Re: gEDA-user: New Column: From the CAD Library

[Colin D Bennett]

On Mon, 7 Feb 2011 13:38:30 -0500
DJ Delorie  wrote:

> 
> > What are people using this software for that requires better than 10
> > millionths of an inch accuracy?
> 
> The problem people see is that a metric grid is rounded to the inch
> units, so a 45 degree line does NOT always end on a grid point, due to
> X and Y being rounded different ways.  So PCB adds an 0.01 mil stub on
> the end to connect the "diagonal" line to the grid point.

So that's why I get the little stubs...  That is actually a more
annoying problem than the length display error since those stray
"stubs" clutter the layout, especially when you want to select and
delete/move/modifiy groups of traces.

Regards,
Colin


___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

Re: gEDA-user: New Column: From the CAD Library

[Colin D Bennett]

On Mon, 07 Feb 2011 13:30:24 -0500
rickman  wrote:

> On 2/7/2011 11:21 AM, Peter C.J. Clifton wrote:
> > On Feb 7 2011, rickman wrote:
> >
> >> The real question is does the current method cause any problems?  
> >> When this was discussed a few months ago the answer was "no".  So
> >> why worry about 1 part in a million error?  Engineering is all
> >> about tolerances.
> >
> > We do have some problems with rounding and 45-degree lines, but I'm 
> > not convinced metric units internally is a magic bullet for that.
> >
> > Display of accurate coordinate in metric IS a problem though.
> 
> How so?  I thought the dimensions in these tools are done with 10 
> microinch resolution.  Isn't that enough for anything on the visible 
> horizon?  Actually, I wouldn't think the display is the problem, but 
> rather the problem would be generation of rounded data for output
> such as Gerber files.  What are people using this software for that
> requires better than 10 millionths of an inch accuracy?

I often see some odd values due to inexact conversion to millimeters in
pcb.  It slows me down when I'm trying to draw a footprint since I have
to constantly round the numbers in my head; while easy, you have to be
constantly doing this every time you click a point and it is a waste of
time.

See a simple example on the screen shot:
.
Here I used the "measure" tool to measure a distance between two points
on a 0.5 mm grid, but the displayed result is not correct.  Sure,
mathematically it is extremely close to the correct value, but for my
brain to convert this each time I glance at the position/measurement
display in pcb, it does accumulate since I refer to the display so
frequently.

Regards,
Colin


___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

Re: gEDA-user: New Column: From the CAD Library

[DJ Delorie]

> What are people using this software for that requires better than 10
> millionths of an inch accuracy?

The problem people see is that a metric grid is rounded to the inch
units, so a 45 degree line does NOT always end on a grid point, due to
X and Y being rounded different ways.  So PCB adds an 0.01 mil stub on
the end to connect the "diagonal" line to the grid point.


___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

Re: gEDA-user: New Column: From the CAD Library

[rickman]

On 2/7/2011 11:21 AM, Peter C.J. Clifton wrote:

On Feb 7 2011, rickman wrote:

The real question is does the current method cause any problems?  
When this was discussed a few months ago the answer was "no".  So why 
worry about 1 part in a million error?  Engineering is all about 
tolerances.


We do have some problems with rounding and 45-degree lines, but I'm 
not convinced metric units internally is a magic bullet for that.


Display of accurate coordinate in metric IS a problem though.


How so?  I thought the dimensions in these tools are done with 10 
microinch resolution.  Isn't that enough for anything on the visible 
horizon?  Actually, I wouldn't think the display is the problem, but 
rather the problem would be generation of rounded data for output such 
as Gerber files.  What are people using this software for that requires 
better than 10 millionths of an inch accuracy?


Rick


___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

Re: gEDA-user: New Column: From the CAD Library

[Peter C.J. Clifton]

On Feb 7 2011, rickman wrote:

The real question is does the current method cause any problems?  When 
this was discussed a few months ago the answer was "no".  So why worry 
about 1 part in a million error?  Engineering is all about tolerances.


We do have some problems with rounding and 45-degree lines, but I'm not 
convinced metric units internally is a magic bullet for that.


Display of accurate coordinate in metric IS a problem though.


___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

Re: gEDA-user: New Column: From the CAD Library

[rickman]

On 2/6/2011 3:00 PM, Peter Clifton wrote:

On Sun, 2011-02-06 at 12:27 -0500, rickman wrote:

Parts should match the rest of a system I think.  The dimensions of a
part can be stored in the units that best suit the system.  It seems
silly to store a part as metric and let the system round it off to
inches on the fly rather than to do the round off when the part is
created.  But that's just my opinion.

Conversely, to me it seems silly to deliberately loose information on
the part's original specification.


It is storing in metric that preserves the accuracy.  So I probably 
didn't say it correctly above.




If the manufacturer specifies some dimension as 1mm, that is how it
should stay represented in our source design files, not 39.3700787 mils
(or however one chooses to round it).

PCB already supports reading units like 1mm in its input files, but it
always saves out the rounded imperial representation.

Sure - it probably doesn't matter if you round everything based on true
position when creating the file, but if the user were to copy+paste part
of that new (rounded) footprint using an imperial grid, to make a longer
part - rounding errors could easily be compounded.


The issue is not the grid, the issue is the internal units used and 
stored.  The user can and should be able to work in what ever grid suits 
the purpose.


The real question is does the current method cause any problems?  When 
this was discussed a few months ago the answer was "no".  So why worry 
about 1 part in a million error?  Engineering is all about tolerances.


Rick


___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

Re: gEDA-user: New Column: From the CAD Library

[Markus Hitter]

Am 06.02.2011 um 16:24 schrieb Peter Clifton:

Imperial parts are not a problem for a sufficiently fine metric  
grid. I
don't think we should remove the option of working on a Mil grid  
though.


I'm wondering what's the advantage of having an internal grid at all.  
Why not just use doubles, describing a position in meter or  
millimeter? About all mechanical CAD and picture drawing applications  
do it that way.



Markus

- - - - - - - - - - - - - - - - - - -
Dipl. Ing. (FH) Markus Hitter
http://www.jump-ing.de/







___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

Re: gEDA-user: New Column: From the CAD Library

[Peter Clifton]

On Sun, 2011-02-06 at 12:27 -0500, rickman wrote:
> 
> Parts should match the rest of a system I think.  The dimensions of a 
> part can be stored in the units that best suit the system.  It seems 
> silly to store a part as metric and let the system round it off to 
> inches on the fly rather than to do the round off when the part is 
> created.  But that's just my opinion. 

Conversely, to me it seems silly to deliberately loose information on
the part's original specification.

If the manufacturer specifies some dimension as 1mm, that is how it
should stay represented in our source design files, not 39.3700787 mils
(or however one chooses to round it).

PCB already supports reading units like 1mm in its input files, but it
always saves out the rounded imperial representation.

Sure - it probably doesn't matter if you round everything based on true
position when creating the file, but if the user were to copy+paste part
of that new (rounded) footprint using an imperial grid, to make a longer
part - rounding errors could easily be compounded.


-- 
Peter Clifton

Electrical Engineering Division,
Engineering Department,
University of Cambridge,
9, JJ Thomson Avenue,
Cambridge
CB3 0FA

Tel: +44 (0)7729 980173 - (No signal in the lab!)
Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me)


signature.asc
Description: This is a digitally signed message part


___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

Re: gEDA-user: New Column: From the CAD Library

[rickman]

On 2/6/2011 10:24 AM, Peter Clifton wrote:

On Sun, 2011-02-06 at 10:55 +0100, Bert Timmerman wrote:


One of Tom's "issues" that is to be kept in pcb are most of the mil grids
because of the bazillion perf board and mil based parts on the market, to be
bought for cheap by hobby-ists, or for "Quick-and-Neat" proto boards (we
don't play or do dirty ;-).

Just my opinion on the subject.

Imperial parts are not a problem for a sufficiently fine metric grid. I
don't think we should remove the option of working on a Mil grid though.
I do it most of the time, even though I realise it is a habit best got
out of.


I don't think you understand the intent.  I don't think anyone is 
suggesting any changes to what a user sees.  The issue is what base 
units are used internally and in libraries.  By using metric values, 
imperial (is that really the right term?) measures (aka inches and mils) 
can be represented and calculated with no loss of precision.  However, 
when inch type units are used as the fundamental measurement, it can be 
harder to get adequate precision to represent metric units.


1 inch = 25.4 mm... exactly
100 mm = 3.937007874016... sort of.

In reality if you use enough precision in the base units you can always 
get "close enough" which is what engineering is about.  But some folks 
have an issue with not working in the "best" possible manner.


I'm not sure this is really an issue.  There would be significant pain 
making the change now.  If the change is to be made at a later time, 
will it really be any more painful?  So if there is no significant 
reason to switch and the pain does not increase if delayed... why bother 
with it now...?




Way forward:

Metric "nm" grid internally, parts defined in whatever units the
vendor's controlling dimensions are in.

This might require relative origins to be used between the part design
coordinate and the board's snap-grid, but that seems to be mandated by
various IPC standards anyway.


Parts should match the rest of a system I think.  The dimensions of a 
part can be stored in the units that best suit the system.  It seems 
silly to store a part as metric and let the system round it off to 
inches on the fly rather than to do the round off when the part is 
created.  But that's just my opinion.


Rick


___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

Re: gEDA-user: New Column: From the CAD Library

[John Griessen]

On 02/06/2011 09:24 AM, Peter Clifton wrote:

Imperial parts are not a problem for a sufficiently fine metric grid.


+1

I heard Tom Hausherr talk last year about grids, land pattern generators.
For his suggestion to just "go metric", it works with old inch spaced
protoboard layouts just fine if enough resolution is used.  It's always
going to be valuable to combine the old with the new, where you might have
very fine line spacings in the new on a metric grid, the base drawing being
a metric grid, and placing the inch stuff on a calculated-from-metric "semi"
inch grid only means some holes are off by 0.025mm or less if using a 0.05mm 
grid.

Meanwhile, any scripted tool we can write to help make parameterized footprints
from datasheets will be one of the biggest aids to pcb layout.  Much of our 
discussion
about ways to vote on or log successful uses of footprints in production of 
boards
seems invalidated by the industry adoption of most nominal and least
density styles of footprints where the IPC specifies extension of pad area 
beyond
a part lead in mm.  The manufacturer's suggested footprint is pretty much 
ignored
by such a standard, yet deriving footprints from lead shapes for 2 terminal 
chips and
gull wing leaded parts is what the usual
commercial tools (and thus the usual commercial layout folks) do.  When it 
comes to
chip scale packages I didn't see any suggestion to change the MFRs land pattern.

Hausherr does suggest BGAs get the three density levels, but the denser
ones are only used on throw away products.  IPC recommendations seem to take
BGA shapes as published by MFRs and make them a little larger for robustness.
The writeup on http://blogs.mentor.com/tom-hausherr goes over the same material
I heard in person last year about "fairways" between microvias put in BGA ball 
lands
planned for many tracks in inner layers of 8 and up layer boards used for dense
products these days.  Figure 20 and 21 are for board layouts, but look like
chip layouts of the 1990's!

Blind buried microvias here we come?

John

--
Ecosensory   Austin TX


___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

Re: gEDA-user: New Column: From the CAD Library

[Peter Clifton]

On Sun, 2011-02-06 at 10:55 +0100, Bert Timmerman wrote:

> One of Tom's "issues" that is to be kept in pcb are most of the mil grids
> because of the bazillion perf board and mil based parts on the market, to be
> bought for cheap by hobby-ists, or for "Quick-and-Neat" proto boards (we
> don't play or do dirty ;-).
> 
> Just my opinion on the subject.

Imperial parts are not a problem for a sufficiently fine metric grid. I
don't think we should remove the option of working on a Mil grid though.
I do it most of the time, even though I realise it is a habit best got
out of.

Way forward:

Metric "nm" grid internally, parts defined in whatever units the
vendor's controlling dimensions are in.

This might require relative origins to be used between the part design
coordinate and the board's snap-grid, but that seems to be mandated by
various IPC standards anyway.

-- 
Peter Clifton

Electrical Engineering Division,
Engineering Department,
University of Cambridge,
9, JJ Thomson Avenue,
Cambridge
CB3 0FA

Tel: +44 (0)7729 980173 - (No signal in the lab!)
Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me)


signature.asc
Description: This is a digitally signed message part


___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

Re: gEDA-user: New Column: From the CAD Library

[Kai-Martin Knaak]

Bert Timmerman wrote:

> I would like to see some, if not all, of these "standards" reflected in the
> pcb-lib some day _and_ these recommendations end up in the (gEDA) pcb
> documentation, just to prevent "error 404" from happening.

copy-paste would require that he releases the text under GPL, or compatible.


> One of Tom's "issues" that is to be kept in pcb are most of the mil grids
> because of the bazillion perf board and mil based parts on the market, to be
> bought for cheap by hobby-ists, or for "Quick-and-Neat" proto boards.

ack.
But the role of mil and mm should be exchanged. The file format should be 
metric at the bottom. Imperial measures should be derived on the fly.

---<)kaimartin(>---
-- 
Kai-Martin Knaak
Email: k...@familieknaak.de
Öffentlicher PGP-Schlüssel:
http://pool.sks-keyservers.net:11371/pks/lookup?search=0x6C0B9F53



___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

Re: gEDA-user: New Column: From the CAD Library

[Bert Timmerman]

Hi all, 

> -Original Message-
> From: geda-user-boun...@moria.seul.org 
> [mailto:geda-user-boun...@moria.seul.org] On Behalf Of Peter Clifton
> Sent: Saturday, February 05, 2011 3:49 PM
> To: gEDA user mailing list
> Subject: Re: gEDA-user: New Column: From the CAD Library
> 
> On Fri, 2011-02-04 at 08:17 -0500, Bob Paddock wrote:
> > "New Column: From the CAD Library
> > 
> > When creating a CAD library, there are dozens of things to consider 
> > that are often overlooked or not even considered that will directly 
> > affect the quality of part placement, via fanout, trace 
> routing, post 
> > processing, fabrication, and assembly processes. This 
> article, Part 1 
> > of a series, introduces aspects that should be considered when 
> > creating CAD library parts."
> > 
> > http://www.pcbdesign007.com/pages/zone.cgi?a=74214
> > 
> > Has tips worth considering.
> 
> Note that the series continues on his blog. It is VERY good, 
> and contains a lot of details I was looking for recently.
> 
> http://blogs.mentor.com/tom-hausherr/
> 
> --
> Peter Clifton
> 
> Electrical Engineering Division,
> Engineering Department,
> University of Cambridge,
> 9, JJ Thomson Avenue,
> Cambridge
> CB3 0FA
> 
> Tel: +44 (0)7729 980173 - (No signal in the lab!)
> Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me)
> 

+1

I've been following Tom Hausherr's achievements (the formerly PCB-Libaries,
now Mentor Graphics) for a couple of years now, and I think that this blog
is mandatory reading before "Getting Started" with pcb design.

I would like to see some, if not all, of these "standards" reflected in the
pcb-lib some day _and_ these recommendations end up in the (gEDA) pcb
documentation, just to prevent "error 404" from happening.

Of course, there are others sources of information to read before putting
the first trace on a pcb, like app notes and part datasheets from vendors.

One of Tom's "issues" that is to be kept in pcb are most of the mil grids
because of the bazillion perf board and mil based parts on the market, to be
bought for cheap by hobby-ists, or for "Quick-and-Neat" proto boards (we
don't play or do dirty ;-).

Just my opinion on the subject.

Kind regards,

Bert Timmerman.



___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

Re: gEDA-user: New Column: From the CAD Library

[Peter Clifton]

On Fri, 2011-02-04 at 08:17 -0500, Bob Paddock wrote:
> "New Column: From the CAD Library
> 
> When creating a CAD library, there are dozens of things to consider
> that are often overlooked or not even considered that will directly
> affect the quality of part placement, via fanout, trace routing, post
> processing, fabrication, and assembly processes. This article, Part 1
> of a series, introduces aspects that should be considered when
> creating CAD library parts."
> 
> http://www.pcbdesign007.com/pages/zone.cgi?a=74214
> 
> Has tips worth considering.

Note that the series continues on his blog. It is VERY good, and
contains a lot of details I was looking for recently.

http://blogs.mentor.com/tom-hausherr/

-- 
Peter Clifton

Electrical Engineering Division,
Engineering Department,
University of Cambridge,
9, JJ Thomson Avenue,
Cambridge
CB3 0FA

Tel: +44 (0)7729 980173 - (No signal in the lab!)
Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me)


signature.asc
Description: This is a digitally signed message part


___
geda-user mailing list
geda-user@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user