Connect with us

Output bounce, can this be fixed?

Discussion in 'Electronic Basics' started by Anthony Fremont, Mar 22, 2007.

Scroll to continue with content
  1. Here (A.B.S.E) is a shot of the output of an ON-semi 74HC14 straight from
    the pin.
    I'm trying to square up my ~10MHz oscillator signal so that I can feed it to
    a PIC (of course ;-) for frequency counting. I imagine the PIC will do
    alright with it, but man this is pretty sad looking with that ringing. Can
    this be cleaned
    up, or do I need to use a different part?
     
  2. jamsan

    jamsan Guest

    Personal proposal. You can follow a resistance after the 74HC14 output
    and connected to ground with a capacitor and adjust the values. Maybe
    it works.
     
  3. Thank you, I will try that. :)
     
  4. Tam/WB2TT

    Tam/WB2TT Guest

    I have done this many times with no problems. Sounds like you may not have a
    VCC to ground bypass capacitor on the HC14. You could also try a 47 Ohm
    resistor in series with the output, or you may be overdriving the device.
    Bias up the input with a 15K to ground and a 33K to VCC; then capacitively
    couple the input. Signal should be about 3 V p-p. You are kind of pushing
    the frequency for an hc14. A 74AC14 will work better.

    Tam
     
  5. Thanks Tam. I grabbed a 470 Ohm resistor laying nearby and stuck it on the
    output and it got rid of the ringing. It's not real pretty still, but it
    looks allot more like a digital signal now than an analog one. ;-)
     
  6. colin

    colin Guest

    hes using breadboard btw ...
    also what scope probe are you using?
    is it a 100mhz 10:1 type ?

    if you look at figure 11 ..
    <http://linear.com/pc/downloadDocument.do?navId=H0,C1,C1154,C1009,C1028,P1219,D4138>

    it shows the effect of a scope probe on a signal and it looks pretty much
    identical to the waveforms you show.
    the scope probe input basically looks like a tuned circuit at some
    frequency,
    wich is dependant on the 1/4 wavelength of the lead wich is probably 50mhz
    or so.
    the 470 ohm resistor effectivly damps this to some extent, but not vvery
    much.

    another trick is that many scope probes have a 450 resistor in series with
    the input,
    if you then put the scope probe on the X1 position and the scope input to
    50ohms then
    you have a neat 10:1 divider with a purly resistive input.

    I dont know if they designer had this in mind ive never seen it mentioned,
    just found it worked surprisingly well by acident.

    Colin =^.^=
     
  7. Tam/WB2TT

    Tam/WB2TT Guest

    The 470 Ohms is going to give you band limiting. Try a smaller value. Did
    you bypass the 74HC14 VCC line?

    Tam
     
  8. Guest

    You probably need to be more careful about grounding your scope probe
    to the Vss power pin of the chip you are looking at, or some 0V
    (ground) connection very close to this point.

    Junior engineers make this mistake all the time, and I once had to
    persuade a customer who was using a Cambridge Instruments electron
    beam tester that he had to ground his circuit to the electron
    microscope inside the vacuum chamber where his circuit was being
    tested - he was very unhappy with the ringing he was seeing, and dead
    embarassed when it went away when I hooked up the earthing lead. We
    were installing the machine - a prototype - at Thompson-EFCIS in
    Grenoble, and the customer was very nervous.
     
  9. Noway2

    Noway2 Guest

    This was alluded to in an earlier response, but are you sure that the
    ringing isn't an artifact of your scope?

    Putting a resistor in series with the signal will slow the edge rate
    making it both more 'scopable' and limiting the bandwidth (won't be as
    as square).

    Tam is correct, make sure the chip is properly decoupled.

    Also, do you have a good solid ground plane or are you running bread
    board or serpentine traces? Without a solid ground reference, you will
    get a lot of bounce that will screw things up.
     
  10. Tam/WB2TT

    Tam/WB2TT Guest

    Since you mention tghe 'scope, I hope he is grounding the probe at pin 7 of
    the '14. Bypass cap means pin 7 to 14 of the 74HC14. Not the other end of
    the board. We used series Rs between 22 and 51 Ohms to get rid of ringing.
    Extreme case was something like 100 Ohms to meet FCC Part15.

    Tam
     
  11. I did that and it did improve the signal some, but not allot.
    That's what I do. I straddle the IC with them.
    I'm only a hobbyist, but I've "scoped" quite a few signals. It's been
    mostly low frequency stuff and I admit that I'm sure I could learn plenty
    about using one. But I've put probes on PIC pins that were unloaded and the
    ground was far from the PIC Vss pin, and I still never saw ringing like that
    on one.

    I just looked at the specs and it looks like the typical rise time on a PIC
    I/O pin is twice as long as the 74HC part. Slowing the rise/fall times down
    a little by loading the output pin eliminates most of the ringing. I guess
    I just haven't been playing with fast enough parts. ;-) I did notice that
    my old probes (that have thicker cables) don't show the effect quite as
    much.
     
  12. john jardine

    john jardine Guest

    I've the same setup as you and plugged a 74HC14 in. Saw no problems. See
    15nS rise and fall with no overshoots. Gate fed from a fast Wavetek function
    generator. Couldn't get what you show whatever I put in.
    Tried square, triangle and mid biased small sines. Always the same clean
    suare out.
    What you show (with it's 40MHz resonance) looks distinctly like a gate
    feeding a few feet of normal coax connected directly to a scope. Sure that
    probe's working?.
     
  13. My rise times were about half that according to the scope. Once I slowed
    them down a bit, they didn't overshoot like that. You tried feeding a DC
    coupled ~10MHz sine wave into it?
    I tried an older probe that I've had for a long time and got pretty much the
    same thing, massive ringing and overshoot. I put it on my old Hitachi
    V650-F analog scope and viewed the same results. I had trouble with the
    chip wanting to break into oscillation, so I guess I have just the right
    combination of circuit capactance/inductance/frequency to create resonance
    problems. When I have some time to tinker around, I will do some testing
    using a 10MHz crystal oscillator and a 74HC14 on a another breadboard and
    see if I see the same thing. BTW, I tied all unused inputs to Vss or Vdd.
     
  14. john jardine

    john jardine Guest

    The gate works OK with a DC coupled 10Mc sine input. Even works down to a
    1Vpp sine input but for this small a signal the sine needs to be sitting
    about an accurate midrail of 2.5V. (Can even get down to 0.8Vpp input if
    using a square wave!.)
    Other than what most non users casually suspect, those protoboard things are
    NOT a problem. The things are good to at least 50Mc.
    To clear another point, I also added 2 foot of extra wire to the scope probe
    ground lead and only got 1/2V overshoot (20Mc ring) , so that's not the
    problem either.
    My chip has a 10uF 'lytic across it's supply pins so 'decoupling' is not as
    problematic as many people would like to believe.
    Then directly connected a 1mtr 75ohm coax test lead to the gate and still
    only saw about 0.8V overshoots (15Mc ring).

    As you note, it's the chip-on-the-edge-of-oscillation that's the root cause
    of the problem. The key probably sits with the reason for that whopping
    great 12Vpp output.
     
  15. Eeyore

    Eeyore Guest

    Surely you mean 10Mc/s or Mcps ? ;~)

    Graham
     
  16. A 0.1 uF ceramic in parallel with the 10 uF might help.
     
  17. Thanks John, I've done plenty of PIC projects running at 8 and 10MHz with no
    issues like this at all. I really don't understand the hatrid of these
    solderless boards. If I was building it dead-bug style with unclipped
    leads, nobody would care. Or on top of a solid copper clad board with
    islands cut out, would be acceptable thru low VHF. But don't use a proto
    board, oh no too much capacitance. ;-)
    I've never had to go to any great lengths to see signals in this frequency
    range without a 50% overshoot.
    The times that I've seen decoupling make a big difference on these
    solderless breadboards have been relatively few. I do put them in out of
    habbit though. The major exception is the some of the nanowatt PICs (16F88s
    come to mind) not wanting to start. Add a cap straddling the proc power
    pins and viola, it starts every time. In that case I believe the cap is
    acting more like a close battery for switching surges than anything else.
    I really suspect that the whole thing is resonant at about 40MHz and
    plinking the chip at approx 1/4 of that frequency is hilighting that. Right
    now the oscillator and the IC get their power straight from a battery. I
    don't have a low ohmage series resistor going to the oscillator or buffer
    stages, perhaps this is part of the problem.

    At first I thought the bias resistors I used on the IC input pin were
    messing up the oscillator waveform. It turns out that the problem seems to
    be from the IC switching. (I posted a shot of it over on ABSE). Do you
    think the IC switching is mucking up the oscillator signal by causing a
    sudden change in impedance and reflecting back into the oscillator buffer,
    or something happening on Vcc that is looping around the power rail and
    getting into the oscillator and buffer?

    Thanks for everything. :)
     
  18. john jardine

    john jardine Guest

    Oops, slipped into local personal usage.
    In full, "Megacycles per second". Could have sworn it used to be just "Mc"
    or the even easier "mc".
    So ... "tc", "gc", "mc", "kc" "cps".
    (Ban those Cuckoo's nest "Hz" characters!.)
    john
     
  19. john jardine

    john jardine Guest

    Beyond a few megs (MHz, Mc) I don't bother with any extra series supply
    impedance. The connecting wires and traces have quality built in inductance
    so I just add a capacitor locally and it forms a free LC low pass filter. My
    working mental pic' is of -all- connecting wires having a notional '1uH' and
    I design accordingly.
    Yes. The nasty little jiggles are normal. The incoming signal takes a
    capacitive feedback hit every time the gate switches through it's threshold
    region. It's those input fets fed from the gate's input pin.
    Can be ameliorated (todays word) somewhat, by adding say 470ohms in series
    with the gate but we're now in a trade off situation, as the gate switching
    starts to slow down due to the input fets Miller capacitance taking longer
    to charge/discharge via the extra inline resistance.
    If I want a sinewave to feed other areas and keep it 'clean' then for any
    digital squaring up I now use an LMV7219 5V comparator.
    john
     
  20. Rich Grise

    Rich Grise Guest

    In the olden days, "megacycles" was sort of "understood" to mean
    "megacycles per second", like "forty-seven kay" meant 47,000 ohms.

    When they introduced the "Hertz", there was much confusion, because
    "megacycles" meant "megacycles per second", people started saying,
    "Hertz per second", which is actually a rate of change. I remember
    some columnist whining, something about the line frequency being
    "60 hurts per second". :)

    Cheers!
    Rich
     
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

-