Connect with us

bias current

Discussion in 'Electronic Design' started by Walter Harley, Sep 26, 2004.

Scroll to continue with content
  1. Bipolar opamps often have substantial input bias current - for instance, the
    LM833 has typical 0.5uA, which is presently causing me a design headache.

    As I was thinking about how to solve my problem (short of using a JFET amp,
    which introduces other problems), it occurred to me that one solution could
    be to put a current sink on each input of the opamp; it wouldn't be exact,
    but I could probably get it pretty close.

    Which in turn made me wonder: why isn't that current sink built right into
    the opamp? Are there any opamps that do it that way, or is there some
    reason it doesn't work?
     
  2. Genome

    Genome Guest

    | Bipolar opamps often have substantial input bias current - for instance,
    the
    | LM833 has typical 0.5uA, which is presently causing me a design headache.
    |
    | As I was thinking about how to solve my problem (short of using a JFET
    amp,
    | which introduces other problems), it occurred to me that one solution
    could
    | be to put a current sink on each input of the opamp; it wouldn't be exact,
    | but I could probably get it pretty close.
    |
    | Which in turn made me wonder: why isn't that current sink built right into
    | the opamp? Are there any opamps that do it that way, or is there some
    | reason it doesn't work?
    |
    |

    It's done.

    AoE? talks somewhere about input bias current cancellation whereby current
    mirrors sampling base current of a cascode transistor above the input stage
    inject cancellation current into the actual input stage to counteract its
    input bias current.

    It must work because they do it.... on the other hand it sounds like a
    wonderful opportunity for positive feedback.

    Can't give any type numbers tho.

    DNA

    My name is not Guy Macon
     

  3. Found an allusion to "bias-cancellation schemes" but I don't see the part
    about how it works. I need GoogAoE.
     
  4. Pooh Bear

    Pooh Bear Guest

    I'm sure there are some examples. It sounds kinda familair. LM101 ? I expect
    they date back several decades at least.
    Then there were the 'super-beta' input op-amps !

    Graham.
     
  5. They talk about doing it in the data sheet but don't go into much
    detail:
    http://www.analog.com/UploadedFiles/Data_Sheets/36971711685OP297_e_.pdf
    This one has a simplified schematic that shows the general concept:
    http://www.bgmicro.com/pdf/ics1023.pdf
     
  6. See page 13 of this datasheet for one idea:
    http://cache.national.com/ds/LM/LM124.pdf




    Best regards,
    Spehro Pefhany
     
  7. John Larkin

    John Larkin Guest

    OP-27, OP-37, LT1028 that I know of.

    And my name is not Fred.

    John
     
  8. Genome wrote...
    Indeed, it's a pretty popular technique.
    Pages 110, 111.
    Typically the datasheet lists the input bias current as +/- 1nA, etc.,
    showing a possible over-correction. If an input is left floating, it'll
    go to either the negative or positive rail, which could be considered
    positive feedback, but for realistic source impedances it's not an issue.
    We mention the OP-07 and LT1012. The rest of the '07 contenders use
    it, like the OP27, OP37 and TLE2027, plus others like NE5232, LT1001,
    LT1024, LT6011, and LT6012, etc. There are likely many more opamps,
    whose datasheets don't include schematics for us to analyze.
     
  9. Tim Wescott

    Tim Wescott Guest

    I assume you know of the technique to set the parallel equivalent of all
    the resistances on the V+ input to the same on the V- input, so equal
    offsets are generated?

    Like this:


    ||
    .-----||---.
    | || |
    R1 | |
    Vin ___ | |\ |
    ------|___|----o---|-\ | Vout
    | >---o------------
    .----|+/
    | |/
    |
    .-.
    | |
    | | R2 Make sure to set
    '-' R2 = R1
    |
    |
    ===
    GND

    created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de
     
  10. Thanks. I was looking in the opamp chapter.

    I guess the other issue is that one needs to make sure the current source is
    itself very low noise. On the other hand, you would avoid the thermal noise
    of the external resistance otherwise needed to balance Ib.

    Thanks for the leads. I had not realized that there were bipolar opamps
    that had the other parameters I need, but also had low Ib. The AD8672 looks
    like a remarkable component, although a bit pricey and available only in
    SOIC. No schematic on the datasheet, unfortunately.
     
  11. Just out of curiosity, why the insistence on bipolar? What are the
    "other problems" that using a JFET or CMOS input op-amp would cause?
    Noise?
    Both part of a trend, which is unfortunate (especially the latter).


    Best regards,
    Spehro Pefhany
     
  12. Jim Thompson

    Jim Thompson Guest

    See "IB-Cancellation-WithTwoOpAmps.pdf" on the SED/Schematics page of
    my website.

    Discussed here several moons ago, initiated, IIRC, by Win, perfected
    by me ;-)

    ...Jim Thompson
     
  13. Ken Smith

    Ken Smith Guest

    Getting 1nV/sqrt(Hz) out of JFETs requires that they be largish and run at
    moderately high currents. Neither is easy to do in an IC op-amp.

    JFETs have larger offset voltages than bipolars.
     
  14. Yes. Can't do it here, as the particular circuit has a varying impedance on
    the V- input (simple audio mixer).
     
  15. No real "insistence"; if I could find a JFET op with the right parameters
    I'd use it. CMOS won't do, though; I need higher voltages.

    The circuit is a very simple audio mixer, summing to a virtual ground on the
    inverting input of the opamp. (It's just a part of a larger circuit.) It
    has varying input impedance on the inverting input, range about 5k to 30k.
    DC coupled on the input and output; gain of about 5.

    The parameters I care most about are offset voltage on the output (input
    offset voltage and input bias current are the main sources), noise,
    distortion, and speed. Supply voltage needs to handle from +/-6V to +/-9V,
    though the ability to do +/-15V is nice. Supply current is important, since
    the device is battery powered; needs <= 5mA per channel, the less the
    better. And output range within two diode drops of each rail, while driving
    1k. Speed-wise, I'm using the opamp's gain to smooth out distortion in a
    current buffer stage on the output, so I need substantial large-signal gain
    out to 30kHz or so - call it >= 1MHz GBW and at least 1V/usec, preferably
    more.

    The LM833 does all this very nicely, except for the bias current, and it's
    cheap as dirt. I tried an OPA2134 but the higher voltage noise (8nV/rt-Hz
    compared to 4.5) was noticeable. I didn't delve as deeply as I could into
    my results: offset voltage was still higher with the OPA2134 than I would
    have expected, and distortion was slightly *higher* despite the much lower
    spec, so something else is going on that I didn't chase down. In any event,
    the OPA2134 is a lot more expensive, so it would need to be a big
    improvement for me to switch.
     
  16. Walter Harley wrote...
    You want a high-quality low-noise audio opamp. I'm puzzled why
    JFETs are out of consideration. For example the opa627 features
    0.00003% distortion. It's 4.5nV noise level is about the same as
    a 1.2k resistor, and is less than the noise from a 5k or 10k pot,
    such as you'd use in a mixer, why is 1nV necessary?

    The opa627's 0.25mV max offset voltage (for the cheaper A version)
    is also very low, and shouldn't be an issue.
     
  17. Winfield Hill wrote...
     
  18. See response to Spehro for particulars of my application... The OPA627
    won't do in my case, because it's got unacceptably high supply consumption
    (I think - TI datasheet summary says 7.5mA per channel, datasheet itself is
    ambiguous, I haven't tried a sample). More importantly, it costs $16 a pop
    even in 100 quantities - this goes into a unit with a retail price of $139,
    so that won't quite work :)

    There are a lot of opamps, more coming all the time. I'm sure there are
    some that I'm missing just because I've not re-run the parametric searches
    for all the manufacturers each time I've revised my expectations for the
    chip! I'm always glad to learn about ones that are new to me.
     
  19. how about MC33078 from Onsemi, not quite as good, but ISTR 4ma/opamp
    about half the consumption of a NE5532,




    martin

    Serious error.
    All shortcuts have disappeared.
    Screen. Mind. Both are blank.
     
  20. I read in sci.electronics.design that martin griffith
    ST Micro LS204 is quite useful and has low consumption. There is also a
    quad version.
     
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

-