) and as pointed out above will give the average thickness only.

another day in paradise, or is paradise one day closer ?

 

[idanbrk]

the client uses an XYZ machine to inspect it , the shape is opened so its not a problem to insert a measuring device inside , it looks like a an egg that was cut in half

 

[ajack1]

What about something like this.

 

[idanbrk]

cant use what ajack1 suggested or mintjulep because the part is very long (about 400 mm ) and i need to measure the thickness threw all the length and this plier shaped devices are not accurate enough when they are long because they bend

 

[rb1957]

an egg shape, or an open (half) egg shape imples something spherical, equal dim'ns . now one dim'n is very long ... so it's more like a channel. can't you measure it section by section (rather than looking for something that will encase the part ? it sounds like you need some laser device that'll scan both surfaces ? i'd've thought a CMM could do this ?

another day in paradise, or is paradise one day closer ?

 

[racookpe1978]

Then you need to make a "non-bending" specialized micrometer that will give you results to the accuracy at ALL wall angles the tips will encounter at all temperatures you are going to use the non-bending micrometer in.

Expensive, but you are going to have to spend it -> if you really, really really need that accuracy.

 

[tygerdawg]

"Expensive, but you are going to have to spend it -> if you really, really really need that accuracy."
My thoughts also. And the laser / non-contact approach, too.

Define your project parameters better. I'm still trying to discern the size of the object you want to measure. I see 8.5 mm thick (variable) and 400 mm long. Give more specifics if you want good answers (W x D x H, material, surface roughness, etc.). AND what is your budget? This can be done, but I suspect it will be a custom gage. Do you have a budget for that? And what throughput do you require for measuring parts? One or two, or twenty per shift?

If you could relax the +/- 0.01 mm resolution requirement, then I would propose a robotic system. Could be done with a used robot.
1. Use the CAD model to develop measurement scanning paths on the top/bottom surfaces. Each path point has a normal surface vector for perpendicularity. Easy with CAD tools available today (<wink> of which I sell).
2. Use a high resolution laser displacement sensor (Keyence is my favorite, but there are many others) to measure top/bottom.
3. Do the 6DOF arithmetic to generate thickness values.
4. The challenge to this is the best robots on the market today have repeatabilities of +/-0.020 mm or so. Darn good for a robot, but won't meet your +/-0.01 mm requirement.

Option2: You could continue with the laser displacement sensor concept, but mount it on a CMM or an articulated arm CMM (if the repeatability is suitable). Fixturing your part repeatably would be critical.

Option3: Look at the Keyence applications guides, they will give ideas of mounting TWO sensors through which a product is passed to determine thickness. Many types of laser displacement sensors available: 1D with a tiny laser spot, 2D with a scanning line, 3D (magic) to provide depth. Contact Keyence Applications, maybe they have a solution already.

Option4: go the ol' skool hard tooled gage route. Can be done, but $$.


TygerDawg
Blue Technik LLC
Virtuoso Robotics Engineering

http://www.bluetechnik.com

 

[racookpe1978]

Yah know, we never asked if he only has one of these eggs - as if for a university or research single-use gadget, or if he expects to make a bunch of them - each of which is going to vary of course by some additional margin.

I didn't know robots could get a positioner within 0.02 mm accuracy in 3D! Nor that a radar/laser/ir sensor could repeat a 0.02 mm reading without hitting the surface.

 

[3DDave]

I wonder what the usage is that is sufficiently sensitive to thickness.

Not for this particular application, but if the use is sensitive to a characteristic, that suggests the usage itself is a means of measuring it.

For example, no one makes AFAIK, a CMM that can track 1/10th wave over a 2 meter mirror as is required for space telescopes, but the optics used in the telescope certainly can - hence that is the method used to measure the precision of the optics rather than mechanical measures.

OTOH translucency is a related characteristic that is very sensitive to thickness. Maybe that's an option, though the OP is looking for a 0.1% or less variation.

 

[JStephen]

I'm not sure how you'd even define the thickness, where it is variable and on an irregular surface.

 

[racookpe1978]

See, if this is not a one-of-a-kind part - or is the first of many that must be built - then you drill numerous holes in enough test pieces to establish (to whatever level of statistical process control you need) that the process is building good parts.

At each of 200 or 300 holes in 3 - 10 - 30 - 40 eggs, you measure the wall thickness at the hole.

 

[idanbrk]

how do i make a "non-bending" specialized micrometer that is at least 300 mm long and still light enough so a man can hold it while he is measuring?

 

[GregLocock]

Engineering?

If you can't figure it out yourself, designing an instrument like that seems possible. I'd be thinking strain gages

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?