Hi All,
A silly question, but I want to make sure I am interpreting this properly. When the problem asks for 100 dB of attenuation at the switching frequency (as in 10.4b), I interpret that as being 100 dB below the converter (in this case roughly -40 dB)... I don't need to be at -100 dB at fs do I? That would be a gigantic capacitor...
Not a silly question... In fact, I'm perplexed myself with some statements in the book.
ReplyDeleteDr. Maksimovic, could you please confirm/correct? -Thanks!
1) Text P400, 1st paragraph, and p404 Figs 10.28 and 10.29: These all state the attenuation of the 330uH/470uF low-pass is 80dB at 250kHz. What's missing is a plot of the alleged attenuation, similar to Fig 10.28 for the circuit of Fig 10.29b. The voltage transfer function for 10.29b actually does simulate per 10.28, and gives the expected 80dB atten (-80dB gain) at 250kHz. When applied to Fig 10.29a circuit, this process gives -112dB gain, not -80dB, at 250kHz. This -112dB gain is also what I get applying the high-frequency asymptote method to Fig 10.29a to determine magnitude of freq resp at 250kHz. Am I missing something?
2. Olga, what I am doing until otherwise corrected:
Use the filter voltage transfer function as the basis for attenuation calculations, as discussed above, with the assumption that something will be corrected for the perceived problem with Fig 10.29a and 80dB vs 112 dB attenuation. Text P378 and problems 10.1, 10.2 speak of attenuation as the reduction of converter input ripple current from filter output to filter input. The filter is linear, passive, so this "reverse" current transfer function is equal to the "forward" voltage transfer function, which is why the voltage transfer function can be used to design for current attenuation. -- Note that for both cases the filter input is zero-ohm (source or short) and the output is open (includes current source).
******
Dr. Maksimovic, can you clarify/correct if the above is mis-leading? Again, thanks!
******
For problem 10.4, the voltage transfer function is just the familiar 2nd order LC low-pass. 100dB attenuation is -100dB gain, and the 250kHz fs is assumed to be well into the -40dB/decade high-freq asymptote response region, so (f1/fs)^2 = -100dB, or f1 = fs/sqrt(10^5). -- I haven't done the ac converter model yet, so I don't know if this choice of f1 conflicts with the converter ZD anti-resonance. Also, I don't see a use for the Gvd and Gvg transfer functions in the initial attenuation calculations. These functions have the switching ripple averaged out. (Of course the main point of chapt 10 is how to preserve these functions while providing converter input ripple reduction.)
"Attenuation" is the attenuation of the filter itself. This has nothing to do with converter transfer functions. Recall that the purpose of the filter is to attenuate switching harmonics from the converter input current to the input port where an external voltage source is connected. Similarly, the filter attenuates any disturbances from the external voltage source to the input port of the converter.
ReplyDelete