Post Go back to editing

CN0359: Float of signal from Instrumentation Amplifier

We have set up EVAL-CN0359 and have the following problem. 
We are measuring the voltage from J5 pins 2 or 3.
These feed AD8253.
We see the expected distorted square wave on the scope but it is floating up and down between the rails.
Page 18 of the datasheet for AD8253 possibly implies that the inputs are not stabilized by a bias current path to ground in the CN0359 layout.
See also fig. 6 of your article by Kitchin concerning "Common Problems When Designing Amplifier Circuits".

Any advice here? T
on Apr 22, 2019 4:37 PM in reply to fsonnichsen389

Hi Fritz,

Do you have a tracking number you can share for the probes?  Also have you tried this setup using precision resistors instead of the conductivity probe?  Its a good exercise to go through to ensure any drift that you are seeing is in fact due to the probe/solution input or related to the hardware/board configuration.

  • Hi all,

    Just wanted to leave a note saying Fritz has also sent me a probe to do some testing at our Wilmington site. 

  • Hi Fritz,

    We have some preliminary testing done with a TOPAC probe that we have available here and we have seen no anomalies on the output signal. We have tested it with varying resistor values in place of the probe and was also able to get correct values with the readings and the output signal looks as expected concluding that the circuit works as expected with the given test cases.

    We have received the probes yesterday and have been able to start the testing of the board with your probe. We are able to see and replicate your test case where the output of the in-amp swings to the negative rail.

    We are now trying to explore the possible instances why this would happen. Lia and our team are sharing our data to gain better traction in solving this issue.

    Please stay tuned on the advancements of our testing.

    Thank you,


  • Hello EZ,

    Just an update on this in case someone encounters this problem: further discussion on this topic has been done directly through email. If you are using the IST LFS1505 sensor and are encountering this drift issue, please connect a 10 µF capacitor in series with pin 1 of J5 and 1 µF capacitors in series with pins 2 and 3. In addition, place 1 MΩ resistors on pins 2 and 3 of J5 to ground. The below diagram shows the connections for this:

    This should stabilize the output signal of the U15 in-amp as shown below:

    Please note that we are still working with Fritz on this; an update may be posted at a later date. Thank you.

  • Hello EZ! Below is a more detailed explanation of the fix above:

    CN0359 setting: Excitation voltage: 0.3V, Excitation frequency: 100Hz~10kHz, Cell constant: 0.68/cm 

    1. I found the drift and DC bias only show on voltage channel of CN0359, current channel work well
    2. LFS1505 electrode have a coating layer look like a kind of heterogeneous catalysis layer similar to platinum black, I don’t know this forested surface is bring by manufacture or corroded by the solution. This coating more like come from manufacture because the electrode buried in resin have same color with outside. I believe there have a primary cell between oxide layer and metal body of electrode
    3. I measured electromotive force in different solution by add another platinum reference electrode. You could repeat this experiment without platinum electrode:
    • Use a disposable paper cup to take 100ml tap water to make sure the solution isolated with any potential
    • Immerse LFS1505 conductivity sensor into the water
    • Use a multimeter with input impedance > 10MΩ at DC voltage function, set multimeter to DC V function
    • Immerse  multimeter black lead pin into the water beside LFS1505, before immersion make sure lead pin is clean and no scratch
    • Connect multimeter red lead to each LFS1505 pin to check the voltage output
    • Use the same way to measure the current generated by LFS1505
    • Repeat above test with 1g salt added solution
    1. I got the following measure result:
    • Tap water:

    300mV~400mV for all LFS1505 electrode

    150nA~300nA for LFS1505 V1, V2 electrode

    400nA~800nA for LFS1505 I1, I2 electrode

    • Salt solution:

    300mV~400mV for all LFS1505 electrode

    500nA~1.5uA for LFS1505 V1, V2 electrode

    2uA~3uA for LFS1505 I1, I2 electrode

    1. Use AC coupling could eliminate the primary cell DC bias, only V1 and V2 electrode required, the DC bias remove filter introduced load to electrode, so the cell constant need to recalibrate.
    2. Fortunately the excitation current path could use DC coupling, no huge capacitor add to system.
    3. Only add a resistor load to V1 and V2 electrode will give help to reduce DC bias voltage, but DC bias voltage will change relate to concentration of solution, and sensor will aging fast.
    4. There have obviously drift in the V1 and V2 output in 5ms half cycle (100Hz excitation frequency) no flat top on the square wave, the primary cell charge and discharge jam excitation signal, you never can get right result in theory.
    5. From search:
    • four- electrode sensor typically used in concentrated solution measurement to remove stray resistance, they have larger cell constant
    • platinum black two-electrode sensor typically used in middle range conductivity solution or high chemical activity measurement, they have middle cell constant, platinum black could resistance to polarization
    • the electrode with coating must to be immerse into purified water to store, it will be damaged in dry state
    • bare platinum two-electrode sensor typically used in dilute solution measurement, they have smaller cell constant


    • Add AC coupling circuit into V1 and V2 channel, careful select R and C to minimize introduced error. Because the primary cell’s character relate to ion concentration, the excitation frequency and setup time should be set carefully.
    • Use exposed metal surface as V1, V2 electrode body, because there is no current through those electrodes and less corrosion in dilute solution measurement, use two-electrode configuration for concentrated solution.
  • Thank you for your extensive comments. Here are some points that apply here:

    (2) The vendor is not forthcoming here but I believe the electrodes are regular platinum with a physical "frosting" to increase surface area. They are shipped this way--it is not due to corrosion. I do not think there is an oxide layer with a new electrode. Operation under 0.3Vpp insures we are well below the electromotive potentials for any ions that could be involved in Redox (OH, H, K , Cl, Pt).  Examination after long term operation does not show physical differences so I expect oxidation is not happening at least to a notable degree.
    (3)Your experiment is interesting. My intuition is that if we are dealing with very small offsets, the typical noise and precision issues for a reference electrode would not reveal this due to inherit inaccuracy. More important I don't think a DC voltmeter will work here. What I see in the lab is a very quick development of a polarization layer/drift  that would upset an accurate measurement.

    • Regarding your suggestions:
      We have been adding 2 AC coupling caps with pulldowns to both of the inner (V1,V2) electrodes and 1 cap in series with the  outer excitation electrode. This has alleviated the rail to rail instability that is being addressed with this incident.  The drift over a long period of time is probably due to the electrode as you imply.
    • The use of an exposed metal (non frosted) surface for the inner electrodes makes sense and is a very good point. Use of a 2 electrode configuration however would probably lead to polarization error. Also there will be some fouling over long periods of time and this would lead to drift which typically is corrected in the 4 pole design.

    Thanks again