Phase detection of RF carriers

sub-contract the design work but looking for an opinion first about practicality.

and report their relative phase with at least 50 nsec accuracy.

are assumed to be perfectly stable. Nominal C/No >50 dB/Hz each carrier.

know is how much one leads or lags the other in time, like measuring the relative zero-crossing times of the two sinusoids. For measurement accuracy it is acceptable to time-average each measurement over as long as 100 msec.

concept is practical. Any recommendations for someone to do the design are appreciated. We are located in Southern Calif.

We're just now finishing up the design of a 5-channel time stamper with picosecond resolution...

but that's probably overkill. A simple FPGA based, counter type time stamper [1] would probably give you all the resolution you need, based on time averaging lots of edges. Clock it at 100 MHz or whatever. RF noise is going to jitter the edges all over the place anyhow.

The data you get will probably be a RAM or FIFO of time stamps from the two channels. Then you'll need some software to figure out what it means.

[1] a time stamper is just a clocked counter feeding a latch. When your input trigger arrives, freeze the latch. The latched data is "time of day" of the edge.

--

John Larkin         Highland Technology, Inc

jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro   acquisition and simulation

sub-contract the design work but looking for an opinion first about practicality.

and report their relative phase with at least 50 nsec accuracy.

are assumed to be perfectly stable. Nominal C/No >50 dB/Hz each carrier.

know is how much one leads or lags the other in time, like measuring the relative zero-crossing times of the two sinusoids. For measurement accuracy it is acceptable to time-average each measurement over as long as 100 msec.

concept is practical. Any recommendations for someone to do the design are appreciated. We are located in Southern Calif.

No. It can't be done with just a carrier. You'll need some form of pulse or modulation to mark the starting point of the phase measurement. Since the two CW signals are not harmonically related, not frequency locked, and probably do not originate from a common source, the relative phase of the zero crossings will be constantly changing.

However, If you provide a staring pulse, as was done with Loran, you can easily measure the time difference by starting a counter with a pulse on one frequency, and stopping the counter with a pulse on the other frequency. See numerous articles on how Loran works for clues.

Also, I'm not sure that "phase" is the correct term. My guess(tm) is that you want the time difference of arrival as used in direction finding. However, I'm not sure because direction finders usually have a single RF source, and you have two. 1.8 to 2.5MHz looks like one of the HF marine bands, but I can't tell if that's your application.

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann     AE6KS    831-336-2558

I have not the foggiest idea what phase means with nonsynchronous or non-harmonic signals.

Jim

sub-contract the design work but looking for an opinion first about practicality.

and report their relative phase with at least 50 nsec accuracy.

are assumed to be perfectly stable. Nominal C/No >50 dB/Hz each carrier.

know is how much one leads or lags the other in time, like measuring the relative zero-crossing times of the two sinusoids. For measurement accuracy it is acceptable to time-average each measurement over as long as 100 msec.

concept is practical. Any recommendations for someone to do the design are appreciated. We are located in Southern Calif.

t to sub-contract the design work but looking for an opinion first about pr= acticality.=20

ange, and report their relative phase with at least 50 nsec accuracy.

and are assumed to be perfectly stable. Nominal C/No >50 dB/Hz each carri= er.=20

to know is how much one leads or lags the other in time, like measuring th= e relative zero-crossing times of the two sinusoids. For measurement accur= acy it is acceptable to time-average each measurement over as long as 100 m= sec.

this concept is practical. Any recommendations for someone to do the desi= gn are appreciated. We are located in Southern Calif.

Yes you're right, but if there's a better word it escapes me.

How about this approach? =20

Find the zero crossings of each (sinusoidal) carrier to use them as time ma= rks. They will run toward/away from each other at a constant rate, being t= he frequency difference. We know a priori the two frequencies precisely so= we can calculate that out over the measurement interval. This is the next= best thing to having sharp edges to work with. Or what am I doing wrong?

What you said above are exactly my thoughts too.

sub-contract the design work but looking for an opinion first about practicality.

and report their relative phase with at least 50 nsec accuracy.

are assumed to be perfectly stable. Nominal C/No >50 dB/Hz each carrier.

know is how much one leads or lags the other in time, like measuring the relative zero-crossing times of the two sinusoids. For measurement accuracy it is acceptable to time-average each measurement over as long as 100 msec.

concept is practical. Any recommendations for someone to do the design are appreciated. We are located in Southern Calif.

Sure, dead simple--you can do that in one cycle. With a CNR of 50 dB, your single-cycle phase uncertainty is

Delta Phi = 1/sqrt(2*CNR) or 2.2 mrad, i.e. about 200 ps for a 2 MHz carrier, so with two equally noisy ones, that's 300 ps rms. I'd just run the two signals into a mixer, digitize the output for about one cycle of the beat frequency, and curve fit.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058

hobbs at electrooptical dot net
http://electrooptical.net

sub-contract the design work but looking for an opinion first about practicality.

and report their relative phase with at least 50 nsec accuracy.

are assumed to be perfectly stable. Nominal C/No >50 dB/Hz each carrier.

know is how much one leads or lags the other in time, like measuring the relative zero-crossing times of the two sinusoids. For measurement accuracy it is acceptable to time-average each measurement over as long as 100 msec.

That isn't how I interpret it, but I am just a farm boy so how would I know, right?

Jamie

f

Please tell me where I'm going wrong here.

Measure the time of occurrence of a series of zero crossings in Carrier #1 = over a time interval. Measure the time of occurrence of a series of zero c= rossings in Carrier #2. Subtract out the differences in times for each ser= ies by normalizing / correcting for their periods which are precisely known= .

So say the two carriers are transmitted from the same location (equal propa= gation times) but have precisely known different frequencies. Subtracting = them as above gives the beat frequency with no remainder.

That sort of correlator sounds like the perfect application for a software defined radio, or "SDR" in RF-speak. If I wasn't up to the gills full of work right now I'd love to take that on. But others (like the guys at John larkin's company) are probably better suited anyhow because this is a lot of software work.

--
Regards, Joerg

http://www.analogconsultants.com/

This needs some more blinkenlights and bells on the front panel :-)

Nah, George wrote the two frequencies are precisely known and stable. So one could downconvert and run it through some very narrowband filter. But with downconversion this all boils down to an I/Q detection instead of zero crossings. Or if money is not such a concern one could have crystals made for those two frequencies and send the RF straight thought a series of those crystals. You can get really low bandwidths that way and fish almost anything out of the soup.

George, I'd keep downconversion and I/Q detection in mind. This is essentially how Doppler works. Except if I understood right you want to measure the phase shift of one versus the other. So then you'd take one carrier as the local oscillator (or phase-lock it to the received signal) and then monitor phase shifts (I/Q shifts) of the other received carrier. Like a waterfall diagram with only one line on there. You'd also see the other lines around it so you could use that feature to see if a potential interference situation creeps up.

--
Regards, Joerg

http://www.analogconsultants.com/

ect to sub-contract the design work but looking for an opinion first about = practicality.

range, and report their relative phase with at least 50 nsec accuracy.

ce and are assumed to be perfectly stable. Nominal C/No >50 dB/Hz each car= rier.

nt to know is how much one leads or lags the other in time, like measuring = the relative zero-crossing times of the two sinusoids. For measurement acc= uracy it is acceptable to time-average each measurement over as long as 100= msec.

ng this concept is practical. Any recommendations for someone to do the de= sign are appreciated. We are located in Southern Calif.

Phil, I was with you right up to the curve fit part.

OK, you beat the RF carriers and get a difference tone. Then, you do a cur= ve fit of the difference tone amplitude against time? How do we get relati= ve phase of the two RF carriers from the curve fit?

Sorry I'm lagging behind.

George

to sub-contract the design work but looking for an opinion first about practicality.

range, and report their relative phase with at least 50 nsec accuracy.

and are assumed to be perfectly stable. Nominal C/No >50 dB/Hz each carrier.

to know is how much one leads or lags the other in time, like measuring the relative zero-crossing times of the two sinusoids. For measurement accuracy it is acceptable to time-average each measurement over as long as 100 msec.

this concept is practical. Any recommendations for someone to do the design are appreciated. We are located in Southern Calif.

fit of the difference tone amplitude against time? How do we get relative phase of the two RF carriers from the curve fit?

You get the phase difference between them at the centre of the measurement time. Frequency is just the time derivative of phase, so if you measure the phase often enough, you get the frequency difference and starting phase as a function of time.

For constant frequencies, Delta phi = phi(0)+ 2*pi*(f2 - f1)*t. You have to pick an instant to be t=0, but after that the measurement is perfectly well defined.

Cheers

Phil Hobbs

-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203 Briarcliff Manor NY 10510 845-480-2058

hobbs at electrooptical dot net

expect to sub-contract the design work but looking for an opinion first about practicality.

range, and report their relative phase with at least 50 nsec accuracy.

advance and are assumed to be perfectly stable. Nominal C/No >50 dB/Hz each carrier.

want to know is how much one leads or lags the other in time, like measuring the relative zero-crossing times of the two sinusoids. For measurement accuracy it is acceptable to time-average each measurement over as long as 100 msec.

assuming this concept is practical. Any recommendations for someone to do the design are appreciated. We are located in Southern Calif.

curve fit of the difference tone amplitude against time? How do we get relative phase of the two RF carriers from the curve fit?

Should have said, "you get the starting phase, which is a constant, and the frequency as a function of time."

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058

hobbs at electrooptical dot net
http://electrooptical.net

to sub-contract the design work but looking for an opinion first about practicality.

range, and report their relative phase with at least 50 nsec accuracy.

and are assumed to be perfectly stable. Nominal C/No >50 dB/Hz each carrier.

to know is how much one leads or lags the other in time, like measuring the relative zero-crossing times of the two sinusoids. For measurement accuracy it is acceptable to time-average each measurement over as long as 100 msec.

this concept is practical. Any recommendations for someone to do the design are appreciated. We are located in Southern Calif.

curve fit of the difference tone amplitude against time? How do we get relative phase of the two RF carriers from the curve fit?

But in the end you have to measure frequency deviations in the beat. Because for absolute phase you'd have to pick a certain point in time and say "This defines my zero point in phase relationship". After that it's either phase versus (super-precise) elapsed time or permanent frequency monitoring that will show phase deviations.

--
Regards, Joerg

http://www.analogconsultants.com/

If the frequencies are known very well, lock two PLLs on and use a series of dividers and multipliers (or in digital, an NCO block) to obtain equal frequencies. This is easier if the signals are related by a rational ratio. Compare the two results with a type 2 phase detector (or better). The resulting angle representation is the phase difference between the signals times the conversion ratio.

Tim

-- Deep Friar: a very philosophical monk. Website:

Requirement: Receive two CW carriers off the air in the 1.8 to 2.5 MHz range, and report their relative phase with at least 50 nsec accuracy.

Assumptions: The CW carriers' frequencies are precisely known in advance and are assumed to be perfectly stable. Nominal C/No >50 dB/Hz each carrier.

"Phase:" The carriers are at different frequencies. What we really want to know is how much one leads or lags the other in time, like measuring the relative zero-crossing times of the two sinusoids. For measurement accuracy it is acceptable to time-average each measurement over as long as 100 msec.

Regarding implementation, we'd like to start work by 1 September assuming this concept is practical. Any recommendations for someone to do the design are appreciated. We are located in Southern Calif.

expect to sub-contract the design work but looking for an opinion first about practicality.

range, and report their relative phase with at least 50 nsec accuracy.

advance and are assumed to be perfectly stable. Nominal C/No >50 dB/Hz each carrier.

want to know is how much one leads or lags the other in time, like measuring the relative zero-crossing times of the two sinusoids. For measurement accuracy it is acceptable to time-average each measurement over as long as 100 msec.

assuming this concept is practical. Any recommendations for someone to do the design are appreciated. We are located in Southern Calif.

curve fit of the difference tone amplitude against time? How do we get relative phase of the two RF carriers from the curve fit?

Time and frequency are about the easiest things in the world to measure precisely.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058

hobbs at electrooptical dot net
http://electrooptical.net

I'm with you here. The basic design parameters don't make sense. Phase implies the same frequency. [You can get stable displays on a scope with harmonically related signals if locked on the lower frequency signal. That can be thought of as phase locked I suppose.]

of

, 1 over a time interval. =A0Measure the time of occurrence of a series of ze= ro crossings in Carrier #2. =A0Subtract out the differences in times for ea= ch series by normalizing / correcting for their periods which are precisely= known.

pagation times) but have precisely known different frequencies. =A0Subtract= ing them as above gives the beat frequency with no remainder.

Yes, if you want the frequency difference between the two signals that works. But you initially said you wanted the relative phase between them. If you want to do a Delta-frequency calculation then I think you are OK.

The simplest form of a "phase meter" is a lissajous waveform on an oscilloscope. If you get a stable pattern on the screen, a phase meter will also show a stable pattern. Try it with your two carriers, and see what you get. Unless they are harmonically related, you will get a constantly rotating pattern. Only when the pattern is stable can you obtain a phase reading. A phase meter will do the same thing and produce a constantly changing reading.

I'm not sure what you're attempting to build, but it kinda looks like a TDOA (time difference of arrival) type direction finder, which requires a phase measurement. That's possible, but not with two random signals. There has to be some correlation between them.

Creating the beat frequency between the two signals is not going to do much. You can extract phase information from a mixer ( D flip flop, double balance mixer, Gilbert cell, etc) but all you'll get is another sine wave at the output port, which is the constantly varying phase difference that a lissajous pattern will show. To get a constant phase reading, the output would need to be a DC value, which again isn't going to happen unless the two signals are somehow related (as in harmonically related).

How about a clue as to what you're trying to accomplish?

(Sorry about the delayed reply. I attempted to downgrade from Forte Agent 7.0 to 5.0 and trashed my news reader).