For my opinion, an "absolute" phase does not exist because we always need a reference. This explains the second case:  A "relative" phase is nothing else than the phase DIFFERENCE between two signals (of the same frequency).

For my opinion, an "absolute" phase does not exist because we always need a reference. This explains the second case:  A "relative" phase is nothing else than the phase DIFFERENCE between two signals (of the same frequency).

I completely agree with Lutz's answer. To define an "absolute phase" you need an "absolute time" reference t=0, which corresponds to phase zero.

As for voltages in curcuits, where actually an arbitrary "reference voltage" 0V is often considered and we normally say "voltage at node A" or "voltage at node B", intending the voltage difference between the node A and the reference voltage 0V, or the between the node B and the reference voltage 0V, a phase reference (e.g. the rising edge of a clock signal, of the carrier,...) can be arbitrarly established for periodic signals (a full period corresponds to 360° or 2*pi phase shift), and the phases of other signals with the same period can be (almost implicitly) considered with respect to such a phase reference.

"These phases" are very important in the signal processing - the phase shift defines group delay tau = - d(phase)/dt, which should be constant. Is often in fact quite a mess, see for example  https://www.researchgate.net/publication/237144726_Depiction_and_models_of_an_operational_amplifiers_phase_characteristic

Article Depiction and models of an operational amplifiers phase characteristic

group delay tau = - d(phase)/dt

Josef - I suppose you mean tau=-d(phi)/dw, correct?

To illustrate what is meant by phase assume that you have a sinusoidal signal x expressed by:

x= A sin wt, this signal has an amplitude A and angular frequency w, wt is the argument of the sine wave which is an angle proportional to t and when it is required to measure it, it is refereed to a zero angle line which may be the positive x direction. So, when we represent this signal in  a vector form as a phasor it has an amplitude A and phase zero. 

Assume that we have an other signal that has the same angular frequency w but shifted in phase by phi, then the waveform is expressed by:

x1= A sin (wt + phi),

Accordingly, the phasor of x1 will have an amplitude A and pase phi,

Comparing the two wave forms we can see easily the that phi is the angle between the the two phasor the one with the angle zero and the one with the angle phi.

So, the phase is the angle between two phasors. one may be chosen as a reference and the the other is measured with reference to it.

As an angle it needs to be  measured to a reference.

This reference may be fixed for all phasores as in the phase shift keying such a in the QPSK

S1= A sin wt,,

S2 = A sin wt + 90 deg

S3 = A sin wt + 180 deg

S4= A sin wt - 90 deg

Here the phase will be measured to S1

There is differential phase shift keying where say DIFFERENTIAL binary phase shift keying DBPSK whwere every sine symbol is phase shifted to the previous symbol which means that the current symbol is phase referenced to its previous one

Say S1 = A sin( wt +phi1)

S2 will be= A sin wt + phi1 + phi,

In the differential BPSK phi will be either zero or 180 deg.

So the detector has only to detect the phase angle between S2 and S1 since S1 will reference for S2.

Naturally this is the same as said by the colleague above only with examples related to the digital phase modulation tecniques.

Because of the reference requirement to detect it the phase shift keying must be synchronous which means that one has to regeneration the carrier at the demodulate.  

In case of differential PSK one do not need to regenerate the carrier since the every symbol is phase referenced to the previous symbol.It is required only to detect the phase difference between them.

Hi Lutz

of course, You are right.

Josef