What could be simpler than inductors? Pick one from a catalog based on saturation, inductance, DCR, etc, and you're done. If you need more than that, get the Trilogy from Wurth, put it on your bed stand, and by morning you can select better.
Well, this is almost the case if you are designing a POL ess than 100W using ferrite. There are several big vendors, reasonable data, and plenty of tutorials. If you graduate from POLs to higher power where you discover powder cores, amorphous, tape wounds, an others, and dare to custom wind it: things get tough quick.
In this blog entry I am not going to review all of magnetics or inductor design. I am simply going to take a few inductors of different types, take some data, and point out some differences. Also, I’ll show you that life is much better if you make your own measurements.
So here are three commercial inductors. A small 47uH ferrite from Wurth, a big 22uH ferrite from TT Electronics, and a medium size 64uH toroid powder core from PE. Not some kind of scientific match, just some inductors laying on my bench.
All the measurements are made with the Bode 100, which is available from Picotest. I have to put in a shameless plug for this instrument, especially when you have the Picotest Injector Kit. If you are working on power, you are blind without this stuff. They also have lots of design notes and are very helpful when you get stuck on a measurement problem.
Let’s start with the small 47uH.
The Bode 100 has an impedance measurement tool built into its PC Application. You can do some simple and cool things with this, like put a current load on the inductor and measure inductance vs. current. This is a good way to see how flat the inductance is and find the saturation current. For this small ferrite, the data sheet is probably pretty good. But if you wind your own, how do you know if you can trust your calculations?
Ok, so now lets look at something more interesting.
From 10kHz to 1MHz we have a nice straight line for the impedance. We have 90° of phase. This is what an inductor does. Then at 2MHz the impedance drops and the phase becomes –90°. This is the point where the inductor switches from inductive to capacitative. No great discovery here, it is in all the books.
This graph shows the Q vs Frequency. If you read the Trilogy it points out that you always want to operate on the left side of the peak. Operating in the right side of the peak creates losses. This is where the ratio of reactance to resistance goes bad.
The nice thing about this inductor is that the peak of the Q is out past the peak of the impedance, so you can use this inductor out to 1MHz.
Now let’s consider the powder core. You would use these when your currents are very large, and ferrite would be too big, or would saturate. So, the 64uH toroid.
We have nearly the same impedance. Nice straight line out 1MHz. The phase changes from +90 to –90 at the peak, but then goes to zero. The material is running out of poop and looking like a resistance. This comes from the core losses, and other AC losses. We won’t operate out that far, so we can neglect them.
But look at the Q. The downside of the Q peak starts at 20kHz! This means you can’t use the full frequency range without working on the lossy side. Ouch.
So can we just fix this with a big ferrite inductor? Well, let’s try. Let’s look at the 22uH. Remember, if we had a 50uH or so ferrite, it would be even bigger than this one.
Another nice straight impedance line out to 1MHz, and a nice phase change to –90. We can see the change to non-reactive losses, but the phase transition is a bit farther out than the powder core.
The peak Q is at 50kHz. This is 2.5 times the toroid, so there is more working room, but still, nothing like our small 47uH ferrite. Thus, just because it is ferrite, does not mean the Q is out past the peak impedance.
Let’s compare the two big guys.
The ferrite is 22uH, Isat is 26A. The powder core is 32uH, Isat is 6.6A. We have to be careful comparing these, as they are clearly not in the same power handling range. You can get the bigger version of the BI part and it can handle 16A, but I don’t have one nearby to take data on.
The 32uH spec has no graphs in the data sheet. Just the max current, the inductance, and the DCR. They have a 2:1 inductance range. Presumably you might use these if you want to prevent DCM mode at light load, as the inductance is higher and low current. The 22uH spec has graphs for inductance vs. current, and temperature vs current, but that is all.
Neither data sheet says anything about frequency effects on inductance or losses.
Yikes, data sheets that don’t really tell you much. What to do?
Your best defense is to buy a bunch of off the shelf inductors, wind some of your own if you are into large power inductors, and measure everything. This is not to say design does not matter. It is just to say you should not completely trust design or make assumptions about off the self products.
There are lots of instruments you can use to measure these, but I suggest the Picotest Bode 100, mainly because it plots everything and you can paste it into your design documentation. (But also because it has great performance out to 40Mhz.
With the injectors and EMI probes, you can measure 80% of all the things a power engineer should measure. Inductance is just the tip of an ice berg.
Of course if you are lucky enough to have some nice expensive bench instrument around, well, perhaps you should dust it off and use it.
After all, it does get lonely.