Les Cargill wrote:> Marcel Mueller wrote: >> On 26.05.18 07.40, Randy Yates wrote: >>> I was miffed initially by this statement since, as far as I know, >>> there is nothing inherent in wavelength that impacts how RF waves >>> travel through space. >> >> If you are talking about vacuum then yes. In all other media the >> velocity of propagation depends on the frequency. E.g. water >> molecules in the air interact frequency dependent. >> >>> But I guess this was just a way (a confusing one, IMO) of referring >>> to the wavelength dependency of antenna aperture, as explained >>> nicely in this article on the Friis equation? >> >> The coupling of the antenna to the free space also introduces a >> frequency dependent group delay. > > All necessary apologies in advance. > > All group delay is inherently frequency dependent: > > " Group delay is the actual transit time of a signal through a device > under test as a function of frequency." > > http://na.support.keysight.com/pna/help/latest/Tutorials/Group_Delay6_5.htm > > A reasonable definition.But unfortunately dead wrong because it ignores causality. Group delay != true delay, in general. Group delay is d phi / d omega, and is useful as a leading-order approximation to how a nice wide smooth pulse propagates through a network. It's exactly analogous with group velocity in radio or optical propagation, which is d(omega)/d k, where k is the wave vector. You can see the distinction in two ways. First, group delay can be negative, which true delay cannot. Second, networks can have group delay without having true delay. You can undo the effect of a 1-pole RC lowpass with an RC highpass, for instance.> > I have the conceit that I'm not picking nits here so much as heading > off one potentially confusing interpretation of that > sentence :) The� "quantifiers" for "a group delay" sort of leaves > the phrase "for all group delay" dangling. > > >> And last but not least a short >> distance link has some frequencies with poor performance due to >> eigenvalues of the overall geometry. >> > > Aka comb filtering/multipath/cosite interference? > >> >> Marcel >Cheers Phil Hobbs
Wavelength Dependency in RF Propagation?
Started by ●May 26, 2018
Reply by ●June 3, 20222022-06-03
Reply by ●June 3, 20222022-06-03
(This weirdly came up as a new message--silly me.) Phil Hobbs wrote:> Les Cargill wrote: >> Marcel Mueller wrote: >>> On 26.05.18 07.40, Randy Yates wrote: >>>> I was miffed initially by this statement since, as far as I know, >>>> there is nothing inherent in wavelength that impacts how RF waves >>>> travel through space. >>> >>> If you are talking about vacuum then yes. In all other media the >>> velocity of propagation depends on the frequency. E.g. water >>> molecules in the air interact frequency dependent. >>> >>>> But I guess this was just a way (a confusing one, IMO) of referring >>>> to the wavelength dependency of antenna aperture, as explained >>>> nicely in this article on the Friis equation? >>> >>> The coupling of the antenna to the free space also introduces a >>> frequency dependent group delay. >> >> All necessary apologies in advance. >> >> All group delay is inherently frequency dependent: >> >> " Group delay is the actual transit time of a signal through a device >> under test as a function of frequency." >> >> http://na.support.keysight.com/pna/help/latest/Tutorials/Group_Delay6_5.htm >> >> >> A reasonable definition. > > But unfortunately dead wrong because it ignores causality. > Group delay != true delay, in general. > > Group delay is d phi / d omega, and is useful as a leading-order > approximation to how a nice wide smooth pulse propagates through a > network.� It's exactly analogous with group velocity in radio or optical > propagation, which is d(omega)/d k, where k is the wave vector. > > You can see the distinction in two ways.� First, group delay can be > negative, which true delay cannot. > > Second, networks can have group delay without having true delay.� You > can undo the effect of a 1-pole RC lowpass with an RC highpass, for > instance. > >> >> I have the conceit that I'm not picking nits here so much as heading >> off one potentially confusing interpretation of that >> sentence :) The� "quantifiers" for "a group delay" sort of leaves >> the phrase "for all group delay" dangling. >> >> >>> And last but not least a short >>> distance link has some frequencies with poor performance due to >>> eigenvalues of the overall geometry. >>> >> >> Aka comb filtering/multipath/cosite interference? >> >>> >>> Marcel >> > Cheers > > Phil Hobbs-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
