Inductor selection dilema SMPS

[gary_01]

When you have a load that could feasibly be run across a wide range of currents, should you design a SMPS for the max load even if that means the typical usage falls into the discontinuous operation region. Or can you design for typical usage and just ensure the inductor is rated to handle the max usage too?

For example, I'm selecting an inductor for a 12V to 5V step down buck using the LM2596. The max load would be slightly below 2.5A if I maxed out the device processing + maxed out its USB current supply.

Doing the calculations for 2.5A max, 12 in, 5V out, I have an E.T of 18.88 which if I use with figure 30 of the datasheet (see attachment) @2.5A puts me slap in the middle of the L41 region a 33uH inductor.

So far, so good. Using figure 31 gives peak-to-peak ripple current of ~600mA and a peak switch current of 2.8A. Which when plugged into the min load before discontinuous operation gives 600mA/2 = 300mA.

Now what if I expect the device to operate generally at 250mA even though in some circumstances it could be used at 1A or up to 2.5A at the extreme. That would cause it to generally be running in the discontinuous region. That would seem like a poor selection of inductor?

In this case would it be safe to instead design the supply using an inductor choice based around a typical usage current rather than max load, resulting in a higher inductor selection due to lower load and a lower discontinuous cut off point (for example 200mA) but still ensuring the inductor was safe to use all the way up to 3A?

Datasheet LM2596]

[JS]

  Even if the inductor can handle 10A, the value of the inductor is a factor on the range of the PS, you need to be sure not only the inductor can handle the current (and any other component) but also that the complete circuit can handle the operating point at the extremes.

  What's the problem with discontinuous operation? efficency? If so, working at 250mA you shouldn't be as closely worried about the efficency as working at 2.5A as 50% efficency wouldn't be anything to worry about there but 80% could already be a problem in the higher end.

JS

[CraigHB]

If you want to keep the converter in CCM for light loads you need more inductance which reduces current ripple and that's good.  The down side is higher value inductance means higher DCR and lower current limits.  So to reduce DCR you need a physically bigger inductor or a lower inductance value.  Lower values result in a higher current threshold for DCM.  So to answer your question, you need to consider both lower and upper current demands in your design.  If you want to keep the converter in CCM under typical loads you need more inductance.  If you want to support higher currents you need lower DCR which calls for less inductance or a physically bigger inductor

There is a real handy feature some converter controllers offer.  They have a pin programmable forced CCM mode which eliminates some concern over inductor selection.  The disadvantage with forced CCM is you lose efficiency for light loads where the controller would otherwise pulse skip.  Though forced CCM keeps output noise at a constant level regardless of load so it's easier to filter out.  Do a simulation and check it out.  It's quite an interesting mode of operation.  Forced CCM allows current to cycle both directions through the inductor.