Miniaturization Help: Measuring the current response of non-ohmic circuit element with expected response in 100μA ~ 100pA range

[FatSeagull]

Hello. I am a graduate student with background in Biology, and I would like to request a sanity check.

Context:

I am currently working with an organic polymer sample which changes its current response in response to changes in its environment.
Tentative experimental data shows that the polymer's current response seems to be non-ohmic in nature.
Currently, I am using this expensive piece of equipment from Zurich Instruments to measure the current response and the phase response:

MFIA 500 kHz / 5 MHz Impedance Analyzer

The Zurich Instruments MFIA is a digital impedance analyzer and precision LCR meter that sets the new standard for impedance measurements in the frequency range

www.zhinst.com

However, due to its cost, We only have one of these machines in the lab. This has become a serious bottleneck in the pace data collection.
Furthermore, as one of the applications of my polymer is use in sensors, there is a need for miniaturizing the setup that is required to collect the current response data.

When 1V 200Hz current is ran through the polymer sample, its current response ranges from 100μA to 100pA, depending on the conditions.
This leads me to believe that the polymer behaves mostly as a resistor, with limited capacitance (Since there is a phase response of 6~20 degrees, depending on the surrounding physical conditions).

I wanted to see if miniaturizing the measurement setup was possible. However, the fact that the lab went out of its way to purchase this $10K machine instead of telling someone to DIY it is giving me some pause. After all, if a DIY solution was so easy, then chances are, someone would have done it by now, right?

Assuming that optimizing for performance and size takes precedence over cost:

Questions:

1) Would you recommend/recommend against using something like an Arduino/RPi for collecting the readout data? If not, what other portable, small solutions can you recommend? My biggest concern is that Arduinos usually output DC signals, but the polymer sample exhibits response only with AC stimuli.
2) Would a Transimpedance Amplifier (TIA) Circuit be a good starting point for the actual 'measurement' circuit?
3) Given the very small possible current range, Signal to Noise Ratio is a big concern for me. Would a ultra low bias current op-Amp like LMP7721 be the best for the job? Or are there better options out there?
4) Given the wide range of expected current values, I am unsure if a single Op-Amp is able to handle my requirements, as the bias current range for a single Op-Amp may not be able to handle the range of expected inputs. Is this a valid concern? How would you best address this?

Thank you very much for your help.

 

[danadakk]

Low level measurement techniques might be helpful :

Keithley Low Level Measurements Handbook - 7th Edition

Learn everything there is to know about low level measurements in this in-depth handbook authored by the electrical engineering experts at Keithley and Tektronix.

www.tek.com

Possible signal conditioning :

https://www.ti.com/lit/an/sboa169b/sboa169b.pdf

1) Consider a PSOC 5LP type of part, with 20 bit A/D and the following onboard resources (multiple copies in most cases) :

Has USBUART or SDCARD or display or other COM to get data.

Board to use CY8CKIT-059 unless you need a ton of I/O. its ~$15

IDE (PSOC Creator) and compiler free.

2) TIA appropriate but INA (instrumentation opamps) series might be better approach.

3) Best to do a sim of noise analysis for overall solution. And yes, lower input noise contributes
less to overall solution end to end error analysis.

4) You start with spec goals, eg. resolution, accuracy, INL and DNL and use that to screen for
parts.

Take a look at this - https://www.eevblog.com/projects/ucurrent/

Regards, Dana.

 

[FatSeagull]

Hello Dana, Thank you for the answer.

This is a lot of information to digest, but to be clear:

1) In the link that you sent me under 'possible signal conditioning', then in my case, the 'LOAD' would be the polymer, and the low-side sensing element would be some resistor of appropriate value that is in series with the polymer to allow the op-amp to measure the voltage drop, right?

2) In the figure 3 of the same file, it seems like the output is voltage modulated (?) as in, the output is equal to the voltage difference between the inverting input and the non-inverting input, multiplied by some gain. If so, am I supposed to measure the current by using Ohm's law (assuming the gain remains constant (?)

3) I did some cursory research and I'm unsure what exactly the difference is between a Transimpedance Amplifier vs the INA. Why is the latter recommended over the former in this case?

Thank you.

 

[D.A.(Tony)Stewart]

TIA is any Op Amp using Vo to Vin- with negative feedback to convert input current to output voltage. CMOS types tend to be rail to rail in/out so this is your go-to for precision current sensing, while an OTA IC is more about dynamic signal manipulation like measuring phase shift of a frequency from 500 to 5kHz or 5kHz to 50 kHz in a PLL recently asked on this forum. It could be modified to measure phase of current or voltage and provide a 1 decade sweep or 2 then something else to measure log of amplitude.,


If you could define the range of excitation , V, I , f then I am sure a custom solution can be designed or find something that exists in your low current range.

Low cost meters called RLC meters usually drive constant current at 3 fixed frequencies to measure precise R or L or C with ESR and R || shunt characteristics. But for servo circuits people want to see the Bode' amplitude and phase response for voltage or current over several f decades. But if you want data collection at a fixed frequency RLC meters are the way to go with a PC or uC board kit but you don't get to change it from the fixed options like 120 Hz, 1kHz, 100 kHz or 1MHz
etc.

There are also simple designs that can generate constant current (Howland CC cct) or like RLC meters for DC or AC to very low values. then measuring the voltage gives impedance and mixing the two (V*I) gives phase as a duty cycle or filtered as a DC voltage.

Dana's solution is elegant, powerful low cost but has a steep learning curve. You might be able to collaborate with sufficient interest. I don't have his unique expertise or patience.

> 1V 200Hz, 100μA to 100pA, 6 to 20 deg

These two plots are related to linear impedance calculations.

You might get better resolution at 1 kHz with low capacitance readings and high resistance. If the frequency slope is 1st order or 90 degmax then the ideal measurement is to find the frequency that matches a 45 deg. phase shift and measure the current. This solves both RC immediately and could be done with a VCO PLL and auto scaled reference capacitors with analog switches.

So if you can make a list of specs like a meter with range specs that may be feasible with ideal parameters of measurement ( impedance ,degrees , conductance ,resistance capacitance half-power point (R=Xc) etc. Series R, Parallel R and error tolerance.

Then finally if you could buy enough for the whole dept. how many and not to exceed $? Protoypes can be expensive.

 

[D.A.(Tony)Stewart]

By raising excitation from 200 Hz to 200 Khz, if relevant, you can lower reactive impedance 3 decades from 20Gig to 20Meg and then use $100 solutions like the EVAL-AD922EBZ

 

[danadakk]

Hello Dana, Thank you for the answer.

This is a lot of information to digest, but to be clear:

1) In the link that you sent me under 'possible signal conditioning', then in my case, the 'LOAD' would be the polymer, and the low-side sensing element would be some resistor of appropriate value that is in series with the polymer to allow the op-amp to measure the voltage drop, right?

2) In the figure 3 of the same file, it seems like the output is voltage modulated (?) as in, the output is equal to the voltage difference between the inverting input and the non-inverting input, multiplied by some gain. If so, am I supposed to measure the current by using Ohm's law (assuming the gain remains constant (?)

3) I did some cursory research and I'm unsure what exactly the difference is between a Transimpedance Amplifier vs the INA. Why is the latter recommended over the former in this case?

Thank you.

Tools - https://www.analog.com/en/resources/interactive-design-tools/adi-diffampcalc.html

1) Correct.
2) The Vref is essentially a bias, to bias output range. Generally not used as AM modulation but could be.
3) Essentially think of INA as nothing more than an V amp, so Ishunt = Vout/( G x Rshunt) for Vref = 0
An INA is at most basic level a V amplifier, a differential one at that. Diff allows rejecting of CM signals or CM bias.

A TIA is typically single ended supply generally for low level currents

Tonys comment :

Dana's solution is elegant, powerful low cost but has a steep learning curve.

Actually I have worked with many different IDEs and found years ago Freescales Processer expert,
a precursor to PSOC Creator, both excellent in terms of learning curve speed and ease of use. Creator
more advanced and capable.

Creator think of a lab workbench with a pile of discrete parts on it. In PSOC case they are all inside
part. Now one just uses drag and drop the resource onto Creator canvass, uses wiring tool to connect
it to other internal blocks or out to pins, right click on resource and a lib of f() calls to manipulate them.
E.g. no need to write drivers (rare). So code generation very rapid. Add to that a bunch of wizards,
like DMA, DSP Filter, State Machine, even an ohmmeter to check out internal wiring if you are working on
the 20bit precision stuff. You mention Arduino, unquestionably one of the worst IDEs on the planet.
I use it often, only when I have to. Open source stuff compounded by machinations in lib development
as well make it tedious to getting things done. So in closing think of PSOC as a pile of chips on a single
die, and a full GUI design interface. Users, because of its available fabric, have created additional libs of
stuff like DDS, 74HC parts, CPLD, Cordic.....you can add to the standard IDE lib. You can do codeless designs
with this part as well (for another day).....

Note, on your design the PSOC A/D going down to uV resolution would mitigate need for INA I think.

And it has USBUART for easy interface to PC, or use any of the other COM capabilities, or
even SDCARD to save data.....also 100's of projects to cut and paste from to speed up
work.