Strange constellation appears on pi/4 dqpsk and dqpsk thru scope_sink

Hello Friends.

Please someone advice me out for my situation.
I faced strange constellation at pi/4 dqpsk / dqpsk.

I am now trying to implement pi/4 dqpsk modulator/demodulator on
GnuRadio3.2.2 with using USRP2(RFX2400).
And mostly is working now.

In these days, I tried to have BER ratio, by PN9 XORed my original bit
pattern.(for spreading frequency)
and ratio shows very poor. If I compete with other systems.
which ratio is like 6 to 10dB more SNR is needed at same BER.

I’ve tried to have constellation of my implementation with using
Now, I found the implementation shows strange constellation as I

The demodulator is using origin as gr_make_mpsk_receiver_cc().

I sought where this strange constellation is come from.
Of course, I doubt this is because of my implementation at the first.

What I tried following things.
-Used rented Agilent VecterSignalAnalyzer 89600/89640 to get
for whether, At least Tx does work fine both pi/4 dqpsk and dqpsk.and
shows beautiful constellation.

-Used already impremented dqpsk as a modulator. and it’s constellation
similar strange plot. which is as
-Not PN9 XORed dqpsk shows sharpen constellation as shown in
-I have read
and in my guess “phase_error_detector_generic” and “decision_generic”
brought this issues.

Is my guess get pointed? or irrelevant?
If it is, I would like to think more how to make
“phase_error_detector_pi_4qpsk” and “decision_pi_4qpsk”.

I used following parameter for gr_make_mpsk_receiver_cc().
costas alpha = 0.15
gain mu = None
def mu = 0.5
omega relative limit = 0.005
fmin = -0.025
fmax= 0.025

and configuration of the test is as
Thank you very much in advance.

Yoshiya K.

æ ªå¼ä¼šç¤¾ãƒ†ãƒ¬ãƒžãƒ³ãƒ»ã‚³ãƒŸãƒ¥ãƒ‹ã‚±ãƒ¼ã‚·ãƒ§ãƒ³ã‚º
åŠ è—¤ç¾©ä¹Ÿ[email protected]
121-0012 東京都足立区青井6-8-11
電話:03-5888-7300 FAX:03-5888-7301
Telemann Communications Co., Ltd.
TEL:+(81)-3-5888-7300 FAX:+(81)-3-5888-7301
Board Member / Technical Director
Yoshiya K. [email protected]