• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 1 answer
  • Subscribers 46 subscribers
  • Views 499 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.

ADS1675: Single-Cycle Settling vs Fast Response

John Kasson
Prodigy 90 points
Part Number: ADS1675

Is the ENOB (Effective Number Of Bits) for the single-cycle settling setting the same as the fast response setting in the low-latency filter mode of the ADS1675? If so then why?

Given what I have read that a single-cycle settling is completely settled in one conversion, that it only uses data from the previous conversion, and that in the fast response setting the first conversion takes the same about of time as the single-cycle settling setting but successive conversions come faster but at the cost of a delay in a step response doesn't that mean that fast response is using past conversion data for each new conversion (like in a moving average filter)? If this is the case then what method is the fast response setting using to approximate the adc conversion sample?

Thank you for your time. 

  • 0
    TI__Mastermind 46870 points
    Hi John,

    Yes, the ENOB will be the same for a given DRATE setting between single-cycle low-latency and fast-response low-latency modes. Both modes use the same digital filter configuration resulting in the same -3dB cutoff frequency. This cutoff, along with the input modulator noise characteristics, determine the overall output noise, or ENOB, of the ADC.

    Another way to think about this is that in single cycle settling, internally the output data rate is the same as fast response, but the ADC throw's away roughly every other output result, to guarantee that the output data is always settled. (It is actually a little more than 1 output data cycle that is discarded, which is why in fast response, you need to get rid of two cycles after a step input, before reading the 3rd result, per figure 42.)

    I hope this helps clear things up. Please feel free to reply to this post if you have follow-up questions.

    Regards,
    Keith N.
    Precision ADC Applications
  • 0 in reply to Keith Nicholas
    Prodigy 90 points
    Thanks Keith.

    Could you describe, in detail, the internal processing difference between the single cycle settling and fast response modes?

    From my understanding of the configurations then if I was monitoring AC signals and I wanted the most accurate readings and as quickly as possible I would go with the single cycle setting mode, would you agree?

    Regards,
    John Kasson
  • 0 in reply to John Kasson
    TI__Mastermind 46870 points

    Hello John,

    Internally, the Low Latency filter can be thought of as a weighted average moving filter, or FIR filter. This filter uses the same topology as a moving average filter, but each sample uses a different weighting factor, verses a moving average filter where each has the same value. Take a look at this Wikipedia link for a detailed description. 

    For the following discussion, I am using the Low Latency filter with DRATE='000' as an example. This filter uses 556 input samples to calculate the result. (If this was a moving average, then each weighting factor would be equal to 1/556.) Both the Fast Response and Single-cycle use the exact same filter with the same constants. The difference is how often the output result is available, or the output data rate.

    For the Fast response, it outputs a conversion result every 256 input samples, thus the conversion result will not only depend upon the previous 256 input samples since the last output, but a total of 556, hence it will not be settled to a step input voltage change that occurs just before the last output sample result.

    For the Single cycle response, it will output a conversion result every 556 input samples, thus the entire result does not have any overlap with the previous output result. This is similar to the differences between a moving average filter and a simple average filter.

    The advantages of the fast response option are the ability to convert higher input frequencies without aliasing, or to more quickly respond to a step input change. The reading would not be settled, but you would see a change sooner due to the higher output data rate.

    In your case, I assume that the input step can occur at anytime during the data conversion period. Assuming worst case, this could require nearly two conversion periods in Single cycle (the current conversion would not be settled since the step occurred during the conversion period, the result will be an 'average' of the input level before and after the step). For the Fast response, worst case will be 3 cycles. For the same DRATE setting, the Fast response will provide the shorter fully settled reading.

    Regards,
    Keith

  • 0 in reply to Keith Nicholas
    Prodigy 90 points
    Keith,

    Thank you for the detailed answer!

    Regards,
    John Kasson