Post Go back to editing

AD4000 reference voltage and power consumption


A couple of questions about AD4000 design:

It is a 1.8V device; however, the current consumption is not mentioned anywhere in the DS (or I am not looking hard enough). On page 6 of datasheet, power dissipation numbers are given for different sampling rates. Can one assume this to be power consumption (vs power dissipation) numbers also?

In user guide for AD4000 UG-1042 a buffered Vref ADA4807-1 is used. Can it be replaced with an alternate like ADR381? If not, what might be a good candidate for reference voltage?


  • Hi Mohi,

    Thanks for your queries. I'll address the current consumption question first:

    Current consumption per supply is unfortunately not spelled out in the data sheet but you can calculate them for VDD and VIO based on their power consumption specs and voltages. You can also take a look at the TPCs on page 15 which show supply current vs. temperature and power consumption vs. sample rate. Note that the current consumption will scale linearly with throughput (sample rate of the ADC), and VDD current is higher when High-Z Mode is enabled. Also, VDD is the only supply which draws more current when High-Z Mode is enabled, so the difference between 

    Just using the spec table power consumption items for now, we can estimate the following:

    VDD Current:

    • 2 MSPS, High-Z Disabled = (9.75mW)/(1.8V) = 5.4mA
    • **2 MSPS, High-Z Enabled = (9.75mW + (16mW - 14mW))/1.8V = 6.5mA

    VIO Current:

    • 2 MSPS = (0.5mW)/(1.8V) = 0.28mA

    **In this calculation, the term (16mW - 14mW) was taking the difference between the total power consumption with High-Z Mode enabled minus High-Z Mode disabled, since VDD is the only supply which draws more power when High-Z Mode is enabled. i.e. at 2 MSPS the VDD supply draws ~2mW more when High-Z Mode is enabled.

    Also as a reminder, all of these numbers scale linearly with throughput so if you sample at 1 MSPS instead of 2 MSPS, for example, these current and power numbers in general will be cut in half.

    I'll respond to your REF drive question in a separate comment.



  • Hi Mohi,

    In this comment I'll respond to the query on the AD4000 REF input drive requirements (i.e. buffered vs. un-buffered voltage reference).

    The short answer is "it depends" and it is a function of the sample rate of the AD4000 and the accuracy/error tolerance of your system. This article is a good way to familiarize yourself with the nature of SAR ADC REF input as a dynamic current load and requirements on the end-system:

    The ADA4807-1 on the EVAL-AD4000FMCZ board was selected due to that amplifiers low noise and low output impedance, plus it's ability to maintain stability with a 10uF load (i.e. the capacitor next to the REF pin). It's simply one of the best options for maintaining really high accuracy of the VREF voltage even while running the AD4000 at the full 2 MSPS.

    Question: What sample rate do you plan to operate the AD4000 at?

    When selecting which device will be driving the SAR ADC REF pin (i.e. voltage reference directly driving vs. buffering it with an op amp), the main consideration should be, how much error in the VREF voltage can be tolerated for the application, and how fast of a sample rate needs to be used.

    From the article provided you'll notice that SAR ADC REF pins behave like transient current loads, where they draw a spike of current for a short period of time, as opposed to being a fixed current over time. Basically whenever the ADC performs a conversion, it needs to pull charge from the REF input for the SAR ADC bit trials, and that's what you'll see in those plots.

    If you would prefer to drive the REF pin directly with a voltage reference device, then I would recommend either the ADR4550 or ADR435 family of voltage references, as they have quite good load regulation specifications.



  • Hi Tyler,

    Your explanation is very through. Thanks for your help. A couple of questions.

    (16mW - 14mW) is taken from the power dissipation table. Correct?

    At the end of Table 1, "Energy per Conversion 7 nJ/sample" is stated. Would this energy draw includes total power consumption?

    In my case, I am inclined to use this number since sampling frequency will be less than 100Hz.

    Thanks again for your help.

  • Hi Mohi,

    Yes, the 16mW and 14mW numbers are from the spec table.

    The "Energy per Conversion" spec is just a measure of the total energy consumed by the SAR ADC core per conversion, and I think this is a combination of VDD and REF current.

    Personally I think you should look at Figure 35 (Power Dissipation vs. Throughput TPC) instead of the Energy per Conversion spec. That TPC shows typical power consumptions down to 10SPS for VDD, VIO, REF. You can see that the power consumption does start to "flatten out" at low sample rates because at that point the standby current/power consumption is a large portion of the total current/power consumption.

    The numbers in that table are obviously in mW, so you'll have to scale them down by the appropriate voltages for current, as described in the previous comments.

  • Thanks, Tyler. Appreciate your comments.

  • Hi Tyler,

    The current draw by REF pin of AD4000 is 100uA at maximum, so ADR4550 seems to be a good fit.

    It can supply 10mA of current and has very low output noise.

    Thanks for your analysis and pointing to the literature.


  • Hi Tyler,

    One last question: Is there any performance concern if we use ADR4540 (4.096Vref or any other reference) instead?

    The input to AD4000 will be 0 to 4V.



  • Hi Mohi,

    From a settling/accuracy perspective, the ADR4540 should behave just as well as the ADR4550.

    Note that the dynamic range/SNR will be slightly less for a 4.096V vs. a 5V VREF, but since your signal is only 0 to 4V anyway, that won't be a problem. Input referred noise will be the same between VREF values anyway.