Measure AC losses of inductor

Hint: remove the core.... Paul Mathews

Low flux-density measurements don't tell much about actual power-level losses.

Measure operating temperature of the inductor. Then shut off the switcher and apply a DC current to the inductor to get the same temperature. Now calculate DC power dissipation.

(I just made that up.)

John

Hello Klaus,

Not sure what you are actually after. If it's saturation effects those can be measured by closely investigating the ramp, to see where it starts it's "shark fin" rise.

Check my website under electronics/inductors.

You can estimate at least eddy current loss by the current offset at turn-on. Other errors may manifest as subtle changes in the current waveform (steeply sloping shallower of course means voltage developed across a resistance (LR exponential shape), while sloping steeper is saturation).

Tim

-- Deep Fryer: a very philosophical monk. Website:

A precision LCR Meter you will easily measure Lac, Rac and Q at the SMPS frequency of operation. This reading is for sin waves and the SMPS is dealing square waves. Maybe someone knows a slick way that the sin measurement can be converted to an equivalent square wave Rac measurement. Maybe, measure Rac at the 3rd, 5th, 7th and ninth harmonic, knowing the duty and rise and fall times of the square wave, then recalculate Rac. A conversion factor from sin measurement to square wave equivalent would be sweet. Measuring both Rac and Rdc, then calculating Rac/Rdc will give insight into proximity losses. Cheers, Harry

Hello Harry,

Or just run the square wave or whatever your desired load is into the inductor via a resistor, and acquire the current and voltage per time interval. That should lead to the real and imaginary parts of the power involved. The real part attributed to the inductor would be whatever it consumes or radiates off.

I do this stuff with an impedance analyzer. You could measure it at the fundamental and then the harmonics up to where your demand for precision would be satisfied. Then weigh them each accordingly and add. My old HP4191 can be programmed to hop the harmonics so I could go get a coffee while it's doing the measurement :-)

But again, in a SMPS what really matters is usually core saturation. That's what causes losses and also the occasional kablouie.

Joerg skrev:

I have control of the saturation level. I need accurate measurements of the losses to be able to choose among several different inductors to find the optimum sample

inductor to the outgoing. But wont that just give me the entire losses of the inductor - including DC losses.

Regards

Klaus

Make a series LC circuit with a high quality cap as you mentioned. Drive the circuit with a sine wave with a known impedance, say 50 ohms. Tune the frequency of the source until resonance is obtained on the series LC circuit as a null or dip in the response. Measure the ratio of the driving voltage to the series of the driving impedance and the LC network as a voltage divider. Solve the voltage divider for the AC resistance of the series network just as you would a simple DC voltage divider. If the cap is sufficiently high quality, most of the measured resistance will be due to the inductor and it's losses. Note, this must be done at resonance where the phase angle is zero or reactive voltages will give a false indication. The resistance at different frequencies can be measured by choosing different resonating capacitors. Bob

Hello Klaus,

Yes, it would. But DC losses can easily be calculated out by measuring the DC values. In case of an inductor that is wound with well insulated wire this is usually only the DC winding resistance. Skin effect should be counted towards the RF loss budget.

WRT to RF losses: These can vary. Cores are manufactured in batches just like chips. Except that they usually aren't date-coded so you won't know whether the inductor you bought in July contains a different batch core than the one from March.

For a really thorough AC loss totalizer I don't think you can get away without an impedance analyzer. At least not if you have to do a lot of them. Before I bought the HP4191A for the lab here I did this kind of stuff with a generator and a comms receiver. Wore out the encoder bearings on one of them and now I can't get a spare :-(

So measure temperature.

John