• State Resolved
  • Locked Locked
  • Replies 10 replies
  • Answers 1 answer
  • Subscribers 47 subscribers
  • Views 813 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.

ADS1278: ADS1278 vs ADS1282 comparison

Karthik Raveendranath
Intellectual 840 points
Part Number: ADS1278
Other Parts Discussed in Thread: ADS1282, LM27762, TPS54061, ADS8900B

Hi,

   I am in search of a precision ADC and came across ADS1278 and ADS1282. My input signal is a true bipolar signal with a low frequency of less than 50 Hz. My objective is to get highest noise free resolution. Comparing both ADCs following key points are observed.

  ADS1278 is having an SNR of 111dB at a throughput of 52KSPS when operated in high resolution mode which corresponds to an ENOB of 18.14 bits (Since SINAD is not given datasheet, I have taken SNR = 6.02 x ENOB + 1.76). Oversampling theorem says for every 4 times the sampling and averaging, 1 additional ENOB is gained. So if my required throughput is 1 KSPS, My OSR of ADC output (different from internal Sigma delta OSR) is 52, which corresponds to an ENOB of 2.85. Thus the total ENOB is 18.14 + 2.85 = 20.99 bits. Please correct me if wrong.

  For ADS1282, at a throughput of 1 KSPS and PGA =1, the SNR is 124 dB equivalent to 20.30 bits.

  Which ADC is the best among both with respect to noise free resolution.

Thanks and Regards

     Karthik R

  • 0
    TI__Mastermind 47750 points

    Hello Karthik,

    I agree with your above analysis.  If you do additional, external, oversampling (averaging), you can further improve the SNR of the ADS1278, and should be able to get about 4dB SNR increase at 1ksps verses the ADS1282.  However, you will consume much more power to achieve this.  The ADS1278 will require about 65mW per channel, plus external input amplifier per channel.  The ADS1282 includes the input amplifier, and will require around 25mW under the same conditions.

    Regards,
    Keith Nicholas
    Precision ADC Applications

  • 0 in reply to Keith Nicholas
    Intellectual 840 points

    Hi Keith,

     Thanks a lot for the reply. As you told due to power and multiple digital supply requirement of ADS1278, I will go ahead with ADS1282. I am planning for MCM realisation using Dice, the device is available in KGD. I am planning to use REF5025KGD for 2.5V ref. However I am having only +/-15V supply. So I need to generate +/-2.5V supply and reference for ADS1282. Can you suggest a solution for the same. My input is bipolar.

    Or I can translate input to unipolar 0-5V and use 5V input and reference. Which scheme will be better.

    In both the cases I assume supply and reference can be commoned. Please correct if wrong

    Thanks and Regards

         Karthik R

  • 0 in reply to Karthik Raveendranath
    TI__Mastermind 44720 points

    Hi Karthik,

    If your signal-source is bipolar, then I'd probably recommend using bipolar supplies for the ADC to avoid introducing additional noise and errors into the signal through the signal conditioning circuitry.

    You might consider the LM27762 to generate the +/- 2.5V supplies for bipolar operation. This device requires a single 2.7 to 5.5V supply, so you'd likely use it in conjunction with a buck converter to efficiently generate a 3.3V or 5V supply.

    Once you have the +/- 2.5V supplies you use them to power the ADS1282 as well as a REF5025 or REF5040 (you'd connect the GND pin of the reference to -2.5V and all of the decoupling capacitors would also be referenced to -2.5V). By using a 2.5V or 4.096V reference source (as opposed to a 5V reference), you wouldn't have to generate an additional 5.3 - 5.5V supply for the reference to accommodate the dropout voltage.

    I generally wouldn't recommend using the reference source like an LDO to power the ADC. For one, this is not the most efficient way to generate a supply for the ADC (the large voltage drop from VIN to VOUT means you'll dissipate more power to generate this supply). Also, most voltage references are fairly limited in how much output current they can drive so they don't make for the best supply sources.

  • 0 in reply to Christopher Hall
    Intellectual 840 points

    HI Chris,

      Thanks for the reply.

    Christopher Hall said:
    You might consider the LM27762 to generate the +/- 2.5V supplies for bipolar operation. This device requires a single 2.7 to 5.5V supply, so you'd likely use it in conjunction with a buck converter to efficiently generate a 3.3V or 5V supply.

      Actually I was planning an MCM with this device but LM27762 doesnt have a Dice version I think. So anyother option. ?

    Christopher Hall said:
    I generally wouldn't recommend using the reference source like an LDO to power the ADC. For one, this is not the most efficient way to generate a supply for the ADC (the large voltage drop from VIN to VOUT means you'll dissipate more power to generate this supply). Also, most voltage references are fairly limited in how much output current they can drive so they don't make for the best supply sources.

     LM27762 is an LDO ryt. So the power dissipation is inevitable I think unless I use Buck converter to generate +/-2.5V from +/-15V.

      I was also thinking to generate 2.5V from reference and then using a good current sourcing /sinking opamp to generate +/-2.5V using inverting and non-inverting configuration to give common supply and reference to ADS1282. Whether it will be fine. ?

    Thanks and Regards

        Karthik R

  • 0 in reply to Karthik Raveendranath
    TI__Mastermind 44720 points

    Hi Karthik,

    I was thinking you'd still use a switching converter (such as the TPS54061) before the LM27762.

    You might try posting the supply question to the E2E Power Management Forum. I'm not aware of what options there are for die versions of power supply devices.

    I've only experience using the LM27762 (integrated charge pump + LDOs), or TPS54061 and TP62125 switching supplies (in conjunction with TPS71725 and TPS7A3301 LDOs to filter the switching supplies). The latter solution required a lot more PCB area though, but I suppose if you filtered the switching converters properly you could probably get by without having to use LDOs. There is a App Note (here: https://www.ti.com/lit/an/snva871/snva871.pdf) which shows an LC filter that performs nearly as well as an LDO at filtering the output noise on the switching supply if you're interested.

    I don't see any issues with using op-amp to generate the +/- 2.5 supplies, but for the ADC's reference voltage I'd recommend taking this voltage directly from the reference source since the op-amps will add additional noise and voltage errors to the reference signal, which in turn will directly increased the observed noise, gain error, and gain error drift in the ADC conversion results.

  • 0 in reply to Keith Nicholas
    Intellectual 840 points

    Hi Keith,

      Recently I came across one more ADC from another manufacturer with 20 bit resolution and 1.8 MSPS throughput. The ADC has a nominal SNR of 101 dB (ENOB is 16.77). So if my required throughput is at 1 KSPS, the OSR will be 1800 resulting in additional resolution of 5.4 bits resulting in total ENOB of 22.17 bits with a power consumption of 15mW at 1.8 MSPS. But for ADS1282, at a throughput of 1 KSPS and PGA =1, the SNR is 124 dB equivalent to 20.30 bits at double the power (which is not a main constraint).

      So again back to the same doubt. Which ADC is better or is these calculations are accurate only in paper.

         Thanks and Regards

            Karthik R

  • 0 in reply to Karthik Raveendranath
    TI__Mastermind 44720 points

    Hi Karthik,

    Is there an online datasheet you could point us to to see additional details? Is the nominal SNR of 101 dB including the the OSR of 1800, if so then you'd need to average an additional 1800 samples to see the additional 5.4 bits of resolution.

  • 0 in reply to Christopher Hall
    Intellectual 840 points

    Hi Chris,

      Its AD4020. The nominal SNR of 101dB is at 1.8 MSPS without accounting for OSR.

       Regards

          Karthik R

  • 0 in reply to Karthik Raveendranath
    TI__Mastermind 47750 points

    Hi Karthik,

    Your calculations are correct with respect to thermal noise.  Please keep in mind that the total SNR for the ADS1282 includes the internal input amplifiers.  The SNR calculated for the SAR converter is just for the ADC, and does not include the external amplifiers, and neither include any noise from the external reference.  Your overall system SNR will depend on all of these additional components.

    FYI, if you do decide to choose a SAR converter and operate at a high sampling rate, and then average the readings to improve SNR at 1ksps, the ADS8900B is also an option.  This part provides 104.5dB SNR at 1MSPS, and includes the reference buffer.

    Regards,
    Keith

  • 0 in reply to Keith Nicholas
    Intellectual 840 points

    Hi Keith,

      Thanks for the reply. We had evaluated ADS1282EVM and got superior results previously not achieved with any other ADCs.. So we have made a custom board for ADS1282 but yet to evaluate it, When we considered the final realization, it requires many supplies for our application and constraints and results in larger power. Anyways I am planning to start a thread on its powering and reference. Thanks for the ADS8900B suggestion. I will study it in detail and see.

       Regards

          Karthik R