Connect with us

Op-Amp Problem

Discussion in 'Electronic Basics' started by Mike, Aug 1, 2004.

Scroll to continue with content
  1. Mike

    Mike Guest

    Hi,

    I'm using an LM358 op-amp to amplify a signal. The signal is AC and has a
    peak-to-peak value of 5mV. The op-amp is setup to to amplify the signal by
    500 (using a 1M resistor and a 2K resistor). However, the output of the
    op-amp is only amplified by a factor of 10 (not 500). If I change the 2K
    resistor to a 1K resistor I would expect the op-amp's output to double, but
    it doesn't, it just stays the same. Does anyone know what I'm doing wrong?
    The op-amp is powered from a +5V and -5V supply. The frequency of the
    signal is around 35KHz.

    Thanks for any help.
     
  2. Look at the 'open loop frequency response' of the datasheet given
    here:

    http://cache.national.com/ds/LM/LM158.pdf

    You note that the open loop gain at 35kHz is less than 40dB, which
    means its less than 100. Op-amps are internally compensated to do
    this, so they don't oscillate. Thats the most gain you can get at a
    particular frequency with this opamp.

    Using a negative feedback scheme like you are using, the open-loop
    gain should be much higher than the final gain, or there will be
    distortion. Thus, if the open loop gain is 100, your target gain
    should be more like 10.

    Regards,
    Bob Monsen
     
  3. CFoley1064

    CFoley1064 Guest

    Subject: Op-Amp Problem
    Hi, Mike. An op amp is a gain block whose total open loop gain is dependent on
    several factors, including the built-in compensation capacitor on the op amp
    die. Because of that, if you set up a circuit like this with an LM358 (view in
    fixed font or M$ notepad):

    Open Loop Gain
    Input Signal |\|
    o---------|+\ Output Signal
    | >-------o
    .---|-/
    | |/|
    |
    ===
    GND

    created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

    and you apply a small amplitude signal at 1 MHz, the output will have a gain of
    about 1. If you apply a small signal at 100 KHz, the gain will be about 10.
    If you apply a small signal at 10 KHz, the gain will be about 100. This
    product of gain and bandwidth for an open-loop op amp will be the same for any
    given frequency, and is called the gain-bandwidth product. For an LM358, the
    gain-bandwidth product is 1 MHz, which means you can't expect more than a gain
    of 30 at 35 KHz, no matter what feedback resistors you use. As a practical
    matter, the theory of negative feedback you're using to set gain requires
    excess gain (minimum of 20 dB or at least 10 times), so you actually can't set
    it up for 35 KHz with more than a gain of 3 and expect to get any fidelity in
    amplification (your distortion will be too great). So, even using both op amps
    of the LM358, you can't really expect to get more than a gain of 10 at 35 KHz
    (gain of 3.2 * gain of 3.2). So your voltage gain of 10 means you've actually
    got an overachiever for an op amp. GBW product is specified min./typ., not
    max.

    If you can find a faster dual op amp (GBW >= 10MHz), you can set it up so each
    op amp has a gain of 22, which will give you a total gain of 500. At that
    gain, you'll also be concerned about noise amplification from different sources
    (such as your resistors and the op amp itself), so choose a low noise op amp
    and try to stay away from those megohm resistors. Also try to do a good job on
    power supply bypassing -- with a gain of 500, power supply noise will become a
    factor. Try something like this:

    35KHz Gain of 500
    ___ ___
    .--|___|--. .--|___|---.
    35KHz | 22K | | 22K |
    Input Signal| | | |
    ___ | |\| | | |
    o--|___|- o---|-\ | ___ | |\ |
    1K | >--o--|___|--o--|-\ |Output Signal
    -|+/ 1K | >----o------o
    |/| -|+/
    |/
    created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

    Good luck
    Chris
     
  4. John Larkin

    John Larkin Guest

    That amp is out of gain-bandwidth. Flat out, with no feedback at all,
    its gain at 35K is only about 30. You need an amp with a lot more
    gain-bandwidth product, or you can cascade two or three stages of
    lower gain/stage.

    A single gain stage, even with a hotter amplifier, will suffer from
    parasitic capacitance across such a big feedback resistor. The
    several-stage thing is a cleaner way to get a lot of gain.

    John
     
  5.  
  6. John Jardine

    John Jardine Guest

    Come what may, should be getting more than X10 out. Are you sure you haven't
    got a massive DC offset at the op-amp output?.
     
  7. You have two problems. The first is that the LM358 has a gain
    bandwidth product of about 500,000. Divide that by 35,000 and you get
    a gain of about 14 or so. Include negative feedback, and you get
    less. You could use several stages with gain of less than 10, each,
    and arrive at a total gain of 500, but then the second problem shows
    up. 500 gain would turn 5 millivolts into 2.5 volts, but assuming
    these are peak to peak values much less than RMS) this implies an
    output rate of change of about .35 volts per microsecond, max. The
    LM358 slews at no more than .5 volts per microsecond, so you are very
    close to having the last opamp not being able to swing its output fast
    enough to keep up with the signal. If the 5 millivolt input is an RMS
    value, you are in big trouble.

    To achieve this 500 gain with a single opamp, it should have a gain
    bandwidth product about 10*500*35,000=175MHz. Such a beast can be
    quite difficult to get working correctly without a good layout and
    some experience or luck.

    I would probably go with a dual opamp that had a gain bandwidth
    product of about 7 to 10 MHz, each. Each amplifier would have a gain
    of about 22. This is based on needing the square root of 500 gain in
    each stage for a GBP of 10*sqrt(500)*35000=7.8MHz. An LM6132,
    perhaps. Its slew rate of about 8 volts per microsecond is also much
    more capable of keeping up with the demands of the 2.5 volt output
    signal.

    http://cache.national.com/ds/LM/LM6132.pdf
     
  8. Activ8

    Activ8 Guest

    On 01 Aug 2004 19:21:17 GMT, CFoley1064 wrote:

    <snip

    And since these buggers have high input impedance, keep the
    conductor length going into the inputs as short as possible (some
    apps require guard rings) or you'll also pick up more noise.
     
  9. BobGardner

    BobGardner Guest

    The op-amp is setup to to amplify the signal by
    The 2k and 1k resistor are loading down the high impedance output of your
    signal source. You need 2 gain stages... first one needs to be non inverting
    (and hi impedance) with gain of 10 or 20, 2nd stage can make up the rest of the
    gain.
     
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

-