DVB-T MODULATOR IMPERFECTIONS

EVALUATION AND MEASUREMENT

Tomáš Kratochvíl, Radim Štukavec and Martin Slanina

Department of Radio Electronics, Brno University of Technology, Purkyňova 118, Brno, Czech Republic

Keywords: COFDM modulator, I/Q error, Amplitude Imbalance, Phase Error, Digital Video Broadcasting, DVB-T.

Abstract: The paper deals with simulation, evaluation and measurement of the DVB-T modulator imperfections.

Modulator imperfections’ and I/Q errors’ influence on the DVB-T signal and its spectrum and

I/Q constellation analysis are presented. Theoretical backgrounds of the Amplitude Imbalance, Phase Error

and Carrier Suppression effects are outlined in the paper. Then the practical results measured in the

laboratory environment are compared to the theoretical assumed impacts on Modulation Error Rate from

I/Q constellation and Bit Error Rates before and after Viterbi decoding in DVB-T signal decoding.

Commented results of the measurements are presented in the paper as well.

1 INTRODUCTION

The DVB-T (Digital Video Broadcasting –

Terrestrial) is already a standard (ETSI, 2004) and

technology for Digital TV distribution.

Technical characteristics make the DVB-T

system flexible to operate in a combination of

 3 modulations (QPSK, 16-QAM, 64-QAM),

 5 FEC rates (Forward Frror Correction),

 4 Guard Intervals (1/32, 1/16, 1/8, 1/4),

 3 modes of carriers - 2k (1705) or 8k (6817),

 3 channel bandwidths (8, 7, 6) MHz.

Using different combinations of the above

parameters, a DVB-T network can be designed to

match the requirements of the network operator,

finding the balance between robustness and capacity.

Terrestrial transmission path is subject to

numerous impacts such as echoes and multipath

reception, AWGN (Additive White Gaussian Noise)

and Doppler shift in case of mobile reception. Apart

from this, the quality of the transmission link is also

determined by the DVB-T modulator and transmitter

parameters. A lower quality signal can be produced

caused by Crest factor limitation, intermodulation,

noise, I/Q errors and interferers (Fisher, 2008).

Figure 1: DVB-T modulator with I/Q errors

and imperfections (Fisher, 2008).

To avoid effects of the transmission link and

modulator imperfections, DVB-T uses COFDM

(Coded Orthogonal Frequency Division Multiplex).

DVB-T modulator (see Fig. 1) can exhibit

imperfections caused by different gains in I/Q

signals (Amplitude Imbalance), imprecise 90 degree

phase shift between I/Q signals (Phase Error) or

residual carrier in the frequency spectrum caused by

DC component in I/Q signals (Carrier Suppression).

Due to the channel estimation and carrier pilots all

these effects result in lower MER (Modulation Error

Rate) in dB from I/Q constellation and according

higher BER (Bit-Error Rate) (ETSI, 2001).

Kratochv

ıl T.,

Stukavec R. and Slanina M. (2010).

DVB-T MODULATOR IMPERFECTIONS EVALUATION AND MEASUREMENT.

In Proceedings of the International Conference on Signal Processing and Multimedia Applications, pages 50-54

 SciTePress

2 EFFECTS OF THE I/Q ERRORS

The effects can be observed only at the center

carrier. The other carriers exhibit noise like

interference in the presence of any AI (Amplitude

Imbalance) and PE (Phase Error).

a) AI presence

b) PE presence

Figure 2: Determining SNR by vector diagrams in case of

I/Q errors (Fisher, 2008).

While determining the SNR (Signal-to-Noise

Ratio) with the AI and PE presence, equations

(1) to (4) apply:

SNR

−

(1)

[]

100/%

100/%2

log20

dBSNR

−

= ,

(2)

⎟

⎠

⎞

⎜

⎝

⎛

−

⎟

⎠

⎞

⎜

⎝

⎛

−

⋅=

90cos

90sin

SNR ,

(3)

[]

⎟

⎠

⎞

⎜

⎝

⎛

⎟

⎠

⎞

⎜

⎝

⎛

−=

90tanlog20

dBSNR .

(4)

The disturbances in DVB-T caused by I/Q errors

can be explained by using vector diagrams (see

Fig. 2). Both mixers of the DVB-T modulator

operate with CS (Carrier Suppression) and SSB

(Single Sideband Modulation) technique. Using this

technique, two sideband vectors are added and two

sideband vectors are subtracted (cancelled band).

If the AI or PE exists, it means that the upper

or lower sideband is no longer canceled completely

and leaves an interference component.

It is clear that all the subcarriers are subject to

noise like interference, with the exception of the

center carrier. This can be shown in the spectrum if

the lower carrier band is switched off. Hence, if the

I/Q modulator is adjusted to produce AI or PE, an

evident crosstalk from the upper to the lower

sideband is clearly apparent.

A DC component in re(t) or im(t) signals after

the IFFT leads to a residual carrier in the I or Q

branch or in both of them. Apart from the

corresponding amplitude, the residual carrier also

exhibits phase angle. A residual carrier at DVB-T

modulator shifts the constellation diagram out of the

centre in I or Q signal direction. The diagram

remains undistorted and it can be verified only at the

central carrier. Insufficient carrier suppression

appears in the centre of the constellation diagram.

Some works related to this paper were published

also by (Bucholtz, 2000) and (Palipana, 2005, 2007).

3 BEHIND THE I/Q ERRORS

The Viterbi decoder can correct bit errors depending

on the code rate selected in the convolutional

encoder. The approximation condition for the QEF

(Quasi-Error Free) reception, which corresponds

to one error per hour, is defined as BER after Viterbi

decoding equal to 2.10

-4

or less. This is the limit

at which the subsequent Reed-Solomon decoder still

delivers an output BER after RS decoding of 1.10

-11

or less. This condition almost corresponds to the

“fall of the cliff” and “blockiness” appearing in the

picture. Slightly more noise or interference suffices

for the DTV transmission to break down.

From the SNR or MER in dB, the BER before

Viterbi decoding (channel BER) can be determined

or at least estimated. Theoretical minimal SNR for

QEF reception depends on the convolutional code

rate, type of the COFDM inner modulation and the

type of the transmission channel - Gaussian, Ricean

or Rayleigh (ETSI, 2001, 2004).

Theoretical CNR (Carrier-to-Noise Ratio) value

for the DVB-T signal transmission analyzed in this

paper (64-QAM, 8k mode, 2/3 code rate, 1/8 guard

interval, non-hierarchical modulation) is equal

to 16.5 dB in the AWGN (Gaussian) channel

for stationary reception. Practical CNR value

is about 18 to 20 dB (Fisher, 2008).

DVB-T MODULATOR IMPERFECTIONS EVALUATION AND MEASUREMENT

Figure 3: Laboratory workplace for the DVB-T

measurements and transmission link.

4 DVB TRANSMISSION SYSTEM

Laboratory workplace for the DVB-T measurements

(see Fig. 3) has been used for the evaluation of I/Q

errors on MER. The devices are:

 SFL-T R&S DVB-T test transmitter,

 DVRG R&S MPEG-2 TS generator,

 MSK200 Kathrein DVB-T test receiver,

 STB Humax F3 FOX-T commercial set-top box

adapted with tuner IF outputs and transport

stream MPEG-2 TS data output,

 OSC Agilent digital storage oscilloscope,

 MMT Metex multimeter,

 TVP Panasonic TV set.

The DVB-T system transmission parameters were

set to the European most common type of DTV

broadcasting. These parameters are the most

characteristic for large DVB-T SFN networks:

 RF level 60 dBuV (medium sensitivity),

 8 MHz channel (bandwidth 7.608 MHz),

 64-QAM modulation (TS 19.90588 Mbit/s),

 8k mode – 6817 subcarriers (fixed reception),

 2/3 convolutional code rate (robust code),

 1/4 Guard Interval (large size SFN),

 non-hierarchical modulation (one TS).

For the Amplitude Imbalance AI, Phase Error PE

and Carrier Suppression CS influence on MER,

BER

(before Viterbi) and BER

(after Viterbi)

evaluation, the DVB-T modulator and test

transmitter parameters were set in these intervals:

 AI - (0 .. 25% with step of 2%),

 PE - (0..10

with step of 1 degree),

 CS - (0 .. 50% with step of 5%).

BER before Viterbi vs. I/Q errors

1,0E-05

1,0E-04

1,0E-03

1,0E-02

1,0E-01

1,0E+00

0 1020304050

AI [%], PE [deg], CS [%]

BER

before Viterbi

[-]

Amplitude Imbalance Phase Error Carrier Suppression

a) BER

before Viterbi

= f (AI, PE, CS)

BER after Viterbi vs. I/Q errors

1,0E-08

1,0E-07

1,0E-06

1,0E-05

1,0E-04

1,0E-03

1,0E-02

0 1020304050

AI [%], PE [deg], CS [%]

BER

after Viterbi

[-]

Amplitude Imbalance Phase Error Carrier Suppression

QEF

b) BER

after Viterbi

= f (AI, PE, CS)

MER vs. I/Q errors

0,0

10,0

20,0

30,0

40,0

0 1020304050

AI [%], PE [deg], CS [%]

MER

[dB]

Amplitude Imbalance Phase Error Carrier Suppression

MIN

c) MER = f (AI, PE, CS)

Figure 4: DVB-T modulator imperfections and I/Q errors

measurements (independent AI, PE and CS).

5 MEASUREMENT RESULTS

The results of the BER before Viterbi, BER after

Viterbi and MER measurements as the functions of

the AI, PE, CS are shown in Fig. 3. The “QEF”

symbol in Fig. 4b) indicates the situation where BER

after Viterbi decoding is equal to 2.10

-4

. This is the

formerly defined condition of practically error-free

signals at the input of the MPEG-2 TS

demultiplexer. The “MIN” symbol in Fig. 4c)

indicates the situation where the DVB-T with

SIGMAP 2010 - International Conference on Signal Processing and Multimedia Applications

modulation 64-QAM, convolutional code rate 2/3

and non-hierarchical modulation has the minimal

required CNR equal to approx. 18 dB. This is the

reference value of possible DVB-T in a no-

interference reception.

The main impact on I/Q errors on BER and MER

is caused by Amplitude Imbalance. Decrease of

10 dB in MER is caused by the AI of approx. 10 %

between I/Q signals. It is evident that the most

serious impact on I/Q errors has the Phase Error.

Decrease of 10 dB in MER is caused by PE of

approx. 6 degrees between I/Q signals. There is

no serious impact and influence of Carrier

Suppression on I/Q errors (see Fig. 5, all carriers

displayed).

Figure 5: I/Q constellation of the DVB-T, 64-QAM, 8k

mode and in case of CS = 50% presence (CS error visible

in the centre of the diagram).

A really interesting evaluation could be the

dependence of AI and PE simultaneously on MER as

it is shown in the Fig. 6. With the AI higher than

approx. 20 % and PE higher than approx. 8 degrees,

the DVB-T signal is not available for decoding.

In Fig. 7, I/Q constellation diagrams of the

DVB-T with 64-QAM, 8k mode are shown. In the

individual pictures the simultaneous influence of AI

and PE on the I/Q constellation is displayed. I/Q

errors of the modulator partially affect all the

carriers as noise-like disturbance (typical shape

of clouds) and can only be identified by observing

the central carrier (again no. 3408 in 8k mode).

6 CONCLUSIONS

From the measured results it is easy to see that both

the AI and PE lead to lower MER from I/Q

constellation analysis (see Fig. 7) and noise like

crosstalk in the spectrum (see Fig. 8 and Fig. 9).

15,0

20,0

25,0

30,0

35,0

40,0

MER [dB]

0246810

PE [deg]

AI [% ]

Figure 6: DVB-T modulator imperfections and I/Q errors

measurements (simultaneous AI and PE).

a) AI = 0%, PE = 0

MER = 35.2 dB

b) AI = 0%, PE = 5

MER = 26.3 dB,

c) AI = 10%, PE = 0

MER = 25.6 dB

d) AI = 10%, PE = 5

MER = 23.5 dB

e) AI = 20%, PE = 0

MER = 20.5 dB,

f) AI = 20%, PE = 5

MER = 19.7 dB,

Figure 7: I/Q constellation of the DVB-T with 64-QAM,

8k mode and in case of AI and PE presence (DVB-T

channel correction and all pilots ON).

DVB-T MODULATOR IMPERFECTIONS EVALUATION AND MEASUREMENT

Figure 8: DVB-T spectrum without suppressed carrier,

C21 (470 – 478) MHz, 8k mode, CS = 50% (CS error

visible at central carrier, 474 MHz).

a) AI = 0%, PE = 0

b) AI = 0%, PE = 5

c) AI = 10%, PE = 0

d) AI = 10%, PE = 5

e) AI = 20%, PE = 0

f) AI = 20%, PE = 5

Figure 9: Simulated spectrum of the DVB-T with 64-

QAM, 8k mode and in case of AI and PE presence (lower

band carriers 0 - 3407 were set to zero to illustrate the

crosstalk).

The effect can be described by means of simple

trigonometric operations which can be derived from

the vector diagram of the signal and noise. In the

case of AI, the opposite vectors of I/Q signals with

noise are not cancelled completely and it results

in a noise vector causing crosstalk from the upper

band to lower frequency band.

A phase error will result in a noise vector with

the length determined by the vector parallelogram.

In both cases the actual useful signal amplitude

decreases by the same amount by which the

crosstalk increases.

In the practical and commercial DVB-T

modulator implementations, usually AI less than 5%

and PE less than 0.5 degree is the aim of the design

and it is verified only very close to the center carrier

(no. 3408 in case of the DVB-T and COFDM 8k

mode) and adjacent carriers.

ACKNOWLEDGEMENTS

The paper was supported by the Research program

of the Brno University of Technology

no. MSM0021630513, “Electronic Communication

Systems and New Generation Technology

(ELKOM)” and the research grant projects of the

Czech Science Foundation GACR no. 102/08/P295

“Analysis and Simulation of the Transmission

Distortions of the Digital Television DVB-T/H“

and no. P102/10/1320 “Research and Modeling

of Advanced Methods of Image Quality Evaluation

(DEIMOS)”.

REFERENCES

Fisher, W. Digital Video Broadcasting, The family

of international standards for digital television

edition). Berlin: Springer Verlag, 2008.

ISBN 978-3-540-76357-4.

ETSI EN 300 744 V1.4.1 (2004-11). European Standard.

Digital Video Broadcasting (DVB); Framing

structure, channel coding and modulation for DTT.

EBU, 2004.

ETSI TR 101 290 V1.2.1 (2001-05). Technical report.

Digital Video Broadcasting (DVB); Measurement

guidelines for DVB systems. EBU, 2001.

Buchholz, M., Schuchert, A., Hasholzner, R. Effects of

tuner IQ imbalance on multicarrier-modulation

systems. Proceedings of the 2000 Third IEEE

International Caracas Conference on Devices, Circuits

and Systems, 2000 , pp. T65/1 - T65/6

Palipana, R.B., Chung, K.S. The effects of receiver

impairments in terrestrial digital video broadcasting.

APCC2003, Asia-Pacific Conference on

Communications, 2003, pp. 1143 - 1146

Palipana, R.B., Chung, K.S. Pilot based correction of iq

cross-talk in the presence of a frequency offset.

APCC2007, Asia-Pacific Conference on

Communications, 2007, pp. 69 – 72

SIGMAP 2010 - International Conference on Signal Processing and Multimedia Applications