Connect with us

Programmable slew rate ?

Discussion in 'Electronic Design' started by Robert Lacoste, Nov 7, 2006.

Scroll to continue with content
  1. Hello,
    We need to reshape a rectangular pulse (say TTL, around 100ns wide) into a
    trapezoidal pulse of the same width, I mean with programmable 2ns to 20ns up
    to low and low to high rise times. Precision can be medium but the rising
    should be quite linear (ie not simply low-pass filtered). Any idea of an
    analog chip that could help to build this function ? Our first idea is to to
    it digitaly in an FPGA with an external quick DAC and to synthetise this
    waveform but may be you will have a simpler idea ? U to I converter then
    capacitor ? Can it be quick enough ?
    Cheers,
    Robert.
     
  2. Guest

    How big a pulse do you actually want? If you were prepared to settle
    for a 0V to 3V pulse and had -5V and +10V rail available, I'd
    definitely go for the current sources and the capacitor (with the
    capacitor buffered by a fast amplifier capable of driving trerminated
    50R transmission lines).

    3V in 2nsec with a 10pF capacitor is 15mA. Farnell stocks the BFR92 and
    BFR93 5GHz wideband NPN transistors, and the BRT92 and BFT93 PNP
    complements. I always needed to put a 27R surface mount transistor in
    series with the base (and right up against the base) to stop them
    oscillating, but - given that - they can be made to provide very nice
    quick steerable current sources at around the 20mA level, not to
    mention level shifters.

    You will need a real resistor - 50R should do - in series with the
    emitters in any long tailed pair. You can set the current with a DAC
    driving a Howland current pump or an instrumentation ampliifer, but
    these outptus will look capacitative at the frequencies of interest, so
    you need the series resistor to keep the fast edges looking more or
    less tidy.
     
  3. Many thanks Bill, that looks like a good path... Of course the initial wish
    list includes a 0/5V only power supply but we already know that a larger
    range of power would be needed with an analog solution. We will check your
    proposal... We also found some RS485 driver chips that have some form of
    configurable slew rate outputs but too slow for us.

    Cheers, and long life to sci.electronics.design.

    Robert
     
  4. Fred Bartoli

    Fred Bartoli Guest

    Robert Lacoste a écrit :
    You may use a diode gate. That's simple but for a say 10pF output node
    capacitance you'll have a permanent 2.5 to 25mA, depending on your
    programmed slew rate.

    I = 2.5 ... 25mA
    v
    |
    |
    |
    Fast 0-5V >---+--|>|--+--|>|---+--->
    pulse | |
    | |
    | |
    '--|>|--+--|>|---+
    | |
    | ---
    | --- 10pF total
    | |
    | ===
    | GND
    v
    I = 2.5 ... 25mA

    For such a high speed and low integration cap you'll need low
    capacitance schottkies.
     
  5. Ban

    Ban Guest


    .---------+-------+---------------+-o+5V
    | | | |
    | >| |< |
    | ??? |- -| |
    | /| |\ |
    | | | |
    | .---+ | HP5082-2755 |
    | | .-. .--+--. |
    | '->| | | | |
    | | | V V |
    |\| '-' - - |\|
    | \ | | | | \
    o---| >--------)----+ +---+------| >-o
    | / | | | |10p | /
    |/| | V V --- |/|
    | | - - --- |
    | | | | | |
    | | '--+--' GND |
    | | | |
    | \| |/ |
    | BFM520 |- -| |
    | <| |> |
    | | | |
    '---------+-------+---------------+-o-5V
    (created by AACircuit v1.28 beta 10/06/04 www.tech-chat.de)
    Similar to Bill, but very simple. I think there is also a complementary
    PNP-pair available, but I'm too lazy to search. The thing with the beads
    applys as well, but since the sources are always on, they are more stable.
    The first buffer is not needed, if the oscillator can sink/source enough
    current, it has to be faster than 2ps. TTL in/out.
    set the pot from 180R to 2k. Eventually you can put small linearizing
    resistors(100R) in the emitters. The diode ring has to be fast and low
    capacity, bulk resistance doesn't matter in this case.
     
  6. Jim Thompson

    Jim Thompson Guest

    Even fixing the DCT's, it won't work. You need emitter padding
    resistors.

    ...Jim Thompson
     
  7. Ban

    Ban Guest


    .---------+-------+---------------+-o+5V
    | 100 100 |
    | | | |
    | >| |< |
    | ??? |-+-| |
    | /| | |\ |
    | +---' | |
    | .---+ | HP5082-2755 |
    | | .-. .--+--. |
    | '->| | | | |
    | | | V V |
    |\| '-' - - |\|
    | \ | | | | \
    o---| >--------)----+ +---+------| >-o
    | / | | | |10p | /
    |/| | V V --- |/|
    | | - - --- |
    | | | | | |
    | | '--+--' GND |
    | +---. | |
    | \| | |/ |
    | BFM520 |-+-| |
    | <| |> |
    | | |
    | 100 100 |
    '---------+-------+---------------+-o-5V
    (created by AACircuit v1.28 beta 10/06/04 www.tech-chat.de)

    I forgot the connections and I wrote about the resistors, here they are.
    These are dual monolitic 9GHz transistors, and inexpensive.
     
  8. "Robert Lacoste" <> a écrit dans le message de
    4550a88c$0$25953$...

    Many thanks to all of you for your postings, now we are going to switch on
    the soldering iron...
    Thanks again,
    Robert
     
  9. joseph2k

    joseph2k Guest

    From about 20 years ago ISTR a programmable pin driver, V(h), V(l), t(r),
    and t(f) all independently programmable. I wonder if you could find and
    use some of those?
     
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

-