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.

ADS8361: Input Overvoltage conversion behavior increases previous conversion

Part Number: ADS8361
Other Parts Discussed in Thread: OPA2192

We tested our conversion circuit recently and have found increased conversion error if the input of the ADC is overvoltage and we had a RC filter on the analog power line.  My question is will the behavior change with increased or decreased temperature or part to part.

Our circuit is shown below:  Note the original circuit is with R137 at 10 ohms, a less sensitive circuit to the overvoltage is with 1 ohm.

Normally the voltage coming in on the Ch A and B is ±10V with a max up to ±12V.  But there are a couple of occasions where the input is very near -15V.   The input Ch A and Ch B come from multiplexers.   Part of the normal sequence is while converting the held input voltage on the ADC the next signal to sample is being selected.  We have found that if the next signal that is to be converted is -15V (will make output of U45.1 5.5 as an example) the current conversion has an increased error.  We have tested the circuit by removing the overlap of converting and changing the input voltage and found the error went away.  We have also changed the ADC reference to 2V, but the error continued.  Below are two charts, one is a chart that shows error of the conversion if the next signal is -15 and the other shows the error of the conversion if the next signal is 0V.  The data that is to be looked at is in red.

     

By reducing the AVDD series resistor to 5, 1, and then zero ohms we decrease the error jumping and the overall error.

    

There is a general worry that the 1 ohm solution might look different as temperature changes and / or as the AVDD supply changes from 4.7 to 5.3.  Are there any comments or suggestions for the design?

Thank you very much for your help.

-john

  • We would be happy to help with this question.

    Here are some questions about your design:

    1. What is the sample rate of your converter?

    2. What is the switching frequency of the ADC Input Mux?

    3. What is the switching frequency of the external Mux?

    4. Are there gain options for the input amplifiers U45A and U47A?  If so, what resistor values are you using?

    Thank you!

  • Bryan,
    Thanks for the help.
    Here are my responses:
    1. The ADC clock rate is 8.333 MHz (120 ns clock period). Conversions are initiated at a rate of 256 kHz (3.9 us interval). The ADC clock stops after each ADC conversion sequence completes.
    2. The ADC mux does not switch.
    3. The external mux is switched at 256 kHz (3.9 us interval)
    4. No gain options on U45a and U47a. gains set at -0.2V/V

    Thanks,
    -John
  • John:
    I have been looking at your circuit some more. All of your answers to the questions above make sense. The external mux is buffered, so it should not be causing any errors. The attenuation of the inverting amplifier stage looks like it protects the inputs from unsafe voltages, so that is okay too.
    You are using an external reference, which is buffered by the ADS8361. That should also be fine.
    Your observations about the analog supply are interesting. I would recommend a low value resistance for R137, because the converter will draw a high current during the conversion, but a low current during the acquisition phase. The average current draw is how we calculate the datasheet specification. With R137 set to 10 ohms, there may be a voltage drop across the resistor during the high-current phase, which might be significant enough to cause problems.
    Along with that, many times we see errors such as in your circuit being the result of the analog drive circuit. The choice of amplifiers, circuit topology, and the input RC values are all important. Because you are using AD8512BRZ’s in your design, it is difficult for me to simulate your circuit behavior, as the Spice models for those devices do not accurately model output impedance. Just by inspection, however, it looks to me like resistors R138, R140, R143, and R144 are very high values. (One common misunderstanding is that these components form an anti-aliasing filter, which is not really the case. The purpose of the capacitor with this type of ADC is to provide charge to the sampling circuit very quickly.)
    With that being said, if you were to replace the amplifiers with OPA2192ID's, I could more accurately simulate your circuit behavior. OPA2192ID’s are pin-compatible with your design, and compared to the AD8512BRZ, are lower cost, have better offset specs, and better offset drift, and higher bandwidth. Using the OPA2192 in your design, you would see better performance by changing R138, R140, R143, and R144 to 38.3 ohms, and C227 and C230 to 390pF. This should allow you to sample as fast as about 460kSPS without errors.
    If you are reluctant to change Op Amps, that is understandable; you might try some similar values with your AD8512BRZ design. With the correct choice of values for the input RC, you will be able to minimize your errors, although with this circuit topology you may not be able to achieve max sample rate with this ADC.
    Let me know if this helps!