According to Pete - PCB Layout Part II


In the last edition of “According to Pete,” our Director of Engineering Pete Dokter talked about his favorite tips and tricks for PCB design and layout. Many of you chimed in with your own helpful comments - if you haven’t checked out that post, it’s worth a read.

In today’s episode, Pete is back for part deux, with even more PCB layout discussion.

As always, we welcome any questions/suggestions/recipe tips in the comments section below. We hope you enjoy this edition of “According to Pete” and we’ll be back with more engineering goodness soon! Cheers!


Comments 17 comments

  • Don’t forget test points at various ground points. People always seem to check for power, but fail to check for ground as well.

  • Great post, I love these things! I wanna note a few things, and please correct me if I’m wrong on any:

    -If you do not have the space for one larger via to handle all of your current, you can place a few smaller ones and the total current capability is cumulative. So you could use three 10 mil vias to handle currents at about 3.3A.

    -If anybody finds that they do need some calculations on via inductance and all that jazz, there is a toolkit called SaturnPCB that has a bunch of calculators just for things that aren’t thought of as much, but could potentially cause problems.

    -It seems that the pours discussed is geared towards microcontrollers so I should mention power supply controllers have two separate grounds (SGND, PGND) as well as Vin and Vout. Check out the LTM4624 from linear technologies. That datasheet has a good example of what we are talking about.

    -And I like saying it, decoupling caps, very important. Put them close to the controllers.

    • Do you think you could go over keepouts and other such mechanical layout issues in the third video? I, personally, have a penchant for placing bypass caps so close to ICs that they become difficult to solder in place! Also, there is another advantage to having several small barrels set instead of a single large via: the paralleled small vias will have less parasitic inductance.

    • Well played, sir.

  • Please post this on Vimeo… for some reason my linux laptop doesn’t play youtube video worth a damn, but Vimeo… Yum!

  • An autorouter is the best way I know to get 13,000 vias to fit on a 2-layer 3" x 3" board.

  • I am very glad to learn a lot from you this meaningful knowledge. From an article describing your unique way , we can see that you are an approachable , humorous person.

    get facebook subcribers get facebook photo likes

  • One case I can thing of for a copper pour other than power or ground is for heat sinking. On some power devices the tab is not electrically isolated within the package, and may not be fixed at ground potential in your circuit. But you may want a large area of copper tied to it to help convey away the heat.

    The other cases I can think of are where you might be using the a copper pad as part of an intentional capacitor, for capacitive touch sensing.

  • It might be useful to have a separate power ground plane for return lines on high current devices (like an H bridge) so return current isn’t adding a bias to your digital or analog ground pins.

  • Go Canada!!! Nice Shirt from the great white north.

  • Those two hole semicircle ground clips are very nice to have on any board. I think they are 0.2" spacing and a probe ground clamps on them real good. I got an arduloidinal device off eBay that had one and thought, “I gotta get me some a them things!” But I have not seen a source. You guys should carry em!

    As for layout, I wasn’t paying real close attention, but did you talk about the advantages of thru hole resistors in 2 layer work? As in not needing vias because you can bridge over a trace? The vertizontal resistors used to be standard in consumer stuff to save space and keep it tight in radio gear. With the 1/8th watt being so small, I have not bothered with that in a long time.

  • Good video. A couple of comments:

    • For test points, a small via (that isn’t covered by solder mask) can make a good one, and the “hole” helps keep the probe from slipping while you’re looking up to read the meter or ‘scope. Also, looking through a complex board and making sure that any trace that already has vias has at least one that isn’t covered by a solder mask will assure that you can get a probe on that trace. (Scraping off solder mask is a real pain, and can easily damage traces, so better to not have to do it.)

    • As for multiple pours, if you have multiple supply voltages (or even just separate analog and digital at the same voltage), they’re each a candidate for a pour.

    • From a mathematical point, the “rule of thumb” sounds reasonable. The circumference (and thus effective trace width) of the hole will be 3.14159… times the diameter, but since we have inside plating, it will be a little less. So 3X sounds good.


This Week

This Month

Heartbleed

Happy Arduino Day!