On 12/3/2009 1:30 PM, Clay wrote:> On Dec 3, 3:01 pm, "Michael Plante"<michael.pla...@gmail.com> wrote: >>> On Dec 3, 11:08=A0am, Jerry Avins<j...@ieee.org> wrote: >>>> Clay wrote: >>>>> On Dec 3, 12:32 am, "gravelhe"<grave...@gmail.com> wrote: >>>>>>> On Dec 2, 7:59=3DA0am, "gravelhe"<grave...@gmail.com> wrote: >>>>>>>> I need an optical FIR for experiment. where can i buy this kind >> of >>>>>> produc=3D >>>>>>> t? >>>>>>>> Thanks! >>>>>>> The hardware store. Get a piece of glass. Let the light reflect off >> o= >>> f >>>>>>> both the 1st and 2nd surface and interfere together. This is a >> simple >>>>>>> comb filter. You may want to look it up under its fancier names >> like >>>>>>> etalon or a Fabry- Perot interferometer. I assume you have a >>>>>>> monochromatic light source. >>>>>>> As Jerry asks, what kind of experiment do you need to do? >>>>>>> Clay >>>>>>> I need the optical transversal filter as an OCDMA Encoder/Decoder >> in >>>>>> time domain. Thanks- Hide quoted text - >>>>>> - Show quoted text - >>>>> Then the piece of glass suffices. >>>> Avoiding interference when summing the taps requires great -- probably >>>> unobtainable -- mechanical stability. >>>> Jerry >>> [...] >>> He >>> still hasn't revealed a whole lot of detail in what he needs >>> experimentally other than "time domain." >>> [...] >> I thought that initially. It turns out the first line of his response had >> an extra greater-than, so it was lumped into your response on Google, and >> it's therefore easy to miss: >> >> (part of your response, but not really) >> "I need the optical transversal filter as an OCDMA Encoder/Decoder in" >> >> (what you probably saw, and what I initially saw) >> "time domain. Thanks"- Hide quoted text - >> >> - Show quoted text - > > Thanks - I did not see the other part. The easiest way to have > multiple simultaneous channels is to use different colors of light > (WDM) and polarization diversity can be added to this as well. That is > what is done now. > > The OP may want to look into spatial filtering to see how optics can > be used to obtain nearly instantaeous fourier transformation and > matched filtering. > > ClayAs Jerry mentioned early on the thread, one can implement an optical correlator/filter with some Fourier Transform lenses and a mask. It winds up being the equivalent of a FIR filter, without having to deal with the wavelength-related issues of the pipe taps, combiner, etc. Not sure why that wouldn't work for this application, but maybe I missed something. -- Eric Jacobsen Minister of Algorithms Abineau Communications http://www.abineau.com
optical FIR product
Started by ●December 2, 2009
Reply by ●December 8, 20092009-12-08
Reply by ●December 9, 20092009-12-09
On Dec 8, 7:53�pm, Eric Jacobsen <eric.jacob...@ieee.org> wrote:> On 12/3/2009 1:30 PM, Clay wrote: > > > > > > > On Dec 3, 3:01 pm, "Michael Plante"<michael.pla...@gmail.com> �wrote: > >>> On Dec 3, 11:08=A0am, Jerry Avins<j...@ieee.org> �wrote: > >>>> Clay wrote: > >>>>> On Dec 3, 12:32 am, "gravelhe"<grave...@gmail.com> �wrote: > >>>>>>> On Dec 2, 7:59=3DA0am, "gravelhe"<grave...@gmail.com> �wrote: > >>>>>>>> I need an optical FIR for experiment. where can i buy this kind > >> of > >>>>>> produc=3D > >>>>>>> t? > >>>>>>>> Thanks! > >>>>>>> The hardware store. Get a piece of glass. Let the light reflect off > >> o= > >>> f > >>>>>>> both the 1st and 2nd surface and interfere together. This is a > >> simple > >>>>>>> comb filter. You may want to look it up under its fancier names > >> like > >>>>>>> etalon or a Fabry- Perot interferometer. I assume you have a > >>>>>>> monochromatic light source. > >>>>>>> As Jerry asks, what kind of experiment do you need to do? > >>>>>>> Clay > >>>>>>> I need the optical transversal filter as an OCDMA Encoder/Decoder > >> in > >>>>>> time domain. Thanks- Hide quoted text - > >>>>>> - Show quoted text - > >>>>> Then the piece of glass suffices. > >>>> Avoiding interference when summing the taps requires great -- probably > >>>> unobtainable -- mechanical stability. > >>>> Jerry > >>> [...] > >>> He > >>> still hasn't revealed a whole lot of detail in what he needs > >>> experimentally other than "time domain." > >>> [...] > >> I thought that initially. �It turns out the first line of his response had > >> an extra greater-than, so it was lumped into your response on Google, and > >> it's therefore easy to miss: > > >> (part of your response, but not really) > >> "I need the optical transversal filter as an OCDMA Encoder/Decoder in" > > >> (what you probably saw, and what I initially saw) > >> "time domain. Thanks"- Hide quoted text - > > >> - Show quoted text - > > > Thanks - I did not see the other part. The easiest way to have > > multiple simultaneous channels is to use different colors of light > > (WDM) and polarization diversity can be added to this as well. That is > > what is done now. > > > The OP may want to look into spatial filtering to see how optics can > > be used to obtain nearly instantaeous fourier transformation and > > matched filtering. > > > Clay > > As Jerry mentioned early on the thread, one can implement an optical > correlator/filter with some Fourier Transform lenses and a mask. �It > winds up being the equivalent of a FIR filter, without having to deal > with the wavelength-related issues of the pipe taps, combiner, etc. > > Not sure why that wouldn't work for this application, but maybe I missed > something. > > -- > Eric Jacobsen > Minister of Algorithms > Abineau Communicationshttp://www.abineau.com- Hide quoted text - > > - Show quoted text -Actually I was the one who mentioned Fourier transform optics, but it sounds like the OP needs temporal and not spatial filtering. The two lenses and mask method work great for spatial filtering. Now he can turn a temporal problem into a spatial problem by letting the light (temporally modulated) enter a light pipe that is sliced at an angle, thus the light emitted from the side of the pipe now has a spatial distribution that corresponds to the temporal one entering the light pipe. Do this with monochromatic light and using spatial masks in the fourier plane between two lenses can do your filtering for you. And you will of course need a space to time convertor at the other end as well. This problem really comes down to becoming adept at building optical things. Clay
Reply by ●December 9, 20092009-12-09
On 12/9/2009 9:01 AM, Clay wrote:> On Dec 8, 7:53 pm, Eric Jacobsen<eric.jacob...@ieee.org> wrote: >> On 12/3/2009 1:30 PM, Clay wrote: >> >> >> >> >> >>> On Dec 3, 3:01 pm, "Michael Plante"<michael.pla...@gmail.com> wrote: >>>>> On Dec 3, 11:08=A0am, Jerry Avins<j...@ieee.org> wrote: >>>>>> Clay wrote: >>>>>>> On Dec 3, 12:32 am, "gravelhe"<grave...@gmail.com> wrote: >>>>>>>>> On Dec 2, 7:59=3DA0am, "gravelhe"<grave...@gmail.com> wrote: >>>>>>>>>> I need an optical FIR for experiment. where can i buy this kind >>>> of >>>>>>>> produc=3D >>>>>>>>> t? >>>>>>>>>> Thanks! >>>>>>>>> The hardware store. Get a piece of glass. Let the light reflect off >>>> o= >>>>> f >>>>>>>>> both the 1st and 2nd surface and interfere together. This is a >>>> simple >>>>>>>>> comb filter. You may want to look it up under its fancier names >>>> like >>>>>>>>> etalon or a Fabry- Perot interferometer. I assume you have a >>>>>>>>> monochromatic light source. >>>>>>>>> As Jerry asks, what kind of experiment do you need to do? >>>>>>>>> Clay >>>>>>>>> I need the optical transversal filter as an OCDMA Encoder/Decoder >>>> in >>>>>>>> time domain. Thanks- Hide quoted text - >>>>>>>> - Show quoted text - >>>>>>> Then the piece of glass suffices. >>>>>> Avoiding interference when summing the taps requires great -- probably >>>>>> unobtainable -- mechanical stability. >>>>>> Jerry >>>>> [...] >>>>> He >>>>> still hasn't revealed a whole lot of detail in what he needs >>>>> experimentally other than "time domain." >>>>> [...] >>>> I thought that initially. It turns out the first line of his response had >>>> an extra greater-than, so it was lumped into your response on Google, and >>>> it's therefore easy to miss: >>>> (part of your response, but not really) >>>> "I need the optical transversal filter as an OCDMA Encoder/Decoder in" >>>> (what you probably saw, and what I initially saw) >>>> "time domain. Thanks"- Hide quoted text - >>>> - Show quoted text - >>> Thanks - I did not see the other part. The easiest way to have >>> multiple simultaneous channels is to use different colors of light >>> (WDM) and polarization diversity can be added to this as well. That is >>> what is done now. >>> The OP may want to look into spatial filtering to see how optics can >>> be used to obtain nearly instantaeous fourier transformation and >>> matched filtering. >>> Clay >> As Jerry mentioned early on the thread, one can implement an optical >> correlator/filter with some Fourier Transform lenses and a mask. It >> winds up being the equivalent of a FIR filter, without having to deal >> with the wavelength-related issues of the pipe taps, combiner, etc. >> >> Not sure why that wouldn't work for this application, but maybe I missed >> something. >> >> -- >> Eric Jacobsen >> Minister of Algorithms >> Abineau Communicationshttp://www.abineau.com- Hide quoted text - >> >> - Show quoted text - > > Actually I was the one who mentioned Fourier transform optics, but it > sounds like the OP needs temporal and not spatial filtering. The two > lenses and mask method work great for spatial filtering. Now he can > turn a temporal problem into a spatial problem by letting the light > (temporally modulated) enter a light pipe that is sliced at an angle, > thus the light emitted from the side of the pipe now has a spatial > distribution that corresponds to the temporal one entering the light > pipe. Do this with monochromatic light and using spatial masks in the > fourier plane between two lenses can do your filtering for you. And > you will of course need a space to time convertor at the other end as > well. This problem really comes down to becoming adept at building > optical things. > > ClayI think Jerry had mentioned that such things were used in radar processing, and then it went from there. My recollections are from my days at Goodyear where the original SAR processors were optical. In that case they turned temporal radar returns into an optical signal and did all the filtering and correlation (in two dimensions) optically with FT lenses and masks matched to the radar pulses. So clearly it's possible, using multiple methods, so, as usual, the tradeoffs must be assessed by the OP. Interesting stuff in any case. -- Eric Jacobsen Minister of Algorithms Abineau Communications http://www.abineau.com
Reply by ●December 9, 20092009-12-09
Eric Jacobsen wrote:> On 12/9/2009 9:01 AM, Clay wrote: >> On Dec 8, 7:53 pm, Eric Jacobsen<eric.jacob...@ieee.org> wrote: >>> On 12/3/2009 1:30 PM, Clay wrote: >>> >>> >>> >>> >>> >>>> On Dec 3, 3:01 pm, "Michael Plante"<michael.pla...@gmail.com> wrote: >>>>>> On Dec 3, 11:08=A0am, Jerry Avins<j...@ieee.org> wrote: >>>>>>> Clay wrote: >>>>>>>> On Dec 3, 12:32 am, "gravelhe"<grave...@gmail.com> wrote: >>>>>>>>>> On Dec 2, 7:59=3DA0am, "gravelhe"<grave...@gmail.com> wrote: >>>>>>>>>>> I need an optical FIR for experiment. where can i buy this kind >>>>> of >>>>>>>>> produc=3D >>>>>>>>>> t? >>>>>>>>>>> Thanks! >>>>>>>>>> The hardware store. Get a piece of glass. Let the light >>>>>>>>>> reflect off >>>>> o= >>>>>> f >>>>>>>>>> both the 1st and 2nd surface and interfere together. This is a >>>>> simple >>>>>>>>>> comb filter. You may want to look it up under its fancier names >>>>> like >>>>>>>>>> etalon or a Fabry- Perot interferometer. I assume you have a >>>>>>>>>> monochromatic light source. >>>>>>>>>> As Jerry asks, what kind of experiment do you need to do? >>>>>>>>>> Clay >>>>>>>>>> I need the optical transversal filter as an OCDMA Encoder/Decoder >>>>> in >>>>>>>>> time domain. Thanks- Hide quoted text - >>>>>>>>> - Show quoted text - >>>>>>>> Then the piece of glass suffices. >>>>>>> Avoiding interference when summing the taps requires great -- >>>>>>> probably >>>>>>> unobtainable -- mechanical stability. >>>>>>> Jerry >>>>>> [...] >>>>>> He >>>>>> still hasn't revealed a whole lot of detail in what he needs >>>>>> experimentally other than "time domain." >>>>>> [...] >>>>> I thought that initially. It turns out the first line of his >>>>> response had >>>>> an extra greater-than, so it was lumped into your response on >>>>> Google, and >>>>> it's therefore easy to miss: >>>>> (part of your response, but not really) >>>>> "I need the optical transversal filter as an OCDMA Encoder/Decoder in" >>>>> (what you probably saw, and what I initially saw) >>>>> "time domain. Thanks"- Hide quoted text - >>>>> - Show quoted text - >>>> Thanks - I did not see the other part. The easiest way to have >>>> multiple simultaneous channels is to use different colors of light >>>> (WDM) and polarization diversity can be added to this as well. That is >>>> what is done now. >>>> The OP may want to look into spatial filtering to see how optics can >>>> be used to obtain nearly instantaeous fourier transformation and >>>> matched filtering. >>>> Clay >>> As Jerry mentioned early on the thread, one can implement an optical >>> correlator/filter with some Fourier Transform lenses and a mask. It >>> winds up being the equivalent of a FIR filter, without having to deal >>> with the wavelength-related issues of the pipe taps, combiner, etc. >>> >>> Not sure why that wouldn't work for this application, but maybe I missed >>> something. >>> >>> -- >>> Eric Jacobsen >>> Minister of Algorithms >>> Abineau Communicationshttp://www.abineau.com- Hide quoted text - >>> >>> - Show quoted text - >> >> Actually I was the one who mentioned Fourier transform optics, but it >> sounds like the OP needs temporal and not spatial filtering. The two >> lenses and mask method work great for spatial filtering. Now he can >> turn a temporal problem into a spatial problem by letting the light >> (temporally modulated) enter a light pipe that is sliced at an angle, >> thus the light emitted from the side of the pipe now has a spatial >> distribution that corresponds to the temporal one entering the light >> pipe. Do this with monochromatic light and using spatial masks in the >> fourier plane between two lenses can do your filtering for you. And >> you will of course need a space to time convertor at the other end as >> well. This problem really comes down to becoming adept at building >> optical things. >> >> Clay > > I think Jerry had mentioned that such things were used in radar > processing, and then it went from there. My recollections are from my > days at Goodyear where the original SAR processors were optical. In > that case they turned temporal radar returns into an optical signal and > did all the filtering and correlation (in two dimensions) optically with > FT lenses and masks matched to the radar pulses.Don't forget the conical lenses. That was a really neat part!> So clearly it's possible, using multiple methods, so, as usual, the > tradeoffs must be assessed by the OP. > > Interesting stuff in any case.Very. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●December 9, 20092009-12-09
> > >gravelhe wrote: > > >> The optical transversal >> filter will be consist of a splitter,3 fiber delay lines(different >> length),and a coupler. optical pulse will be split into 3 parts, eachpart>> as the input of one fiber delay line, and then they will be combined >> together by the coupler after different delays in 3 fiber delay lines. > >There is a little problem here: you can add beams, but you can't >subtract them. There is no "negative" light. > >> There is still one problem. The optical pulse initially is not >> monochromatic, and the split pulse is still multi wavelength, so thefiber>> delay line will process several wavelengths pulse simultaneously. Insuch>> case,what should i pay more attentions to make such equipment? > >Well, if you want to do coherent combining of the beams, that makes >things much more difficult even if you use single mode fibers. > >VLV >Thanks a lot! What I need is incoherent combining!
Reply by ●December 9, 20092009-12-09
>On Dec 7, 10:55=A0pm, Vladimir Vassilevsky <nos...@nowhere.com> wrote: >> gravelhe wrote: >> > The optical =A0transversal >> > filter will be consist of a splitter,3 fiber delay lines(different >> > length),and a coupler. optical pulse will be split into 3 parts, eachp=>art >> > as the input of one fiber delay line, and then they will be combined >> > together by the coupler after different delays in 3 fiber delaylines.>> >> There is a little problem here: you can add beams, but you can't >> subtract them. There is no "negative" light. >> >> > =A0There is still one problem. The optical pulse initially is not >> > monochromatic, and the split pulse is still multi wavelength, so thefi=>ber >> > delay line will process several wavelengths pulse simultaneously. Insu=>ch >> > case,what should i pay more attentions to make such equipment? >> >> Well, if you want to do coherent combining of the beams, that makes >> things much more difficult even if you use single mode fibers. >> >> VLV > >It depends if his 3 taps all have positive coefs. If not, he can look >into an AM radio sort of thing where the modulation is offset so the >multiplication is always by positive values. If you want to get really >fancy, you can subtract light using either Kerr or Pockels cells. > >Clay >Actually, the 3 taps all have positive coefs "1". Thanks again!
