I know words like "ultra" aren't helpful, but anyway I need to use a CO2 sensor that requires "... the impedance of amplifier should be within 100—1000GΩ，Its testing current should be control below 1pA." That's "ultra" in my book!

Here's the datasheet for the sensor:

http://www.parallax.com/Store/Senso...ortField/0/Level/a/ProductID/616/Default.aspx

I believe all I need to do is convert a 30-50mV signal at that high impedance, to a "low" impedance, something I could measure with a DMM or other data acquisition scheme. Voltage amplification would be a plus.

Any alternative ideas or advice highly appreciated. The long-run goal is to make a rapid-response feedback to blood oxygenation level, as measured by exhaled CO2.

Here's a similar device:
http://www.alphasense.com/alphasense_sensors/co2_carbon_dioxide_sensors.html

Its datasheet refers to a signal conditioning box. That's what I need to do.

 

AD8641

1 pico-Amp input current, affordable.

 

There are quite a few CMOS op amps available with input current on the order of a few to tens of femptoamps, but if you also want low offset voltage, they get pretty pricey.

 

So, the parameter I'm looking for is a low input bias current?

Then something like this would also be an OK choice (assuming it could be found)?

http://www.national.com/pf/LM/LMP2234.html#Overview

EDIT: It looks like Jameco offers these. Suitable? (It even says "ULTRA low input bias current".)
http://www.jameco.com/webapp/wcs/st...1994617&productId=1994617&keyCode=WSF&CID=GMC

http://www.national.com/ds/LM/LMC660.pdf

 

So, the parameter I'm looking for is a low input bias current?

Then something like this would also be an OK choice (assuming it could be found)?

http://www.national.com/pf/LM/LMP2234.html#Overview

$5 @ DigiKey

Though the slew rate is very low, for your application, it may be fine. 10uV offset isn't too bad.

You'll go into the double digits looking for something better, I'd guess.

 

So, the parameter I'm looking for is a low input bias current?

Then something like this would also be an OK choice (assuming it could be found)?

http://www.national.com/pf/LM/LMP2234.html#Overview

National has a pretty good variety of the LMP series precision op amps. Have you checked prices, though?

 

A word of caution when working with ultra-low currents in the femto-amps. - Keep in mind that the circuit board material may have more leakage than the component. Also, extra caution is needed to remove flux and other contaminants that will can conduct small currents. You may want to use Teflon rather than FR4 to hold the critical components.

 

A word of caution when working with ultra-low currents in the femto-amps...

Noted, good points. I may end up connecting the op-amp directly to the pins of the device's socket. I'd love to get a look at the commercial box that I'm thinking of replicating.

 

National has a pretty good variety of the LMP series precision op amps. Have you checked prices, though?

I'm narrowing in on the LMC660, $1.98 at DigiKey

If anyone sees a problem with that choice, speak now!

 

I'm narrowing in on the LMC660, $1.98 at DigiKey

If anyone sees a problem with that choice, speak now!

Input offset voltage is as high as 6mV. With a 30mV input, you could have a 20% error. The offset voltage is basically independent of the input voltage, so with 60mV input, you would still have 6mV maximum offset.
If you add gain with a feedback divider, the offset voltage gets amplified along with the signal.
Of course, your offset will probably be less. Typical is only 1mV, but 6mV is the guaranteed max.

 

I'm narrowing in on the LMC660, $1.98 at DigiKey

If anyone sees a problem with that choice, speak now!

Look for one rated for medical use, in the event you go mass production.

As Ron H stated, keep the offset voltage under 1mV, preferably <100μV. When dealing with somebody's life, the little things become important, such as board type mentioned above. I'd go so far as using a 4 layer PCB for power and ground planes due to the sensitivity of this unit. Grounded metal enclosure, etc.

 

Check the OPA1641. 10^13th impedance. Digi-Key carries them, along with 1% resistors up to 5000 meg.

 

Be careful to observe anti-static precautions especially rigorously when handling devices with such low input current specs.

Their input impedances are of course very big, and some may have little or no input protection, to minimise leakage current.

Finally, very slight damage may result in leakage which would be insignificant in other applications, but critical for this work.

 

Check the OPA1641. 10^13th impedance. Digi-Key carries them, along with 1% resistors up to 5000 meg.

20pA max bias current.
3.5mV max offset voltage.

 

I've just discovered that the sensor supplier also supplies a buffer board ...
http://www.parallax.com/Store/Senso...ortField/0/Level/a/ProductID/598/Default.aspx

...with schematic!
http://www.parallax.com/Portals/0/Downloads/docs/prod/sens/27929-CO2GasSensorModuleSchematic.pdf

They chose to use the LMC6035 op-amp, which is SMD but otherwise very similar to the LMC660 I was considering.