I use a PWB layout program where I can see the internals. Seems if you inp ut the data in inch units you can convert to metric perfectly. 1 inch = 25.4 mm. But if you input data in metric and try to store it in inches it doesn't convert so well and round off has to happen. 1 mm = 0.0393700787 4016... inches. As a consequence, this package stores everything as intege rs in nanometers. At one time trace widths were convenient in thousandths of an inch because they were typically whole units, 6/6 mil was a common trace and space. Now trace and space are commonly fractional thousandths, so metric is just as good. Some BGAs require 3.5/2.5 mil or 0.09 mm/0.0635 mm or 90um/63.5um. I think on my last layout my basic grid was 0.325 mm which was a bit more t han 12 mil for 6/6 mil trace and space which turned into 0.15 mm if I recal l. This was close enough to 12.5 mil that it worked ok for routing to TSSO Ps and such. Routing a 0.5 mm QFP I had to switch to a 0.5 mm grid. Does anyone still use inch measurements for anything other than mounting ho le locations on PWBs? Rick C.

Despite living in a strongly metric country, I've often found mils to be a handy unit for traces - as you say, they are typically whole units. I can imagine how 6/6 mil looks, but 0.16/0.16mm doesn't have the same feel. Other than that, metric all the way.

Metric, good or bad?

S

Steve Wilson 7 years ago

I remember a serious problem when slivers got between traces and shorted them out. Very hard to spot and required reworking every board. Painful.

However, now with computers, you can make even bigger mistakes much faster!

Vote

W

Winfield Hill 7 years ago

Yes, sometimes those were from poorly heat-treated tin-coat overhand breaking after etching, it could still happen now.

Thanks, - Win

Vote

D

DecadentLinuxUserNumeroUno 7 years ago

Winfield Hill wrote in news: snipped-for-privacy@drn.newsguy.com:

Way back then it was usually from bad art and bad QA (not) looking at the art before sending it to production.

Vote

G

gnuarm.deletethisbit 7 years ago

Not because they want to be different, because they don't want to break surveying.

"Any data expressed in feet derived from and published as a result of geodetic surveys within the United States will continue to bear the following relationship as defined in 1893 : 1 foot = 1200/3937 meter"

Notice the date they defined this formula!

Rick C.

Vote

P

Phil Hobbs 7 years ago

I still use Gaussian units for electromagnetics. (Gaussian is a rationalized version of the CGS ESU system.) My fave EM textbook is Jackson's second edition--for the third he switched to SI, the dog. ;)

Cheers

Phil Hobbs

Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com

Vote

T

Tauno Voipio 7 years ago

Ouch!

With 4 * pi hidden into the units at random ...

-TV

Vote

P

Phil Hobbs 7 years ago

When you're doing theory, it matters a lot more that the equations are simpler. You can convert to SI trivially at the end of the calculation.

Cheers

Phil Hobbs

Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com

Vote

G

George Herold 7 years ago

I'm reading.. well slowly chewing.. 'atomic physics,...' by D. Budker

formatting link

He uses cgs because he finds it easier. Well, he also sets h-bar = 1 sometimes...

George H.

*I like this book, but it's graduate level physics.. and I find myself going back to lower level sources to review things.... I don't remember the eigen values of the (L dot S) operator, with hydrogenic wavefunctions, for instance. GH

Vote

J

John Larkin 7 years ago

Just apply a lot of current!

We make very few mistakes on PC boards. We don't prototype. Rev A is built by production and we expect to sell it.

We don't get PCB shorts either; all board are bare-board tested. One short or open instantly disqualifies a PCB vendor.

John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com

Vote

J

John Larkin 7 years ago

We use ENIG, gold over nickel, and we don't see shorts any more.

John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com

Vote

B

bill.sloman 7 years ago

them

er!

When your average new-product development time is two weeks - as John Larki n has told us is his norm - there's not a lot of difference between success ive printed circuit layouts. More adventurous development does offer more o pportunities for things to go wrong, but careful design reviews at every st age can catch quite a lot of bugs.

If they claim to do bare-board testing before they ship it certainly ought to.

It's nice when you have enough suppliers to let you be that rigorous.

Bill Sloman, Sydney

Vote

Metric, good or bad?

Join the Discussion

Didn't find your answer?