The common class d output filter inductance overheating problem.

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have read in a few threads about overheating of the inductor in the output filter.

I have found using an inductor with a much higher current rating helps here.

I also noticed with a semi-discrete class d amplifier that the class d frequency also made a difference. I guess this is because a series LC is a short circuit at the resonant frequency and the further you are away from the resonant frequency the less short circuit current there.

At 100KHz I found my inductor got to 200 degrees C.
At 150KHz it got to 170 degrees C.
The resonant frequency of the output filter was 40KHz.
I can clearly see a drop in heat at the higher frequencies.

I need a faster op amp in my triangle wave generator to get the frequency higher for further tests.
It is true but you seem to include only the "electric heat" and not the "magnetic heat".

Total heat = core loss (caused by ac) + copper loss (caused by ac&dc).

Core loss reduces with frequency (which permits smaller core in SMPS than tfr) and increases with ampere turns (resulting in larger cores for higher power applns).
Copper loss increases with frequency due to skin effect (increase in AC resistance with freq) and reduces with thickness of wire.

Hence, a good inductor is one in which both losses meet.
Also, higher current choke => larger core & heavier wire => lesser heat.
Good, I was looking for such a dedicated thread that would discuss LC filters of ClassD amp.
Can we discuss this spiting in smaller topics ? Like, 1) when Ideal Inductor and Capacitor is used with Ideal Resistive load. 2) Understanding and discussing copper losses. 3) Then the magnetics. Together this thread can be A to Z solutions and understanding on the subject.

We can use Spice modeling for each so easier to visualize.
Recently i've found that the same core (MPP 60u toroidal) produced a lot of heat with 1,2mm wire wound lightly loose (50°C) than the same core with 1mm wire wound very tight on the core (no space from wire to core) and epoxy glue on the outer surface (i've done that with the hope to limit vibration of the wire), the latter works nicely at 38-39°C in the same amplifier. Obviously this needs further investigation but if someone wanna try this i would be glad to recive a feedback.

It is core heating, some spot of the core side has over density.
I have a way but many people not accepting this. Normally over density happening at near the coils and less at the center of inductor. To make it distributed well, use high permeability core at the center and lower at the edge, and make more flux space area (wider inductor). Avoids to use toroids.

Here is my noheatatall inductor picture.
Last edited:
Some 10yrs ago my employer had asked me to design SMPS and I had no knowledge of magnetics domain. Also the product was to go in huge quantity when commercialized. So had a big pressure. during this time I had some readouts on net and some discussion with the ferrite core dealers. but after that I never even spell ferrite. my sharing is based on that knowledge gathering.

Firstly heating of inductor does not necessarily mean its a problem. Heating could be core losses the coil is allowed to. First step would be knowing the frequency of operation, based on this core material is selected. Then you need to know the operating voltage. waveform, like Sine wave or square wave.. etc. current flowing through the inductor. based on this electrical parameters the capacity / size of core would be determined. We have to remember that when core saturates rest of the energy is converted to heat.
Although the above are only a basic lines of design understanding. we can still design a good efficient inductor using ready tools available on net. Here is one of the links.
I bought an Ebay Class D amp which uses IRS2092 chip & IRFB4227 mosfet.
As per the specs listed it is rated 350w 4Ω DC supply +55v -0- -55v.

Used a 480VA 40v-0-40v transformer with DC filtering of 4x 10,000uf. Test speaker is a subwoofer rated 400w 4Ω.

On connecting the unit the speaker is silent. When the finger is touched at the audio input noise can be heard correctly through the speaker, however the inductor coil is turning really hot

An externally hosted image should be here but it was not working when we last tested it.
The test speaker is 400w 4Ω from Peerless, India

Alas ! After running the fingertip test at audio input for continuous 3 minutes the inductor coil burned up ! Guess the VA with speaker load is too much for the amplifier board to handle.

Photos with the coil removed -

An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.
Checked the coil. It is 22uH. The spec list describe it as 20A capable, I doubt it can handle more than 3A. Am not aware of the oscillating frequency of the amp since currently I don't have access to an oscilloscope.

480AV is indeed overwhelming for the amplifier because initially I did a test with
a 20W 8Ω with a series resistance of 20Ω 2W (to reduce load) & still the coil was getting hot.

Intend first to replace the coil to get the board working. If not successful, try replacing the mosfets.

Am planning to replace coil with 2x 22uH 6A toroid coils connected in parallel.
Though effective induction should decrease by the equation
L (total) = (1 / (1/L1)+ (1/L2)) ≈ 11uH, it should suffice for the 4Ω subwoofer speaker.

There is a shunt capacitor 0.047uf also to suppress high frequency. Since the application is a subwoofer / low pass amplification I'll try & increase it to around 5uf non-polar polypropylene see whether there is any adverse effect / low frequency <500Hz audio suppression.
Last edited:
Am planning to replace coil with 2x 22uH 6A toroid coils connected in parallel.

Besides current load there is another effect which causes heat in the choke.
AC flux density. ==> Core losses. Darkfenriz pointed to this in one of the early posts of this thread.
Your description would fit to this mechanism.

In case your parallel connection shows similar choke heating, then you can be sure that the choke is not suffering from the current stress but from AC flux density in the core.
In this case you would need a choke with larger air gap and more turns.
For experimentation you may then also try the 2x22uH in series instead of parallel.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.