Very high impedance Op Amp / follower

Discussion in 'The Projects Forum' started by sage.radachowsky, Jun 18, 2010.

  1. sage.radachowsky

    Thread Starter Member

    May 11, 2010
    241
    38
    Hello,

    I have an engineering challenge of measuring a very small current produced by microbes... it may be about 10 or 100 pA of current that I need to measure.

    The measurement doesn't have to be super-accurate, but accuracy is desirable. Basically, I have to be able to discern between no current and some current, and the ability to quantify the current to the nearest 10 pA would be excellent.

    I have found this TI part OPA129, which is a DiFET input for 100 femtoAmp input bias current! The offset error is 2 mV max, but I think that will be okay because my scheme to charge up an extremely low leakage capacitor and read the voltage from it, buffered by this femtoAmp op amp.

    I was just reading a Bob Pease essay on capacitor leakage... he is showing that polypropylene capacitors have extremely low leakage, such that they will drop a few mV per year.

    The microbes that I want to gain electricity from can probably produce 0.4V or so, and at least 0.1V, so if I can see that capacitors are charging up to 0.1V when connected to the microbes, and do not charge up when everything is the same but there are no microbes, then perhaps this is good enough proof in the pudding.

    What I am wondering here are people's input about doing extremely low current measurements like this.

    It is possible to make a Faraday cage, and I also expect to lay out a PCB with traces very far apart to eliminate leakage by the board or anything else.

    Has anyone here done anything with picoamp quantification? Any pointers or ideas?

    Thanks!
     
  2. Ghar

    Active Member

    Mar 8, 2010
    655
    72
    Very far apart traces are more susceptible to noise pickup. [err, far apart signal and return is bad, separation between different signal lines is good]
    One method for avoiding leakage is by forming guard rings, shown at the bottom of this page:
    http://www.analog.com/library/analogDialogue/archives/39-09/layout.html
    The TI datasheet for this part also discusses guard rings on page 6 and 7.
     
  3. Bychon

    Member

    Mar 12, 2010
    469
    41
    Another way to keep leakage low is to use teflon stand-offs..as in, no circuit board at all.
     
  4. sage.radachowsky

    Thread Starter Member

    May 11, 2010
    241
    38
    Thank you, Ghar and Bynchon, for the tips. Those are both great inputs.

    I just was reading the Bob Pease article more, and I found that he mentioned a unity-gain follower made by his old employer (National Instruments) with 3 femtoAmp input bias current -- can you believe it? And 3 mV max offset error.

    That is 30 times better than the TI part I had found. Amazing!
     
  5. sage.radachowsky

    Thread Starter Member

    May 11, 2010
    241
    38
    So, perhaps I should avoid the PCB leads altogether and use a twisted-pair of wires from the capacitor to the LMC662, so that noise becomes mostly common-mode and the voltage of the capacitor gets buffered very well with 3 femtoAmps of input bias current... unbelievable performance in a very simple design!

    Am I missing something here? Any other things I should be watching out for? Thanks for all input even if it's going out on a limb. I have never worked with femtoAmps before. It's a little intimidating.

    Apparently 1 femtoAmpere equates to 6,000 electrons per second. So few electrons! It's amazing.
     
  6. Ghar

    Active Member

    Mar 8, 2010
    655
    72
    I don't know if you'd gain anything from twisted pair.
    If you use very tight traces it's probably comparable.. think of the thickness of the insulation on the wire and how tightly you can actually manage to twist it.
    Using twisted pair also makes the wires longer than a good tight trace.

    If you're going with two sided or multi layer boards you can try those techniques in the Analog Devices link I gave, mainly the microstrip and stripline traces.
     
    sage.radachowsky likes this.
  7. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    That looks awesome on paper. In practice, it's not that hard. It's more significant to control noise so the amplifier signal is clean. Using 1 to 10 Gohms and a good op amp with higher input impedance works nicely.

    Yes, of course, and you'll need one. Just what environmental noise is present may surprise you. Got a radio station on campus? Is there a pager system running? You might get by with aluminum screen mesh.

    What is worse is the crud on the power line. Think about using a dedicated supply line and an Isobar line filter. Place the power supply in an enclosure separate from the main instrument. Do your analog amplification in one shielded box, and digitize in another. If you have to use a Dataq DI-720, you will be amazed by the channel-to-channel crosstalk.

    Good luck!

    I make stuff to monitor piercing-and-sucking insect feeding activity. With an OPA134 headstage amp, I can get nice signals corresponding to internal muscle activity through a lead glued to the insect's thorax. We see 50 mv potential drops when the insect pierces the plant cell membranes.

    I have made a successful single electrode voltage clamp using CA3140 op amps with only 10 Gohm input impedance. Good grounding and noise control will get you there.
     
    sage.radachowsky likes this.
  8. sage.radachowsky

    Thread Starter Member

    May 11, 2010
    241
    38
    Thank you to both.

    Beenthere -- do you do continuous or discontinuous single electrode voltage clamps? That is fascinating stuff.

    Regarding a power supply, my goal is to make it battery-powered and the data transmission wireless. I think that a 12-bit ADC will be sufficient.

    Regarding noise control, since this system is measuring the charging of a poly capacitor, I suspect that this will stabilize it right at the entry to the femtoAmp-input buffer. That makes me wonder whether EMI will even be a problem at all. In effect, the system is a super-low-pass filter, I think. Rates of change will be seconds if not minutes. Do you think that a Faraday cage is necessary even so, if I place the entry of the buffer very close to the capacitor lead? I guess it will still be good to enclose that, given the super low current levels.
     
  9. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    That single electrode clamp was done back around 1980 or so. It started as a copy from a published paper, but the circuit had suspicious values & did not function. I did a successful redesign of the circuit. That was part of being in a major university electronic shop - you got to see and do everything.

    Any time you have a high impedance, you have an opportunity for noise to show up. Take a look at Bud Industries cast aluminum boxes to see if your experiment might fit in one - hard to get better shielding than that, and the metal is easy to work with.
     
  10. atferrari

    AAC Fanatic!

    Jan 6, 2004
    2,645
    759
    What about copper in this case? I always believed it is the best shielding screen. And it is easy to work too.
     
    Last edited: Jun 25, 2010
  11. John P

    AAC Fanatic!

    Oct 14, 2008
    1,632
    224
    I know nothing about this kind of stuff, but I'd bet that the microbes have (in electrical terms) a very high source impedance. I wonder if it's a good idea to have a capacitor connected directly to that source: you'd see the cap charge very slowly, and you might be affecting the microbes by connecting a voltage to them, where if the voltage produced by the microbe changed, it would be subject to current coming back from the cap. My instinct is to say that the microbes should be connected to an electrical system that's as un-invasive as possible. But then, I don't know anything about microbes.
     
Loading...