On Jan 14, 2018, snipped-for-privacy@gmail.com wrote (in article):
I don?t recall this thread, but it sounds interesting. Before my time? Does anybody recall the subject and date range, so I can recover this thread?
What is it that rendered your position wrong?
Joe Gwinn
Some detail rather than platitudes would be appropriate. ...Jim Thompson
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| STV, Queen Creek, AZ 85142 Skype: skypeanalog | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
It's what you learn, after you know it all, that counts.
On Jan 14, 2018, Tauno Voipio wrote (in article ):
How do the non-feasible approaches work?
Yes, we do use this approach, but only in narrowband radars, where the resulting waveform distortion is tolerable.
Yes, I was after the device physics of the phase shifter itself. A parallel is a SAW (Surface Acoustic Wave) device used to implement chirp generation and compression - the SAW device is completely passive, but the insertion loss is 10 or 15 dB, so there is always an amplifier nearby. Joe Gwinn
A counterexample that demonstrated nonzero group delay and zero true delay.
Group delay is -d(phi)/d(omega). An RC lowpass has nonzero group delay, but its effect can be undone by an RC highpass, with some additional gain. True delay can't be undone by anything other than a time machine.
(Or some wiseguy HFT outfit building an as-the-crow-flies microwave link from the Chicago Board of Trade to the New York Stock exchange, and they only managed to reduce it.)
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
are most
,lay
te
s a
tple
An example of an LC delay line at 4-8Ghz:
Of course, Lumped capacitors are employed in the filter networks, but the inductors are really just transmission lines of high characteristic impedan ce.
Any finite pole-zero system cannot be truly said to have delay, but perhaps fortunate for your question, there are no truly real pole-zero systems; all real systems have delay sooner or later.
The phase shift of a real system may be exaggerated due to pole-zero-like behavior, indeed to the extent that we prefer to analyze them exclusively that way (RLC circuits), but it's important to keep this in mind at frequencies and scales where it matters.
So even though you might analyze a physical passive variable phase shifter as a pole-zero system, it will contain true delay (which will probably be variable).
Is that a satisfactory answer?
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: https://www.seventransistorlabs.com/
Not true of systems with zeros, because a zero can go anywhere, but poles can only go in one half plane for a stable system. So if you have a function with a zero in the unstable half plane, when you invert it to get the inverse operator, you wind up with a pole in the unstable half plane, i.e. an attempt at a time machine.
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
Ah, true -- well, to be fair I didn't stipulate that the system had to be stable as well, but that's obviously relevant to the "real physical" (and add "practial" as well) case. :)
More reading:
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: https://www.seventransistorlabs.com/
If you start with both sin(w*t) and cos(w*t), it's just two potentiometers and a sum junction (nothing active there). The adjustable components are two coefficients, not 'delay elements'.
That's true as far as it goes. But I have a question. How do you get both sin(wt) and cos(wt) without one being a time delayed version of the other?
NT
+1 mark
ems are most
e
X-Band is 10GHz territory, 1GHZ is L-Band. Bands do not refer to *bandwidth
*, those radars may be quite narrow band operating at a high frequency that is tightly controlled. You're not likely to find a "wideband" phase shifte r at mm-wavelength frequencies.
The radar receiver characteristics must mach transmitter signal characteristics for best range. In the simplest case that would be the transmitter at a fixed frequency and a receiver at the same frequency with a few NHz bandwidth to get somewhat accurate distance measurements.
At least for military radars, you do not want to use a fixed frequency, but move around both the transmitter as well as the receiver to avoid jamming. If you can move the frequency across the whole X-band helps a lot. For this reason the front end should be wideband, while a single frequency is used at a specific time. If some form of spread spectrum is used, the total spectrum can be quite wide, while the despread receiver signal can be quite narrow.
Are you asking about oscillators with more than one output? You build 'em with that in mind. It doesn't use a time-delay element in general, an LC tank can be outfitted with a voltage follower for sine, and a current transformer pickoff on the inductor current gives cosine.
Whether sine and cosine ARE time-delayed equivalents, is a philosophical question. Whether they're generated by using a time-delay, is an engineering question. Which do you intend?
Or a "quadrature oscillator", which uses two integrators to generate the sin and cos outputs as part of the oscillator feedback. If you're careful, you can get some pretty low distortion with this sort of oscillator, too.
on the contrary it fails to understand what's going on. Integrator output consists of 2 signals mixed, history and now. Differentiator ignores the summed history and just responds to the now element.
NT
meters
s are two
th sin(wt) and cos(wt) without one being a time delayed version of the othe r?
em with
that certainly answers nothing.
ted
e inductor
interesting, hadn't thought of that approach
question. Whether
ich do you intend?
the latter. AFAIK all oscillators are based on time delay one way or another, but your v versus i outputs are interesting. I was thinking more of a classic ring o scillator with however many outputs at different phases - there each output is a time delayed version of the previous one. I suppose you could say tha t output A is time delayed B, B is time delayed C and C is time delayed A!
NT
Or, instead of a time delay, you add a phase delay. ;-)
e:
iometers
nts are two
both sin(wt) and cos(wt) without one being a time delayed version of the ot her?
'em with
itted
the inductor
l question. Whether
Which do you intend?
r v versus i outputs are interesting. I was thinking more of a classic ring oscillator with however many outputs at different phases - there each outp ut is a time delayed version of the previous one. I suppose you could say t hat output A is time delayed B, B is time delayed C and C is time delayed A !
Hmm OK I'm seeing things different. If there's no time delay in an RC then there's no delay in an L/R. and there are lots of LC oscillators.
And if there's no delay in an RC there should be no delay in a collection of them... something like a phase sequence filter... an RC network.
However I think all these things have a transient response. So steady state you get nice stable oscillators or phase shift.
If you want a phase shift with no transient response, then do a time delay. At least that my view, as always I welcome corrections.
The first time I really bumped into this was trying to make acoustic impedance matching waveplates... I'd have to look up the details, but Brass to plastic with aluminum in between. (or something like that.) With a single wavelength of the 'right' frequency.. nothing. (no improvement) But as I added more wavelengths, (I love digital sig. gens.) I could see more transmitted power and less reflected. (Except at the beginning and end.)
Waveplates, work in the steady state. George H. I've enjoyed this thread.
I suggest taking Calc 1 over again.
Cheers
Phil Hobbs
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.