This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

OPA197: Input impedance vs output voltage deviation

Part Number: OPA197

Hi all,
My customer has some questions about OPA197.

They evaluated OPA197 based circuit as below;

When the value of R1 is 22k and when it is 160k, the result which is different from the expected output comes out.






Test conditions

Ambient temperature: 25 ℃

VS(Supply voltage ( V + -V- )) :10.8 V / 24 V / 30 V (single supply)

R1 Resistance value : 22k / 33k / 47k / 62k / 82k / 100k / 160kΩ

Input voltage: 0.20 V / 0.65 V / 1.10 V / 1.55 V / 2.00 V / 2.45 V / 2.90 V / 3.35 V / 3.80 V / 4.25 V / 4.7 V

Expected output value for each applied voltage:

                       1.0 V / 1.4 V / 1.8 V / 2.2 V / 2.6 V / 3.0 V / 3.4 v / 3.8 V / 4.2 V / 4.6 V / 5.0 V

Formula:     Vout ≒0.8889*Vin+0.8222 [V]

When R1 = 22k and the input is 0.2V, the expected output is as low as 20mV with respect to 1V, and the input becomes 0.65V or more almost the expected output value.
When R1 is 100k, are close to expected values from 0.2V to 4.7V.
Also, when R1 is over 160 k, the output value will vary.

Questions;

1. What are the possible causes?
2. What is the appropriate R1 value? 

3. We believe that OPA197 is designed to be suitable for power saving operation as an FET structure based on its characteristics.
    We believe that it is unnecessary to make the equivalent of making the input resistance of two differentials of the Bi-polar circuit equivalent.
    However, this malfunction occurred with the equivalent resistance value trial calculated this time.

    As shown in the data sheet, we are incorporating proprietary technologies such as input protection, suppression of leakage current, EMI filter,
    phase reversal protection function, temperature protection function ... ....
    Is there a possibility that the above phenomenon will occur due to the influence of such unique function?

As my TINA based Simulation result, there is no problem on DC results, but Phase margin is less than 30-deg.
So I suspected oscillation, but the output waveform was normal on the actual circuit.

Best regards,
Toshi
 

opa197_original.TSC 

  • Hi Toshi,

    hmm, strange... Is the 5V absolutely stable and free of noise?

    I would not connect the voltmeter directly to the output of OPA197. Voltmeter can show considerable input capacitances which can make the OPA197 become unstable. I would mount a low ohmic resistor of 100R, or so, directly at the output of OPA197 and connect the voltmeter to the "cold" end. See also table 3 of datasheet.

    Kai
  • Hi Toshi,

    Since the OPA197 has input bias current in the range of a few picoAmps, you should not need to compensate for the bias current.

    As Kai has mentioned, when connecting the external meter, using an isolation resistor may help isolate the capacitive load and help with stability as shown on table 3 of the datasheet.

    A couple of questions:
    Is your V+ voltage supply well bypassed and the supply voltage stable/noise free?
    What generates the +5V supply, is it stable/noise free?
    What type of meter is used to measure the output signal?

    Thanks and Regards,
    Luis
  • Toshi

    We have not heard back from you so we will close this thread. Please reply if further discussion is needed.

    Thanks
    Dennis
  • Hi, Dennis-san,

    Sorry for my late response.

    My customer is trying to reproduce the same phenomenon and is trying improving it with an isolate resistance, but it was impossible to reproduce it because the measuring instruments etc are not exactly the same and it took time to reply .

    However. regarding the output waveform, observation was made using a different voltmeter and an oscilloscope in parallel, but no oscillation waveform was observed even in the state without insulation resistance.

    When 1000 pF capacitor is connected to the output end for testing, the oscillation waveform is confirmed.

    Anyway, I would like to answer your questions and to update the circuit information.

    Q1. Is your V+ voltage supply well bypassed and the supply voltage stable/noise free?

    A1. Yes, it is. The power supply is Keysight E36312A.

    Q2. What generates the +5V supply, is it stable/noise free?

    A2.  Generation of 5 V uses the linear regulator TOREX XC 6702. The output current 300 mA and ripple rejection ratio is 65 dB @ 1 kHz.

           Regarding the 5 V power supply, there are machine equipments such as a compressor and a pressure generator around, and it is in an    environment where noise is easy to ride.

           Noises with amplitudes on the order of a few mV will get on the waveform if filtering is not done.

    Q3. What type of meter is used to measure the output signal?

    A3.  Multimeter: KEITHLEY 2100 or Keysight 34970 A + 34901 A

    OPA197 is used for sensor output, and its output stage has a protection circuit, so its circuit diagram is attached as below.
    The output stage capacitor is the load capacitance to check the oscillation waveform.

     O

    When I simulated using TINA, peaking was seen in the AC characteristics when R1 = 0.
    I thought that the input bias current is not affected so much because it is on the order of pA, but why did the characteristics change?

    Best regards,
    Toshi

  • HI Toshi,

    From the schematic point of view, I can not see anything obvious as to why the issue is occurring.

    One possibility, is that high frequency noise or ripple  is being injected to the +24V supply or the +5V supply, or at the input signal point producing noisy/fluctuating measurements.  Is it possible to obtain oscilloscope plots of the output of the amplifier, the +24V Supply and the+5V supply and input signal when the issue occurs?  Please place the oscilloscope probes right at the supply pins, using a very short connection to the OPA GND.  Attached is a pdf file with oscilloscope probe placement recommendation.  Please refer to slide 8-10 of the pdf file, where a short scope lead is used to connect the probe to the local circuit ground.

    Noise Free Scope Measurements.pdf

    A second recommendation is to ensure that the PCB board has been properly cleaned for Flux contamination... Residual flux left on a PCB assembly can manifest itself as unwanted parasitic capacitances and resistances.  High impedance/low current circuits are extremely sensitive to flux contamination. Normal hand washing or cleaning of PCB assemblies with isopropyl alcohol or deionized water are usually NOT sufficient to remove all flux residue.  Flux gets trapped underneath ICs where it can form undesired, parasitic capacitances and resistances. The proper PCB cleaning method will depend upon the flux type you use.   Attached are some additional materials regarding flux contamination you may find useful.

    clean-pcb-assemblies.pdf

    Thank you and Best Regards,

    Luis

  • Hi Toshi,

    as Luis I think that high frequency noise and interence could be the cause of trouble. If a sensor is mounted to the input I miss the low pass filtering. It's no good idea to give the sensor signal totally unfiltered to the +input of OPA197. So, add a low pass filter to the input. You could add a cap to signal ground behind the 1k resistor, for instance.

    Then, it's not wise to have high impedance resistors at the +input and in the feedback path of OPAmp. Make these resistors as small as possible and as high as necessary. Is there any reason to put a 86.6k resistor in front of the +input of OPA197 at all? Shall this be a input bias current cancelation scheme? But then you overlook the 10k pot in the feedback path.

    Then, it's not wise to have no phase lead capacitance in the feedback path of OPAmp (across the 10k pot). Even only 22p...33p can help a lot. In all my OPAmp circuits I have such a phase lead capacitance.

    Last but not least, it's not wise to have no isolation resistor at the output of OPA197. Even only 10...47R can help a lot. In all my OPAmp circuits I have such an islation resistor.

    Keep in mind, that HF intereference at the input and output pins of an OPAmp can show very strange effects, when the HF interference is demouldated across internal pn-junctions that are not fast enough to handle the HF interference properly. So, it's of essential importance to do everything to prevent HF interference from entering the OPAmp. -> Use extensive filtering when the signal comes from a sensor and the sensor is mounted in a rough industrial environment.

    Kai
  • Also, the circuit will have stability issue because of the high value of the gain resistors used. The total input capacitance, 1.6pF and 6.4pF, will interact with the parallel combination of the gain resistors creating a zero in the close-loop transfer function. If the zero occurs within the effective bandwidth of the configuration, it needs to be cancelled by adding appropriate value feedback cap or it will diminish the phase margin leading to oscillation. Since the zero will occur at f=1/(2*Pi*86.6k||10k*8pF) = 2.22MHz, it is well within the bandwidth of the configuration and thus you must add a feedback cap across 10k resistor with a value of C=10k||86.6k*8pF/10k =~7.2pF.

  • Hi Toshi,

    I'm curious. Could you solve the issue?

    Kai
  • Kai-san, 

    Thanks for your attention. 

    I have provided the above answers and my analysis with TINA based simulation to my customer, 
    My customer has responded that he will reexamine the circuit based on these information.
    However the situation is that he can not respond immediately, so we closed this for now.

    If there are any additional questions from my customer, I would like to ask for your support.

    Thanks your kind support.

    Best regards,
    Toshi