On Tue, Sep 15, 2015 at 6:41 PM, Hrvoje Lasic lasich@gmail.com wrote:
On Tue, Sep 15, 2015 at 7:22 PM, Christopher Havel laserhawk64@gmail.com wrote:
On Tue, Sep 15, 2015 at 1:18 PM, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
[...] even i was unable to move the micro-desktop board (which is only 4in x 4.5in) forward because i had designed it as a 4-layer PCB - costs are around $400 for qty 5 4-layer prototype boards on a 3 week turnaround by complete contrast, a 7-day turnaround for qty 5 2-layer prototype (bare copper) boards with larger vias is around $40 for qty 2, and around $100 for qty 5.
Imbecile question: is it at all reasonably possible to redesign the micro-desktop board to be a 2-layer board?
If board have high speed design it is very unlikely that you can do it properly (because you have to be careful about routing to ground, you have many lines that you have to match etc etc and when you have only 2 layers it is difficult to keep it all correct). In theory maybe it could be possible but it is a hell of a work and then you risk a lot that PCB will not be good.
i took a look at a gigabit ethernet board that phil kindly sent me a while back. the layout of the differential pairs was absolutely fascinating [and the board had, obviously, passed FCC tests].
the layout involved putting ground vias exactly... something like 20mil from the differential pairs, spaced out at exactly something like... 100mil, right the way *both* sides. there were no components permitted either side for some distance either. there were no vias in the actual differential pairs, either.
there were twenty sets of differential pairs like this - all with exactly the same very very clearly and meticulously laid out arrangement, with the spacing between each differential pair also very meticulously laid out.
so it can be done.
in the micro-desktop board, however, the actual distance that the USB differential pairs has to travel is well under 1cm. i am arranging the connectors *directly* in front of where the signals come out. i have also deliberately arranged the EOMA68 interface so that the pairs come out directly and do not require a via to "cross over" each other.
so it is much less of a concern than might otherwise normally be. these aren't 10cm traces, where EM radiation would definitely be a major concern, they're literally 1cm long, and i intend to surround them with ground vias.
of slightly more concern is the VGA interface (RGB/TTL being converted to analogue), which will be operating at around 75mhz (or so). however that's not 480mhz, so i am not hugely concerned. i am again getting in as many GND vias as can fit, and keeping the traces very very short. the buffer ICs actually straddle the PCMCIA interface on the other side of the board, so that the RGB/TTL signals can, with vias, go left or right, routing as appropriate, maximum trace length about... 3mm.
l.