Help to find utilisable core for class D amplifier output stage inductor

[nn20]

Hello

I have registered in this forum to add to my knowledge from your experiences.
I have made a class D amplifier using ir2110 and 2 pairs of MOSFET half bridge topology.
switching frequency is 250 KHZ and symmetrical power supply voltage is about 60 volts per voltage rail. output load is 4 ohms and Maximum Peak RMS power should be about 400 watts. 0.47 uF CAP and 22uH inductor is needed for LC low pass filter,but core losses,frequency response and THD varied by trying different cores.
NB.circuit is based on IRS1500-D made by Detex Audio. **broken link removed**
My question is what kind of core (material,shape and size) should be used for class D output stage with following power ratings?

Regards
nn20

 

[goldsmith]

0.47 uF CAP and 22uH inductor is needed for LC low pass filter

Hi nn20
Why such values ? it will deliver 49KHZ cut off frequency ! i think you're referring to the audio amp . hence you should design your filter for 20 KHZ ! ( or 22 KHZ )
Well , for selecting the core , i think a PQ4040 will handle your aim as well .
Best Wishes
Goldsmith

 

[nn20]

I know cut-off frequency is above the standard frequency you have told, but they set these values that seems incorrect! maybe their ears could hear the ultrasonics! :razz:

20-22 KHZ would be suitable for full range audio amplification,but i am to use this amp for subwoofer and input FREQ range from 15 up to 200 hertz filtered by an op-amp LPF circuit. so do you prefer to limit the cut-off frequency for 200 hertz, assuming that the input frequency range of amplifier is below 200 hertz or not?
if yes how to choose values for C and L (according to L/C ratio)?
as the optimum value for the inductor is L= RL/2πfC (RL= speaker impedance ) then we could calculate C by C= 1/((2πfC)^2) • L)
Q factor should be checked to be at the range of 0.6-0.8 by Q= RL√(C/L)
what's your idea?

 

[FvM]

49 kHz cut-off doesn't sound completely wrong for a second order filter, if you'll want to avoid noticeable magnitude and phase deviation in the audio range.

Another point is size minimization of filter components, particularly inductors. This might suggest not to reduce the cut-off frequency, even for a subwoofer. Ferrite core inductors always involve a certain amount of non-linearity that can easily dwart the THD figure of a good class-D amplifier.

 

[goldsmith]

I know cut-off frequency is above the standard frequency you have told, but they set these values that seems incorrect! maybe their ears could hear the ultrasonics!

I hope you know about dissipation and and HF components and sidebands of SPWM signal . and about response of your loud speaker . your speaker didn't design for such frequencies !

ou prefer to limit the cut-off frequency for 200 hertz, assuming that the input frequency range of amplifier is below 200 hertz or not?

Of course .
But you should normalize the impedance of your filter . as you probably know such a filter has an out put impedance . you should design it for your desired range of impedance .
Do you know how to design it for a special impedance ?
best Wishes
Goldsmith

 

[nn20]

Of course .
But you should normalize the impedance of your filter . as you probably know such a filter has an out put impedance . you should design it for your desired range of impedance .
Do you know how to design it for a special impedance ?
best Wishes
Goldsmith

what you mean by output impedance? you mean load impedance which is speaker(s)? if yes , it's 4 ohms.

another question
what's your idea about minimising cut-off frequency below 22 KHZ? this may put core on pressure to force to filter a high power bandwidth which may increase core losses and THD.
i have seen many class d subwoofer amplifiers in half or full bridge modes which their LPFs were designed for full range audio but input FREQ were filtered by an active low pass filter to reach the desired frequency response for driving subwoofers.

I wrote the formulas in my previous post for calculating L and C for a specific load impedance(RL).

if I could ascertain about cut-off frequency , I could calculate the values and check them with you.

 

[goldsmith]

what you mean by output impedance? you mean load impedance which is speaker(s)? if yes , it's 4 ohms.

Hi again
I hope you know about theorem of a 2nd order LC filter and it's responses instead of different loads ! your filter should be able to support RL=4 ohms . do you know about it ? you can't simply put any value for your filter with your desired range of cut off frequency ! normalized impedance of that filter is pretty important .

what's your idea about minimising cut-off frequency below 22 KHZ? this may put core on pressure to force to filter a high power bandwidth which may increase core losses and THD.
i have seen many class d subwoofer amplifiers in half or full bridge modes which their LPFs were designed for full range audio but input FREQ were filtered by an active low pass filter to reach the desired frequency response for driving subwoofers.

The best idea is using a filter to allow to pass all of the audio range in out put and then using a low power N order filter in your preamplifier ( before comparator )

this may put core on pressure to force to filter a high power bandwidth which may increase core losses and THD.

Do you know what issue will increase THD in a class D amplifier??! why you think so ?

I wrote the formulas in my previous post for calculating L and C for a specific load impedance(RL).

I told you before ! you should design your filter to be able to support special value of out put load !
And it's design procedure is different ! ( consider value of zeta= 1 in your design process , then your filter will be normalized in your desired impedance ) .
Good Luck
Goldsmith

 

[nn20]

Hi again
I told you before ! you should design your filter to be able to support special value of out put load !
And it's design procedure is different ! ( consider value of zeta= 1 in your design process , then your filter will be normalized in your desired impedance ) .
Good Luck
Goldsmith

Hi
could you tell me why you consider damping factor in calculation ? what if we consider quality factor? for instance quality factor of 1/√2=0.707 would be critically damped which is ideal for this type of filter but damping factor would be=0.707 !
but when we consider damping factor=1 , Q would be=1/2
now could you clarify what should I do for it?

 

[goldsmith]

Hi
could you tell me why you consider damping factor in calculation ? what if we consider quality factor? for instance quality factor of 1/√2=0.707 would be critically damped which is ideal for this type of filter but damping factor would be=0.707 !
but when we consider damping factor=1 , Q would be=1/2
now could you clarify what should I do for it?

Hi
I though you know about filters hence i didn't describe it and i've preferred to notice about damping factor ! but now i think i've to tell you more .
A 2nd order LC filter will have different responses instead of different loads .
For instance , consider , please , that you've designed such a filter and then you don't have any load in your out puts . do you know what will happen ? out put voltage will become higher and higher ( something around some hundred or some thousand volts for high power systems ) and when you put load you will see your out put voltage is lower than your desired value . it is why you should consider damping factor in your design process .
the best responsibility for a 2nd order LC filter for your purpose is Zout= 0.5 sqrt ( L/C ) .
Best Luck
Goldsmith

 

[nn20]

Hi again

I Know these you said Mr goldsmith. but considering damping factor or quality factor was problematic for me.however,thanks for your advice.

please take a look at this paper (shared link) and then tell me your idea.

link has been changed due to technical problems.

new link: **broken link removed**

Regards
nn20

 

[goldsmith]

Hi again
Well , as i saw your literature it seems you are familiar with somethings in filter design and it is inspiring !
But A question , why you've tried to fine Q value ? if doesn't require in this process !
Your calculated values are correct . congratulations for that !

but considering damping factor or quality factor was problematic for me

A guidance about damping factor :
If you consider value of zeta , around 1 , then zout will be 0.5sqrt(L/C) i hope you are able to write relation between zeta and zout ( it's responsibility ) .
Any where , now your filter is with good values , L=57uH and C= 900nF .
So , is your problem solved now ? any other question ?

Best Wishes + Good Luck
Goldsmith

 

[nn20]

Hi again

now your filter is with good values , L=57uH and C= 900nF .
So , is your problem solved now ? any other question ?

Best Wishes + Good Luck
Goldsmith

Hi.alright?

calculations seem to be done correctly, but I am waiting for the others to share their experiences about core selection except pq4040 which you told before to know more about practical cores.

so any other suggestion for inductor core used in class D half bridge 2nd order LPF would be appreciated.

thanks bro
nn20

 

[goldsmith]

For core selection , i've took these pictures which dealing with core selection , however they are for SMPS but they can be used in this case too because a class D amplifier is like an SMPS ! ( i hope you knew it )





Good Luck
Goldsmith

 

[nn20]

once I tried ETD 30 with 1mm air gap and it was nice. unfortunately it was dropped down and broken.

what about RM cores ? for example RM12?

what about Iron powder cores? like T157-2?

nn20

 

[goldsmith]

once I tried ETD 30 with 1mm air gap and it was nice. unfortunately it was dropped down and broken.

what about RM cores ? for example RM12?

what about Iron powder cores? like T157-2?

nn20

Notice : Core selection has some limitations ! for example frequency of operation - loss - permeability - allowed magnetic flux and ... etc . how you are selecting your cores ? what are your criteria ?

 

[nn20]

Notice : Core selection has some limitations ! for example frequency of operation - loss - permeability - allowed magnetic flux and ... etc . how you are selecting your cores ? what are your criteria ?

I tried it without any criterion! now i'm here to ask which parameters should be considered more for core selection .

as I told before , switching frequency is 250 KHZ so other considerable parameters could be core loss ,and behaviour (linearity or not)

because the output sound Sinusoidal waves' fidelity is important and as you surely know, no other harmonics should not be added to output sound, just this.

 

[goldsmith]

(linearity or not)

for being sure about linearity of an inductor , a gap space will guarantee it !

because the output sound Sinusoidal waves' fidelity is important and as you surely know, no other harmonics should not be added to output sound, just this.

I hope you know your sound or a music , isn't a sine wave ! according to the fourier series each un sinusoidal wave is consists gang of sine waves .
About adding unwanted harmonics yea you're right . but THD is most related to the triangle wave and it's precision . and it also depends on how much dead time you've created . and of course it depends on your mosfet driver and your mosfets too . but effect of filter on THD isn't as high as mentioned parameters . but most of the created THD is because of dead time and triangle wave .
To avoid complexity , i think those tables will handle your aim as well .
Best Luck !
Goldsmith