Logarithmic Picoammeter

[staticshock]

For my Senior Design course. my team is required to design and build a Logarithmic Picoammeter. So far, we've decided on 3 op amp designs: shunt resistance, feedback, and logarithmic diode feedback. Which one is best suited for our project? Also, we need to purchase a low level current source to be used for testing. Which device would you recommend? Our budget is capped at $3000.

[tggzzz]

For my Senior Design course. my team is required to design and build a Logarithmic Picoammeter. So far, we've decided on 3 op amp designs: shunt resistance, feedback, and logarithmic diode feedback. Which one is best suited for our project?

That is a principal part of your coursework: stating why you chose the approach and its advantages and disadvantages relative to the specification and other approaches.

It would be improper for us to answer that - and you would be unwise to rely on our answers.

Also, we need to purchase a low level current source to be used for testing. Which device would you recommend? Our budget is capped at $3000.

Now that is a perfectly reasonable question. You may find you have to make one

Apart from that, have a look at the Tektronix "Low level measurements handbook".

[David Hess]

Does it need to support positive and negative input currents?

What kind of precision do you need?

With a low input bias current operational amplifier, say something like an LMC6001 or selected LMC6081, measurements down to 1 picoamp is possible using a simple transistor based log amplifier which converts an input current into a voltage.  I would hold the logging transistor at a constant temperature instead of using a 3300ppm/C resistor but that is a matter of taste.

At picoamp current levels, shielding will be needed to prevent AC rectification from various sources of interference.

[danadak]

This might help -


http://download.tek.com/document/LowLevelHandbook_7Ed.pdf


http://www.electronicdesign.com/test-amp-measurement/whats-all-femtoampere-stuff-anyhow






Regards, Dana.

[danadak]

Also there is a Metrology forum here that may be useful -


https://www.eevblog.com/forum/metrology/


Regards, Dana.

[TimFox]

The old Keithley picoampere current sources are actually a calibrated voltage source followed by a switchable high-value resistor, of the same resistor construction used in their electrometers.  These work well into a low-impedance load, such as the virtual ground of an I-V amplifier.  The calibration routine requires shorting the input protection resistor in the electrometer, but I find it easier to just tweak the voltage to include that resistor in series with the source's internal output resistor.  With proper shielding, etc., it is straightforward to do this with a good voltage source (more than 1 V, preferably more than 10 V) and high-value resistors.

[staticshock]

Thanks for all the help. Does anyone have the schematics for the Keithley 26000 or any logarithmic picoammeter?

[David Hess]

Thanks for all the help. Does anyone have the schematics for the Keithley 26000 or any logarithmic picoammeter?

There is a very simplified one in Bob Pease's book Troubleshooting Analog Circuits.  He just used an uncompensated transistor based log amplifier to drive a meter movement.

[staticshock]

Alright so we did decide on the logarithmic way. I've attached the circuit schematic and its output. Basically we have three op-amps (LMC6001). Input current comes into the left op-amp , a reference current comes into the right op-amp, and the output comes out of the bottom op-amp. David Hess you suggested on holding the transistor at a constant temperature instead of using the 3300 ppm PTC resistors, which we included in our design and they are the 1K resistors in the bottom op-amp. We aren't exactly sure of how to do that. However, we are having trouble in finding those resistors. Can someone point us into the right direction. In addition, we are losing some accuracy in the low range. Other than calibration, is there anything we can do to improve it?

[Kleinstein]

For those 3300 ppm/K resistors one can use PT1000 (around 3800 ppm/K) in combination with a fixed resistor. Some other relatively linear PTCs work too.

[David Hess]

In addition, we are losing some accuracy in the low range. Other than calibration, is there anything we can do to improve it?

At the picoamp level:

1. Leakage due to surface contamination is a serious problem.  Special substrates, air wiring, and guard rings on the printed circuit board are often used.  DIP and through hole parts have an advantage due to greater lead separation.  Air wiring sensitive nodes works well because air is an excellent insulator but Teflon insulated terminals are sometimes used.  My preference is to pull the inverting input off of the board and air wire that node.

Test for surface contamination by blowing onto the circuit through a straw.  The water vapor in your breath will alter the conductivity of ionic surface contaminates.

2. Accuracy falls off at high currents due to resistance and at low currents due to leakage.  I did not find much discussion online about compensating for these errors in the transistor.

[staticshock]

Update on the project:
We've built the circuit and the output is pretty consistent with the simulated results until 10^-10. There will be some calibration done there. However, I'd like to ask if there are ways we can increase the response time of the output especially for the very low currents. It really does take a long time for the output to stabilize after 10^-10.