Connect with us

PCBs correct first time?

Discussion in 'Electronic Design' started by Leon Heller, Nov 25, 2003.

Scroll to continue with content
  1. Can I ask where? That's pretty reasonable, especially for such a large
    board. Solder mask too?

    Best regards,
    Spehro Pefhany
  2. Brian

    Brian Guest

    I'll check Monday, but I believe came from Sierra Proto Express, No Touch,
    for those. They had silk and mask.
  3. Oh, right. I get a slightly higher price ($122 plus shipping) from
    their web site but fair enough. Thanks!

    Best regards,
    Spehro Pefhany
  4. Brian

    Brian Guest

    I went and looked at the site, still said $51 each (min. 2) for the No
    Touch. That's $102 plus shipping (what I paid) .
  5. I use , About the same price. No hidden
    tricks with drillsizes, mask, silk, and contour routing, internal
    and external, just add a router layer. Quality is top. 8mil tracks
    and spaces. 0.2mm (20mil) smallest hole.
  6. budgie

    budgie Guest

    If you adopt MicroShaft's philosophy of "it's time to ship, who cares
    how ready the product is". My take is to try just as hard to get the
    software ready as the hardware.
  7. Brian

    Brian Guest

    Frank, How do you know this? Are you one of those "slop" engineers like me?

    Sloooww (15 working days for the cheapest service vs. 4), and more
    expensive for big boards, but cheap enough for something small like
    100mm x 100mm.

    Best regards,
    Spehro Pefhany
  9. Russell Shaw

    Russell Shaw Guest

    How? I only equate bug-free work with a) simple designs
    b) recycled cut-n-paste designs, c) bounded designs where
    a simulator handles *all* variables (such as in an asic),
    d) final results from prototyping.
    I make mostly bug free designs by proper prototyping of
    things that aren't practical to calculate. I'm always
    working on tools to solve these problems. It's no fun
    paying US$50k for a field simulator when i can write
    my own.
    In a way, i already do. Doing production pays for the next
    great product, and opens new areas to research how to make
    them faster, cheaper, and better performance.
    I work on my own because a) i do things covering more areas
    than is available in any 9-5 job, b) everything i do is aligned
    for my long-term (10+ year) goals, c) everything i do is interesting
    and challenging and totally unviable for a salary employee to do,
    d) i own all the ip, e) can work whenever i want, f) free time
    to do research for my own things is plentiful, g) i never need to
    rush a design (mainly because of long-term planning), h) knowledge
    and know-how is far better than a short-term buck just to rush to
    a design deadline (which is what most business does)
  10. Depends on the numbers. 3-10 days max for 1-9, above that
    5-15 days. The waiting can be spend on those other projects
    that run parallel ;)

    Their schedule is very reliable, the number of days include
    the shipping. Often one day earlier. I think I have used
    this service about 30 times, also for small production runs
    of 50 pieces. They have never dissapointed me.

    I also used Olimex twice, quality is good but less than Eurocircuits,
    and they are less flexible with their online prototype deal. For one or
    two, their price is excellent. But they estimate 5 days, and then it
    takes another 5 for shipping. So that's also 10 ;) Their order system
    is a bit of hassle too, you have to print & sign their order confirmation
    and fax(!) it back. But still a very good deal for just one or two.

    I've also used a local PCB maker, for larger prototypes
    that consisted of a set of 13 different PCB's, but without silk/solder.
    That was much cheaper than entering 13 PCB's into the eurocircuit
    website, and faster too, 4 days perhaps. These guys are also cheaper
    for production runs, and when you give them 4 weeks time to produce
  11. me?

    I guess so. The bold ones goes straight to production. I don't
    trust myself (nor my tools) enough to justify that. History has
    proven it is the best method for me. In my case it often saves
    time too, time that I prefer to spend to writing better software.
    I hate production boards with blue wires on the back.

    Okay, who has PCB's in the field with blue patch wires on
    the back ? Now don't be shy...

  12. John Larkin

    John Larkin Guest

    Engineers design, production builds. The last thing I want my
    engineers (or me!) doing is hunting down parts or stuffing boards. Or,
    even worse, changing something to make it work without fixing the
    documentation. Another advantage of releasing the first unit to
    production as a formal release is that it debugs the documentation,
    too. Intel has always insisted that its IC designers have their chips
    made on a regular production line; same idea, it verifies the entire

  13. John Larkin

    John Larkin Guest

    Sure, we do that. I want to use red wires (sort of a Scarlet Letter
    thing) but my production people ignore me and use blue wires on blue
    boards, so it hardly shows. Heck, they're so good you often can't see
    their ECOs.

    An occasional kluge is OK as long as it's not hideous and the engineer
    (which includes me) is at least a little embarassed.

  14. Fred Bloggs

    Fred Bloggs Guest

    This all comes under the rubric of LRIP- low rate initial production- to
    mature the manufacturing processes- widely applied and useful technique-
    but you can't go into as cold as lead others to believe.
  15. I had one with a about a hundred[*] "yellow wires" ~25 years ago.
    It was my prototype. I came in at 10:00AM, after a 2:00AM debug
    session(finally got it working) and found "they" had shipped the
    only working prototype to the customer (a replacement was
    delivered a couple of weeks later)! At least they saved my PLA

    [*] The layout person used the bottom drawings as the top for
    several devices.

    The last board I did (I don't design boards all that often) was
    rather complicated (8"x11", 10-layer, everything from PTH thru
    FG680 packages, several >200MHz 64-bit data busses). It had
    perhaps a dozen yellow wires. The AVdd and AGnd weren't detailed
    on the synthesizer (four) datasheets, and I switched EPROMs at
    the last minute for supply issues. The replacement part was
    twice as large, so I decided to make use of the extra size by
    connecting it to a port on the 8051. Bad move, that pin was
    already in use in the CS logic cone. Essentially I wasted the
    bottom half of the EPROM (that I really didn't need anyway).
    Since the board was never leaving our lab, I simply fixed the
    problems on the next version (two versions of the same basic
  16. Mike Page

    Mike Page Guest

    Being consultants, we often deliver a working prototype to the client
    who will then do at least 1 more board turn for production. We meet our
    spec (really!), they do the rest. In a typical project we will *plan in*
    at least 2 board turns. I always try to get Issue A 95% correct or
    better. You have to fix the bugs sometime, why not up front ? Issue B I
    like to have 100% correct. I rarely have to add a board turn to fix a
    problem I caused - the Issue A board gets worked on till it's right,
    then we copy it.

    I'm mostly a software guy, I aim to be writing production software from
    the start. I wouldn't deliberately code badly because I'm not near a
    deadline. Why should hardware be any different ? Your field may be more
    challenging, but I don't find the hardware I do difficult enough to need
    to experiment too much. Mistakes I make are of the D'oh! variety. I make
    use of modelling, but only small blocks, as required.

    We're finding that hardware is shrinking and software is growing, so
    that's where my priorities are headed. We're using cheaper, nastier,
    simpler, production-friendly hardware to do the same job as something
    esoteric a decade ago. Customers love it.

  17. 95% correct? 5% errors? How much % is forgotten input pin? 12.5% if
    there is only one 8 pin IC on the board? Rev. A gets worked on until
    it's right? Why is that needed, when bugs are fixed up front?

    I thought this thread was about designing a PCB without errors
    is practically impossible, for the very first proto.

    It seems that everyone 'suffers' from that, some more than
    others, no doubt. But then we see these 'holier than the Pope'
    lectures, with disclaimers such as 'gets worked on until it's
    right', or 'a little rework is okay' or 'productions uses blue
    wire instead of red'.

    I know enough.
    Ah, the D'oh! variety. That doesn't count.
    And we love the customers.
  18. maxfoo

    maxfoo Guest

    Learn to focus on your job better instead letting your mind wonder, is a start.

    Remove "HeadFromButt", before replying by email.
  19. I read in that maxfoo <[email protected]
    On the contrary, allow time for your mind to wonder every day.
  20. I read in that Mike Page <[email protected]
    CAPSweb.BLAMEco.SWENuk> wrote (in <.
    uk>) about 'PCBs correct first time?', on Sat, 29 Nov 2003:
    Hardware and software are VERY different in this context. Coding badly
    probably just gives you a speed penalty, whereas a track routing fault
    or a wrong dimension means that the board doesn't work at all. Even
    something like careless grounding, which might be more analogous to bad
    coding, may cause the board not to work.
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day