Connect with us

CD4000 Series device with high output drive

Discussion in 'Electronic Design' started by Klaus Vestergaard Kragelund, Jun 13, 2004.

Scroll to continue with content
  1. Hi

    For an ultra cheap buck converter living off a 15V supply I'm looking for a
    inverter/buffer to replace the combination of P-MOSFET / freewheeling diode
    / level translation resistors

    I looking for a device that can deliver say 100mA of current. For example a
    CD40106 can do this with six gates in parallel, at least the current
    sinking, unfortunately not the sourcing part :-(

    Or perhaps - anyone know how much can be drawn with respect to the graph on
    page 8 (figure 5) of this spec?

    If I respect the total power limitation of the device can I get away with
    drawing so much current in a standard 40106 device?

    Or perhaps there's another device out there costing as low as the 4000
    series and delivering the current I need?


  2. Joerg

    Joerg Guest

    Hi Klaus,

    That sounds a bit stressful to a CD chip. Anyway, a BSS84 is well under
    10 cents and if you use discontinuos mode you might get away with a fast
    silicon diode for 3 cents or less.

    Regards, Joerg
  3. Hi Joerg

    Thats exactly what I'm using currently - a BSS84 and a BAV74.

    Allthough the price for the 4 components (MOSFET, diode and two
    resistors) is below 4 cent (high volume) we have a figure for the cost
    of production that adds about 1.6 cent per placed SMD component. Thats
    why I'm looking for a one chip solution


  4. Klaus Kragelund wrote...
    I assume that's the *incremental* cost, not simply obtained
    by dividing the total assy cost by the number of parts, but
    obtained from the difference in cost if parts are added?
    Can you tell us what quantity is in each production run?

    - Win

    (email: use hill_at_rowland-dot-org for now)
  5. How about a 4511, with all segment outputs in parallel?

    Set /BL to 0, and toggle the /LT input.
  6. What I can't understand is the '4000' series requirement?. Normally it is
    only necessary for a part to be 4000 series _compatible_. For instance,
    Texas do the UC3708, which has 4000 compatible inputs, and can drive a
    couple of amps on each output, with push-pull drive. Allegro also do power
    drivers like this. Remember also, that the MOSFET pairs in the 4066, if
    running on 15v, can sink/source this sort of current, but only for a limited
    time before the maximum power dissipation for the switch is reached. The
    same limit is likely to affect other 4000 family devices in this regard.

    Best Wishes
  7. That should be an incremental cost figure, but I have discussed some with
    the production engineer responsible for that figure. A 44 pin
    microcontroller costs the same as a 0603 SMD cap. to place and that does not
    correspond to the real world. But anyhow - as a mean for the SMDs is should
    be correct to a certain extend


  8. Probably a bit too expensive, but the Fairchild NDC7001C is a nice 50V
    500mA/300mA P/N channel pair in an SOT-6 package. Under 20 cents in

    Best regards,
    Spehro Pefhany
  9. Klaus Vestergaard Kragelund wrote...
    Presumably much of the cost is in inventory pulling, machine
    setup, teardown, restocking reels, cleanup, paperwork, etc.
    These are relatively fixed costs, so the length of the run must
    be a big factor. What's the quantity for your numbers? Roughly
    how long does a run of that size take after pressing START?

    - Win

    (email: use hill_at_rowland-dot-org for now)
  10. Joerg

    Joerg Guest

    Hi Klaus,

    1.6 cents placement costs for SOT23 seems a bit high. Is that just a
    lumped number for the "average package complexity"? If so that may not
    reflect the true placement cost for a SOT23 or 0603. But anyway, I'd
    guess the two resistors wouldn't go away as I assume these are for
    feedback. So you are only looking at the FET and diode plus their
    placement cost.

    It may be possible to find a motor driver chip that can do this. 12V
    Stepper motor drivers could be candidates as they are often rated above
    12V. These would have four driver sections on one chip. But it is
    doubtful that it would end up lower in cost and they can be long leadtime.

    Regards, Joerg
  11. My sentiments too - I think I need to push for a better model
    Actually I always need the feedback resistors, so they are left out of the
    comparison. The resistors I mention is to turn on the high-side P-MOSFET
    making sure not to exceed Vgs limitations
    Thanks - I'll try to find some parts in that area


  12. The second and succeeding identical parts should be cheaper than the
    first.. it is not much more expensive to put down 10 bypass caps than
    just one. But sometimes the assembly houses use silly models too,
    especially for smaller quantities, so matching their model might be
    more important than matching their cost if you are outsourcing the

    Best regards,
    Spehro Pefhany
  13. Joerg

    Joerg Guest

    Hi Spehro,

    And then there is another huge cost saver. I always try to obtain the
    jelly bean list from the manufacturing operations, with cost data. Not
    easy in all cases, "why do you need this...?". But then the cost goes
    down because we can try our best to design with what they have rigged as
    standard or where there is Kanban floor stock.

    Regards, Joerg
  14. Joerg

    Joerg Guest

    Hi Klaus,
    Seriously, if you guys are designing extremely cost sensitive stuff
    these cost models need to be refined into much greater detail. The
    higher the cost sensitivity the more detail the models must provide.
    Using just one lumped figure for placement costs will lead to
    dangerously wrong business decisions.

    I know pressing for details is hard. You'll hear things like "but we've
    always done it that way and the company is doing just fine".
    Nevertheless, persistence is key here. You can quote me on that if it
    helps. I ran a biz and know this stuff from a lot of exposure to it.

    Regards, Joerg
  15. The quantity for this product is 500k/year. The production line is split
    into two lines because we have been able to split the product into two
    boards - one with only SMD parts and one with leaded parts. The calculations
    I'm doing is on solely SMD because the reported cost of a leaded placement
    is set to 10 times the cost for an SMD (and that makes we stay well clear of
    leaded parts)

    The product includes over 500 SMD parts and something like 20 leaded ones.
    As far as I recall the product is something like 30 seconds on the
    production line during SMD pick&place. We calculate ~0.4$ per second on the
    line. I don't know the entire run time - I'll try to get some more numbers
    at work tomorrow and fill the gaps in another post

    As a side story we just dismantled a vacuum cleaner to look inside to get
    ideas. I think it was a Korean manufactor. The nozzle power was 630W and the
    inlet power was 1600W. The price of that product is about the same as that
    of ours, but it is entirely leaded parts. With our models for the cost
    calculation we could never target that price range. I however sucspect the
    rason is that low-cost workers are used to mantle it manually and thats the
    reason for the low price/leaded product


  16. Joerg, I agree with you totally. I tried to ask more specific questions last
    year when the new "model" was released but I could not get any numbers

    I'll try again because as you say (and we have discussed this internal to
    our department) we might be chasing the wrong places to cut costs due to a
    failed model


  17. Joerg

    Joerg Guest

    Hi Klaus,
    Very likely it's not even assembled in Korea but some country with a
    much lower average wage. Possibly China. What Winfield mentioned in
    terms of other costs that require personnel, this is what also drives
    SMT rigging prices up in western countries like Denmark. Somehow your
    name sounds Danish :). Somebody has to get the parts, fill out
    paperwork, rig the machine, maintain the machine and so on. Then the
    government puts all kinds of costs on top and soon you are at twice or
    more of what a Chinese contract fab would have to face in terms of cost.
    In California we have the same problem, for example workers compensation
    (accident) insurance adding up to 25% or more on top of an hourly wage.
    In China they don't have that, so most of the circuit board fab work
    goes there now.

    Another thing to look at is these 500 SMT parts. How large is the
    variety? Is it possible to shrink that variety by using 10k, 100k and 1M
    resistors wherever you can, or use just two types of transistors so you
    have less different part numbers on there? Can you link up with other
    design teams and agree on one "jelly bean parts list" that everybody
    tries to stick to? That reduces rigging times a lot.

    Regards, Joerg
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