WWalker wrote:> Hi Hardy, > > A (FIR) linear phase filter will phase shift the modulation a small amount > without distorting the signal in the pass band. As I mentioned in a > previous post. I am trying to measure a 3 degree shift of a 50MHz > modulation, 500MHz carrier signal. > > But, I should mention, that the following technique does work. Fourier > Transform the signal. Replace the higher harmonics mixer terms with zeros, > and then inverse Fourier Transform back to the time domain.Phase shifts don't matter, only relative shifts. A symmetric FIR has none. You are fighting a phantom. Do the math. Jerry -- Discovery consists of seeing what everybody has seen, and thinking what nobody has thought. .. Albert Szent-Gyorgi �����������������������������������������������������������������������
How to get envelope from AM signal without phase shift
Started by ●March 7, 2010
Reply by ●March 21, 20102010-03-21
Reply by ●March 22, 20102010-03-22
On 22 Mar, 00:55, "WWalker" <william.walker@n_o_s_p_a_m.imtek.de> wrote:> Hi Rune, > > Although the system is dispersive, provided the phase and amplitude reponse > of the system are linear over the bandwidth of the signal, the signal will > propagate undistorted. This is satisfied in my system with a 50MHz > Modulation, 500MHz Carrier AM signal. I simply want to measure a predicted > 3 degree phase shift of the Modulation. In order to do that I need to > extract the modulation and compare it to the modulation before the > propagation. I do not know if this can be done. This is why I am asking.Why didn't you say that first time around? You asked the wrong question. If you want to measure dispersion, ask about how to measure dispersion, not about how to measure obscure quantities that are sensitive to most minute variations in initial conditions or environment parameters. In order to achieve what you want, the ideal set-up is to measure the input and output simultaneously and synchronously. If this is possible, estimate the cross spectrum of the twu signals, and estimate the phase. If it is not possible to measure input and output but you can use an array to measure the output, you can still come up with an estimate of the phase velocity. The one thing *not* to do , is to work directly with phase. But even if you want to do that, it is perfectly possible to use a linear phase FIR filter, and subtract the corresponding delay from the filtered data. But none of this matters until you have answered the real question: Why would you want to verify these predictions? What purpose could it possibly serve? If you don't trust the maths - why not verify it by measuring a quantity that is actually possible to measure with any accuracy, is robust to environmental variation, and can be processed? Rune
Reply by ●March 22, 20102010-03-22
Hi Rune, The question you are asking is complicated but I will try to explain. I am trying to measure the speed of information transmission in the nearfield of a dipole source. This can be done by measuring the time delay of the envelope of an AM signal between two dipole antennas. Theoretical calculations show that the envelope should deviate about 3 degrees from light speed for a 50MHz modulated, 500MHz carrier signal. I would like to capture the transmitted and received signal on a 1GHz digital scope, extract the modulation envelopes, and measure the time delay. In order to measure the speed of information propagation, I need to also include the time it takes to demodulated the signal. Since the modulation light speed deviation is expected to only occur within a fraction (<1/10) of the carrier cycle, I need to come up with a demodulation technique that does not use a filter. This is because a filter takes more than a fraction of a carrier cycle to filter out the unwanted signals. If you want more detailed information you can refer to a paper I wrote: http://xxx.lanl.gov/pdf/physics/0603240 William>On 22 Mar, 00:55, "WWalker" <william.walker@n_o_s_p_a_m.imtek.de> >wrote: >> Hi Rune, >> >> Although the system is dispersive, provided the phase and amplitudereponse>> of the system are linear over the bandwidth of the signal, the signalwill>> propagate undistorted. This is satisfied in my system with a 50MHz >> Modulation, 500MHz Carrier AM signal. I simply want to measure apredicted>> 3 degree phase shift of the Modulation. In order to do that I need to >> extract the modulation and compare it to the modulation before the >> propagation. I do not know if this can be done. This is why I amasking.> >Why didn't you say that first time around? > >You asked the wrong question. If you want to measure >dispersion, ask about how to measure dispersion, not >about how to measure obscure quantities that are >sensitive to most minute variations in initial >conditions or environment parameters. > >In order to achieve what you want, the ideal set-up >is to measure the input and output simultaneously >and synchronously. If this is possible, estimate >the cross spectrum of the twu signals, and estimate >the phase. > >If it is not possible to measure input and output >but you can use an array to measure the output, you >can still come up with an estimate of the phase >velocity. > >The one thing *not* to do , is to work directly >with phase. But even if you want to do that, it >is perfectly possible to use a linear phase FIR >filter, and subtract the corresponding delay from >the filtered data. > >But none of this matters until you have answered >the real question: Why would you want to verify >these predictions? What purpose could it possibly >serve? If you don't trust the maths - why not >verify it by measuring a quantity that is actually >possible to measure with any accuracy, is robust >to environmental variation, and can be processed? > >Rune >
Reply by ●March 22, 20102010-03-22
On 22 Mar, 11:55, "WWalker" <william.walker@n_o_s_p_a_m.imtek.de> wrote:> Hi Rune, > > The question you are asking is complicated but I will try to explain. I am > trying to measure the speed of information transmission in the nearfield of > a dipole source. This can be done by measuring the time delay of the > envelope of an AM signal between two dipole antennas. Theoretical > calculations show that the envelope should deviate about 3 degrees from > light speed for a 50MHz modulated, 500MHz carrier signal."Deviate 3 degrees from light speed" ??? Again, one of the best ways to measure the effects of the system is to measure both the input and the output and then examine the cross correlation bewteen the two. This standard approach will extract the relative changes through the system while at the same time avoiding questions about absolute phase, which depends on all kinds of details you couldn't possibly track down anyway. Get a copy of the book "Random Data" by Bendat and Piersol. Rune
Reply by ●March 22, 20102010-03-22
Rune Allnor wrote:> On 22 Mar, 00:55, "WWalker" <william.walker@n_o_s_p_a_m.imtek.de> > wrote: >> Hi Rune, >> >> Although the system is dispersive, provided the phase and amplitude reponse >> of the system are linear over the bandwidth of the signal, the signal will >> propagate undistorted. This is satisfied in my system with a 50MHz >> Modulation, 500MHz Carrier AM signal. I simply want to measure a predicted >> 3 degree phase shift of the Modulation. In order to do that I need to >> extract the modulation and compare it to the modulation before the >> propagation. I do not know if this can be done. This is why I am asking. > > Why didn't you say that first time around?Amen! This underscores the consultant's dilemma: give the client what he asks for, or what he needs. I understand how we got here. Walker embarked on an inappropriate method for getting a result and ran into difficulties. He asked about resolving those (unnecessary) difficulties, rather than about solving the real problem. It didn't help that a number of unwarranted assumptions blocked his understanding of the suggestions we made, but it didn't hurt much either because we were talking at cross purposes anyway. I feel rather silly for not having figured it out. Jerry -- Discovery consists of seeing what everybody has seen, and thinking what nobody has thought. .. Albert Szent-Gyorgi �����������������������������������������������������������������������
Reply by ●March 22, 20102010-03-22
Hi Rune, The cross correlation technique does not seem to work very well with sinusoidally modulated AM signals but it does seem to work with pulsed AM signals. It appears that if the signal is not windowed properly one gets leakage effects, whereas a pulsed AM signal is automatically windowed properly.>"Deviate 3 degrees from light speed" ???ph=wt=360*f*d/c where: ph=phase in deg, f=freq, c=speed light, r=wave propagation distance. For a light speed propagating signal, at r=20cm, the carrier will phase shift 120deg [360*500MHz*20cm/(3E8)] and the modulation will phase shift 12 deg [360*50MHz*20cm/(3E8)]. I expect the modulation to arrive 3 degrees earlier (i.e at 9 deg). William>On 22 Mar, 11:55, "WWalker" <william.walker@n_o_s_p_a_m.imtek.de> >wrote: >> Hi Rune, >> >> The question you are asking is complicated but I will try to explain. Iam>> trying to measure the speed of information transmission in the nearfieldof>> a dipole source. This can be done by measuring the time delay of the >> envelope of an AM signal between two dipole antennas. Theoretical >> calculations show that the envelope should deviate about 3 degrees from >> light speed for a 50MHz modulated, 500MHz carrier signal. > >"Deviate 3 degrees from light speed" ??? > >Again, one of the best ways to measure the effects of the >system is to measure both the input and the output and then >examine the cross correlation bewteen the two. This standard >approach will extract the relative changes through the system >while at the same time avoiding questions about absolute phase, >which depends on all kinds of details you couldn't possibly >track down anyway. > >Get a copy of the book "Random Data" by Bendat and Piersol. > >Rune >
Reply by ●March 22, 20102010-03-22
On 22 Mar, 00:48, glen herrmannsfeldt <g...@ugcs.caltech.edu> wrote:>�After a signal goes through a dispersive > medium (such as optical fiber), it then goes through a phase > conjugation device. �That reverses the effect such that passing > through the same amount of fiber restores the original signal. > That is, dispersive fiber+phase conjugation+dispersive fiber > is, overall, not dispersive!I remember reading some time in the mid / late '90s about a phase conjugation tecnique used in a multipath scenario, in the context of active sonars. Since phase conjugation in time domain amounts to time reversal, these guys suggested to 1) Emit a known waveform into the water 2) Record the echo reflected off the target (which suffers from reverberation, multipath and what not) 3) Reverse the recorded signal and emit 4) Record the reflection from the time-reversed recording I never understood what the purpose of all this might have been.In 'standard mode' there are all kinds of problems detecting the reflection of interest inbetween all the multipaths and distortions. If you already know these factors, you also know the reference time around which to flip the signal. If you are unable to untangle the recieved signal, you don't know the key references, and effectively emit a random signal. Even if the idea works, and you recieve something that is close to the original pulse, you have no idea which part of the emitted signal interacted with the target. In the end, one have spent an awful lot of effort for no gain at all. Rune
Reply by ●March 22, 20102010-03-22
WWalker wrote:> Hi Rune, > > The question you are asking is complicated but I will try to explain. I am > trying to measure the speed of information transmission in the nearfield of > a dipole source. This can be done by measuring the time delay of the > envelope of an AM signal between two dipole antennas. Theoretical > calculations show that the envelope should deviate about 3 degrees from > light speed for a 50MHz modulated, 500MHz carrier signal. > > I would like to capture the transmitted and received signal on a 1GHz > digital scope, extract the modulation envelopes, and measure the time > delay. In order to measure the speed of information propagation, I need to > also include the time it takes to demodulated the signal. Since the > modulation light speed deviation is expected to only occur within a > fraction (<1/10) of the carrier cycle, I need to come up with a > demodulation technique that does not use a filter. This is because a filter > takes more than a fraction of a carrier cycle to filter out the unwanted > signals. If you want more detailed information you can refer to a paper I > wrote: > http://xxx.lanl.gov/pdf/physics/0603240You aren't making sense. "3 degrees from light speed" has less meaning than "3 degrees from south." Earlier, you wrote of group- and phase velocities that vary with distance. That is possible, but unusual. Is it really so? It does not matter how much delay there may be in a filter, just so long as you know how much it is. It simplifies matters greatly if the delay is independent of frequency, and symmetric FIRs have that property. Removing the imagined restrictions on what techniques can you use will make it much easier to fine a solution that you can implement woth your equipment. jerry -- Discovery consists of seeing what everybody has seen, and thinking what nobody has thought. .. Albert Szent-Gyorgi �����������������������������������������������������������������������
Reply by ●March 22, 20102010-03-22
WWalker wrote: ...> ph=wt=360*f*d/c where: ph=phase in deg, f=freq, c=speed light, r=wave > propagation distance. For a light speed propagating signal, at r=20cm, the > carrier will phase shift 120deg [360*500MHz*20cm/(3E8)] and the modulation > will phase shift 12 deg [360*50MHz*20cm/(3E8)]. I expect the modulation to > arrive 3 degrees earlier (i.e at 9 deg).A misconception. The frequencies of the signals carrying the modulation are 450 and 550 MHz. Together with the carrier, they produce the beat pattern seen as an envelope. ... Jerry -- Discovery consists of seeing what everybody has seen, and thinking what nobody has thought. .. Albert Szent-Gyorgi �����������������������������������������������������������������������
Reply by ●March 22, 20102010-03-22
On 22 Mar, 14:27, "WWalker" <william.walker@n_o_s_p_a_m.imtek.de> wrote:> Hi Rune, > > The cross correlation technique does not seem to work very well with > sinusoidally modulated AM signals but it does seem to work with pulsed AM > signals. It appears that if the signal is not windowed properly one gets > leakage effects, whereas a pulsed AM signal is automatically windowed > properly. > > >"Deviate 3 degrees from light speed" ??? > > ph=wt=360*f*d/c where: ph=phase in deg, f=freq, c=speed light, r=wave > propagation distance. For a light speed propagating signal, at r=20cm, the > carrier will phase shift 120deg [360*500MHz*20cm/(3E8)] and the modulation > will phase shift 12 deg [360*50MHz*20cm/(3E8)]. I expect the modulation to > arrive 3 degrees earlier (i.e at 9 deg).First of all - you are wrong. The phase shift of the *demodulated* 50 MHz signal depends on all kinds of details in the demodulating system, details you have no way of knowing with sufficient accuracy. Again: The only way you *might* come close, is to measure both the input and output, run both through as similar processing stages as possible (watch out for effects of variables in the physical implementations!) and then run a cross correlation analysis. The spatial phase you talk about should be measured at 550 MHz, which is the signal that actually propagates down the physical channel. And again: You haven't said anything about *why* you want to do this. Relying on phase meaurements is very poor way of doing anything. There is almost certainly a better way of doing whatever it is you are up to. Rune
