Linear division in early 20th century

[Michael GilliganParticipant @michaelgilligan61133]

Bazyle,

Returning to your original question

Was there a specific item that you had in mind?

MichaelG.

[BazyleParticipant @bazyle]

Tte purpose is immaterial to the problem. A length that is not a round figure in inches or mm is to be divided into a non round figure number of divisions whose increment seems to be a specific but not round figure in tenths of thousandths of inches. Possible to machine in a number of ways.

The puzzle is why the designer chose an obscure number when round figures could be used and still achieve the end of being non compatible with competing equipment. So did they choose the number and just make it, or is there a technique that drops this increment out? One proposal that comes quite close is a rack so I wondered if this was a likely technique.

[Michael GilliganParticipant @michaelgilligan61133]

Bazyle,

To this reader at least; your last posting seems rather contrary.

We still don't know the size & shape of the component, the number of holes, or the quantity in which it was to be made. All of these are relevant to the discussion … so why the secrecy?

One point worth making: There is nothing special about "round figures in tenths of thousandths of inches [or mm]" … there are plenty of base-units to choose from; fractions are just as valid as decimals; and polar co-ordinates are as valid as rectangular.

Consider the mounting-hole-pattern typically quoted for a NEMA 23 motor.

MichaelG.

[MICHAEL WILLIAMSParticipant @michaelwilliams41215]

For instance dividing an 11 inch length into 19 sections each of which sections then has a length which cannot be described easily in common measurements .

The super precision way is toolmakers buttons with specially made slips in between .

The poorest method is dividers and trial and error .

In the period about 1940 to 1965 there were hand operated devices available for setting out holes in one and two dimensions to good accuaracy . One of these was a big engineers square with rulings and a vernier reader on each leg . Together with some attachments it was possible to set out hole centres to about one thou accuracy .

The method chosen really depends on level of accuracy and availability of skills and equipment in the workshop .

[BazyleParticipant @bazyle]

Michael, this thread has thrown up some interesting tangents. Enumerating the specific problem would just produce pointless instruction on how to use a dividing head. It has been discussed over decades on other forums and media resulting in the rack based suggestion that seemed a little odd though accurate to tenths. So I just wondered how likely it would have been.

It is human nature to think in round figures. When people divide up an inch the normal result is 8 or 12, exceptionally 9, even 8.5 but not pi. If something unround like pi is forced into the equation they then twist it to make it round again by inventing module and DP. Sometimes there is an oddity introduced by metric /imperial conversion if using the 'wrong' units. Nobody ever says they'll make a 25.4 DP gear. (this is an interesting tangent worth discussion.

[TerrydParticipant @terryd72465]

Posted by John Haine on 03/04/2012 13:05:07:
Why is everyone talking about the early 19th century when the question was about the early 20th century?

Because to quote the Victorian philosopher Philarete Chasles:

“…..every great idea shall be a resumè of the past and a germ of the future.”

Best regards

Terry

Edited By Terryd on 04/04/2012 10:12:14

[MICHAEL WILLIAMSParticipant @michaelwilliams41215]

Hi Bazyle ,

I can't say that a rack has never been used but it certainly has never been a common way of doing things . It certainly has little merit compared to a screw based system or modern DRO .

The only common use of a rack for precision measurement is in the common dial gauge and dial caliper but these are really very small lightweight devices .

Just for interest in the very different world of mechanical computing all sorts of weird racks and gears have been used :

Pi racks , curved racks , Pi gears , snail gears , oval gears – the list is endless .

Regards ,

Michael Wiliams .

[TerrydParticipant @terryd72465]

Hi,

To divide odd numbers by other odd divisions mechanically, again, would not be difficult for the Victorians. Maudslay showed examples of very accurate threads cut on his precision lathes with up to 100 tpi. At that time it was also within their capability to produce extremely accurate rotary scales with any number of divisions using dividing engines such as those used by Ramsden.

It would not have been very difficult to combine these two techniques in order to measure linear scales. Whether or not they did is another matter. However the Screw Generating Engine shown earlier seems to suggest that the techniques were combined even at that early age. If it was known by the Victorians surely it was known by the Edwardians.

Regards

Terry

[MICHAEL WILLIAMSParticipant @michaelwilliams41215]

There were all sorts of measuring machines available to scientists and specialist engineers from very early on but very few of these ever got used in jobbing workshops .

[TerrydParticipant @terryd72465]

Hi Michael W,

Bazyle didn't define the workshop in his question. There were very many large factories involved in volume production of accurate work, not least in the field of munitions and weapons. We are after all talking about a period around WW1 where there were many large manufactories in a very industrialised society where the need for accuracies involved in interchangeability etc. were long established and well understood. Obviously not the sort of work for a back street jobbing workshop of one or two employees at that time.

Regards

Terry

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Bazyle on 04/04/2012 09:34:45:

Michael, this thread has thrown up some interesting tangents. Enumerating the specific problem would just produce pointless instruction on how to use a dividing head. It has been discussed over decades on other forums and media resulting in the rack based suggestion that seemed a little odd though accurate to tenths. So I just wondered how likely it would have been.

It is human nature to think in round figures. When people divide up an inch the normal result is 8 or 12, exceptionally 9, even 8.5 but not pi. If something unround like pi is forced into the equation they then twist it to make it round again by inventing module and DP. Sometimes there is an oddity introduced by metric /imperial conversion if using the 'wrong' units. Nobody ever says they'll make a 25.4 DP gear. (this is an interesting tangent worth discussion.

Bazyle,

You say "Enumerating the specific problem would just produce pointless instruction on how to use a dividing head."

I must disagree; but it is your perogative to maintain a position of "non-disclosure" if you wish. However; in the absence of any information about the component, we are all guessing.

My reference to the NEMA 23 motor was meant as a very strong hint towards what I suspect to be the underlying reasons for the "strange" [albeit undisclosed] dimensions that you found.

A cryptic question deserves a cryptic answer.

MichaelG.

[BazyleParticipant @bazyle]

Nice one! Nema 23 is 'strange' to a computer kid with DROs on his mill but "bleedin' obvious" to an old time draughtsman. Nema 42 must have been designed by his son. Both would throw a Frenchman.

My original question was answered early on – rack dividing not common – but I'm enjoying the links this is bringing up.

MW I would PM you but I think you need to have a profile for that.

[Anonymous]

Posted by Bazyle on 04/04/2012 14:27:01:

Nice one! Nema 23 is 'strange' to a computer kid with DROs on his mill but "bleedin' obvious" to an old time draughtsman.

Nice theory, but sadly disproved by the facts. I'd class myself as a computer kid with DROs, but nevertheless it's "bleedin' obvious" to me.

Andrew

[Gordon WParticipant @gordonw]

Well I'm an old time draughtsman, never heard of NEMA 23 or 42 ! What is it please?

[MICHAEL WILLIAMSParticipant @michaelwilliams41215]

Bazyle ,

My personal Email address is in your PM's .

Regards ,

Michael Williams

[MICHAEL WILLIAMSParticipant @michaelwilliams41215]

47.14 MM times root 2 divided by 25.4 = 2.625 inches .

The odd pitch comes from being four holes equispaced on a 2 5/8 inch PCD .

Michael Williams

[Michael GilliganParticipant @michaelgilligan61133]

Gordon,

NEMA specifies the "frame size" of a motor [typically, but not exclusively, a Stepper Motor].

23 is the size commonly used by hobbyists, for CNC conversions.

Datasheets and Engineering Drawings typically show rectangular coordinates for the four mounting holes. … and the dimensions are tricky, whether specified in inches or mm.

The reason being that the mounting was DESIGNED in fractional inches, with the holes on a pitch circle.

MichaelG.

[TerrydParticipant @terryd72465]

NEMA standards, only in the USA could it happen

[TerrydParticipant @terryd72465]

Hi Michael G,

Those are very interesting papers, I have long known about Ramsdens work but not in such detail. Fascinating stuff. Thank You for the guidance,

Best regards

Terry

[Michael GilliganParticipant @michaelgilligan61133]

Terry,

You're welcome … it's good to find another Ramsden appreciator.

MichaelG.

[John McNamaraParticipant @johnmcnamara74883]

Hi All

On Precision Racks

Google "precision machine hydrostatic rack"

In this case High accuracy and stiffness.

Cheers

John

[Gordon WParticipant @gordonw]

Thankyou, Michael. Yes know what you mean now. Just didn't do fract . h/p. Lots of fun with wheel stud spacing etc. tho'.

[Sub MandrelParticipant @submandrel]

I'm always impressed by all those electronic components with legs at 2.54mm and 1.27mm spacing

Letters To A Grandson was a series in ME by an ex telecoms chap who knew an awful lot (and found out even more) about some of these topics (such as accurate measurement of distance using 'etalons' and diffraction gratings.

 

Watt's micrometer he had the precision, but not the accuracy…

Neil

Edited By Stub Mandrel on 05/04/2012 22:29:31

[TerrydParticipant @terryd72465]

Posted by Stub Mandrel on 05/04/2012 22:28:24:

I'm always impressed by all those electronic components with legs at 2.54mm and 1.27mm spacing ………………………….

……….Watt's micrometer he had the precision, but not the accuracy…

Neil

Edited By Stub Mandrel on 05/04/2012 22:29:31

The components have to be backwardly compatible with components which were unfortunately developed in one of the four Imperial countries left in the World, and it wasn't Burma .

But the Grand Chancellor was accurate as well as precise. and you have to be accurate as well as precise to produce a 12" long nut to fit on a 2" diameter 50 tpi thread (600 threads in engagement.)

Regards

Terry

P.S.

ac·cu·ra·cy     [ak-yer-uh-see] Show IPA

noun, plural -cies.

1.

the condition or quality of being true, correct, or exact; freedom from error or defect; precision or exactness; correctness.

Terry

Edited By Terryd on 06/04/2012 08:08:15

[Anonymous]

Posted by Stub Mandrel on 05/04/2012 22:28:24:

I'm always impressed by all those electronic components with legs at 2.54mm and 1.27mm spacing

Not any more; legs are so yesterday! I've just finished designing a fairly complex board and only one IC had a lead pitch of 1.27mm, and none at 2.54mm. The majority of ICs didn't have any legs, being either BGAs (ball grid array) or LCCs (leadless chip carriers). The only big ICs that had legs were the Ethernet MAC/PHY and Ethernet switch at, I think, 0.65mm, which is of course 25 thou. All the other ICs with legs were 0.5mm. The use of BGAs is pretty much universal now, and not just for the larger ICs.

BGAs are used for several reasons. For large numbers of connections the packages are much smaller than equivalent leaded components. For example the processor we used had 423 connections, and that's a fairly low number. Even at 0.5mm lead spacing that's a big component. A large leaded package is more expensive to make and requires stamped out lead frames. Also, a large leaded package has significant time delays and parasitics associated with leads in the corners of the package.

If you think components are an odd mix of units, then PCBs are worse. Thickness and size can be metric or imperial, or both, depending on the final customer. Via holes are normally metric, but track and gap design rules are often imperial. To cap it all the copper on each layer is specified by weight. On the PCB mentioned above all layers used 0.5oz copper. That indicates the weight of copper used to cover 1 square foot of PCB, in our case the copper is 17.5 microns thick.

Regards,

Andrew

PS: Sorry about the off topic 'lecture'; it's been a hard few weeks!

[Author]

Posts