• State
  • Locked Locked
  • Replies 7 replies
  • Subscribers 49 subscribers
  • Views 1590 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Low bias current amplifiers

Scott Hamilton
Prodigy 110 points
Other Parts Discussed in Thread: LMC6041, LMC6001

For measurement of very low currents e.g. atto or femtoamps (in particular using transimpedance amplifiers), the input bias current (Ib) of the active device becomes critical. In trying to achieve the best signal-to-noise ratio it is necessary to know the variation of Ib over temperature and over a batch of devices (variation within a production lot and from lot to lot) to determine whether it may be beneficial to undertake selection of devices, and whether lowering temperature produces significant decrease. There is little published information on these matters in the literature or in datasheets so this is a request to the community to solicit any practical information. The recent post by Bruce Trump on 'Input bias current of CMOS and JFET amplifiers' is a starting point.

Scott Hamilton.

  • 0
    Guru 21975 points

    TI peeps--  In offline communication Scott was referring specifically to the LMC6041 which is well-suited to his application.

    Regards, Bruce.

  • 0
    TI__Mastermind 49966 points

    Hello Scott,

    For the LMC60xx series, from 25°C to about 90°C the bias current change is about 1.8x to1.9x for every 10°C. Above 90°C, the slope may increase to 2x up to 130°C.

    Theoretically, below room temp, the bias currents will decrease at 1.8x every 10°C. But at this point the currents are so low (a few fA's) that the "bias" current is dominated by internal die and external resistive leakages, which causes the curve to "flatten" out at low temps.

    And that is in a very, very dry environment at cold, otherwise condensation will ruin any chance of low currents.

    Does anyone else find the fact that we can match diode leakage currents to within femtoamps in mass production incredible??

    There are no "published" specs because then we would be held to them and have to guarantee them! With the exception of the LMC6001, the LMC CMOS devices are marketed as general purpose devices that just happen to have incredibly low bias currents. These bias currents can and will vary over production lots - and testing at these currents takes a lot of time so it is not economical to test every batch. The LMC6001 is tested for almost a minute each - which is why it is $9. We cannot do that for every $0.60 part.

    If you are seriously talking about aA resolution - you will need to individually calibrate each system, anyways. Every high precision instrument needs to go through a calibration routine. You will also need environmental control (temperature and vibration), too. At these levels, you are also building a sensitive thermometer. I have played down at these levels, and EVERYTHING you do affects the measurements.

    Bruce's article covers the basics if the input current, and I covered some of the external influences in my EDN articles:

     http://www.edn.com/design/analog/4368681/Design-femtoampere-circuits-with-low-leakage-part-one

     http://www.edn.com/design/analog/4375459/Design-femtoampere-circuits-with-low-leakage---Part-2--Component-selection

     http://www.edn.com/design/analog/4395651/Design-femtoampere-circuits-with-low-leakage---Part-3--Low-current-design-techniques

    The floating input "test" circuit that Bruce showed is a good way to find the "sweet-spot" where the bias current is zero. This will be vary device to device by a few percent, and will be sensitive to temperature and supply voltage.

    Regards,

  • 0 in reply to Paul Grohe
    Prodigy 110 points

    Paul,

    Thanks for the comments regarding my query. I have been involved in this area for very many years now and agree that EVERYTHING matters in this region. I have  read through your EDN articles and it is useful to be reminded of the difficulties, and the comments will also be of great help to others. My amplifiers are housed in an evacuated and temperature controlled (15C) housing to realize the optimum performance and I will try reducing the temperature to say 0C to see if the noise decreases. The S/N ratio  improvement as a function of increasing feedback resistance (maximum 100Tohm at present as that is the maximum resistor value presently available) is finally limited by the shot noise of the bias current so any further improvement depends on this. The matter of absolute calibration is not covered by standard instrumentation but that is at present not too important as it is relative sensitivity between a number of amplifiers that is appropriate. There is however, a more difficult measurement in determining the transient response since settling time is critical as measurements have to be taken in a few seconds since the signals decay with time. The development of a suitable very low (proper i.e. with very high source resistance) current source that has a known short response time (or much shorter than the amplifier) remains a considerable problem. This is required to allow adjustment of the compensation arrangements required to achieve much faster response than from the 'standard'  transimpedance circuit configuration. So if there are any readers with experience in such a current source I would like to hear about  it.

    Kind regards, Scott.

  • 0 in reply to Scott Hamilton
    Intellectual 330 points

    What do you think about a Keithley 6430?

  • 0 in reply to Bela istvanfi
    Prodigy 110 points

    Bela,

           Thanks for the suggestion. I have previously looked at this instrument but it does not quite get to the required performance. The source settling time for the lowest current range is too long (compared with the amplifier settling times) and the effective current-source resistance is too low. Since the amplifier feedback resistance is up to 100Tohm (10^14), the spec value for the instrument of 1Tohm (10^12) is far too low.  The amplifiers measure (high voltage) ion beam currents so the effective source resistance is very high indeed.

    Regards, Scott Hamilton.

  • 0 in reply to Paul Grohe
    Prodigy 110 points
    Dear Paul,
    Having made a number of very high gain (100T) transimpedance amplifiers using the LMC6041 as input IC I find the output shows random transient deviations well above the noise level (like the old popcorn noise?). There have been suggestions that these may be due to cosmic ray ionizations around the input but leaving the system running for long periods tends to show some decrease in the occurrences, which may suggest they are internal to the LMC6041. So I would like to ask if there is any information on 'noise' of this sort. Particular devices show significant differences in the frequency of such events. The old 'popcorn noise', as far as I recall, applied to early bipolar devices so I wonder if there may be a related effect in CMOS amplifiers. I would appreciate your views on this phenomenon.

    Regards, Scott Hamilton.
  • 0 in reply to Scott Hamilton
    TI__Mastermind 49966 points
    Hi Scott,

    Popcorn noise will have a very distinct "high" and "low" transition (assuming an appropriate time constant) - like Morse code. Popcorn noise also occurs in CMOS, but the charge-state transitions tend to be smaller since CMOS bias currents tend to be lower than bipolar (fA's vs nA's). Popcorn can "pop" every few seconds, or once a day, or continuously in small, fast pulses.

    Ionization hit signatures will differ depending on the application (integrator or transimpedance [I/V converter]), since it is a deposition of charge.

    For an integrator, the ionization pulse will cause a jump-step in the integration (like a sudden change in offset), but the integration slope returns to normal afterwards.

    For a transimpedance amp, it generally looks exponential spikes that settle back to the baseline.

    Recently, EMI noise (from Cell phones) has been mistaken for popcorn noise, as it tends to create occasional random amplitude "bursts" as the phone chats with the network. But the give away is that there is usually a distinct repetitive pattern between the individual pulses. This can be tested by making a phone call or using a data-intensive application (like scrolling Google maps - over the air - not over WiFi).

    The "fix' is to minimize and guard the active input area. A thin aluminum shield will block most all Alpha and Beta radiation. Gamma will still get through, but with much lower occurrences.

    Stick your circuit inside a thick, metal box and see if the pulses go away. Coffee cans, paint cans or "gift' cookie tins work well. A thick safe may work for Gamma rays..

    Regards,
    Paul Grohe