Finding sufficiently crappy inductors

High DC current rated bead. Sounds like something Coilcraft or

A parallel resistor should kill the Q just as well as a series resistor. That's the first thing I'd try.

Jon

But it'll totally screw up the filtering.

Cheers

Phil Hobbs

Your output filter is inside the feedback loop? Why not place the filter after the control loop.

"Another issue that needs to be dealt with is compensation. It may be count er intuitive, but it is almost always better to put the filter inside the f eedback loop. This is because putting it in the feedback loop helps damp th e filter somewhat, eliminates dc load shift and the series resistance of th e filter, and gives a better transient response with less ringing. Figure 5 shows the Bode plot for a boost converter with an LC filter output added t o the output."

It is after the control loop, but the resonant load seems to confuse the controller.

I could maybe do some lead/lag thing with a series RC across the second cap of the pi network, but I wouldn't have too much confidence in it unless the lead/lag cap were much bigger than the filter cap, so that the impedance to ground would be mostly resistive over a pretty wide frequency range. Seems like a bit of a waste, but might be the only good way out. Some nice silicon-iron bead would be the ticket. ;)

Cheers

Phil Hobbs

Seems to me that a bad inductor has to have either ESR or EPR, so real resistors can do whatever a bad inductor can do.

You could add a parellel R to get loop stability, then lowpass some more, ferrite bead+cap maybe.

They're pretty good inductors down at 100 kHz, which is usually a good thing but not in this case. :(

Murata makes some parts whose impedance is rated at 1 MHz. I'd really like a tiny choke with a silicon steel core. ;)

I might try finding a ~47 uF tantalum with about 0.7 ohm ESR to parallel the 4.7 uF output cap of the filter. That might keep the Q down over a wide enough range--the reactance of the 1-uH inductor is about an ohm at

Cheers

Phil Hobbs

A second winding (basically, use a transformer instead of an inductor) with a resistive load on it could damp oscillations, but not be in the DC current path. Less heating, at least.

In true esprit d'escalier style, it occurs to me that a series-connected transformer with resistor load on the secondary is the cleaner solution.

What's wrong with damping the filter with R+Cs? If you can't afford space for a resistor then a tant will do, or if you're allergic to tant, polymers can be found in adequate ESRs.

sqrt(1uH/4.7uF) ~= 0.46 ohm so shop for something around there. A 1 ohm DCR on an inductor will do the same thing but costs a lot of power; a bigger choke can be put in parallel with the resistor for the same effect (the lossy inductor you were looking for).

I don't know of any "inductors" quite that crappy. Powdered iron (specifically mix #26) is bad but still manages a Q of 5ish at that frequency. Ferrite beads are about that bad (Q in the 1-3 range is possible), but they saturate very easily.

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Design Website:

That's equivalent to just hanging a parallel resistor on the inductor. It'll de-Q it all right, but it'll also trash the filter performance at high frequency.

Cheers

Phil Hobbs

What I really want is something made of silicon steel. ;)

I'm going to try a 47-uF tant with ~1 ohm ESR on the output side and see. I can get that in a 1411 package, which isn't horrible.

Cheers

Phil Hobbs

rd

A 1 ohm

Maybe time for some flower arranging wire. It's insulated iron or steel. People used to make audio transformers with it in the 20s :)

NT

It's the core that needs to be silicon steel, suitably subdivided to prevent shorted-turn effects. So something like wire rope made of black-oxidized Si steel might work. They don't exist, of course.

Cheers

Phil Hobbs

ford

nt,

? A 1 ohm

is

NT

Except that core losses are frequency-dependent, so that the ESR goes up with frequency in lossy cores. That's why ferrite beads are so useful in stabilizing low-noise, moderate-frequency amplifiers--the series resistance pretty well goes away at frequencies you care about.

Seems like I can almost hear the voice of a pal from San Francisco saying "lotsa parts". ;)

I'm going to try to do a lead/lag thing with just a biggish tantalum across the last cap. It should work if I get the capacitance and ESR right, but it's somehow inaesthetic to have to waste that much capacitance to bypass a 4.7 uF X7R. :(

Cheers

Phil Hobbs

Tantalum on a power rail? Lotsa smoke.

Your inductor is already too low in resistance to be bothered about a parallel resistor. I suggest a 2uH inductor and a shunt 2 ohm resistor.

Then you're totally screwed.