group delay parabolic
anybody knows,please!
Group delay is group delay, ie Δphase/Δf.
When you plot group delay on a graph, some group delay's look linear, some look parabolic (certain filters), most is some combination of the two. Usually communications systems can stand a fair amount of parabolic group delay, but can only tolerate less linear group delay. In other words, the shap of the graphed group delay matters.
Lets say you measure some rf component and its group delay looks mostly "linear" with a negative slope. That means the higher frequency energy gets much more phase shift per Δf than the lower frequency energy. If it is a phase modulated signal, think how that would screw up the modulation constellation. Higher frequency components rotate more, while lower frequency components rotate less. If it was instead a flat group delay (like an ideal transmission line), all frequency components rotate by the same amount, and the decision constellation is unchanged. You could only tolerate so much group delay slope before you started to make bit errors.
thank you,but do you think the group delay graph look likes linear or parabolic ,so it is called.
Most filters are almost constant in the center and parabolic-like at the edges of the passband.
You can easily calculate the delay from the phase of pole pairs. Then add the delays from all of the pole pairs of a filter. This is easily done with a spread sheet program.
Find attached a very good document on Group Delay Explanations and Applications
but where is document,thanks!
Or Download from here...
www.radiolab.com.au/DesignFile/DN004.pdf
I have download the document,but it's useless for my questions.thanks!
http://206.223.8.10/linksite/manuals.../delaydeg7.pdf
thanks a lot,when i measure a BPF with vector net analyzer,can anybody tell me the measured group delay is linear,parabolic or ripple?if it is ripple ,how can i measure the linear and parabolic?thanks!
I am not sure what a linear looks like other than linear, but as for ripple or parabolic. In a practical sense it goes like this.
You measure the peak to peak variation over the band of interest, typically the pass band.
Added after 5 minutes:
there are several methods that supply these responses. In a gausian filter for example, each section grabs at a range of frequency and so on down the filter
this gives a ripple but a narrow range of variation on a narrow band filter.
In a ladder network type filter, the flat part of the pass band gives theflatest response, whereas the 3 dB points give a parabolic shape, basically if your pass band is flat, your response is flat, if you have a narrow filter you must implow some scheme to grab at the frequenceies so that they are alll delayed about the same over the band of interest
is this any help, oh and if you have a low pass lets say then it will be pretty much linear over the whole pass band
thank you,but whether your meaning is that the BPF have not linear or parabolic group delay ?