The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Measurements >> RF Measurements >> EVM of MSK and HalfSine-Pulse-Shaped-OQPSK
https://designers-guide.org/forum/YaBB.pl?num=1195226387

Message started by pancho_hideboo on Nov 16^th, 2007, 7:19am

Title: EVM of MSK and HalfSine-Pulse-Shaped-OQPSK
Post by pancho_hideboo on Nov 16^th, 2007, 7:19am

Hi.

Halfsine pulse shaped OQPSK is one of the method to generate MSK signal.
So Halfsine pulse shaped OQPSK is same as MSK.

If I evaluate EVM of MSK signal, there can be three options.
Here my question is difference of results of the following three EVMs:
(1) EVM measured as MSK with Tsymb=Tbit, reference_filter=none
and measurement_filter=none.
(2) EVM measured as QPSK with Tsymb=Tbit, reference_filter=none
and measurement_filter=none.
(3) EVM measured as OQPSK with Tsymb=2*Tbit, reference_filter=halfsine
and measurement_filter=none.

Here Tbit mean bit period and Tsymb mean symbol period.

As far as my understanding, the constellations of MSK and QPSK are identical although their trajectories are different.
So ideally EVM of (1) and (2) have to be same.

Anyone please teach me. :-[

Title: Re: EVM of MSK and HalfSine-Pulse-Shaped-OQPSK
Post by didac on Nov 16^th, 2007, 11:04am

Hi pancho,
I think that all arises from the fact that QPSK and MSK are two different modulations, I mean their spectrums are different regardless the way they look their constellations. The MSK has the energy more concentrated so I think that it's more noise tolerant than QPSK(if not bandwidth limitation it's present cause MSK main lobe is wider than QPSK modulation).
hope it helps,

Title: Re: EVM of MSK and HalfSine-Pulse-Shaped-OQPSK
Post by pancho_hideboo on Nov 16^th, 2007, 8:53pm

didac, thanks for append.

But my question is not difference between QPSK and MSK.
Their trajectories are different. And MSK have constant envelope.

My question is EVM's difference of MSK signal when I use three different measurement setups.
As far as my understanding, (1) and (2) has to be identical.
On the other hand, I'm not sure about (3).

Title: Re: EVM of MSK and HalfSine-Pulse-Shaped-OQPSK
Post by didac on Nov 17^th, 2007, 3:30am

Hi pancho,
Sorry for my misunderstanding of your question. I'm not sure but according to Agilent when you specify your reference signal it's not generated from zero at the test equipment, in fact what they seem to do is to demodulate the signal under test make a kind of "correction" and then modulate again(http://forums.tm.agilent.com/phpBB2/viewtopic.php?t=1175&sid=b5ed7a67b31bdb3ee0c59f5d71e9881d,see
the picture-it's the same than in some app notes-), so maybe is due to the process of demod/mod and "correction" of the signal that it's different for each modulation.
Hope it helps,

Title: Re: EVM of MSK and HalfSine-Pulse-Shaped-OQPSK
Post by pancho_hideboo on Nov 18^th, 2007, 10:39pm

didac, thanks for append.

You are correct.

In most actual vector signal analyzer, different demodulation algorithms are used for generating the reference signals for MSK and QPSK.
So I can't use QPSK mode for measuring EVM of MSK signal even if their constellation points are completely same.
See attached figure first is MSK's Constellation as MSK, and second is MSK's Constellation as QPSK.
Note : EVM value in "Signal Quality Window" is not correct.

In IEEE802.15.4, MSK signal is generated as OQPSK with Halfsine pulse shaping.
See pp.47-48 of the following.
http://standards.ieee.org/getieee802/download/802.15.4-2003.pdf

So appropriate measurement method of EVM is (3)
But I don't have this measurement setup condition in my instruments.

What is different between (1) and (3)
(1) EVM measured as MSK with Tsymb=Tbit, reference_filter=none
and measurement_filter=none.
(3) EVM measured as OQPSK with Tsymb=2*Tbit, reference_filter=halfsine
and measurement_filter=none.

Here Tbit mean bit period and Tsymb mean symbol period.

Title: Re: EVM of MSK and HalfSine-Pulse-Shaped-OQPSK
Post by pancho_hideboo on Nov 18^th, 2007, 10:53pm

This is second figure.

Title: Re: EVM of MSK and HalfSine-Pulse-Shaped-OQPSK
Post by didac on Nov 19^th, 2007, 1:00am

Hi pancho,
Thanks for the pictures(it's good to see something graphical and not just my idea) and the reference. I was thinking if it is would be possible to generate the desired testbench using Labview(you will have to implement the block diagram of the EVM equipment), I've used in the past to generate VI's like an spectrum analyzer but I've never tried a EVM so maybe it's a silly idea,and most probably the data should be processed in baseband.

Title: Re: EVM of MSK and HalfSine-Pulse-Shaped-OQPSK
Post by pancho_hideboo on Nov 19^th, 2007, 2:04am

The following instruments have OQPSK(with Halfsine pulse shaping) mode.

http://wireless.agilent.com/zigbee/zigbee.shtml
http://www.rohde-schwarz.com/www/dev_center.nsf/frameset?OpenAgent&website=com&navig=/www/dev_center.nsf/html/nav,10,11&content=/appnote/RSI03.html
Tektronix RSA3408A also have OQPSK(with Halfsine pulse shaping) mode.

Although my tektronix RSA6100A is very expensive instruments,
this doesn't have OQPSK(with Halfsine pulse shaping) mode.
So I have to use MSK mode.

Since my RSA6100A can save I/Q vs time signal as MAT file,
so maybe I can analyze EVM as OQPSK(with Halfsine pulse shaping) using MATLAB or Agilent Ptolemy.

Title: Re: EVM of MSK and HalfSine-Pulse-Shaped-OQPSK
Post by didac on Nov 19^th, 2007, 5:01am

Hi,
Yeah I think you are right, with Matlab or Ptolemy it would easier than my approach with Labview.

Title: Re: EVM of MSK and HalfSine-Pulse-Shaped-OQPSK
Post by pancho_hideboo on Dec 25^th, 2007, 10:34pm

I compared the following two measurement setups for MSK signal(Tbit=0.5usec).
(1) EVM measured as MSK with Tsymb=1/2MHz=0.5usec, reference_filter=none
and measurement_filter=none.
(3) EVM measured as OQPSK with Tsymb=1/1MHz=1.0usec, reference_filter=halfsine
and measurement_filter=none.

Here Tbit mean bit period and Tsymb mean symbol period.

MSK signal was generated by Agilent E4438C signal generator.
In (1), RMS_EVM=0.77%, Peak_EVM=2.11%
In (3). RMS_EVM=0.74%, Peak_EVM=2.13%

It seems EVMs in both cases have no remarkable difference.

Attached pictures are results.
First one is (1) and second is (3).

Title: Re: EVM of MSK and HalfSine-Pulse-Shaped-OQPSK
Post by pancho_hideboo on Dec 25^th, 2007, 10:35pm

This is second picture.
Here OQPSK demodulation mode is used.

Title: Re: EVM of MSK and HalfSine-Pulse-Shaped-OQPSK
Post by didac on Dec 29^th, 2007, 3:28am

Hi pancho,
Thanks for posting the results,then option (3) seems to be the correct one. I wil take note, oncemore thanks for sharing this.