chebyshev fc isnt 3db point?

[fousekis]

Hello, i simulated a Chebyshev LC low pass filter in ADS with passband ripple=0.5dB
At the designed fc, the measure S21 was 0.5dB instead of 3dB

Then i simulated an LC low pass filter in ADS with passband ripple=0.1dB
At the designed fc, the measure S21 was 0.1dB instead of 3dB


But when i use TLINs or MLINs, the measurements show 3dB in relative close frequencies to the designed frequency. (as it should)

Seems that analog filters always show their 3dB point equal to the passband ripple.
Please can you help me understand this weird behavior?

 

[snafflekid]

It is a matter of definition. A filter with a flat response needs a cutoff defined somewhere and the half-power point, -3dB, is useful.

In the case of the Chebyshev filter, because you can set the passband ripple, defining the cutoff frequency at the passband ripple makes sense. The objective is to set a maximum amount of ripple in the passband and -3dB would not meet that requirement.

 

[fousekis]

thank you for the reply, but

he signal itself doesnt know the type of filters we design.
I mean, lets say that rejection band starts at the frequency which is equal to the passband ripple. If i actually wanted to attenuate that freq and some others after that, what should i do?

Design the filter with sligthly higher cutoff?

Wouldnt it have been easier for the chebyshev filters to go with the flow?? In most of the books ive read, noone even mentions a thing about this, maybe you could send me some links?

 

[snafflekid]

One could have a requirement that a single pole RC filter have a maximum 0.1dB "ripple", and you can solve for this condition at the needed frequency to find what RC is needed.

here is the link to Wikipedia https://en.wikipedia.org/wiki/Chebyshev_filter
It says "(note: the common definition of the cutoff frequency to −3 dB does not hold for Chebyshev filters!)"

 

[LvW]

thank you for the reply, but

he signal itself doesnt know the type of filters we design.
I mean, lets say that rejection band starts at the frequency which is equal to the passband ripple. If i actually wanted to attenuate that freq and some others after that, what should i do?
.............

Snafflekid is 100% right.
Be careful, some authors and some filter tables define the -3dB point and some (in fact most of them) the ripple as the end of the passband.
But realize, that the rejection band does NOT start at the end of the passband.
There is a band between called "transition band". The attenuation starts at a frequency and a rejection (in dB) defined by the user. These specifications define the minimum order of the filter.