X-Authentication-Warning: delorie.com: mail set sender to geda-user-bounces using -f X-Recipient: geda-user AT delorie DOT com Message-ID: <4F9C3F46.2000404@schinagl.nl> Date: Sat, 28 Apr 2012 21:04:38 +0200 From: Oliver Schinagl User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:11.0) Gecko/20120416 Thunderbird/11.0.1 MIME-Version: 1.0 To: geda-user AT delorie DOT com Subject: Re: [geda-user] Metric vs Imperial, Grid snapping References: <4F9BFF5D DOT 9010407 AT schinagl DOT nl> <1335634972 DOT 2539 DOT 13 DOT camel AT AMD64X2 DOT fritz DOT box> In-Reply-To: <1335634972.2539.13.camel@AMD64X2.fritz.box> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Reply-To: geda-user AT delorie DOT com On 04/28/12 19:42, Stefan Salewski wrote: > On Sat, 2012-04-28 at 16:31 +0200, Oliver Schinagl wrote: >> Hi list, >> >> this time, somewhat unrelated question. >> >> Someone on this list posted a link to Tom Hausherr's blog about PCB >> design and that was an incredible interesting read. >> > > Thanks -- I have never heard about that blog, I will look at it when I > have some spare time... > >> A few things he mentioned that where extremely interesting was his post >> about metric usage vs imperial usage of parts etc [1]. >> >> Here it is mentioned that metric is the way forward in PCB design and >> far it allows for far higher accuracy eventually. I'm sure hobby usage >> won't need/'allow' such high accuracy, most hobby PCB houses don't have >> equipment that does these high tolerances, but even so. >> >> Why is gEDA/PCB by default using the imperial system? It's quite easy to >> change, absolutely, but if it is actually recommended to use >> millimeters, why 'force' a default of mil's? Even if it is somewhat >> autodetected from the system's Language, mm should be the default, no? > > Beginners often start with DIP8 packages, that has imperial spacing, so > default to mils should be ok. Advanced users should be smart enough to > select a reasonable grid, based on used parts. Ah, well my next PCB will be mm everywhere but may use a DIP8 package; shouldn't cause any issues though, right? > > Accuracy should be not worse for imperial units in principle -- it was > limited for pcb to 0.01 mil in the past due to pcb's internal units, but > now internal unit is nm. Should be fine for metric and imperial grid. > >> >> Also, he mentions using 0.05 mm routing grid as most component lead pin >> pitches are 0.05mm. > > Using a very fine grid is like no grid at all. Indeed size and base > (imperial/metric) is not too important, because ends of traces can snap > to pins. Do you really mean 0.05mm? That is very fine. If I remember > correctly I used parts with 0.5mm pitch for my last layout, and I worked > with 0.25mm grid and 0.2mm width traces. > > > Not sure what I ment :) My concern was, if I snap to 0.5mm, where the fab house lists its limits in mil's; is there some problem? I can imagine that 10, 20 years ago, this could be a huge issue, but in effect, it's a print/cnc operation and the tolerances are from drill bits and the photo sensitive operation, not placement of parts? Or is there some grid size to keep in mind? I know the DRC check keeps in minds the photosensitive bits and drill size bits, but other then that? Oliver