Op-Amp minimum input signal - differential mode

Discussion in 'Analog & Mixed-Signal Design' started by eyesee, Aug 15, 2016.

  1. eyesee

    Thread Starter Member

    Oct 19, 2013
    33
    3
    What parameters should I be looking at in a datasheet to determine the minimum input signal level an Op-Amp can handle in differential mode?

    Thanks.
     
  2. ramancini8

    Member

    Jul 18, 2012
    442
    118
    input offset voltage, bias current, input offset current
     
  3. hp1729

    Well-Known Member

    Nov 23, 2015
    1,948
    219
    How about some suggestions for how to interpret these characteristics for a minimum input voltage?
    Input offset voltage? A change in input still gets noted, doesn't it? Same with input bias current or input offset current???
    How small can a change in input be before it does not register at the output? Speed not being the question.
    How about noise figure? If the input signal is down where noise is it would not be noticed. I say look at noise level.
     
  4. OBW0549

    Well-Known Member

    Mar 2, 2015
    1,308
    884
    Not sure what you mean by "differential mode" but the following parameters all have some effect on an opamp circuit's performance with small signals, irrespective of how the amplifier is used:

    Input offset voltage and its behavior over temperature;
    Input bias current and its behavior over temperature;
    Input offset current and its behavior over temperature;
    Input voltage noise density, together with system bandwidth
    Voltage noise 1/f corner frequency;
    Input current noise density, together with system bandwidth and input source resistance; and
    Current noise 1/f corner frequency.

    Those are the major factors to consider. Also, if there is any significant noise on the opamp's supplies, the power supply rejection ratio could be important.

    EDIT: add input offset voltage long-term drift to the above list; in some cases, it can be an important parameter to consider.
     
    Last edited: Aug 16, 2016
  5. hp1729

    Well-Known Member

    Nov 23, 2015
    1,948
    219
    Noise figures I can see. The others, I don't see how they prevent a change in input signal from being recognized.
     
  6. #12

    Expert

    Nov 30, 2010
    16,261
    6,770
    The question is rather vague. That leaves a lot of options open for what could possibly interfere with his signal. For instance, if the input offset voltage is 5 mv and the input signal is 1 mv, the input signal might not cause a change of state because it does not overcome the input offset. The fact that you don't know this (and several other scenarios) is not a helpful answer.
     
  7. hp1729

    Well-Known Member

    Nov 23, 2015
    1,948
    219
    Re: Input offset voltage
    This will come out as an error in the true representation of the voltage but a change at the input of 1 mV will still be reflected in a change at the output according to the gain of the circuit. Right? The 5 mV is just an "offset" from the true representation of the input, isn't it?
    Thermal drift, same thing.
    Now if I want an op amp to operate at very low input levels I would pick one with noise levels below the expected input voltage. How much noise does the op amp generate at the expected input frequency?
     
    Last edited: Aug 16, 2016
  8. OBW0549

    Well-Known Member

    Mar 2, 2015
    1,308
    884
    I guarantee you, you'd see it real quick if you had an opamp with significant Vos, Ib and Ios temperature coefficients operating in an environment with fluctuating temperature. Even in a controlled room-temperature environment, like home or office, those tempcos can cause apparent input shifts that far exceed the opamp's voltage noise, often by more than an order of magnitude.

    If a circuit has the amazingly good fortune to operate in an environment where the temperature never changes, or if the input offset voltage, input bias current and input offset current were somehow magically constant over temperature, you'd be correct; in that case, obviously, the only thing left to consider would be the opamp's noise specs.

    But in a real-world environment in which the temperature continually changes, and with real-world components whose input characteristics are temperature-dependent, those characteristics MUST be considered when designing systems that must process small signals.
     
  9. eyesee

    Thread Starter Member

    Oct 19, 2013
    33
    3
    Thanks for all the replies so far.

    I am interfacing an un-amplified bridge pressure sensor with the output from the sensor connecting directly to the inverting and non-inverting inputs of the Op-Amp.

    The sensitivity of the pressure sensors I am looking at are typically quoted in the µV or low mV range.

    The Freescale MPX2300DT1 is one of the sensors I am considering.
     
  10. OBW0549

    Well-Known Member

    Mar 2, 2015
    1,308
    884
    Connecting the outputs of the bridge sensor directly to the opamp inputs won't work, because opamps have extremely high voltage gains that are way higher than what your application will need, plus they are highly unpredictable. They're meant to operate in a circuit in which feedback resistors are used to precisely set the circuit's voltage gain.

    If you want a 1-chip solution that will allow you to connect the sensor directly and have a precise voltage gain, you need an instrumentation amplifier, such as an LT1167. It has excellent input specs and is designed for bridge amplifier applications.

    Linear Tech also has an excellent application note on bridge measurement techniques that is well worth reading.
     
    Last edited: Aug 16, 2016
  11. hp1729

    Well-Known Member

    Nov 23, 2015
    1,948
    219
     
  12. OBW0549

    Well-Known Member

    Mar 2, 2015
    1,308
    884
    It doesn't necessarily mean that an input signal would not be seen; but it does mean that unless the input signal is significantly larger than all the other stuff (noise, temperature-induced shifts, interference, etc.), you might not be able to say for certain that "Yes, THAT is a signal. It is definitely NOT noise or some other garbage."

    For high-quality audio work, obviously the only thing we would care about very much is noise. We certainly don't care about DC errors (e.g., offsets), nor do we care much about temperature shifts because those occur over a time span of seconds, minutes or hours, not milliseconds or microseconds. In other words, temperature fluctuation effects occupy a portion of the spectrum far, far below audio so they're of no importance.

    But for instrumentation and measurement applications, such as processing the millivolt-level signal from a strain gauge, thermocouple, RTD, Wheatstone bridge pressure sensor, etc., those factors can be critically important, just as much as (or even more than) noise.

    I spent most of my career doing that kind of work, so I learned to pay close attention to all of the opamp characteristics that affect performance in circuits with small signal levels.
     
  13. ramancini8

    Member

    Jul 18, 2012
    442
    118
    This was a poor question, and my short terse answer was meant to indicate that. I should have ignored the question. Now that I have reviewed the question I answer with confidence that there is no minimum signal that a differential op amp can handle, but you should review all op amp parameters..
     
  14. eyesee

    Thread Starter Member

    Oct 19, 2013
    33
    3
    Thanks for the Linear Tech. suggestions.

    I intend to use feedback and not operate the Op-Amp in open-loop mode. What I wasn't sure about is if the specifications state an input offset voltage in the mV range whether the Op-Amp would be suitable for use with a sensor that outputs a µV signal? For example, if I decided to use the LT1167 Instrumentation Amplifier, the input offset voltage is typically between 15-20µV however, the Freescale sensor mentioned above outputs a low µV signal.

    Could you clarify what you mean by this?
     
    Last edited: Aug 17, 2016
  15. OBW0549

    Well-Known Member

    Mar 2, 2015
    1,308
    884
    Keep in mind that the amplifier you choose is not the only source of offset error: according to its data sheet, the MPX2300DT1 itself has a zero-pressure offset of +/- 750 μV, more than ten times the +/- 60 μV maximum offset of the LT1167.

    One way or another, offsets are an unfortunate fact of life and must be dealt with: they can either be nulled out by a strategically-placed trimpot or, if an A/D converter and microcontroller are involved, they can be calibrated out in software, with correction factors stored in EEPROM.
     
  16. hp1729

    Well-Known Member

    Nov 23, 2015
    1,948
    219
    LT1167? Good choice. Eve at maximum gain of 1,000 your offset error would be 0.06 V. Your input signal would ride on top of that. AC coupling, looking for a change in voltage, omits the offset error. Or the offset can be handled in software. Noise below 10 Hz is down around 0.28 uV. No problem there even at 5 V or 3.3 V power. Maximum output signal might be about 1.5 mV, plus offset.
     
    Last edited: Aug 17, 2016
  17. OBW0549

    Well-Known Member

    Mar 2, 2015
    1,308
    884
    AC coupling is almost never an option in pressure measurement, unless what's being measured is pressure variations.

    Read the data sheet. The LT1167 is specified to operate only down to +/- 2.3 volts supply, or 4.6 volts total, and will not work at 3.3 volts. If the LT1167 could operate from a 3.3 volt supply, the manufacturer would have said so.
     
    Last edited: Aug 17, 2016
  18. hp1729

    Well-Known Member

    Nov 23, 2015
    1,948
    219
    Super!. No problems at those voltages.
     
  19. OBW0549

    Well-Known Member

    Mar 2, 2015
    1,308
    884
    I don't understand what you mean by that. Explain, please?
     
  20. hp1729

    Well-Known Member

    Nov 23, 2015
    1,948
    219
    I see no problems with your design working at +/- 2.3 or so volts.
     
Loading...