Connect with us

schmitt trigger

Discussion in 'Electronic Design' started by Yzordderex, Jul 14, 2004.

Scroll to continue with content
  1. Yzordderex

    Yzordderex Guest

    I am not very good with digital parts so I could use some help.

    I need to set up some turn-on and turn-off delays to run some mosfets
    in 1-2 MHZ region. I have a whole string of logic running into some
    elantec/Intersil gate drivers. Hopefully I can tweak it all in within
    a few ns.

    I am thinking of using a pair of 4148 steering diodes in series with
    resistors, and one cap to ground. Basically setting the charge time
    and discharge time of the capacitor independently. I am hoping I can
    then run this signal into a schmitt trigger type of circuit and then
    output of that goes to elantec driver.

    I have 8 channels, so octal part would suit me very well. I don't
    care about enable or tri-state, but could use something in the 10-20ns
    range. Problem is I don't have a digital library. Somebody please
    point me to a part that might be good.

    If there is a better way then I am all ears.

  2. Ken Smith

    Ken Smith Guest

    I think the 4148 will result in too much temperature sensitivity. I
    suggest a high frequency schottky.
  3. John Jardine

    John Jardine Guest

    Standard CMOS such 74HC132's and 74HC14's work very, very well with diode-RC
    nets. 10-20ns is close to the edge but in practice no real problem. Make
    sure the steering nets do actually 'steer' and that you're not relying on
    diode back resistances and circuit strays to hold a logic level.
  4. CFoley1064

    CFoley1064 Guest

    Subject: schmitt trigger
    Hi, Bob. I looked for an octal schmitt trigger buffer before without results,
    and ended up punting and using a couple of 'HC14s (hex schmitt trigger
    inverters). Gate delay is 11ns typ., 26 ns. max.

    If the boss can't stand seeing unused gates, try going with four NC7WZ14 Tiny
    Logic dual schmitt-trigger inverters. They have a gate delay of a little more
    than 8ns typ.

    You'll usually find a fairly wide variation in hysteresis from lot to lot when
    using digital logic hysteresis. The switching points are usually specified to
    be within a range of values, which can be amazingly wide. If you need precise
    switching points, I'd go with comparators.

    The other post is correct about the diodes -- watch your temp.

    By the way, many of the IC manufacturers have their standard logic catalogs
    available as .pdf files (if you want to download them at work overnight), and
    will celebrate the issuance of an updated version by making them available for
    free or minimal cost on CDs. If they're not free, you can ask a friendly
    distributor and they'll be happy to ship you one for free (as long as they have
    them in stock).

    Good luck
  5. Yzordderex

    Yzordderex Guest

    Ok, thanks to all. Sound advice all the way around.

    I will use some small schottky diodes. I think I would like to stick
    with the schmitt trigger part. I need about 20ns turn-on delays, and
    maybe a tad more. The Elantec gate drivers actually have a shorter
    turn-on time than turn-off. I am basically trying to synchronize the
    8 channels. Tune it up so that the fets are dancing to the same drum
    so to speak. I'm not too concerned about unused gates, and may in
    fact be able to put some of them to use.

    regards to all
  6. Rich Grise

    Rich Grise Guest

    Do you have half a clock cycle to spare? A resynchronizer would do
    this with a lot less futzing around. Just pipeline your critical
    signals through a set of D FFs, clocked on the inverse clock,
    i.e. clock it midway between the normal system clock pulses.
    They'll be guaranteed to be aligned within the delta prop.delay
    of the chip.
  7. Tim Shoppa

    Tim Shoppa Guest

    Do you need 10-20ns consistency in delays, or do you need to be able
    to accurately tune the delay in the 10 to 20 ns range?

    "Regular" 74HC14 type Schmitt triggers will probably get you to within
    10-20 ns consistency.

    If you're needing to tune the delay in the 10-20ns range, then a very
    common approach is to use ECL differential line receivers/transmitters
    and varying lengths of cable between them to set the delay. 1ns is about
    a foot (details like cable velocity factor are quite relevant).

  8. Yzordderex

    Yzordderex Guest

    No 1/2 cycle available. I'm actually trying to un-synchronize em.
    What I'm trying to do is run a full-bridge 2kW power supply at 2MHz.

    Each of the channels will drive a gate drive amplifier. Two
    amplifiers, feed into a gate drive transformer on each fet to drive
    the fets in a bipolar fashion - I have to sweep the charge out pretty
    fast to get the bridge to spin at 500ns. I have to introduce about
    20-25 ns of commutation delay with each gate driver. The delays in
    the driver chips are all over the place (fortunatly not too drifty
    with temperature) so I have to trim each turn-on and turn-off delay
    independantly. See?

    Thanks again
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