I am trying to caracterize a system consisting of a 1) pseudo random generator (binary pulses, "on-off modulation") --> 2)laser diode --> light beam --> reflection from surface --> 3)photodiode detector --> 4)amplifier I would like to formulate at a model from measurements that mathematically describe how the binary pulses are "shaped" thru the chain 2) to 4). I need to do this because I want to convolve my original binary on-off sequence with the composite response of the signal chain 2) to 4). I want to use that shaped signal as the pattern to look for in a correlator for detecting time differences. I am fairly new to DSP techniques so I would like to ask for some advice on the following: Q1) How do I set up a measuring system to measure the complete response of the 2) to 4) chain? Do I for example use a step or a pulse input? Should measure the response with a digital oscillocope that can store data to file? Please give your suggestions. Q2) When I have gathered measurement data in discrete sample form (file on my disk), how do I calculate the convolution coefficients to use on the original binary sequence? Q3) How do I set up a correlator for testing this in Matlab? (What functions do I use? Do you have a particular procedure or setup you use for correlation experiments?) Mike
How to caracterize transfer function of signal chain
Started by ●July 18, 2009
Reply by ●July 18, 20092009-07-18
On 18 Jul, 13:42, "Dsp1000" <rezycle....@gmail.com> wrote:> I am trying to caracterize a system consisting of a 1) pseudo random > generator (binary pulses, "on-off modulation") --> 2)laser diode --> light > beam --> reflection from surface --> 3)photodiode detector --> 4)amplifier> Q2) When I have gathered measurement data in discrete sample form (file on > my disk), how do I calculate the convolution coefficients to use on the > original binary sequence?This is a non-trivial problem. The reason is step 2, where the light beam is emitted into some space and reflected off some surface. The exact behaviour of this step depends on a number of factors, like the beam pattern of the diode, the surrounding environment, the geometry of the set-up, the geometry of the reflecting surface, the scattering properties of the reflecting surface, and so on. So you have two options: Either use a simple correlator of whatever type you would have used in an optical fiber, or set up a fully-fledged simulator where you simulate the whole set-up in all its gory detail. Rune
Reply by ●July 18, 20092009-07-18
>On 18 Jul, 13:42, "Dsp1000" <rezycle....@gmail.com> wrote: >> I am trying to caracterize a system consisting of a 1) pseudo random >> generator (binary pulses, "on-off modulation") --> 2)laser diode -->light>> beam --> reflection from surface --> 3)photodiode detector -->4)amplifier> >> Q2) When I have gathered measurement data in discrete sample form (fileon>> my disk), how do I calculate the convolution coefficients to use onthe>> original binary sequence? > >This is a non-trivial problem. The reason is step 2, >where the light beam is emitted into some space and >reflected off some surface. > >The exact behaviour of this step depends on a number of >factors, like the beam pattern of the diode, the surrounding >environment, the geometry of the set-up, the geometry of the >reflecting surface, the scattering properties of the reflecting >surface, and so on. > >So you have two options: Either use a simple correlator >of whatever type you would have used in an optical fiber, >or set up a fully-fledged simulator where you simulate >the whole set-up in all its gory detail. > >Rune >Hello Rune, thank you for replying. The reflection from the reflective surface is assumed to be constant and not varying (other than amplitude). What I need to do is to characterize (not simulate) the whole transmission / reception chain by empirical experimentation. In this way I dont need to simulate anything, but meaasre. How do I set up a measuring setup? Do I use step response? How do I arrive at a mathematical characterization of what the "channel and whole system" does to my original on-off bit pattern. (lets look at one bit first only). What function do I use from Matlab for this correlator? Mike
Reply by ●July 18, 20092009-07-18
On 18 Jul, 15:03, "Dsp1000" <rezycle....@gmail.com> wrote:> >On 18 Jul, 13:42, "Dsp1000" <rezycle....@gmail.com> wrote: > >> I am trying to caracterize a system consisting of a 1) pseudo random > >> generator (binary pulses, "on-off modulation") --> 2)laser diode --> > light > >> beam --> reflection from surface --> 3)photodiode detector --> > 4)amplifier > > >> Q2) When I have gathered measurement data in discrete sample form (file > on > >> my disk), how do I calculate the convolution coefficients to use on > the > >> original binary sequence? > > >This is a non-trivial problem. The reason is step 2, > >where the light beam is emitted into some space and > >reflected off some surface. > > >The exact behaviour of this step depends on a number of > >factors, like the beam pattern of the diode, the surrounding > >environment, the geometry of the set-up, the geometry of the > >reflecting surface, the scattering properties of the reflecting > >surface, and so on. > > >So you have two options: Either use a simple correlator > >of whatever type you would have used in an optical fiber, > >or set up a fully-fledged simulator where you simulate > >the whole set-up in all its gory detail. > > >Rune > > Hello Rune, thank you for replying. > > The reflection from the reflective surface is assumed to be constant and > not varying (other than amplitude).The assumption might be good enough, but since you talk about a surface, and not a point reflector, the net effect is that the recieved signal might experience interference bewteen rays reflected off the near end of the surface, and rays reflected off the near end of the surface. Rune
Reply by ●July 19, 20092009-07-19
>On 18 Jul, 15:03, "Dsp1000" <rezycle....@gmail.com> wrote: >> >On 18 Jul, 13:42, "Dsp1000" <rezycle....@gmail.com> wrote: >> >> I am trying to caracterize a system consisting of a 1) pseudorandom>> >> generator (binary pulses, "on-off modulation") --> 2)laser diode-->>> light >> >> beam --> reflection from surface --> 3)photodiode detector --> >> 4)amplifier >> >> >> Q2) When I have gathered measurement data in discrete sample form(file>> on >> >> my disk), how do I calculate the convolution coefficients to use on >> the >> >> original binary sequence? >> >> >This is a non-trivial problem. The reason is step 2, >> >where the light beam is emitted into some space and >> >reflected off some surface. >> >> >The exact behaviour of this step depends on a number of >> >factors, like the beam pattern of the diode, the surrounding >> >environment, the geometry of the set-up, the geometry of the >> >reflecting surface, the scattering properties of the reflecting >> >surface, and so on. >> >> >So you have two options: Either use a simple correlator >> >of whatever type you would have used in an optical fiber, >> >or set up a fully-fledged simulator where you simulate >> >the whole set-up in all its gory detail. >> >> >Rune >> >> Hello Rune, thank you for replying. >> >> The reflection from the reflective surface is assumed to be constantand>> not varying (other than amplitude). > >The assumption might be good enough, but since you talk >about a surface, and not a point reflector, the net effect >is that the recieved signal might experience interference >bewteen rays reflected off the near end of the surface, and >rays reflected off the near end of the surface. > >Rune >Certainly, but thats where the correlator comes in. The transmitted signal will be amplitude modulated and if there are two reflections, a little delayed in time, the correlator will be able to detect those two signals and will be giving out two correlation peaks. I may use both or just one of them as the received peak. There is no need for extremely high time resolution. Still. I wonder if somebody could help me with how to measure the transfer function of the system? Mike
Reply by ●July 19, 20092009-07-19
On 19 Jul, 12:03, "Dsp1000" <rezycle....@gmail.com> wrote:> >On 18 Jul, 15:03, "Dsp1000" <rezycle....@gmail.com> wrote: > >> >On 18 Jul, 13:42, "Dsp1000" <rezycle....@gmail.com> wrote: > >> >> I am trying to caracterize a system consisting of a 1) pseudo > random > >> >> generator (binary pulses, "on-off modulation") --> 2)laser diode > --> > >> light > >> >> beam --> reflection from surface --> 3)photodiode detector --> > >> 4)amplifier > > >> >> Q2) When I have gathered measurement data in discrete sample form > (file > >> on > >> >> my disk), how do I calculate the convolution coefficients to use on > >> the > >> >> original binary sequence? > > >> >This is a non-trivial problem. The reason is step 2, > >> >where the light beam is emitted into some space and > >> >reflected off some surface. > > >> >The exact behaviour of this step depends on a number of > >> >factors, like the beam pattern of the diode, the surrounding > >> >environment, the geometry of the set-up, the geometry of the > >> >reflecting surface, the scattering properties of the reflecting > >> >surface, and so on. > > >> >So you have two options: Either use a simple correlator > >> >of whatever type you would have used in an optical fiber, > >> >or set up a fully-fledged simulator where you simulate > >> >the whole set-up in all its gory detail. > > >> >Rune > > >> Hello Rune, thank you for replying. > > >> The reflection from the reflective surface is assumed to be constant > and > >> not varying (other than amplitude). > > >The assumption might be good enough, but since you talk > >about a surface, and not a point reflector, the net effect > >is that the recieved signal might experience interference > >bewteen rays reflected off the near end of the surface, and > >rays reflected off the near end of the surface. > > >Rune > > Certainly, but thats where the correlator comes in. The transmitted signal > will be amplitude modulated and if there are two reflections, a little > delayed in time, the correlator will be able to detect those two signals > and will be giving out two correlation peaks.No. There will not be two rays. There will be one beam, that reflects off the surface. Light scattered from the whole footprint of the beam will interfere at the receiver. Rune
Reply by ●July 19, 20092009-07-19
On Jul 19, 3:03�am, "Dsp1000" <rezycle....@gmail.com> wrote:> >On 18 Jul, 15:03, "Dsp1000" <rezycle....@gmail.com> wrote: > >> >On 18 Jul, 13:42, "Dsp1000" <rezycle....@gmail.com> wrote: > >> >> I am trying to caracterize a system consisting of a 1) pseudo > random > >> >> generator (binary pulses, "on-off modulation") --> 2)laser diode > --> > >> light > >> >> beam --> reflection from surface --> 3)photodiode detector --> > >> 4)amplifier > > >> >> Q2) When I have gathered measurement data in discrete sample form > (file > >> on > >> >> my disk), how do I calculate the convolution coefficients to use on > >> the > >> >> original binary sequence? > > >> >This is a non-trivial problem. The reason is step 2, > >> >where the light beam is emitted into some space and > >> >reflected off some surface. > > >> >The exact behaviour of this step depends on a number of > >> >factors, like the beam pattern of the diode, the surrounding > >> >environment, the geometry of the set-up, the geometry of the > >> >reflecting surface, the scattering properties of the reflecting > >> >surface, and so on. > > >> >So you have two options: Either use a simple correlator > >> >of whatever type you would have used in an optical fiber, > >> >or set up a fully-fledged simulator where you simulate > >> >the whole set-up in all its gory detail. > > >> >Rune > > >> Hello Rune, thank you for replying. > > >> The reflection from the reflective surface is assumed to be constant > and > >> not varying (other than amplitude). > > >The assumption might be good enough, but since you talk > >about a surface, and not a point reflector, the net effect > >is that the recieved signal might experience interference > >bewteen rays reflected off the near end of the surface, and > >rays reflected off the near end of the surface. > > >Rune > > Certainly, but thats where the correlator comes in. The transmitted signal > will be amplitude modulated and if there are two reflections, a little > delayed in time, the correlator will be able to detect those two signals > and will be giving out two correlation peaks. I may use both or just one of > them as the received peak. There is no need for extremely high time > resolution. > > Still. I wonder if somebody could help me with how to measure the transfer > function of the system? > > Mike- Hide quoted text - > > - Show quoted text -You want to get a step response for each components in the chain, I guess? You can send a fairly good (good rise time) signal to the input and measure the output, plot them, and interpolate the plot to the a step response.
Reply by ●July 21, 20092009-07-21
>On Jul 19, 3:03=A0am, "Dsp1000" <rezycle....@gmail.com> wrote: >> >On 18 Jul, 15:03, "Dsp1000" <rezycle....@gmail.com> wrote: >> >> >On 18 Jul, 13:42, "Dsp1000" <rezycle....@gmail.com> wrote: >> >> >> I am trying to caracterize a system consisting of a 1) pseudo >> random >> >> >> generator (binary pulses, "on-off modulation") --> 2)laser diode >> --> >> >> light >> >> >> beam --> reflection from surface --> 3)photodiode detector --> >> >> 4)amplifier >> >> >> >> Q2) When I have gathered measurement data in discrete sampleform>> (file >> >> on >> >> >> my disk), how do I calculate the convolution coefficients to useon>> >> the >> >> >> original binary sequence? >> >> >> >This is a non-trivial problem. The reason is step 2, >> >> >where the light beam is emitted into some space and >> >> >reflected off some surface. >> >> >> >The exact behaviour of this step depends on a number of >> >> >factors, like the beam pattern of the diode, the surrounding >> >> >environment, the geometry of the set-up, the geometry of the >> >> >reflecting surface, the scattering properties of the reflecting >> >> >surface, and so on. >> >> >> >So you have two options: Either use a simple correlator >> >> >of whatever type you would have used in an optical fiber, >> >> >or set up a fully-fledged simulator where you simulate >> >> >the whole set-up in all its gory detail. >> >> >> >Rune >> >> >> Hello Rune, thank you for replying. >> >> >> The reflection from the reflective surface is assumed to beconstant>> and >> >> not varying (other than amplitude). >> >> >The assumption might be good enough, but since you talk >> >about a surface, and not a point reflector, the net effect >> >is that the recieved signal might experience interference >> >bewteen rays reflected off the near end of the surface, and >> >rays reflected off the near end of the surface. >> >> >Rune >> >> Certainly, but thats where the correlator comes in. The transmittedsigna=>l >> will be amplitude modulated and if there are two reflections, a little >> delayed in time, the correlator will be able to detect those twosignals>> and will be giving out two correlation peaks. I may use both or justone =>of >> them as the received peak. There is no need for extremely high time >> resolution. >> >> Still. I wonder if somebody could help me with how to measure thetransfe=>r >> function of the system? >> >> Mike- Hide quoted text - >> >> - Show quoted text - > >You want to get a step response for each components in the chain, I >guess? You can send a fairly good (good rise time) signal to the input >and measure the output, plot them, and interpolate the plot to the a >step response. >Verictor, thank you for your suggestion. What do you mean by a fairly good rise time? Is that a fast rise time? What I am actually looking at making is a laser range finder that will work against leaves & vegetation. For example, measure the range to the top of a tree. Does anyybody on this group have any experience with laser range finders and applied DSP techniques? Mike
Reply by ●July 21, 20092009-07-21
On Jul 21, 4:52�am, "Dsp1000" <rezycle....@gmail.com> wrote:> >On Jul 19, 3:03=A0am, "Dsp1000" <rezycle....@gmail.com> wrote: > >> >On 18 Jul, 15:03, "Dsp1000" <rezycle....@gmail.com> wrote: > >> >> >On 18 Jul, 13:42, "Dsp1000" <rezycle....@gmail.com> wrote: > >> >> >> I am trying to caracterize a system consisting of a 1) pseudo > >> random > >> >> >> generator (binary pulses, "on-off modulation") --> 2)laser diode > >> --> > >> >> light > >> >> >> beam --> reflection from surface --> 3)photodiode detector --> > >> >> 4)amplifier > > >> >> >> Q2) When I have gathered measurement data in discrete sample > form > >> (file > >> >> on > >> >> >> my disk), how do I calculate the convolution coefficients to use > on > >> >> the > >> >> >> original binary sequence? > > >> >> >This is a non-trivial problem. The reason is step 2, > >> >> >where the light beam is emitted into some space and > >> >> >reflected off some surface. > > >> >> >The exact behaviour of this step depends on a number of > >> >> >factors, like the beam pattern of the diode, the surrounding > >> >> >environment, the geometry of the set-up, the geometry of the > >> >> >reflecting surface, the scattering properties of the reflecting > >> >> >surface, and so on. > > >> >> >So you have two options: Either use a simple correlator > >> >> >of whatever type you would have used in an optical fiber, > >> >> >or set up a fully-fledged simulator where you simulate > >> >> >the whole set-up in all its gory detail. > > >> >> >Rune > > >> >> Hello Rune, thank you for replying. > > >> >> The reflection from the reflective surface is assumed to be > constant > >> and > >> >> not varying (other than amplitude). > > >> >The assumption might be good enough, but since you talk > >> >about a surface, and not a point reflector, the net effect > >> >is that the recieved signal might experience interference > >> >bewteen rays reflected off the near end of the surface, and > >> >rays reflected off the near end of the surface. > > >> >Rune > > >> Certainly, but thats where the correlator comes in. The transmitted > signa= > >l > >> will be amplitude modulated and if there are two reflections, a little > >> delayed in time, the correlator will be able to detect those two > signals > >> and will be giving out two correlation peaks. I may use both or just > one = > >of > >> them as the received peak. There is no need for extremely high time > >> resolution. > > >> Still. I wonder if somebody could help me with how to measure the > transfe= > >r > >> function of the system? > > >> Mike- Hide quoted text - > > >> - Show quoted text - > > >You want to get a step response for each components in the chain, I > >guess? You can send a fairly good (good rise time) signal to the input > >and measure the output, plot them, and interpolate the plot to the a > >step response. > > Verictor, thank you for your suggestion. > > What do you mean by a fairly good rise time? Is that a fast rise time? > > What I am actually looking at making is a laser range finder that will > work against leaves & vegetation. For example, measure the range to the top > of a tree. > > Does anyybody on this group have any experience with laser range finders > and applied DSP techniques? > > Mike- Hide quoted text - > > - Show quoted text -Idealy a step response needs the input to be rise immediately so that later on you can get the impulse response, right? But in real life, it is not an easy job to find very short rise time stimulus. What is a laser range finder? I don't know about that. You mean use laser's reflection to determine the arrival time so that you can decide the object's dimension? Just a side question to your application. You don't have to answer my question.
Reply by ●July 22, 20092009-07-22
>On Jul 21, 4:52=A0am, "Dsp1000" <rezycle....@gmail.com> wrote: >> >On Jul 19, 3:03=3DA0am, "Dsp1000" <rezycle....@gmail.com> wrote: >> >> >On 18 Jul, 15:03, "Dsp1000" <rezycle....@gmail.com> wrote: >> >> >> >On 18 Jul, 13:42, "Dsp1000" <rezycle....@gmail.com> wrote: >> >> >> >> I am trying to caracterize a system consisting of a 1) pseudo >> >> random >> >> >> >> generator (binary pulses, "on-off modulation") --> 2)laserdiode>> >> --> >> >> >> light >> >> >> >> beam --> reflection from surface --> 3)photodiode detector-->>> >> >> 4)amplifier >> >> >> >> >> Q2) When I have gathered measurement data in discrete sample >> form >> >> (file >> >> >> on >> >> >> >> my disk), how do I calculate the convolution coefficients touse>> on >> >> >> the >> >> >> >> original binary sequence? >> >> >> >> >This is a non-trivial problem. The reason is step 2, >> >> >> >where the light beam is emitted into some space and >> >> >> >reflected off some surface. >> >> >> >> >The exact behaviour of this step depends on a number of >> >> >> >factors, like the beam pattern of the diode, the surrounding >> >> >> >environment, the geometry of the set-up, the geometry of the >> >> >> >reflecting surface, the scattering properties of the reflecting >> >> >> >surface, and so on. >> >> >> >> >So you have two options: Either use a simple correlator >> >> >> >of whatever type you would have used in an optical fiber, >> >> >> >or set up a fully-fledged simulator where you simulate >> >> >> >the whole set-up in all its gory detail. >> >> >> >> >Rune >> >> >> >> Hello Rune, thank you for replying. >> >> >> >> The reflection from the reflective surface is assumed to be >> constant >> >> and >> >> >> not varying (other than amplitude). >> >> >> >The assumption might be good enough, but since you talk >> >> >about a surface, and not a point reflector, the net effect >> >> >is that the recieved signal might experience interference >> >> >bewteen rays reflected off the near end of the surface, and >> >> >rays reflected off the near end of the surface. >> >> >> >Rune >> >> >> Certainly, but thats where the correlator comes in. The transmitted >> signa=3D >> >l >> >> will be amplitude modulated and if there are two reflections, alittle>> >> delayed in time, the correlator will be able to detect those two >> signals >> >> and will be giving out two correlation peaks. I may use both orjust>> one =3D >> >of >> >> them as the received peak. There is no need for extremely high time >> >> resolution. >> >> >> Still. I wonder if somebody could help me with how to measure the >> transfe=3D >> >r >> >> function of the system? >> >> >> Mike- Hide quoted text - >> >> >> - Show quoted text - >> >> >You want to get a step response for each components in the chain, I >> >guess? You can send a fairly good (good rise time) signal to theinput>> >and measure the output, plot them, and interpolate the plot to the a >> >step response. >> >> Verictor, thank you for your suggestion. >> >> What do you mean by a fairly good rise time? Is that a fast rise time? >> >> What I am actually looking at making is a laser range finder that will >> work against leaves & vegetation. For example, measure the range to thet=>op >> of a tree. >> >> Does anyybody on this group have any experience with laser rangefinders>> and applied DSP techniques? >> >> Mike- Hide quoted text - >> >> - Show quoted text - > >Idealy a step response needs the input to be rise immediately so that >later on you can get the impulse response, right? But in real life, it >is not an easy job to find very short rise time stimulus. > >What is a laser range finder? I don't know about that. You mean use >laser's reflection to determine the arrival time so that you can >decide the object's dimension? Just a side question to your >application. You don't have to answer my question. >Hi again. Thanks for your post regarding step response. A laser range finder is a laser distance (range) measuring device. It will give you distance to an object you shine light on as a binary or analog output. There are two main methods to use: Time of flight of the laser beam and phase difference detection. The latter is better on short distances but has some challenges associated with it. It is a more advanced version of the latter I am looking at. All experts on this list: anybody with laser range finder experience? Mike