Maker Pro
Maker Pro

Phase modulation

  • Thread starter Rene Tschaggelar
  • Start date
R

Rene Tschaggelar

Jan 1, 1970
0
While I think to have understood what phase modulation is
about, the finer details of implementation have escaped me.

The phase shift of a signal compared to a reference signal
can be coded by a few bits depending on the number of phase
steps. This is understood.
However what happens when the reference signal is not available ?
This can happen when the transmitting and/or receiving side
has a generator that is not stable to a radian per day,
PLL-ed up from some lousy temperature dependent quartz with
some jitter on it.
How am I expected to get my bits back ? Running a PLL would
get me a bitstream as the derivative of the phase or so.
However since lacking the reference signal the received coding is
arbitrary, just a phase differential. I'd have to check all
possible combinations and somehow select the right phase.

Any hints, where I can gleam the phase-zero from ?
And how can I preserve the phase-zero over time.

Rene
 
S

Scott Stephens

Jan 1, 1970
0
Rene said:
However what happens when the reference signal is not available ? ...
How am I expected to get my bits back ?

Derive it with a PLL that locks onto the phase-modulated input signal.
The PLL low-pass filter averages out the phase modulation.

Running a PLL would
get me a bitstream as the derivative of the phase or so.
However since lacking the reference signal the received coding is
arbitrary, just a phase differential. I'd have to check all
possible combinations and somehow select the right phase.

You use I and Q (quadrature) component mixers.
Any hints, where I can gleam the phase-zero from ?
And how can I preserve the phase-zero over time.

Then you know an arbitrary complex phase, you lock onto that the same
way as Manchester coded signals, as discussed a few days back.

--
Scott

**********************************

DIY Piezo-Gyro, PCB Drill Bot & More Soon!

http://home.comcast.net/~scottxs/

**********************************
 
T

Tim Wescott

Jan 1, 1970
0
Rene said:
While I think to have understood what phase modulation is
about, the finer details of implementation have escaped me.

The phase shift of a signal compared to a reference signal
can be coded by a few bits depending on the number of phase
steps. This is understood.
However what happens when the reference signal is not available ?
This can happen when the transmitting and/or receiving side
has a generator that is not stable to a radian per day,
PLL-ed up from some lousy temperature dependent quartz with
some jitter on it.
How am I expected to get my bits back ? Running a PLL would
get me a bitstream as the derivative of the phase or so.
However since lacking the reference signal the received coding is
arbitrary, just a phase differential. I'd have to check all
possible combinations and somehow select the right phase.

Any hints, where I can gleam the phase-zero from ?
And how can I preserve the phase-zero over time.

Rene

I'm a bit behind the times on the latest techniques, but basically what
you do is (a) encode your signal so you don't need an absolute phase
reference and (b) PLL your reference so you always hit the center of the
expected "bin" you're aiming for.

The simplest example of this is differential binary PSK, where a '1' is
transmitted as a phase change and a '0' is transmitted as none. You
take the received signal and square it, which causes the 180-degree
changes to wash out to nothing, and gives you a nice steady tone that
you can PLL to. You do, and mix it with your incoming signal. Then you
lock onto the inherent clock signal with another PLL to get the raw bit
scheme, and finally you exclusive-or todays bit with yesterday's to get
the original bit stream.

You can also do this with a gizmo called a "Costas Loop", where you find
the relative phase of the signal at the same time that you decode it,
then you use the value of the raw decoded bit to multiply the detected
phase error by +1 or -1. It has a theoretical advantage over the
squaring scheme because you're limiting the noise in one of the branches.

Similar things can be done with higher-order phase modulation schemes,
of course.

As an alternative way to deal with the 180 degree phase ambiguity is to
use PSK with forward error correction, with an FEC scheme that helps you
detect when your phase is backward. This looks really good on paper,
but tends to be _very_ sensitive to an inserted or deleted bit (which
can happen when the bit-clock PLL gets confused).
 
R

Rene Tschaggelar

Jan 1, 1970
0
Scott said:
Derive it with a PLL that locks onto the phase-modulated input signal.
The PLL low-pass filter averages out the phase modulation.

Assuming there is no preferred phase, equally spread bit pattern permit
phase jumps in both directions
Running a PLL would



You use I and Q (quadrature) component mixers.

Sure, but what phase does the LO have ? The IF output will just
have the difference phase on its output.
Then you know an arbitrary complex phase, you lock onto that the same
way as Manchester coded signals, as discussed a few days back.

I'll look it up.

Thanks

Rene
 
R

Rene Tschaggelar

Jan 1, 1970
0
Tim said:
Rene said:
[snip questions]

[snip differential PSK]

You can also do this with a gizmo called a "Costas Loop", where you find
the relative phase of the signal at the same time that you decode it,
then you use the value of the raw decoded bit to multiply the detected
phase error by +1 or -1. It has a theoretical advantage over the
squaring scheme because you're limiting the noise in one of the branches.

Similar things can be done with higher-order phase modulation schemes,
of course.

As an alternative way to deal with the 180 degree phase ambiguity is to
use PSK with forward error correction, with an FEC scheme that helps you
detect when your phase is backward. This looks really good on paper,
but tends to be _very_ sensitive to an inserted or deleted bit (which
can happen when the bit-clock PLL gets confused).

Thanks

Rene
 
S

Scott Stephens

Jan 1, 1970
0
Top