Interfacing to ADCs: Power Supplies, Part 2

Thus far, we've looked at the various power supply inputs of a typical ADC and one method for driving these inputs with LDOs. In part 1 of this series, we specifically looked at using separate LDOs to drive each power supply input. This method provides the best isolation and, in most cases, the best noise performance. As the math demonstrated in the example in part 1, the noise at the LDO output can be much lower than the ADC noise and thus is not a major factor in the overall noise.

However, there are some disadvantages when driving low input supply voltages, where multiple LDOs may be required. Before we take a look at that, I would like to address a reader comment that I thought deserved some attention. The comment was about having too many LDOs in the previous example (one for each power supply input). An alternative approach is to use a single LDO that would fan out to multiple power supply inputs to the ADC. This approach is shown in the figure below.

Driving multiple ADC power supply inputs from a single LDO.

Driving multiple ADC power supply inputs from a single LDO.

This example moves to the opposite extreme and sources most of the ADC power supply inputs from a single LDO. Let's now look at some of the advantages and disadvantages of this approach and maybe try to examine some middle ground. As you can see from the diagram, this is a rather simple approach that uses fewer components. Reducing the number of LDOs also reduces the overall system cost.

First and foremost, the cost is lower because there is one LDO to purchase, instead of three (in terms of analog, digital, and driver supplies). Secondary to that, with fewer LDOs, there are fewer SMD components for the LDOs: resistors, capacitors, etc. There is cost associated with the new ferrite bead components, but the cost is much less than the cost of the LDO. Currently on the Digi-Key website, the 1.5k unit price for an ADP1741 is $1.53. By contrast, the 1k unit price for a typical ferrite bead on the Digi-Key website is about $0.029. This does not count the savings that can be achieved by using less board space.

This all seems great, doesn't it? If only it were that simple.

This may not be the best solution from a performance standpoint. Care must be taken to select a ferrite bead that provides sufficient isolation without a large DC resistance (DCR). In cases where small supply voltages (1.2 V) are required along with higher input currents (500-1,000 mA), the voltage drop across the ferrite bead could result in performance issues. For example, a ferrite bead with 150 mΩ of DCR on a 1.2V supply requiring 750 mA would have a voltage drop of 150 mΩ × 750 mA = 112.5 mV. That is nearly 10% of the supply voltage. In addition, it may not be possible for one LDO to provide enough current or handle enough power to drive all these power supply inputs.

Let's take another look at the example from part 1, where we calculated the power dissipation in the ADP1741 on the AVDD supply of a typical 14-bit ADC, which required 1 W of power. In that example, the ADC's total power was 2 W. In the same example, if used a total power of 2 W (since we are using a single LDO), the picture wouldn't look as good. The ADP1741 would be required to dissipate an approximate power of (6 V – 1.8 V)*1110 mA = 4.662 W. This would push the maximum junction temperature (Tj ) of the ADP1741 to TA + Pd x Θja = 85°C + (4.662 W x 42°C/W) = 281°C, which exceeds the maximum rating for the LDO by over 100 degrees.

(Note: In part 1, the equation for the power should have been (6 V – 1.8 V)*0.5556 = 2.33 W. This was the right power dissipation, but the error was in the presentation of the equation.)

As you can see, there is a balance to be achieved between cost, power, and performance. Does that seem familiar? I think we all face this tradeoff in most of our designs. Next time, we will continue looking at the power supply inputs to the ADC and how we can use multiple LDOs or a combination of LDOs and ferrite beads to mitigate the power dissipation dilemma. This does come with some increased cost, as you can deduce from my comments in this blog, but performance does not come for free, as we all know. I welcome and encourage the comments and questions. There is always time to stop along the way and dive a little further into different areas that interest our community members.

4 comments on “Interfacing to ADCs: Power Supplies, Part 2

  1. ecoguy
    August 6, 2014

    Don't blow-off the decoupling capacitors on both sides of the ferrite bead. You're attempting to create islands of good, clean power, the decoupling capacitors are necessary.

  2. paulfl
    August 7, 2014

    Adding inductors (ferrite beads) adds voltage transients with load steps – value -Ldi/dt. Keeping a capacitor on the ADC side mitigates this

  3. etnapowers
    August 7, 2014

    The capacitors are useful to stabilize the input supply voltage, then I agree with you on the need of adding this elements to the circuit.

  4. etnapowers
    August 7, 2014

    That's a good solution, provided that the capacitors have a good robustness to the voltage spikes.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.