AN-1798 LMP7721 Potentiostatic Circuit

Hello NGinEar,

This is the "generic" circuit that biases the sensors in the "zero bias" mode.

The circuit is based on the recommended circuit contained in Alphasense Appnote AN105-03.

The Alphasense appnote goes into more detail about biasing the sensors and circuit operation. Basically, in the first stage, the sensor is inside a loop that maintains the counter electrode at the reference electrode potential. This stage requires a very low bias current amp as the reference electrode has a very high impedance output. The second stage is a basic current to voltage converter that converts the working electrode current into a voltage.

 http://www.alphasense.com/industrial-sensors/pdf/AAN_105-03.pdf

You would apply the reference voltage into the non-inverting inputs ("GND" end of R3 and R5 - which you really do not need R3 and R5 since this is a fA CMOS device and bias current cancellation is not needed).

Also see the LMP91000 - a compete gas sensor front end that contains this basic circuit.. An eval board is available, too.

http://www.ti.com/product/lmp91000 

http://www.ti.com/tool/lmp91000evm

Regards,

Hello Paul,

Thank you for your detailed reply.  So what you are saying is that the circuit does not generate a bias reference voltage as this is not required with the intended gas sensors?

I will be using the circuit with electrodes that require a bias reference voltage so I should apply the reference voltage in place of R3? 

What is the purpose of R5 in the TIA?  Why do they show resistors connecting the non-inverting inputs to ground if they are not required?

The circuit is powered by a dual voltage rail of 2.5V.  At present I am seeing approximately -1.70V on the Counter electrode with respect to the Reference and Working electrodes?  Why would I be measuring -1.70V in zero bias configuration?

The 'reference" voltage is 0V in this case. The circuit is designed for "split" (±2.5V) supplies.

The sensor does not generate a "bias voltage". Don't confuse reference voltage with bias voltage....they can be different.

The sensor "bias" voltage would be an externally set voltage difference between the reference electrode and the working electrode. Some sensors work better with a slight bias (~200-500mV) between RE and WE.

In reading the Alphasense appnote, when they mention "ground" potential, they are referring to the set potential on the reference electrode. It is 0V in this case. In a non-split supply, the "reference" electrode potential would be set above ground at the first stage reference voltage. All the voltages on the other terminals would be referred to the reference terminal.

The circuit as shown is a "zero bias" circuit, where RE and WE are expected to be at equal potentials since both non-inverting inputs (R3/R5) are grounded. Because it is a "zero bias" circuit, there is only one "reference" voltage - which happens to be ground (±0V).

If you needed to apply a bias voltage across the sensor, then you would need two different reference voltages at R3 and R5. In this case, you would then have two "reference" voltages, that would be separated by the required sensor "bias" voltage. Remember that the amplifiers will servo the attached inverting input node to the applied voltage on the non-inverting input.

The -1.7V is the voltage required to maintain the Reference electrode at the potential set by the non-inverting input of the amplifier. The voltage on the counter electrode will vary and change to whatever is needed to maintain the reference electrode at the set reference bias voltage. As the appnote says, the actual voltage on CE is not important as the current out of the working electrode is what we are really interested in.

As for R3/R5, in the original reference design, they were using bipolar op-amps with nA bias currents - which generally need the "current cancelling" resistors. By using fA bias current amplifiers, this allows better low-end sensitivity and resolution since the nA's of bias currents do not "muddy" up the low-end measurement. The resistors do not hurt anything and can be left in, but they do introduce extra un-needed broadband thermal noise.

Regards,

Paul,

Thank you again for your in-depth reply.  You have clarified my misunderstanding between bias and reference voltages.  It is indeed a reference voltage that I require and not a bias voltage.

The Alphasense application note indicates that a bias voltage should be applied to the non-inverting input of the control Op-Amp via two 10K Ohm resistors.  If a reference voltage is applied to this input instead, wouldn't the Counter electrode potential be at the reference since the Counter electrode is the output of the Op-Amp?

My understanding is that the reference voltage should be present at the Reference electrode.

Just to add, the circuit is drawing approximately 50mA which seems rather high for two Op-Amps?  I can't see how in the configuration as per the app. note, the resistors around the Op-Amps would contribute significantly to supply loading with no electrodes attached?

The supply current per Op-Amp in the datasheet is listed as 1.75mA maximum at 2.5V.

I am completely puzzled - hopefully someone can answer my queries from this and the previous post.

Thanks.

What sensor are you using?

Is anything getting hot? Is it oscillating? Are you using split supplies or single supply? Are you trying to split the references?

Are you sure the resistor values are what you think they are? Are the amplifier outputs railed? Are you using very low value resistors to divide the voltages?

To get 50mA, there is either an oscillation, or a low-Z path from the servo loop output to ground (through the sensor), or through the input clamping diodes to the supply.

Please post a full schematic of what you are doing and a voltage map (measured voltages on the various nodes). A photo of the setup is also helpful.

This is not an easy circuit. Though it looks simple, implementation can be difficult, particularly the Reference Electrode path.

Regards,

The circuit is powered by a 12V regulated supply.  The 2.5V precision reference zener is fed from the 12V supply via a 330 Ohm series resistor (not shown).  Previously with a lower value resistor the current consumption was 50mA.  Now it has dropped to 29mA with the 330 Ohm resistor.  It is this resistor that is getting hot.  Even though I have swapped it for a 0.5W one?  There is 9.6V dropped across the resistor so it should be dissipating 0.28W?

The negative rail is generated using a MAX1044 IC.  I have checked all the voltages and the component values.  There is no oscillation of either Op-Amp.  I have verified this with an oscilloscope.

The p-type JFET had to be removed as it was not switching off fully with +2.5V on the Gate.  It isn't really required for this system anyway.

The electrodes used are made of carbon strips.

Unconnected, the voltages on the terminals are as follows: Working electrode (0.00V), Reference electrode (0.05V), Counter electrode (2.48V).  Once the electrodes are connected and placed in solution, the output of the TIA reads a negative voltage.  I would have expected the output of the TIA to be a positive voltage?

One thing which I found odd was that the power indicator LEDs are dim when the circuit is operating open-loop with no electrodes attached.  As soon as the electrodes are attached and placed in solution, the LEDs are lit to normal brightness?

Hi EnGinEar,

What is your output connected to? Is the next stage prepared for a negative output? Is it a 0-2.5V input ADC??

What is the voltages on the outputs of each amp when the electrodes are submerged? Could the strips be generating large voltages?

If your resistor is dissipating 0.28W, you should at least have a 1/2 watt resistor as an absolute minimum. The minimum is to have a wattage rating at least 2-3x the expected max wattage. The bigger the better for reliability.

How are you measuring the voltages on the test points? Touching a DMM on the REF-ELECT terminal will not work since it is a high-impedance node.

Remember that the TIA is inverting by nature...A positive current into the input will cause the output to go negative (as the output pulls current out of the node through the feedback resistor to maintain zero on the node).

If the TIA is railing, then it may be too sensitive. Decrease the value of R11. Right now you are at 4.7V/mA

Have you measured/calculated the current from the strips? Is it in the mA's?

If you are not having output drive issues, then I still think you may be having oscillations - particularly around U3. IF you are using a digital scope, be sure to sweep from the fastest time/Div to the slowest, as digital scopes will "hide" high frequency (~50-100KHz) oscillations at their low sweep settings due to aliasing. Also make sure averaging is off ,and it is best to use peak detect display setting.

Regards,

Hi Paul,

Thanks for your advice so far - I'd really like to be able to get the circuit working!

Here is some further information in response to your last post:

The Output is currently connected to a DAQ device with upper and lower measurement capability of ±2.5V.

The present 330 Ohm resistor is ½W.  My calculations (see previous post) indicate that the resistor should only be dissipating 0.28W?  So it shouldn't be getting hot.

Here are the voltages measured using an oscilloscope with reference to ground:

No electrodes connected

DC Coupled

Counter electrode (2.5V), Reference electrode (0V), Working electrode (0V), Output of TIA (0.03V pk-pk 278kHz)

AC Coupled

Counter electrode (0.1V pk-pk 2500Hz), Working electrode (0.025V pk-pk 555Hz), Reference electrode (0.11V pk-pk 2631Hz), Output of TIA (0.030V pk-pk 278Hz)

Electrodes connected and submerged

DC Coupled

Counter electrode (1.5V), Reference electrode (0.4V), Working electrode (0V), Output of TIA (-0.7V)

AC Coupled

Counter electrode (0.02V pk-pk 2777Hz), Working electrode (25mV pk-pk 833Hz, 278kHz), Output of TIA (26mV pk-pk) 278kHz

Most of the waveforms look like low level noise and are sawtooth in shape.  The MAX1044 oscillator frequency is 2631Hz.

I have previously tested similar electrodes using a 2-electrode potentiostat that I built using an LM358 (Ref. voltage) and LF356 (TIA).  The voltage on the reference electrode was -0.5V and it produced a positive response on the inverting TIA output.  The current on the Reference electrode was in the region of µA.

I would like the output voltage from the TIA on this circuit to also be positive.  The potential divider on the non-inverting input of U3 (4k7 & 1k2) produces +0.5V.  This would presumably give +0.5V on the CE too when the loop is closed with the electrodes connected and submerged?  Hence, the output of the TIA is negative?