why can a GMSK modulated system use non-linear PA?

[rickleicn]

gmsk ber graph -bpsk

why can a GMSK modulated system(such as GSM) use non-linear Power amplifier?
if a signal has been input to a non-linear power amplifier,then the output signal's shape will change from the original one's due to the non-linear,therefore,the spectrum of output will differ from input.

is there any explanation for the problem?

 

[sophiante]

gmsk linear power amplifier

Because GMSK is a constant envelope type of frequency modulation

 

[rickleicn]

gmsk constant envelope

sounds reasonable.
but no matter what modulation used,the amplitude of modulated carrier will keep on changing versus time,so the shape of the carrier will have a distortion due to non-linear power amplifier.

maybe the root cause is that power amplifier doesn't amplify the amplitude but the power,is that right?

 

[RFDave]

constant envelope pa

Because GMSK is a constant amplitude signal, the power of the modulated signal is constant. Because of that, you can use a compressed, non-linear amplifier for GMSK.

 

[gied]

gmsk pa

quote:"maybe the root cause is that power amplifier doesn't amplify the amplitude but the power,is that right?"

well, power is proportional to amplitude^2, so you will end-up in amplifying amplitude as well.

 

[rickleicn]

so why the GMSK is a constant amplitude signal?
envelope is not same with amplitude.
i think the GMSK signal's amplitude is always various versus time.

 

[gied]

Let's look at definition of an amplitude of sinusoid:
amplitude is max deflection.

Now looking from modulation point of view MSK is constant amplitude modulation, however GMSK is MSK with Gausian impulse filter applied. This gausian filter distorts amplitude, and GMSK is NOT constant amplitude modulation. If you look at IQ graph of GSMK, you would seek that the graph is not a circle as it should be in constant amplitude case.

Data is kept in phase/frequency rather than amplitude in MSK and GMSK. So distortion of amplitude with non-linear PA will not spoil data. Intermodulation products introduced by non-linarities end-up out of band and are filtered out in receiver.

Hope this helps

 

[RFDave]

That's interesting, because when I look at GSM I & Q in ADS, I see a perfect circle, showing that the amplitude stay's constant. Are you saying ADS & all of the texts on GSM are wrong?

Dave

 

[marlboro_x]

As i know , GMSK can be generated using FM method , then the evelope is constant.

 

[bcl]

Modulation schemes such as MSK, GMSK and Offset-QPSK have constant envelope characteristics because their constellation does not cross the origin, i.e., no zero crossing, and the information is embeded in the differential phase and not the amplitude change. Since there is no (minimal) amplitude chanage, th enon-linear PAs do not generate spectral regrowth problem and degrade channel performance.

 

[jallem]

It is not that the envelope is constant, it is that the envelope does not goes to zero,
so the amplifier does not have to be of large signal.

 

[bcl]

It is called constant envelope because the envelope is not constant? And the POWER AMPLIFIER "does not have to be of large signal?" Care to elaborate?

 

[Haytham-Sief]

GMSK is a constant amplitude signal while the data is loaded on the phase of the signal so any variation in the amplitude of the signal will not affect the information in the signal as the information is exctracted from the phase of the signal. Non-linear PA will only affect the Amplitude Modulation signal such as ASK or whatever.

 

[marlboro_x]

offset-QPSK could have constant evelope if not pulse-shaped.

In Lap's book ,about O-QPSK, to explain why use offset , he says when the constellation variation is pi, the evelope may go across zero , and at that time the signal is small which would be hard to amplify linearly, then spectrum regrows, which contributes to ISI and degrades BER.

Very often, should we consider that small constellation variation is better?

That's my own comprehension. Is it right ?

 

[bcl]

Constant envelope waveform allows the transmitter power amp to be run into compression region for better efficiency, but the demodulation process could be a little more complicated, compare to, for example, QPSK. I don't think there is a standard answere to this trade off, it is truly application dependant.

 

[rickleicn]

yes,because the envelope is constant, so in every power level there is only one point works,so although the pa is non-linear,the shape of amplified never change,in consequence,the information will not be lost.

 

[lincolndsp]

The envelope is constant and because gmsk is a non-linear modulation