Advertisement

Blog

Interfacing to ADCs: Power Supplies, Part 3

After reading through the comments and having a few discussions with some of my coworkers, I thought it would be good to continue looking at the example I gave in my last blog (Interfacing to ADCs: Power Supplies, Part 2) where we looked using fewer LDOs and combining power supply rails on an ADC while maintaining isolation with ferrite beads. One very important item that I’ve left off thus far has been proper power supply decoupling. The focus has been more at a high level and looking at the topologies that can be used for ADC power supplies.

I want to take a moment to go back to that example and add in the decoupling. The size and value of the decoupling capacitors (shown as n capacitors in Figure 1) will depend on several factors, such as power supply voltage, frequency of operation, ADC power consumption, LDO characteristics, etc. There are many items to look at, but for the purpose of this discussion we will assume that the proper decoupling capacitors have been chosen. I’d ask that the reader keep in mind that it is always good design practice to decouple the power supply inputs to the ADC properly.

Figure 1

Driving multiple ADC power supply inputs from a single LDO. (Proper decoupling.)

Driving multiple ADC power supply inputs from a single LDO. (Proper decoupling.)

Let’s now take a look at a topology that will mitigate some of the power consumption issues we have looked at in the last few blogs. In many cases there is a higher voltage supply available in the system, but a lower supply voltage is required from the ADC. Many of the ADCs available today use a 1.8V power supply voltage. In many systems a higher supply voltage such as 6V or 12V is available (and could be higher in some cases). Let’s take a look at an example where a 6V power supply voltage is available and the ADC requires a 1.8V power supply input. For the purposes of this discussion we’ll focus mainly on the analog, digital, and driver supply inputs of the ADC. The input buffer supply is often a higher voltage such as 3.3V and is not a high current supply input, so the drop from 6V to 3.3V can be accomplished with a single LDO.

Figure 2

Stepping down a high input voltage for lower ADC power supply inputs.

Stepping down a high input voltage for lower ADC power supply inputs.

Let’s look at an example using the 14-bit 250MSPS dual channel AD9250. The typical total power consumption listed in the data sheet for the AD9250 is 711mW. This ADC has three power supply inputs which are the analog (AVDD), digital (DVDD), and driver (DRVDD) supplies. Let’s use the topology shown in Figure 1 and calculate the power consumption and junction temperatures. For this example, we will use two ADP1741 LDOs — one configured for a 3.3V output and the other configured for a 1.8V output — so that we produce the supply voltage needed as shown in Figure 1.

To begin, let’s look at the total current draw from the AD9250. Summing the current requirements from the three supplies, the total current requirement of the AD9250 is 255mA (IAVDD ) + 140mA (IDRVDD + IDVDD ) = 395mA. First let’s look at the case for the ADP1741 generating the 3.3V from the 6V supply input. In this case, the ADP1741 will be required to dissipate (6V – 3.3V) x 395mA = 1.067W. The means the maximum junction temperature Tj would be equal to TA + Pd x Θja = 85o C + 1.067W x 42o C/W = 129.79o C, which is less than the maximum rated junction temperature of 150o C for the ADP1741.

This is the larger of the two voltage drops on the supply rail so it means the second ADP1741 is ok as well, but let’s look at the calculations. We have the same current since the second ADP1741 as in the first ADP1741, which is 395mA. For the case of the second ADP1741, we have a voltage drop of 3.3V – 1.8V = 1.5V. Calculating the power dissipation, we arrive at (3.3V – 1.8V) x 395mA = 0.5925W. Now, calculating the maximum junction temperature we get 85o C + 0.5925W x 42o C/W = 109.89o C, which is once again less than the maximum rated junction temperature of the ADP1741. Assuming we have properly selected the ferrite beads and decoupling capacitors, we have arrived at a nicely functioning power supply for the ADC. Stay tuned as we continue to look at driving ADC power supplies.

16 comments on “Interfacing to ADCs: Power Supplies, Part 3

  1. vasanjk
    August 29, 2014

    JH

     

    I was advised by a circuit protection component company FAE to use a ferrite bead in series with the decoupling cap. He claimed that it improves the performance. What is your opinion on this?

  2. jonharris0
    August 29, 2014

    I think I understand, but let me make sure.  The circuit protection FAE was suggesting to place a ferrite bead in series between decoupling caps and the device.  I can understand this logic to some extent.  This is based on a circuit protection mentality.  I am looking at these components more from an isolation/noise suppression angle.  I see where the FAE is trying to go since (in the past I worked as an AE on circuit protection components).  A lot of folks will use some amount of series inductance to help mitigate ESD events.  However, this is generally used where there is not good protection otherwise.  In the case of an ADC (particularly an ADI ADC) there is ESD protection on the pins of the device.  In addition using a ferrite bead for ESD on every power supply pin would be cost prohibitive (component cost, layout complexity, board area, etc.).  Another point is the fact that most ADCs are designed into some form of equipment that does not regularly come into contact with people or machinery so HBM or CDM discharges would only be at the time of assembly for the most part.  It helps to have the FAE's recommendation in context to know what application he is referring to.  I would not recommend such a topology in most ADC applications but there could be other applications for different components that might warrant considering the design approach…however, with good ESD protection designed into the component the external ferrite beads should not be necessary.  Apologies for the lengthy answer, but I hope this helps shed some light for you.

  3. etnapowers
    August 29, 2014

    I think that the costs of the protection circuitry have to be aligned with the standard margin of the IC.

  4. dates
    September 1, 2014

    I don't know any applications where you'd want to cascade two LDOs and dissipate this much power.

    One alternative is TPS54120, which combines a dc/dc converter and a high performance LDO. You could take the 6V and convert to a useful 3V3 which can be used elsewhere in the system, then use the LDO to get a quiet 1.8V.

  5. goafrit2
    September 1, 2014

    >> I was advised by a circuit protection component company FAE to use a ferrite bead in series with the decoupling cap.

    I do not think that makes a lot of sense. It is better you run it in parallel if you want a better performance. A series connection will depress circuit performance.

  6. goafrit2
    September 1, 2014

    >> I think that the costs of the protection circuitry have to be aligned with the standard margin of the IC.

    That is exactly what happens in the industry. You need to evaluate how much this IC is fetching you in dollars to what the cost of the protection will be. It is the way it works in practice because for a sub-dollar chip, you can do as you want. But when you are paid top-dollar, you need to deliver better value.

  7. vasanjk
    September 1, 2014

    goafrit2,

    I may disagree.

    Circuit protection provides value on the whole for a product. If a sub-dollar IC is used without protection, it can impact product performance. One needs to rationally decide how the protection offers performance enhancement in stringent conditions.

     

    Rgds

     

    JKVasan

  8. etnapowers
    September 2, 2014

    Agreed. during the evaluation phase of the profitability of a new device the costs for protection circuitry have to be taken into account together with the gain, only if the gain margin is good the final decision can be taken.

  9. etnapowers
    September 2, 2014

    The integration of an LDO and a DC DC converter in a chip is a good solution, very effective because the LDO output is very stable and the DC/DC output may feed more blocks inside the IC.

  10. jonharris0
    September 2, 2014

    This is a great discussion guys.  Indeed, one has to weight advantages/disadvantages of using ESD protection components.  One thing I'd point out in that regard is the response time of the external protection.  Even if I device is rated for a higher discharge (say 20kV) and the IC is rated at 1kV, if the response time of the external device is less than that of the IC, then the IC will take the brunt of the ESD discharge rendering the external device virtually useless.

    On to the comments on the DC/DC converter and LDO combo…you guys are seeing where I am going with this discussion, very good!  Indeed this combo can make a lot of sense, but there are things to think about there as well.  Stay tuned! 🙂

  11. fasmicro
    September 2, 2014

    >> Circuit protection provides value on the whole for a product. If a sub-dollar IC is used without protection, it can impact product performance. 

    No one is saying that you do not need protection in a cheap IC. The point made is that the cost could determine the quality and resources deployed to make that protection.

  12. fasmicro
    September 2, 2014

    >> Agreed. during the evaluation phase of the profitability of a new device the costs for protection circuitry have to be taken into account together with the gain,

    Absolutely. That explains why you cannot use the same circuit used for a pacemaker for a toy because in the pacemaker you may be commanding $200 per chip while the toy is 66 cents. Understanding the cost is important to stay in business.

  13. RedDerek
    September 2, 2014

    I have seen many datasheets and app notes talk about adding in ferrite beads on the supply. Is there a simple approach as to the bead and capacitor selection? There is a wide range of beads as well as capacitors.

  14. vasanjk
    September 3, 2014

    Hi

     

    Beads are specified based on their frequency of operation. It is important to refer,the manufacturer's datasheets and decide upon the part number.

  15. Victor Lorenzo
    September 3, 2014

    @RedDerek, perhaps this app note from Murata could be usefull for you too. It requires some special and asian font sets for Adobre Reader.

    http://www.murata.com/~/media/webrenewal/support/library/catalog/products/emc/emifil/c39e.ashx

  16. etnapowers
    September 11, 2014

    Agreed. The expensive device must be protected effectively by a circuitry, that require components, accurate simulations and tests , whose number is proportional to the cost of the device. The profits related to the sold of the device can be used for investing in simulators and test equipments to further increase the quality of the expensive product.

Leave a Reply

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