ADS127L21: Functional Safety AFE

Part Number: ADS127L21
Other Parts Discussed in Thread: OPA2192, OPA2325

Tool/software:

Hi TI Team

I have two questions.

1. We plan to use the ADC in a safety PLC certified according to the IEC 61508 standard. We need to achieve SIL2 with a proof test interval of 20 years. Is this possible with just using a single ADS127L21 and the build in safety features (CRCs, STATUS1 & STATUS2) and by checking the DRDY signal frequency? Or is mandatory to have some sort of control method (second ADC, reference measurement)?

2. Is the OPA2192 a suitable driver for the inputs?

Regards Theo

  • Hallo Theo,

    thanks a lot for your interest in our ADS127L21 for your functional safety PLC AI module design.

    Without performing an actual FMEDA it is a little hard for me to determine what additional measures might be required to meet the metrics for SIL 2. Following just some thoughts:

    • Communication faults can be detected using the SPI CRC.
    • In addition you can detected a couple of device or communication faults when SDO is permanently low or high. I usually call this measure "host loss of communication detection". This can for example happen when the CSn pin cannot be drive low anymore or when the device went through an unintended reset, etc.
    • Bit flips in the register map can be detected using the register map CRCs.
    • Faults in the clock tree can be detected by monitoring the DRDYn output with a window watchdog. The conversions, and thus the DRDYn indication, are directly proportional to the clock frequency.
    • Offset errors in the ADC can be detected by measuring the device offset. This can be done by leveraging the MUX[1:0] = 10b register setting.
    • In order to detect gain errors or errors in the external VREF you would have to inject a separate test signal into the ADC in my opinion, which would require an external MUX.
      Faults in the external VREF could of course also be detected by measuring the VREF with an independent ADC, e.g. an ADC within the MCU.
    • The test signal measurement might also be useful to detect faults in the Mux.
    • In case you can create a variable test signal, e.g. using a DAC or PWM with RC filter, then you could use that also to check for any major stuck bits in the output codes.
    • You might also want to monitor the supplies, but I assume you will do that anyway.

    Let me know if that is any helpful.

    I'll let Keith in our team comment on your input driver question. He is the main application engineer for the ADS127L21.

    Viele Grüße,
    Joachim Wuerker

  • Hello Theo,

    Yes, the OPA2192 can drive the ADS127L21 with a few things to consider.

    1.  If operating from the same supply voltage as the ADC analog (5V typical) you will get crossover distortion.  This will result in a small offset voltage shift when the input voltage is within 3V of the positive supply rail.  This is usually acceptable if you are measuring low frequency signals, but will increase distortion (higher THD) if you are digitizing AC signals.

    2.  You can increase the supply voltage of the OPA2192, but then you will need to protect the ADC inputs in case the amplifier output exceeds the ADC maximum input rating.

    3.  Another option is to use the OPA2325 which is optimized for 5V operation and does not exhibit cross-over distortion.  Offset voltage and noise is higher than OPA2192.

    Regards,
    Keith Nicholas
    Precision ADC Applications