On 01:31 AM 2/08/2002 +1000, John Haddy said:
>John,
>
>If the base units are metric then there will be NO rounding errors in
>conversion!
>
>e.g. if the base unit is one micron then 0.1 thou = 2540 base units
>EXACTLY - there's
>no error introduced.

Exactly - by choosing the base unit as 1 um (micro-metre) there is no 
truncation/rounding issue for back and forth swapping between imperial and 
metric *view* of the data.  I do not think anyone is suggesting forcing 
people to work in one unit or the other.  By making the change suggested 
(first suggested by Geoff Harland, to my memory, about 3 or 4 years ago), 
into micro-metres as the base unit, everyone can be completely happy.

With 32-bit arithmetic we can still represent massive boards - and still 
have some left over to allow full four quadrant design.

The problem for Protel may be the conversion of all the old libraries 
etc.  But I for one would have worn any small conversion problems.  And all 
conversion roundings can be solved by simply changing the base unit to a 
small one and reducing the maximum representable board.  With 32-bit 
numbers we can got to a base unit of 1 nano-metre and still represent 
boards metres across.

Someone else raised the issue of whether 25.4 mm to the inch is an 
approximation.  No it is exact.  Internationally (ISO I thik and accepted 
by all even the US to my knowledge) is that the inch is defined in terms of 
the metre (being 25.4 mm exactly).  The metre is actually now a variable 
quantity - it varies minutely as we determine the speed of light to a 
better precision.  I have not worked it out but I suspect we would need 
something like 40 bit arithmetic before we should start worrying about our 
components going off-grid when some scientist improves our measure of the 'c'.

Bottom line - as you say John.  25.4 is not a repeating decimal.  1/25.4 
may not be repeating ( I haven't tested) but it sure does need a lot of 
decimal places.  This is the root cause of the problem and why the (hidden) 
base units should have been converted with the DXP release.  This would not 
have affected anyone using imperial and would have improved life for those 
of using metric or mixed metric/imperial.

DXP should have been changed IMO.

Ian Wilson


************************************************************************
* Tracking #: 6E6958E2B65C5947BB1CED71F4425CE69C262F97
*
************************************************************************

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* To post a message: mailto:[EMAIL PROTECTED]
*
* To leave this list visit:
* http://www.techservinc.com/protelusers/leave.html
*
* Contact the list manager:
* mailto:[EMAIL PROTECTED]
*
* Forum Guidelines Rules:
* http://www.techservinc.com/protelusers/forumrules.html
*
* Browse or Search previous postings:
* http://www.mail-archive.com/proteledaforum@techservinc.com
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to