• State TI Thinks Resolved
  • Locked Locked
  • Replies 6 replies
  • Answers 3 answers
  • Subscribers 47 subscribers
  • Views 1369 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

ADS131E08 Measurement Errors

Vinod Babu
Prodigy 40 points

Other Parts Discussed in Thread: ADS131E08

Problem:

We are using an ADS131E08 on our board. It is observed that when the input frequency to analog input is slightly above or below 0.125fMod, the ADC exhibits Alias like behavior and there is an output amplitude of about -15dB, whereas the datasheet (figure 35) shown better than -40dB attenuation for above 55KHz.

Steps to Recreate Problem:

ADC Channel 3 used for measurement, both P and N lines biased at 1.5V DC. 
Sinusoidal signal from a source is connected with AC Coupling to P input of ADC. 
ADC is configured for external Clock and DR bit are set as “100”. 
ADC clock is about 2.2Mhz, AVDD = 3V, AVSS=GND, DVDD=3.3V, VREFP=2.5V and VREFN = GND. 
The sampling rate is 4320 Samples/Sec.
The internal gain of ADC is set as 1.
The data read from ADC by the MCU for each conversion is output on a DAC of the MCU immediately. 
The DAC uses a 3.3V supply and level change when output on DAC is 3.3/5. 
The input signal, DAC output and DRDY signals are observed on oscilloscope.
The signal amplitude from signal source is kept at 325mV rms at about 139Hz and the output of DAC is observed to make sure the test setup and gains are as expected
The signal amplitude from source is changed to frequencies near 138KHz (frequency = 0.125_MOD). 
The input signal amplitude is kept at about 500mV rms.
Alias like behaviour is observed on the oscilloscope at input frequencies slightly above and below = 0.125_MOD. The waveforms are captured

First picture below is for reference when input signal is 139Hz (very low frequency compared to Sampling frequency).

The second picture below is to show the input at 139KHz. The screen is zoomed to see the High frequency and DAC output is not visible properly (see next picture)

The third picture below is to show the output of DAC. The screen is zoomed out to see the output of DAC at low frequency. 

Can the experts guide us on why this is happening?

Regards,

Vinod

  • 0
    TI__Mastermind 38745 points
    Hey Vinod,

    I simulated the filter and passed your high frequency input through it. The output was beneath the noise floor. To rule out a problem with passing the data between the SPI module and the DAC, could you collect a series of points, send them to a text file, and send them to me to review?

    Regards,
    Brian Pisani
  • 0 in reply to Brian Pisani
    Prodigy 40 points

    Hi Brian,

    Thank you for looking into it.

    I have captured the ADC output (decimal) for about 200ms for multiple input frequencies. Capture for each frequency is in different files with filename showing the frequency. The first column is the index and second column is the data.

    The captures are done near the frequencies observed with errors.

    All the files are zipped together. Please let me know if you prefer any other format..

    Thanks again..

    Regards

    Vinod

    multipleCaptures.zip

  • 0 in reply to Vinod Babu
    TI__Mastermind 38745 points
    Hey Vinod,

    What you are seeing is not aliasing. Delta-sigma modulators are standard control systems where an error signal is added to the inputs in the form of negative feedback. Take a look at Figure 43 in the datasheet. What is happening when you input such a large out-of-band signal is that the error that is pushed to higher frequency by the modulator is being fed back to the inputs and the modulator circuitry saturates.

    This converter is not intended to measure out-of-band signals. Are you concerned that large out of band signals (hundreds of mV rms) will appear at the inputs? If so I recommend simple anti-aliasing with an RC filter.

    Regards,
    Brian Pisani
  • 0 in reply to Brian Pisani
    Prodigy 40 points

    Hi Brian,

    Thank you for the detailed answer. 

    I had a look at the Fig 43. But compared to that, isn't the noise we are seeing in our case really high? We are seeing a noise of about -15dB with respect to input strength, as you can see in the previously attached text files. In the Figure 43 of datasheet, near 0.125fMod shows about -60dB noise. Can you help me understand why the difference? Do you see similar results in your lab?

    >> This converter is not intended to measure out-of-band signals. Are you concerned that large out of band signals (hundreds of mV rms) will appear at the inputs? If so I recommend simple anti-aliasing with an RC filter.

    We faced some issues in the measurement on our board due to strong high frequency noise signals and that is the reason for the detailed experiment. We decided to use anti-alias filter. But we wanted to check if this is an expected behavior and properly understand it so that we can make sure there will not be a surprise later.

    Thanks again..

    Regards

    Vinod 

  • 0 in reply to Vinod Babu
    TI__Mastermind 38745 points
    Vinod,

    Yes that figure shows the noise level for an input short configuration, but that noise will add to the signal at the given frequencies and cause the modulator to saturate. I referred you to that image to show how the modulator can behave very differently with different frequencies. I was able to replicate the behavior you see in the lab. This is expected behavior for this device and is a characteristic of the delta-sigma architecture.

    Regards,
    Brian Pisani
  • 0 in reply to Brian Pisani
    Prodigy 40 points

    Hi Brian,

    We will go ahead and use an anti-alias filter. 

    Thank you very much for the clarification.

    Regards

    Vinod