Class D - Output Inductor Calculations

[ElecDesigner]

I'm look at an existing design for a high power Class D audio amplifier type circuit. Its not for driving speakers but other than that the circuit is the same.
I am getting failure if the load is higher than a certain value (e.g. less resistance).

At the moment I'm trying to work out the current rating of a custom made toroid that a past engineer has designed. I am having trouble understanding which formula I should use. I thought the general EMF equation - but its got to be rated to current in some way? Also I assume we use the PWM frequency for calculations, not the output frequency?

 

[KlausST]

Hi,

the most problematic with those inductors is: saturation.
You need to avoid it.

Are you aware that you gave almost no usable information?
* No values, no circuit,
* "audio, but not for driving speakers",
* "I am getting failure" but you don´t say what failure
* and so on ....

It's like a riddle ...

If you need help, you need to provide informations first.

Klaus

 

[ElecDesigner]

I agree that I did not provide much information. I was just looking for a general formula for calculating under what scenario the inductor will saturate.

The circuit is full bridge switching, differential LPF. Output on LPF drives a step up transformer that in turn drives the load.
Getting a FET failure - could be any number of things causing it, but at the moment I am just wondering a what current the inductor will saturate.

The core is a magnetic inc MPP. I would go by the AT graph on the datasheet, but that is for DC.......

 

[treez]

is that a non linear inductor core?....it sounds like it. They give you a graph of AL vs NI, and you have to use magnetic theory to calculate the inductance, and indeed the inductance actually varies as the current in the coil varies.

Basically , the crux of the matter, as you know, is that you have to make out your own B vs H graph...so firstly put a list of H values in rising order...and then calculate the B value for each H value.....but remember that at first, at each H value, you cannot immediately calculate B, but you can calculate dB/dH , because it equals uo.ur, the ur value having being pulled out of the table of AL and ur values that you made for yourself previously. You knew which ur value because it came from the AL value which corresponded to the relevant NI value on the datasheet supplied AL vs NI graph.

You have to work it through..remembering that at every NI value, there is an AL value, which you read from the graph......now AL = L(nH)/n^2
....now Reluctance = N^2/L.

...And also reluctance = [core length]/ [uo.ur.CoreArea]

So you can work out the ur value at each AL value.

So anyway, if you want to work out inductance at a current of 3 Amps say, then you find N.I for 3 Amps.....you then read off the AL value corresponding to this.....and then get the inductance from that.

The attached shows how i calculated the B value at any I value in the coil.

Claculting B and L for a particular current in the N turns of a non linear inductor is a bit of jiggery pokery....but if you keep churning the magnetic equations, then youll get there..........i included a maths doc which lists the equations needed

 

[BradtheRad]

Also I assume we use the PWM frequency for calculations, not the output frequency?

As a guideline, select a Henry value that starts to impede AC at your desired rolloff frequency. You want output to swing up and down in a middle zone between the supply rails.
Too small a value allows strong signals to reach the supply rails, which is like clipping.

Too large a value has a 'choke' effect on your AC output, restricting its travel up and down. You don't get full use from your supply voltage.

Do you also have a capacitor across the load? Select its value so it supplements the filtering role. It may also be needed to perform an amount of power factor correction, depending on your switching topology and inductor value.

 

[FvM]

The core is a magnetic inc MPP. I would go by the AT graph on the datasheet, but that is for DC.......

Peak current (Idc + Iac,pk) has to be checked against saturation limits. In case of a class D amplifier or AC inverter, Idc corresponds to the low frequency peak current.

MPP cores have rather high AC losses. Maximum core flux will be probably set below saturation to acceptable losses.

 

[Easy peasy]

If you have failure at light load, this is likely due to a resonance in your output filter causing "capacitive" switching in the mosfets, leading to pretty quick overheat & fail...

 

[FvM]

"Poking around in the dark" could be an appropriate title for the thread up to now. Too few information about the circuit operation conditions, no significant calculations or even measurements.

To observe possible saturation, you'll best monitor the inductor current with a current transformer. Saturation will show as an increasing and non-triangular current.

Observing "failure if the load is higher than a certain value" can be due to simple overload, dV/dt rating exceedance, shoot-trough or many other parasitic effects.

Presuming the filter cut-off frequency is sufficiently below PWM frequency, you get inductive output current at light load and respectively soft switching with low losses. If the DC part exceeds the inductive current, bridge operation changes to hard switching for one edge, causing more losses and EMI.