• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 46 subscribers
  • Views 348 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.

ADS1672: Analog input single-ended electrical characteristics

PPera
Prodigy 40 points
Part Number: ADS1672

Hello,

I would like to know what are the single-ended electrical characteristics for the analog inputs, because in the datasheet only the typical values for differential mode and common mode are specified. I would like to know what are the single-ended requirements.

In particular, in my design the reference voltage Vref is +3V and the analog power supply AVDD is +5V. I would like to know if one of the inputs (either the positive or the negative side) can exceed Vref. Since in the absolute maximum ratings section it's stated that the "Analog I/O to AGND" should be between -0.3V and AVDD+0.3V, I would say that the inputs can exceed Vref as long as the voltage is less than AVDD, whereas Vref is only a reference for the differential voltage drop. But this is only a guess. I would like to know if it's true and if it can impair the performance.

  • 0
    TI__Mastermind 24408 points

    PPera,

    For Vref = 3V and AVDD = 5V you will have the following input ranges:

    Linear Response Range = ±Vref . The range of input voltage for linear response of digital output codes is ±Vref. It is marked as typical because there will be some very small errors in this range due to offset, gain error, INL, and DNL. This is basically the range of the transfer function (Code = 223 Vin_dif /Vref). See example below.

    Common mode =2.5V: (see page 25, analog inputs section) In order to measure signal on this device the common mode voltage must be held constant at 2.5V. This is generally done using a fully differential amplifier, FDA (see figure 40). This type of input is called fully-differential. If you are not familiar with this kind of requirement I suggest you review ADC Input Types. Also, if you are not familiar with FDA amplifiers please review FDA Amplifiers

    The absolute maximum range = -0.3V to AVDD+0.3V = -0.3V to 5.3V. This is the range that the input signal can move without damaging the device. So, for your example, the device input can range from 0V to 5.3V without damaging the device, but it will not get valid output data unless the input is in the linear range of the device(±Vref), with the common mode at 2.5V.

     

    Example: Code conversion assuming Vref =3V, and Vcm =2.5V

    Vin = 2V

    Code = 223 (2V)/(3V) = 5592405 = 0x55 54 F1

    Vin = -2V

    Code = 223 (-2V)/(3V) = -5592405 =0xAA AB 0F

    Vin = 3V

    Code = 223 (3V)/(3V) = 8388608 = truncated to 8388607= 0x7F FF FF

    Note: we really can’t read 3V, but rather read 3V – 1LSB

    Vin = -3V

    Code = 223 (-3V)/(3V) = -8388608 = 0x80 00 00

    The accuracy of this calculation is impacted by Vref accuracy, Offset, INL, and DNL.

    The absolute maximum range = -0.3V to AVDD+0.3V = -0.3V to 5.3V

    I hope the above explanation helps.  Don't hesitate to ask follow up questions.

  • 0 in reply to Art Kay
    Prodigy 40 points

    Art,

    thank for your quick response!

    I know the best solution would be to use a fully-differential ADC driver and actually that is what we already do in other projects with great results. This time, my main goal is to enhance the long term stability, i.e. to reduce the long term output drift. In particular, I'm interested in keeping the output drift as low as few parts per million (ppms). After testing many fully differential ADC drivers, I found that they all lacked in this feature. The output drifted by tens/hundreds of ppm in one week. I managed to get in the specs by designing a custom front-end with few OPAMPs. With this front-end, I manged to have the inputs in the correct range. Of course, the input common voltage was neither constant nor close to 2.5V, but that was not a problem: the accuracy was very good. Now I want to increase the input range because previously I was not exploiting the entire full-scale range. If I exploit the entire full-scale range, one of the inputs (either P or N side) can have a voltage greater than Vref. For instance, with Vref=3V and AVDD =5V, I can have the following inputs:

    Ain_P = 3.6V and Ain_N = 0.8V, which means that Vin_diff = (3.6 - 0.8) = 2.8V and Vcm = 2.2V

    So the differential input is compliant since it is less than 3V. I would like to know if this condition, i.e., Vref < Ain_P < AVDD, is a problem, or, in other words, if I am obliged to have Ain_P < Vref.

    Regards,

    Pietro

  • 0 in reply to PPera
    TI__Mastermind 24408 points

    Pietro,

    1.  Based on your comments I believe that your FDA solution has poor "long term stability" because of offset temperature drift.  Most FDA type devices are optimized for AC characteristics and not for good DC performance (i.e. low offset, offset drift).  Normally, the term "long term stability" refers to a change in the device performance over time at a fixed temperature.  I would not expect the actual long term stability of the FDA to be better than the amplifier.  Nevertheless, I see your point and would actually expect the drift on the FDA to be worse than most amplifiers.

    2. An op amp drive circuit is a good solution to improve the temperature drift.  Also, I would make sure that any resistors you use are low drift precision resistors.  There are some amplifier typologies that can be used to generate a fully differential input. See http://www.ti.com/lit/an/sbaa265/sbaa265.pdf as one option.  Is your original signal input differential or single ended?  Ultimately, to meet the data sheet requirements for this device you need to force Vcm =2.5V.  If you do not do this your performance is not covered by the data sheet.  Even if it seems like it is working in your initial lab prototypes, it is not advisable to violate the data sheet spec for Vcm.

    3.  To answer your question directly, the input signal can be above Vref.  In fact, if you look at a full scale input signal of +3V, Vinp = 2.5V +3V/2 = 4V, and Vinn = 2.5V - 3V/2 = 1V.  To verify, Vcm = (4V+1V)/2 = 2.5V.  The point is that in order to meet the common mode requirement, the input voltage will need to be larger than Vref.  Nevertheless, as I mentioned previously, I would not use this information to conclude that you can use your circuit as is and violate the common mode specification.

    4.  There are many ADC that do not have this common mode requirement, and operate in what is called a "true differential" mode.  That is, the common mode can change from 0V to FSR.  You may want to consider changing ADC.  Let me know if you need help selecting an alternative device with true differential operation.