Connect with us

LM2907 chattering?

Discussion in 'Electronic Design' started by Glenn Ashmore, Feb 15, 2004.

Scroll to continue with content
  1. I have come up with another problem with the RPM switch section. I have
    tried to distill it down to the basics. Using the speed switch circuit
    on page 8 of the datasheet and controlling a relay drawing 43mA it
    starts to chatter about 10% below the set point and increases rapidly
    until it settles down about 10% above set point. I have tried just
    driving an LED and get the same thing.

    The specifics: Switch point is 116Hz. Vcc is a very clean 12.02V. Pin
    7 Vref is a steady 6.01V The chatter happens when Pin 3 going to the
    comparator is within about .10V of Vref. It seems to be bouncing around
    about .10V. Putting the scope on pin2 between C1 and the charge pump I
    get a clean fairly square wave up to the point it starts chattering and
    then the wave starts skipping pulses.

    Looking at pin 2 on channel A and pin 3 on channel B I can see that the
    variations on pin 3 are in sync with the skips. It continues to chatter
    until the skips out number the pulses and the output of pin 3 settles down.

    I have tried various combinations of C1 R1 (.01 to .1uF and 43K to 430K)
    with the same results. Changing C2 (.5uf to 50uF) effects the time it
    takes to settle but does not seem to effect the charge pump skipping.
    Adding a resistor for hysterisis across pin 5 & 7 does change the on/off
    spread but does not effect the charge pump skipping.

    Where should I look next?

    Glenn Ashmore

    I'm building a 45' cutter in strip/composite. Watch my progress (or lack
    there of) at:
    Shameless Commercial Division:
  2. Glenn Ashmore wrote...
    Set up your LM2907 up as a V-F converter, and use two comparators
    (with hysteresis) on its output. You'll solve the problem, and
    you'll also be able to use just one LM2907 rather than two. Use
    an LM393 (dual) or LM339 (quad) for the comparator, rather than
    an opamp, like a LM741, etc. (I glanced at your schematic.)

    - Win

  3. CFoley1064

    CFoley1064 Guest

    From: Glenn Ashmore
    OK. You've gotten most of the way, and you've got your first check. Look at
    what's going on carefully, and the problem will probably start talking to you.

    You need to start out with the understanding that the LM2907 will have ripple
    at the output -- that's the nature of the beast. You should use a scope to
    determine the ripple, and you will have to make some intelligent choices as to
    how much ripple you can afford in this circuit, and adjust the component
    selections accordingly. This will of course mean that your response time will
    be slower. You may want to add a separate low-pass filter at the output to
    knock it down more.

    Once you've addressed that, you should start looking at your comparator
    circuit. With a couple of extra resistors, you can add hysteresis to the
    comparator, so once it starts switching, the switching point will snap past
    your ripple to provide a steady output. Your solution will probably require
    both of the above -- low pass and hysteresis.

    Having looked at these, start looking at the rest of the circuit. Try removing
    the relay, and replacing it with a light resistive load. Does this solve your
    problem? If so, you might have issues with the comparator itself, the relay
    coil (you have placed a diode across the coil, right?), and the relay load, if
    it's being powered by the same power source that's driving the circuit.

    I dislike using the LM311, or any high speed comparator, for driving relay
    loads, no matter what the data sheets or app notes claim. All comparators are
    meant to switch quickly -- a lot faster than relay time (milliseconds). If
    you're still having problems, try replacing the comparator with a single supply
    dual op amp like the LM358 (use the 1st for the low pass filter, and the second
    as a comparator driving a small NPN transistor like the 2N3904, which drives
    the relay). Does this solve the problem? Op amps are made to be stable when
    changing output voltages, usually at the expense of reduced speed (if you're
    interested, google op amp compensation caps or read some app notes). By the
    way, if you're going to use an op amp, the hysteresis becomes even more
    important, to avoid stressing the transistor by having it linger in a
    non-switching voltage, which could cause high power dissipation (1/2 * 13.8V *
    43 mA is pretty close to maximum Pd for a small signal transistor -- if you
    stay there for any time, you might pop the 2N3904).

    Another thing you might want to look at, especially if you're using a resistive
    voltage divider as the switching voltage reference, is replacing it with a
    separate voltage reference which has some power supply rejection. The LM385 is
    good for this type of thing.

    If you're heading in this direction, you might want to look at the LM10, which
    has an op amp and reference built in. This IC can kill two guano dropping
    birds with one stone.

    The important thing when faced with anomalies of this type is not to panic.
    *Look* at your prototype circuit creatively, trying to find ways to make minor
    changes in the circuit, or even just making a real measurement, in such a way
    that the circuit itself will tell you what's wrong.

    If this doesn't do it for you, you'll have to give a little more information.
    All of the above has been very general advice, and your specific app may have
    difficulties peculiar to the circuit.

    Good luck
  4. Fred Bloggs

    Fred Bloggs Guest

    If you're referring to that atrocity of a pump control schematic, then
    my advice to you is LAY OF THE DAMNED BOOZE while you are working on
    this apparently life-long endeavor. Both of your 2907's show the ripple
    capacitor on pin 4 the output transistor emitter- no charge pump current
    averaging that way!-you are driving the output with high ripple!- and
    the "high RPM" pin 4 output is shunted to GND through the Q1
    BE-junction- so I just don't see that LED driving inverter doing much
    beyond stuck on 1. Where the "F" does K2-1 go/come from anyway?!
    Nevermind- you must be that Burridge in disguise.
  5. Don't worry. With the help of others here that has all been corrected.
    I have the whole thing breadboarded and except for a couple of
    things it is working pretty good.

    Glenn Ashmore

    I'm building a 45' cutter in strip/composite. Watch my progress (or lack
    there of) at:
    Shameless Commercial Division:
  6. Now THAT is an idea! Thanks

    Glenn Ashmore

    I'm building a 45' cutter in strip/composite. Watch my progress (or lack
    there of) at:
    Shameless Commercial Division:
  7. Thanks, I tried most of that with no luck so I started looking at the
    input frequency. I had built a little circuit to emulate the input from
    the proximity sensor and it was introducing some noise. Cleaned that up
    and it is working fine. Now I have to worry about keeping noise out of
    the run to the sensor.

    CFoley1064 wrote:

    Glenn Ashmore

    I'm building a 45' cutter in strip/composite. Watch my progress (or lack
    there of) at:
    Shameless Commercial Division:
  8. Fred Bloggs

    Fred Bloggs Guest

    Don't worry, I won't, liar boy. I just read your piss-ant thread and not
    a single person caught your ripple capacitor on pin 4, the output
    transistor emitter.
  9. Fred, we are not debating the Iraq war any more. This is a simple
    little analog circuit. :)

    I caught that problem with C2 myself but GPG's recommendations cleaned
    up the output side very nicely.

    Glenn Ashmore

    I'm building a 45' cutter in strip/composite. Watch my progress (or lack
    there of) at:
    Shameless Commercial Division:
  10. GPG

    GPG Guest

    Look at pin1, input. The input thresholds are given as +- 15mV,
    so interference on.this scale will affect the input comparator.
    Remove the hex inverter section and AC couple the input signal so that
    the -ve threshold is reliably reached.
  11. Glenn Ashmore wrote...
    My suggest was comparator hysteresis, enough to more than exceed
    the usual F-V ripple.

    - Win

  12. Jamie

    Jamie Guest

    you need a Sample and Hold circuit added to it.
    the Itel data sheet has an example for smoothing it out.
    you may also want to increate the capacitor value for
    longer charge time.
    you do say something about a comparator, if you are
    usng an extra op app for a switch then put alittle feed back
    from the Op output to the + input of the op all and do not shut
    the input to where the loop back does no good.
    use the - input as your input from the 2907 , that is will
    give you a schmitt trigger and latch it.
    i am sure others here can help you with that also.
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