How to determine the cut-off frequency of lowpass filter? It appears on textbooks suggesting around 30-60 kHz, so does the google. But, what is the method or calculation to find the actual optimum cut-off frequency?
Last edited:
Any frequency above audio frequency and based around the oscillator frequency and second harmonic is used.
F=2pi RC.
F=2pi RC.
Hi JonSnell, to my understanding, the majority of output filter is passive LC circuit, rather than RC. Do you agree?
To get the cutoff frequency, multiply the inductance (H) by the capacitance (F) and take the square root of that product. Then take the reciprocal and divide that by (2*PI) - the result is the cutoff frequency in Hz.
Example - 22uH and 680nF gives ~41kHz.
Example - 22uH and 680nF gives ~41kHz.
Hi abraxalito, sorry to make everyone confused with my unclear question. I was intended to ask that how to select the cut-off frequency? In brief, why 41 kHz was selected?
Its selected so that there's a minimum amount of loss at 20kHz but also so there is sufficient attenuation of the carrier frequency.
Another factor to consider is - how wide a range of speaker impedance is expected? Because the speaker resistance provides damping.
Another factor to consider is - how wide a range of speaker impedance is expected? Because the speaker resistance provides damping.
Last edited:
If we consider on a range of speaker's impedance, isn't it good to have the cut-off frequency as highest as possible? So, the result of damping at the knee or cut-off frequency won't much effect the audio spectrum. What's your opinion?
Yes, I agree with that view. But higher cut-off frequency means lower efficiency as more of the carrier power is dissipated in the speaker. So there is a trade-off.
Sorry Jon but that equation is incorrect. You're missing the inversion.
Jan
Inductor for class D, must be set meet fsw, then cap to follow to get cut off of desired sonic frequency.
Most of paper out there mostly only suitable for small, full range, high speed fsw typically 300-400kHz.
Class D inductor is so critical, which inductor sizing is key of robustness of class D. If you follow textbook mostly tell you inductor 22uH and cap 470n, your class D will be long last if your fsw at around +/- 300kHz.
If your fsw i.e. 250kHz or less, your inductor will soon over-temperature, melt your email wire and loss your inductance, degrading sonic quality and soon blow your mosfet. Surprisingly no book tell you. That is why everyone make different inductor sizing (without tell you why).
Most of paper out there mostly only suitable for small, full range, high speed fsw typically 300-400kHz.
Class D inductor is so critical, which inductor sizing is key of robustness of class D. If you follow textbook mostly tell you inductor 22uH and cap 470n, your class D will be long last if your fsw at around +/- 300kHz.
If your fsw i.e. 250kHz or less, your inductor will soon over-temperature, melt your email wire and loss your inductance, degrading sonic quality and soon blow your mosfet. Surprisingly no book tell you. That is why everyone make different inductor sizing (without tell you why).
Hi kartino,
Did this statement refer to the second-order filter? How about the fourth-order filter? What's the method to construct it?
Does it occur only if fsw is set to lesser than 250kHz? Do you mean that the higher fsw is the better? Is there any disadvantage if fsw is set too high?
I cannot thank you enough for your kindly help.
Did this statement refer to the second-order filter? How about the fourth-order filter? What's the method to construct it?
Does it occur only if fsw is set to lesser than 250kHz? Do you mean that the higher fsw is the better? Is there any disadvantage if fsw is set too high?
I cannot thank you enough for your kindly help.
You do not need 4th order in Class D. Some manufacturer want to cut off totally ripple from output. Some people say ripple is not good and audible but that is not necessary. 2nd order filter is enough.
If your inductor too much means your fsw too high, you will get cool mosfet and inductor, but when you turn amp at higher load, badly pumping effect will happen. Your supply POS and NEG voltage become very bad get unbalance voltage, which can blow elcos and mosfet low side will be hotter than high side and also can blow easily.
I have did so much trial and make a table for myself for inductor vs fsw. I have been looking for any formula from internet, but so far only tell about cut off.
If your inductor too much means your fsw too high, you will get cool mosfet and inductor, but when you turn amp at higher load, badly pumping effect will happen. Your supply POS and NEG voltage become very bad get unbalance voltage, which can blow elcos and mosfet low side will be hotter than high side and also can blow easily.
I have did so much trial and make a table for myself for inductor vs fsw. I have been looking for any formula from internet, but so far only tell about cut off.
Last edited:
I made tutorial for indonesian. My brother OVI Raj, translate to english.
This is based on trial only which finally you must fine tuning. But the point is: Inductor size is critical. Also material of core. Beside I did trial only for Pro Audio.
This is based on trial only which finally you must fine tuning. But the point is: Inductor size is critical. Also material of core. Beside I did trial only for Pro Audio.
Last edited:
The inductor likes high fsw because of low ripple current.
However, switching losses will increase because it does require power to turn off/on a mosfet gate.
You may find these helpful:
https://e2e.ti.com/cfs-file/__key/c...-files/6/Class_2D00_D_5F00_LC_2D00_Filter.pdf
LCFILTER-CALC-TOOL Circuit design | TI.com
However, switching losses will increase because it does require power to turn off/on a mosfet gate.
You may find these helpful:
https://e2e.ti.com/cfs-file/__key/c...-files/6/Class_2D00_D_5F00_LC_2D00_Filter.pdf
LCFILTER-CALC-TOOL Circuit design | TI.com
According to Nyquist theory, it recommended that the sampling rate is pointed at 96 kHz for a high sound quality. What if I were to set the cut-off frequency of the output filter at 96 kHz? As stated above in post #8, will the efficiency be reduced too much? And will it effect the tonal balance of the sonic?
That 96kHz is for digitize sampling in DAC recording.
In class D is not a digital mechanism but it is modulated squarewave converted to sonic by inductor and capacitor. Inductor block, store and release energy while the cap cut the ripple and shaping better the sonic output meet audio signal requiremen.
This process including the switching itself require optimum switching speed, considering switching losses, magnetic losses and proper transient timing to keep energy not too leading or lagging. So that class D is very complex mechanism.
In class D is not a digital mechanism but it is modulated squarewave converted to sonic by inductor and capacitor. Inductor block, store and release energy while the cap cut the ripple and shaping better the sonic output meet audio signal requiremen.
This process including the switching itself require optimum switching speed, considering switching losses, magnetic losses and proper transient timing to keep energy not too leading or lagging. So that class D is very complex mechanism.
I have read this document in very early designing class D, but no such information as I mentioned above.
Most document only tell about corner frequency and peaking.
Peaking is actualy the big problem, cause unstable and damage specially for Big Class D which most of people fail. Eliminate peaking is rarely discussed, but I found already eliminate peaking:
1. Carefully sizing inductor as mentioned above.
2. Use higher dummy load will safe amp during speaker disconnected.
3. Add anti peaking diode after inductor
4. The most important: implement post filter feedback