Spurious Accuracy

[Neil WyattModerator @neilwyatt]

One of the things that regularly gets discussed is the accuracy that we should expect from our machines and that we should work to.

In practice, if you look at the specifications of various fits, it is extremely unusual to come across a requirement to work to better than 0.01mm or about half a thousandth of an inch (from here forward I will just use metric).

Even when testing tools, Schelsinger's limits don't rely on better measurement than 0.01mm, with precision being achieved my measuring deviations over an extended length of a bar, for example.

But lots of folk still chase 'tenths of a thou' so let's put the dimensions we work to into a context:

1mm = Typical size of a grain of sand, half a pinhead. About as small an object as you can handle without tweezers.

0.1mm = The thickness of a piece of paper.

0.75mm = roughly the smallest size difference you can see with the naked eye at a distance of about 300mm, normal reading distance. Diameter of a coarse hair.

0.02mm = diameter of a very fine hair.

0.01mm = diameter of a fungal hypha, length of a human chromosome.

0.005mm = diameter of spider silk.

0.005mm = about the size of a red blood cell.

0.001mm = the diameter of an E. coli bacteria, smallest eukaryote cells

0.0006mm = 600nm = wavelength of red light

200nm = smallest free living bacteria.

100nm = resolution of top-end Mitutoyo micrometer (accuracy 500nm)

180nm = deviation from round of the roundest man-made object

50nm = deviation of a top quality telescope mirror from perfect*.

13nm = the finest pattern you can feel with your fingertips.

1nm = how much your fingernail grows in a second.

0.1nm = size of a phosphorus atom.

Neil

*presumably this is better than the roundest object because it's relatively easy to measure and correct a mirror's accuracy by optical means.

[Neil WyattModerator @neilwyatt]

[Brian HParticipant @brianh50089]

Absolutely correct Neil, and that's from an ex aerospace Quality Manager!

[RainbowsParticipant @rainbows]

I recently purchased an Etalon micrometer that measures in 0.00005" or 0.00127mm as part of a job lot. Anyone got some bacteria cultures they want measuring?

[Michael GilliganParticipant @michaelgilligan61133]

Just out of curiosity, Neil … Why did you choose to jump straight from 'milli' to 'nano' ?

'micro' is a perfectly respectable, and convenient, SI prefix.

MichaelG.

[Neil WyattModerator @neilwyatt]

Posted by Michael Gilligan on 03/11/2017 17:57:43:

Just out of curiosity, Neil … Why did you choose to jump straight from 'milli' to 'nano' ?

'micro' is a perfectly respectable, and convenient, SI prefix.

MichaelG.

Because I wanted to use the units we use in our workshop, micrometers, calipers and machine dials give readings in decimal parts of a millimetre. Using micrometres would partly defeat my object of relating the sizes to everyday workshop experience.

Neil

[jimmy bParticipant @jimmyb]

I mostly work to units of “that’ll do”……

Jim

[Alan VosParticipant @alanvos39612]

Posted by Neil Wyatt on 03/11/2017 17:43:06:

One of the things that regularly gets discussed is the accuracy that we should expect from our machines and that we should work to.

<snip>

0.0006mm = 600nm = wavelength of red light

More like orange/yellow. Another comparison. For those who can afford them, the difference between a 1.001 and 1.0005mm gauge block (500nm), is one wavelength of green light.

[Steve WithnellParticipant @stevewithnell34426]

Oh well, I'd better ditch my '10ths' clock then…

[Tony Pratt 1Participant @tonypratt1]

Posted by Rainbows on 03/11/2017 17:49:07:

I recently purchased an Etalon micrometer that measures in 0.00005" or 0.00127mm as part of a job lot. Anyone got some bacteria cultures they want measuring?

How does it achieve that? Measuring in .0001" is problematic at the best of times [from a Toolmaker who walked the walk not talked the talk]

Tony

[larry PhelanParticipant @larryphelan54019]

How about 12"=1 foot

3ft=1 yard

51/2 yards= 1 perch [or something ]

Am I getting warm?

[Martin ConnellyParticipant @martinconnelly55370]

And who amongst us has a workshop with temperature controlled to the same standard as a calibration lab and can leave everything for three days to achieve the correct temperature for accurate measurements.

Martin C

[jimmy bParticipant @jimmyb]

I can leave stuff for weeks….still measures wrong…….
Jim

[colin hawesParticipant @colinhawes85982]

Absolute machine accuracy is unnecessary for hobby work: about the only things that matter are relative tightness between components and any old worn machine or micrometer can achieve that with care. Tolerances are usually only of interest for interchangeability of spare parts. Colin

[HOWARDTParticipant @howardt]

But how many get there measuring kit calibrated and work in a temperature controlled 20C space.

[Anonymous]

It's amazing the precision you can achieve in an armchair.

Andrew

[duncan webster 1Participant @duncanwebster1]

Once had a panic call from a contractor who had machined a stainless steel forging which cost many thousands and the inspector had failed it beacuse it was about 0.001" too small on a dimension of over 3 feet. Told them to measure it again and keep on measuring it until it passed. When it warmed up a bit in the afternoon it was OK. as far as I know the machine it was part of is still going strong after at least 30 years operation.

[Alan VosParticipant @alanvos39612]

Posted by Michael Gilligan on 03/11/2017 17:57:43:

Just out of curiosity, Neil … Why did you choose to jump straight from 'milli' to 'nano' ?

'micro' is a perfectly respectable, and convenient, SI prefix.

Micro is the only SI prefix which uses a greek rather than roman letter. Correct rendering of greek letters is less reliable than roman. Substition with roman 'u' can cause offence.

[vintagengineerParticipant @vintagengineer]

Having worked in Africa laying large size spun iron water pipes, we used to have the opposite problem with the heat when were connecting two ends together!

Posted by duncan webster on 03/11/2017 19:43:44:

Once had a panic call from a contractor who had machined a stainless steel forging which cost many thousands and the inspector had failed it beacuse it was about 0.001" too small on a dimension of over 3 feet. Told them to measure it again and keep on measuring it until it passed. When it warmed up a bit in the afternoon it was OK. as far as I know the machine it was part of is still going strong after at least 30 years operation.

[ChrisHParticipant @chrish]

"I mostly work to units of "that'll do"……"

Sounds much like mine – "she'll be right"……

 

Edited By ChrisH on 03/11/2017 21:00:21

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Neil Wyatt on 03/11/2017 18:02:52:

Posted by Michael Gilligan on 03/11/2017 17:57:43:

Just out of curiosity, Neil … Why did you choose to jump straight from 'milli' to 'nano' ?

'micro' is a perfectly respectable, and convenient, SI prefix.

MichaelG.

Because I wanted to use the units we use in our workshop, micrometers, calipers and machine dials give readings in decimal parts of a millimetre. Using micrometres would partly defeat my object of relating the sizes to everyday workshop experience.

Neil

.

I will leave you to make your point then, Neil … and will try to avoid further comment.

Except to say thay my everyday experience of small measurements is mainly microscope-related, and I therefore feel comfortable with instruments that have focus mechanisms scaled with one micrometre divisions.

MichaelG.

[RainbowsParticipant @rainbows]

Posted by Tony Pratt 1 on 03/11/2017 18:31:52:

Posted by Rainbows on 03/11/2017 17:49:07:

I recently purchased an Etalon micrometer that measures in 0.00005" or 0.00127mm as part of a job lot. Anyone got some bacteria cultures they want measuring?

How does it achieve that? Measuring in .0001" is problematic at the best of times [from a Toolmaker who walked the walk not talked the talk]

Tony

Here is my understanding of it, as you can see its not a standard micrometer to begin with.

Say you want something 0.5" in diameter. First you set the micrometer barrel to 0.5" then you slide in a gauge block or other standard of 0.5" inbetween the anvils (one is sprung). The lower dial on the clock face lets you callibrate the dial so you would set it to read 0. After that you put your work piece in and it will show how under or oversize it is compared to the standard.

This would all be done with it in a micrometer stand, don't want to warm it up at all.

[Neil WyattModerator @neilwyatt]

Not relying on feel is one of the things with these. The £1,800 Mitutoyo has clutch giving a repeatable force of 7-9 newtons.

But these things are for lab conditions, not workshops. A change in temperature of ONE degree will expand your 0.5" steel gauge block by more than 0.00005"…

Neil

[GeorgineerParticipant @georgineer]

I once had a junior engineer working for me who calculated the dimensions of a profiled rod (for a hydraulic damper) to about twelve places of decimals. He was surprised and rather put out when I told him that the last few places were smaller than the wavelength of light, and asked how he proposed to machine them. In the end we settled on 0.1 mm changes of diameter at 1 mm intervals, and it performed perfectly.

George

[HacksawParticipant @hacksaw]

Posted by jimmy b on 03/11/2017 18:07:55:
I mostly work to units of "that'll do"……

Jim

And "Tads"…..?

[Author]

Posts