This is my first "real" design, a (more or less) universal USB serial interface board with two RS-232 ports, one RS-485 and one CAN interface. Hi-res schematics, top and bottom PCB layers with bottom copper pour turned off. Each FT2232 channel has two interface drivers (a RS-232 and either RS-485 or CAN) that can be connected to it with jumpers. Jumper wires can also be used for e.g. SPI or JTAG modes supported by this chip.

I used a FT2232 breakout board from dangerousprototypes.com (can't post a link) as a reference (not for layout though). Layout is what I'm most concerned with here. I tried to implement most of the good advice I found on this site, such as the ground plane (disabled), power trace along the perimeter on the bottom layer, power polygon under the chip, short traces for oscillators and decoupling capacitors etc. The area of the FT2232 seems much more complicated and cramped than on the dangerousprototypes board. I wonder if that's or because their schematics is simpler and crystal / capacitors are placed farther from the chip, or because I suck at this. (In hindsight, placing some components on the bottom layer would probably simplify the layout.)

I tried to comply with requirements of the seeedstudio.com manufacturing service.

(There's a discrepancy between schematics and layout in pin header placement - RS-485 and CAN headers are swapped.)

I'll appreciate any critique or advice.

  • \$\begingroup\$ The board would look so much better with only 45 degree tracks. \$\endgroup\$ – m.Alin Nov 20 '11 at 23:15

Looking briefly at the layout, it looks OK. You seem to have generally followed the advice. The following are things that I would do differently, but are not actually problems with your board:

  • I prefer to err on the side of thicker tracks, especially on a board with this much free space. E.G. For most chips, I would make the tracks as thick as the SMD pads.

  • Where decoupling caps connect to chips, I like to make the tracks as thick as the capacitor pads. Sometimes you need to use two or three tracks in parallel to achieve this.

  • I tend to spend a couple of days fiddling with the board, tidying up tacks, pushing things around, making things symmetrical, especially where crystals connect to chips. I like to make those perfect if I can. No reason, I'm just anal.

  • I think the bottom side can be tidied up a little, especially under the main chip. Some of the tracks there are taking quite indirect routes.

  • One last word of advice. Check and check again all of the connectors. Check that you are using the correct gender, and that everything is the right way round. Even after laying out PCBs for more than 10 years, this is still something I get wrong.

Aside from those, I don't immediately see any reason why it wouldn't work. The length difference between the USB data lines is probably too short to be a problem. I haven't used USB before to know.

| improve this answer | |
  • \$\begingroup\$ BTW, I haven't mentioned that this IC is USB 2.0 (480 MHz!). Most sources say that you have to use a 4-layer PCB for that, but the dangerousprototypes board is 2 layers and seems to work (at least I haven't seen any comments that say othwerise). The roundabout trace on the bottom is ground which will be poured. There're indeed places where traces can be widened with no complications. Thanks for the helpful advice. \$\endgroup\$ – Thorn Nov 20 '11 at 15:38
  • \$\begingroup\$ But not much of the PCB will be operating at this speed. Only the signal coming off the USB connector. The crystal is only 12MHz. \$\endgroup\$ – Rocketmagnet Nov 20 '11 at 17:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.