How big is core losses in an inductor at 40kHz?

Hi. I'm using two 22uH inductors in parallel as a filter on the output of a class D amplifier. The signal is a 40kHz sine and the RMS current through the inductors is about 3A. Their parallel DC resistance is 17.8mOhm which should lead to about 3^2*0,0178=160mW of heat loss, but I'm getting something close to 2W of heat. I know that the losses increase as frequency increases due to eddy currents and hysteresis losses, but how can they be that great at only 40kHz? The switching frequency of the amplifier is 350kHz, but it's S/N ratio is about 40dB which means that the RMS current of the switching frequency is 30mA.

Inductors: http://no.farnell.com/coilcraft/mss1...z/dp/2288504RL

How did you calculate 30 mA switching frequent current? Has the class-amplifier only 3 V power supply?

In any case, core losses for 40 and 350 kHz are probably the dominant factor.

Hi,

how do you know about the 3A RMS inductor current?

You need to take care abot the 350kHz switching frequency and it′s overtones. You need a good measurement tool to get a correct RMS value.

Maybe it is 3A RMS load current...then "RMS_load_current = RMS_inductor_current" is only true if there is no filter capacitor involved.

Klaus

Is the current 3A through each inductor?

Is there any DC bias current present?

You've got 3A rms at 40kHz in each of the two inductors as you have written it

depending on the core material and how close the plain winding wire is to the gap in the chokes, you could easily get 1W in each choke ....

unfortunately real world application losses are not often spec'd for a "DC" choke...

also 3A rms is close to 5.5A peak which is close to the saturation level for the chokes ( B > 0.4 T)

pushing DC chokes this hard at 4okHz AC will definitely give the losses you are seeing...

We often design with RM, 5,6,8,10 etc in low loss ferrite to get low loss chokes ....

for the part you chose the core material is not specified, as it is not of the highest quality, and not for high pk-pk AC ...

The calculation in post #1 suggests 3A rms for both inductors, 1.5 A each.

The specifications given by Coilcraft don't allow to estimate the core losses in the class-D filter application. We are also lacking the information about DC link voltage respectively 350 kHz PWM voltage across the inductor.

Thank you for all the answers :)

The output of the amp was measured with a network analyzer, it showed the switching frequency at -40dB, I kind of then assumed that the switching frequency current would be -40dB through the inductor as well, but that might not be the case. I remember testing the current draw of the amp loaded and unloaded without any signal on the input. The difference in current was mini, 5-10mA or so if I remember correctly so I don't think the switching frequency can be making the inductors hot.

The 3A RMS signal current is for both inductors, so 1.5A through each. There's a DC blocking cap in series with the inductors so there is no DC component.

Is there a way to get an idea of the ratio between core loss and I2R losses for an inductor? I've been toying a little with the Coilcraft core loss calculator (https://www.coilcraft.com/apps/power_tools/compare/). I'm not exactly sure of what parameters to put in as it requires a DC current input, but at 40kHz I'm always getting I2R losses as the dominant factor. I really didn't think that core losses would be more than ten times the I2R losses at 40kHz, could these inductors be really shitty at handling AC? The efficiency of this amp is important so it would be nice to find a way to improve it :)

I assumed so far that the inductor is connected between the amplifier output node and the filter capacitor, with the full PWM magnitude across the inductor. Is there an additional inductor that drops the 350 kHz square wave? Please clarify.

yes, as there are no losses stated in the data sheet for AC

C2 is matched to the transformer inductance so that they are resonant at 40khz. I forgot the 100uF series capacitor.

Looking at other SMD power inductors at farnell, I can't find AC losses or core material information about any of them. What type of inductor should I be looking for for this application?

O.K., +/-6V switching voltage, gives maximal 120 mA AC current at 350 kHz.

I'm however confused about missing AC coupling. The inductor current has a large DC component. How did you measure inductor AC and DC current?

Hi,

Still missing information about the true inductor currents.

*****
Thus time (and nerve) consuming assumptions:
I assume there are two (or three?) halfbridges. Each has it's own inductance.
All inductances are connected to the same node with the capacitor.
From halfbridge to halfbridge there are just two inductors in series. Giving an R_DC of 2x17.8mOhm = 35.6mOhm.
You just need 100mV of voltage to get 3A.
At 12V this means a duty cycle deviation of 0.1V/12V = 0.008
With your switching frequency of 350kHz this is a timing deviation of just 24ns.

What is the timing specification of your halfbridges?

Klaus

I forgot to draw the 100uF series capacitor.

I've been measuring the current through the transducer as well as calculating the efficiency of the amplifier. The whole system is impedance matched at 40kHz so the inductor current should be 27.5 times the transducer current. I tried to put a shunt resistor in series with the inductor to get an idea of the inductor current. While this scewed the system a bit and with the unknown inductance of the resistor I can't be that sure of my measurements, but I got around 3A rms, which was in the ballpark of my previous calculations.

I'm looking at some power inductors. They really don't specify much, neither the AC resistance, core loss or core material used. I found some info about the WE-HCC series from Wurth, they are bragging about low core losses at higher frequencies, saying that they are using a ferrite material that has significantly lower core losses than iron powder material that are usually used in power inductors. The datasheet of these particular inductors gives no AC or core loss information either though. I'm a little bewildered as I don't know what type of inductor I should be looking for. It should be able to handle about 6A rms, but probably only for short bursts.

Hi,

Still it's unclear where and how you measured the 3A.
Also you didn't show a complete circuit.

Thus it's just a waste of time to "guess this or guess that".

Klaus

He told it's not measured but calculated from secondary current. But he didn't tell how 2W power dissipation have been determined. So it could be

- higher primary current
- lower actual power dissipation

AC losses of this old inductor series like Coilcraft MSS or Wuerth PD are known to be high, there are better types like Coilcraft SER or Wuerth HCF. The present application should however work well with MSS1278 - if the reported parameters are correct.

KlausST: I posting the complete circuit is a bit irrelevant as the topic of this thread was really only about core loss vs. frequency and whether core loss could be a substantial factor at 40kHz, a question I have gotten a satisfactory answer to :)

I appreciate the replies! I will order a few different types of inductors and do some tests, perhaps even reduce the inductance a bit :)

Hi,

It's useles to give incomplete schematics, too. It takes time and thus it demotivates the ones who want to help you.

For sure we don't need to see how your microcontroller XTAL circuit looks like. But we need to see the complete and correct circuit that is responsible for the inductor currents.

You said you measured the amplifier output with a network analyser.
Voltage or current or both?
What is the output?
Directly at the half bridge, after the inductor, before or after the 1,2uF capacitor....

Klaus