Shop OBEX P1 Docs P2 Docs Learn Events
Layout issue with USB, need feedback — Parallax Forums

Layout issue with USB, need feedback

pedwardpedward Posts: 1,642
edited 2012-04-25 14:55 in Propeller 1
So I got the PropKey boards back today, and after figuring out that I mirrored the USB pinout, I rewired everything. I started getting enumeration problems and so after looking into things, I'm pretty certain its because the flying wires I used are unequal length and it's messing up the USB comms.

I revisited the layout to fix my errors and I adjusted the trace lengths to be equal. I looked at the Quickstart and found a little jog that was put into a trace to do the same thing, so I feel I'm on the right track.

I have posted a picture of the artwork below and wanted to get feedback on whether this looks okay or will cause me problems.

The trace lengths are now exactly the same length from the connector pads to the chip with this layout.

attachment.php?attachmentid=91944&d=1335251878
propkey-usb_large.jpg
304 x 320 - 13K

Comments

  • jmgjmg Posts: 15,183
    edited 2012-04-24 00:27
    3mm of FR4 is ~ 20ps of delay. or barely 1% of a High Speed USB spec bit time.
  • pedwardpedward Posts: 1,642
    edited 2012-04-24 00:38
    I don't follow your point. Are you saying 3mm lineal length? I'm sure my flying wires are the problem with my current board, but I want to make certain I've got the design dialed before I waste another 2 weeks waiting for boards.
  • TubularTubular Posts: 4,706
    edited 2012-04-24 01:43
    You could try it with a USB extension cable.

    Also are you packing up the space below to make the board thicker? Otherwise could be intermittent contact. You could solder a cable to eliminate this.
  • TubularTubular Posts: 4,706
    edited 2012-04-24 01:49
    The 4D systems uUSB do a similar thing to what you're doing, they don't seem to worry about equalizing the traces
    http://www.4dsystems.com.au/downloads//micro-USB/uUSB-CE5/Docs/Pdf/uUSB-CE5_DataSheet_rev1.pdf
  • Mark_TMark_T Posts: 1,981
    edited 2012-04-24 05:03
    jmg wrote: »
    3mm of FR4 is ~ 20ps of delay. or barely 1% of a High Speed USB spec bit time.

    Actually unless the trace is buried there are two dielectrics involved, FR4 and air, so the air signal will take 10ps for the 3mm and the FR4 will take 22ps. Neither seems worth worrying about.

    What is worth worrying about is the impedance of the transmission lines involved and whether there is cross-talk between the two signals - sort this out first before worring about 0.02ns of delay.
  • pedwardpedward Posts: 1,642
    edited 2012-04-24 11:01
    Tubular wrote: »
    You could try it with a USB extension cable.

    Also are you packing up the space below to make the board thicker? Otherwise could be intermittent contact. You could solder a cable to eliminate this.

    I put 22mil UHMW tape on the bottom of the board where it goes into the connector. This makes the overall thickness the same as a USB cable.
  • jazzedjazzed Posts: 11,803
    edited 2012-04-24 11:07
    How do your power and ground rails look on a scope? I'm a little worried by lack of a ground plane.
    Do you have a .pdf of your schematic that we might check for problems?
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2012-04-24 11:20
    pedward,

    I'm sure the problem lies elsewhere. That little bit of length discrepancy couldn't possibly lead to problems with enumeration. It's not something I've ever bothered with, and I've never had enumeration issues with my USB circuits. IOW, be on the lookout for a horse, not a zebra.

    -Phil
  • pedwardpedward Posts: 1,642
    edited 2012-04-24 11:24
    The *real* problem was that I got the pad layout completely backwards. I just added a couple of short wires to reverse the DP and DM, but it gives me trouble. I'm going clean up and remove the wires/traces and redo the leads to see if that solves the problem. If the problem persists, I will contact FTDI about the problem.
  • jazzedjazzed Posts: 11,803
    edited 2012-04-24 12:05
    pedward wrote: »
    The *real* problem was that I got the pad layout completely backwards. I just added a couple of short wires to reverse the DP and DM, but it gives me trouble. I'm going clean up and remove the wires/traces and redo the leads to see if that solves the problem. If the problem persists, I will contact FTDI about the problem.

    I've done that too, but the board worked anyway until i over-stressed my solder job.
  • tonyp12tonyp12 Posts: 1,951
    edited 2012-04-24 12:39
    Unless you are running at hi-speed (480mbps) I don't think the longer trace will really have any effect.
    But it could not hurt to make it equal traces anyway.
    Of course it's important that D+/D- is correct and most microusb is suppose to be on reversed pcb side

    If you want single sided mounting of components, you could do this trick
    attachment.php?attachmentid=91959&d=1335296367
    Micro_USB.jpg
    561 x 230 - 38K
  • pedwardpedward Posts: 1,642
    edited 2012-04-25 14:55
    The problem turned out to be the LED. The LED was consuming too much current and was causing the chip to not work properly. I removed the LED and it would enumerate.

    I'll have to look at it when I get home.
Sign In or Register to comment.