Electromagnetic Radiation from Video

Display Units: An Eavesdropping Risk?

1. introduction

It i3 w2ll knowi th2t electronic equipment pro-

duces electrom~yxtic field5 which ma;” caux in-
terference to radio and television reception. The
phenomena underlying thih have heen thoroughI>
studied ovc‘r the pabt f~ &C&S. These studies
hai2 rzs~rltzd in internationally agreed methods
for measuring the inrzrierence pr~~duced b>
equipment. These are needed because the masi-
mum interference ir1~1.i. u hich equipment ma!
generate have been laid dou-n h\ law in rnobt
countrx~.
MoweLer. intzrfcrencz i> not the only pwblsm
cau& by electromagnetic radistiCjn. It is possible
in some ~‘a~5 to obtain information on thz signals
used inside the equipment M-hen the radiation is
picked up and the received signals are decoded.
Esgeciall> in the case of digital equipment this
possibiiit> constitutes a problem. becaube remote
reconstruction of signals insid, the equipment ma>
enable reconstruction of the data the equipment is
p:ocessins.
This problem is not a new one; defence special-
ists have been awxe of it for over t\venty >ear>.
Information on the \vay in kvhich this kind of
“eavesdropping” can be prevented is not free14
available. Equipment designed to protect military
information will probably be three or four times
more expensive than the equipment likelq- to be
W’im van Eck wrij born in Z&t used for processing of non-military information.
(Sicthrrlands). Hr was graduated from
Until recently it \vas considered very difficult to
Tuents Univerblt> of Technology in
1981. on his rrearch subject: “.~uto- reconstruct the data hidden in the radiated field,
matx on-line Eserctse Electrocardiog-
and it \vas therefore helkved that eavesdropping
raph\ in patients unable to perform
leg sxsrciss.” He was a member of the on digital equipment could onI>- be performed bq
Bio--nginering Group of the Ekc-
-:_ic
tromcs Department of the Tuentr
professionals Lvith access to very sophisticated de-
~-
CniLersit>. of Tcchnolog>. In Januar?. tection and decoding equipment. -1s a result. dig-
1982. he Joined the Propagation and
ital equipment for processing information requir-
p I Electromagncti; Compatibiht> De-
&&_;
partment of the Dr. Nrhrr Laboraro- ing medium or low level protection. such as private
~tes of the Xethsrlands PTT. He IS in charge of several E>lC
reedrch projects. rangin g from NEMP protection to emission
and business information. is not protected against
and wsceptibility aspects of telecommunications equipment. eavesdropping of this kind.
This report gives the results of a research pro-
gramme carried out by the dr. Neher Laboratories
Sorth-Holland of the Netherlands PTT. These results prove that
Cdmpursrs & Sscurit) 4 (1985) X19-286 the abow assumptions are wx-ong. Although the

0167-X~8,‘YS,‘S3.?0 I_ 1985. Else\ier S;~snce Publishers 8.1’. (North-Hall~nci~

 di!igi!,ii >i,s!:.lC s’u,f-h 2s the ~S.33 signal.
(‘+r,.a.‘.
;‘irx >. :o ,?‘IIZ:s?
’ broadband signals inside 3 \-ide,o
disp!,i> unit. ‘.?;I iid,0 signa! i> amplified from
tral~sjsi~r-tran~:st~~r iogic‘ unit (TTL) 1~~11 to
~~r.~ral hundred volts bzforz it is fed into the
iini\ to study the problem itself. but also to find cathod, r,lv tube (CRT). Thr radiation originating
iv;t>‘ of preventing this kmd of information theft. fr<Tm :hs \.idw Gyal u-ill therzforz bz thz domi-
.An additional aim was the definition of a nt~x~urz- nant comporxnt of the broadband field gznzrritrd
mznt method which could be LI& to check the by thr video display unit in most caszs.
>2xt-<ne>s of thr prohlzm with individual vi&o Exh (radi2tctd) hxmonic of that video signal
di,pl,~> units [i-DL:s !] ipoGbly for typ+approvaI bhov a rzmarkJblz rz>cmblance to a broadcast
purpws). The solutions found are dewrib I,ltzr. Ti. signal. as is shoed in thz technical appzndi.s of
this paper. It is thrrefors possible to rrconstruct
thr plcturz di>pla)ed on thr video &plsq unit
2. Cauve and Effects irt Brief from the radi:;tzd zmission by mrans of a normal
tzlsi-ision rsi&i.;r.

The application of squars \v:av2 Ggnsli and high

.\~itching frzquznciz~ in digital equipment Ictads to
a--is--d b;. the TV receiver does not
Thit sigr,al rLL,,.,
ths radiation of Axtromaynetic fisldi containing
contnin ~*kn~hri~:~i~ation information. This mean5
f:<\iuenc> components up into thr C’HF r+on.
that rhz picturs displayed on thz T1. screen whilr
A!th <)ugh thr power spectral densit> of thzss sig-
’ rttc,-i\inJ‘ radiation from a video displ‘*l; unit \vill
nA dzcrsaxs with increasing frequency. this is
bs n:o\iig over thz swxn in both the horizontal
compznsatzd for in the radiated field, txcau~e the
and vertical directions. unless thz synchronization
radiation cffzctivznrss of the zlzctronic circuits
freqwnsieb in the ridso display unit and thz TC
inside thz equipment inctxascs with frrqurncy.
recei\.tr arz thz same. .-\ithough thi: latter is true
This means that the radiation lzvcl produced by
for man> types of video display units, the picture
digital equipment may br constant up to ssvsral
rzczivsd will not bz \wy stable and therefors not
hundred MHz.
easily rzadabls. The yu;llity of reception can bz
In some cases. xsonancrs in circuits ma>- I?ad
improwd by e.xtzrnall> generating thz neczssart
to higher radiation levels at some frequencies in
svnchronization signals and feeding them into the
ths radiated spectrum. Evrn circuits not d&gned
iI’ recei\.er.
to carrv a certain signal maq radiate part of this
‘ct.ith this esrznaion to the normal TV receiver
signal due to cross-talk and bzcauss ths circuits
(the cohts are appronimately 315). almost an>- type
arz rssonant for some of the signal’s frequrnc>
or ma.k of video display unit can be easvesdropped
comporxnts. .A striking example of such a radiat-
on. provided it scntxx:ts a hufficizntlv high radi-
i2: circuit is th2 main po:ver cable of a pizcz of
ation lzvsi. The rstension can be designed and
equipment.
constructed bq- any e!sctronic amateur within a
_._.
’ ’ Picl’eo Displa! C’nits few da>s.

If NT limit ourselves to video display units. it can

bs rasil! rscognizzd that the field radiated by such
If no prz\-2nti\-s mrasurss arz taken, eavesdrop-
ping on 2 video displav unit i.5 possible at sc\eral
hundreds of metros dist.mce. using onlv a normal
 It is evident that this possibility hai impli.x-
tions with regard to the protection ai inf~wn;lti+n.
This is especially relevant to cases u-here prk>tecti:-e
measures have already been taken, buch a\ encr;\p-
tion and/or physical protection. In any &in c\f
measures taken to protect information. the Lteake~t
link may well be the video display unit r;tdiating
information around. And as everybody know;. a -2.I. I. TAtz E.~tr!-i!c:l Osc~~ilo:or St,iii:!/jil
chain is never stronger than its aeakest link. The easiest and cheapest u.i> of reconstructing
As it is relatively easy to reconstruct informa- the synchronization in the TV receiver is the ~1st oi
tion from the field radiated by video displa! units. a device containing t!vo oscillators:
the phenomenon may have consequence> for in- l one adjustable oscillator for th? t‘iquency rang?
formation security even Lvhere a IOU- or medium 15-N kHz to generate the horizontal synchro-
level of data protection is required. It should be nization signal (line synchronizdtian). and
borne in mind that the easvesdropping possibilit> @ one adjustable oscillator for th= frequency range
r-nay affect telebanking and other activitieh carried 40-80 Hz to generate the ver:i<;il synchroniz,l-
out with the aid of a personal computer. It is tion sipnal (picture synchroniza:ion).
pobsible for a neighhour to copy information &- Both signals can be combined ind fed into the
ptnyed during these activities (e.g. data on the synchronization heparator (Fig. 1 I ot the TV re-
financial situation) using his own TV receii-er. ceiver. I: ih rather difficult to ad;iu..; ‘both oscil!J-
In some cases reception of private infoimation tors to the video display units ;‘r :=rminal> shn-
of neighbours may even happen accidentsily. In- chronization frequencies becau;s ‘52th have to be
formation can be displayed on the receiving tele\ i- adjusted conhtantly during reception.
sion set during normal operation. This is not an It is well knoun that the wrticll Jnd horizontal
imaginary occurrence: The Radio Control Service synchronization frequencies are :elxted accordins
of the Netherlands PTT - which is in charge of to:
spectrum management - has had several com-
plaints from persons who were receiving informa-
tion from a nearby travel agency. where X- is the number of display lines on the CRT
.A11this means that this eavesdropping problem or screen. it is therefore practical to generate onl!
necessitates measures over the entire range of in- the horizontal synchronization frequency. and to
formation security levels, ranging from .( topjsecret‘ obtain the vertical synchroniza;i+n frequent!
to ‘privacy-sensitive. through division of Jhor by k. .A programmable
digital frequency divider bvhich CXI be used ior
this purpose can be bought for about 510. Once
3. Electromagnetic Eavesdropping
the number of screen lines has been determined.
hiany video display units or terminals are based the synchronization can be re‘b;+:ed by adjustir,g
on the same principles as black-and-white t&vi- only one oscillator.
sion. The free-running synchronization oscillators Fig. 1 shows an eavesdroppin; sst-up in which
in a TV receiver can therefore sometimes generate this type of synchronization rewLer> is used.
nearly the same frequency as the one used in the
VDL’. If this happens the displayed information 3.1.2. Recocerv from thr Receir.ri Siprlal
can easily be reproduced on the TV screen. and The horizontal and the verric;i synchronization
this can even occur accidentally. frequencies are available in the spectrum of the
:rm beam in the CRT -. As the frequen+ camp+
nwts are not axxilable in the ‘format’ espected b\
a T\’ receiver. it is necessary to d&n a 5>nch:o- The first measurements of the ele<trom.~;netic field
nization recover\- circuit. .A straightfor\vard ap- strength generated by various types oi L.DUs were
proach in this reipect is extraction of the horizw- carried out on a measuring site as described in
tal synchronization frequency from thz line fxd CISPR Puh!~wtion 16 ‘. Fields strength measure-
[LF] signal. ubing a narrow bandpass filter. The ments according to the forthcoming CISPR recom-
signal obtained is a sinusoidal \vave of 15-20 kHz mendation on data processing ;ind office
with a fair amount of phase noise. This noise can equipment (DP ‘OE) shoued that none of the
be easily removed using a very slow phase locked VDUs under test produced electromagnetic inter-
circuit. A puke shaping circuit can turn the ference beyond the proposed limits.
sinusoid into a square wave (synchronization GE- In spite of this. it ~vas still posGbl< to obtain a
nal) and the vertical synchronization frequenck clear picrux of the displaysd information on a
can easily be obtained again by division of this normal TV receii’er at a distance pi about 50
frequency by the number of screen lines. In order mztrrs from thz L.ideo display unit or terminal.
t17obtain a stable synchronization signal it is nec- For video display units or tsrmina!s in metal
sssarv to hax,e either a high signal-to-noise ratio in covering the maximum reception distance u’as
the rkeived signal or to include a narrow band- about 10 mews.
pass filter in the circuit. In the latter case it is These measurements lvere carried out Lvithin
the TV broadcast bands: field strength measure- clamped (10 dB gain). The received signal was fed
ments. however, showed that the maximum radia- through an antenna. amplified (18 dB) and dis-
tion level produced by a video display unit or played on a television screen inside the van. For
terminal was. always located between the TV obvious reasons we cannot give information on the
broadcast bands. Consequently. the maximum re- data picked up during the experiment. The results
ception distance may be expected to be larger than can be summarized as follows:
the distances mentioned in the aforegoing ‘. Also, l It is possible to eavesdrop on the video display
since measurements were carried out using a di- units or terminals in buildings from a large
pole antenna for reception, the use of a directional distance, using a car fitted up for the purpose.
antenna may provide at least 10 dB extra gain, l Although the experiment was carried out in
thus leading to another increase in maximum re- broad daylight and many people watched us,
ception distance. Sometimes a video display unit nobody asked what we were doing.
or terminal is placed near reflecting objects. This
may in the worst case lead to a transmission gain
of about 3 dB. 4. Solutions
Taking account of all these factors, it seems
justified to estimate the maximum reception dis- 4.1. Decrease Radiation Level
tance using only a normal TV receiver at about 1
km for a video display unit or terminal in plastic There are various techniques for decreasing the
covering and around 200 m for one in metal amount of radiation from electronic circuits. They
covering. include:
l DO not use a family of digital components which
3.3. Experimental Eavesdropping switches faster than necessary for the operation
of the circuits. This limits the high cut-off
To prove that eavesdropping is feasible in a practi- frequency of the radiated spectral intensity.
cal situation using this simple set-up, the following l Keep the radiating area of an electric circuit
experiment was carried out. The equipment (di- loop as small as possible. This can e.g. be done
pole antenna, TV receiver. and synchronization by providing a return lead as near as possible to
oscillators) was put in a car, which was placed in each signal lead on the printed circuit board.
the car park of a building in which a word 0 Keep interconnecting leads as short as possible.
processor was being used. An attempt was then More details and additional techniques can be
made to copy the information from this word found in the various publications on electromag-
processor’s video unit by taking photographs of netic compatibility design, such as [j] and [6].
the screen of the receiving television set. The pho- The above measures will decrease the total
tographs convinced even the most sceptical people amount of radiation from the printed circuit boards
in our organization of the threat of this possibility in the equipment. They cannot decrease the radia-
to information security. tion from the electron beam in the CRT. Thus
additional measures are required.
3.4. Further Experiments If the entire video display unit or terminal
system is electromagnetically shielded. the radia-
In February, 1985, we carried out an eavesdrop- tion can be almost eliminated. A metal shield will
ping experiment in London, in cooperation with keep the electromagnetic energy inside the unit or
the British Broadcasting Corporation. Part of the terminal. The shielding effectiveness (in dB) of a
results were shown in the programme “Tomorrow’s metal shield is almost proportional to the thick-
World.” ness of the shield in the frequency range from
A small van was equipped with a 10 metre high several hundred kHz up to several hundred MHz
pump mast to which a VHF band III antenna was (depending on the size of the video display unit or
terminal).
4 In the future, TV receivers will be able to receive signals at If a metal shield could be constructed round the
frequencies between the TV broadcast bands since these video display unit or terminal, the radiation level
frequencies will be used for cable television. outside the equipment would be determined by the
27-1 W. mn Eck / Electromognrtrc Radtutlon /rum Video Dtsplq Lntts

thickness of the shield and the radiation level or terminals is possible, because each unit has
before the shield was installed. Unfortunately. the different (resonance) frequencies at which the radi-
construction of such a shield is not feasible be- ation is dominant.
cause:
part of the shield would have to be optically 4.3. Cryptographic Display
transparent to be able to see the screen:
The basic factor leading to the detection of the
cables would have to penetrate the shield to link
information displayed on a video display unit or
the unit or terminal to the outside world (inter-
terminal by means of a normal TV receiver is the
facing. power supply);
similarity between the two systems as regards image
the keyboard would have to be reachable for the
build-up. Therefore, a simple and adequate solu-
operator: and
tion to this problem is to change the sequence in
in most cases ventilation openings would be
which the successive display lines are written on
needed.
the screen. A TV receiver expects the picture
To allow all the functions mentioned above, a
build-up to start at the top line and to end at the
vast range of shielding materials and aids are
bottom line in a natural sequence (1. 2, 3, 4,. . , k).
available on the market. including:
It is comparatively easy to change the sequence
metal (gold) coated CRT screens;
of the pattern of the digital image build-up of the
wire mesh nettings to be placed before a CRT
video display unit into a semi-random one. The
screen;
sequence obtained can be made dependent on a
honey comb gratings for ventilation;
code key which can be fed into the units circuitry.
shielded cables for interconnection of VDU and
If the radiated signal is now picked up by a TV
keyboard;
receiver, the information is not readable, and it is
electric filters to prevent radiation from
very difficult to ascertain whether information is
penetrating cables; and
being received at all. The information can only be
special material to join the different parts of the
obtained from the received signal when the se-
shield, etc.
quence is known or when sophisticated decoding
The total of measures which can be taken to
equipment is used. In order to prevent detection of
reduce the radiation level is rather expensive, and
the information by “trial and error” (with k dis-
may double or even triple the price of a video
play lines there are only k! possibilities), the code
display unit or terminal depending on the final
key can be made to change semi-randomly after a
radiation level accepted.
preset time interval. The design of a video display
unit or terminal with such a cryptographic display
4.2. Increase Noise Level
is relatively simple and the total costs are esti-
mated at about $20 extra per terminal.
From a radio-interference point of view, this type
This system does not provide full protection
of solution is the worst imaginable, but it is a
against eavesdropping but it is adequate in most
possibility. Manufacturers already have many
cases. This is especially true where a low or medium
problems in complying with statutory radio inter-
security level or privacy is required. such as in
ference limits, and it is therefore virtually impossi-
home applications and in most office applications.
ble to equip a unit or terminal with such a noise
The costs of the system are realistic in relation to
source.
the required security level. This solution was found
The only solution one might think of to prevent
as a result of our studies in this field. Patents on
eavesdropping in this way is to place a lot of
this method are pending.
equipment in one room, e.g. a large number of
terminals. Experiments have shown that this is not
a real solution. As noted earlier, the radiation 5. Measuring Methods and Requirements
pattern of a terminal is largely determined by
resonances at some frequencies. These resonances 5.1. Existing Standards
occur at different frequencies, even if two units of
the same type are chosen. This means that It can be safely assumed that equipment for mili-
eavesdropping on a cluster of video display units tary and government applications (security
 W. mn Eck / electromognet~ Radiation jrom Vtdeo Dqlay L’mts 275

services) is tested according to stringent standards. 5.2. Usability

Apparently two types of standards exist:
l the NACSIM 5lOOA .Tempest Standard The aforementioned standards are clearly not very
(U.S.A.), and suitable for non-military and non-government ap-
l the AMSG 720B Compromising Emanations plications. especially in cases where a low or
Laboratory Test Standard (NATO). medium security level is required. This is not only
Both standards are applicable to all types of the case because the methods and requirements are
equipment, not only to video display units or not freely available, but also because the require-
terminals. Measurement methods and require- ments are probably too stringent for these applica-
ments in the NACSIM standard are unknown to tions, resulting in unacceptable costs. We have
non-Americans. According to the scarce informa- therefore developed a simple method for testing
tion available [7] equipment is tested under the video display units or terminals in this respect.
surveillance of a special committee (TQSC). If
approved. the equipment may be placed on the 53. Simple Measuring Set-Up
“Tempest preferred product list” (PPL). Once
listed on the PPL, the equipment may not be The aim of the measurement set-up is to check the
exported or sold to the public without U.S. reconstructability of information displayed on a
governmental approval. unit or terminal by means of a normal TV re-
Not long ago (1982), NATO defined its own ceiver. Since various sources of radiation occur,
“Tempest” standard. This AMSG standard is in this reconstructability is not determined only by
special cases used for both military and govern- the radiation level produced [l]. Therefore a nor-
mental applications in NATO countries. As docu- mal TV receiver should be used as a measuring
ments relating to the standard are classified, the instrument. The measuring distance from the unit
information is not freely available. It is unknown or terminal under test will in that case be the
how this AMSG standard relates to the NACSIM variable which determines the stringency of test-
standard but the measurement procedures and re- ing: the information on the screen is required not
quirements defined in the former are known to be to be reconstructable on a normal TV receiver at a
mainly of U.S. origin. distance larger than d metres, where d is de-

Distance on calibrated
measuring site

VDU

Sync.
Combined synchronization signals (via optical fibre)
recovery
circuit

Number of
screen I ines

Fig. 2. Measurement set-up.

 termined by the tester. This type of measurement transformer in most cases is driven at the horizon-
set-up has disadvantages. including: tal synchronization frequency of the display unit
0 reconstruction (reception) is only possible within or terminal as in a normal TV receiver.
the TV broadcast bands, and As described in the Technical Appendix, the signal
l the measurement sensitivity is dependent on the is picked up and filtered, and stabilized in a phase
type of TV receiver used. locked loop. The vertical synchronization
Therefore a measurement set-up according to Fig. frequency is obtained from the horizontal synchro-
2 was used. nization frequency by division through the number
The video display unit or terminal under test is of screen lines on the video unit or terminal. Both
placed 1 metre above the earthed conductive signals are combined and fed into the synchroniza-
ground plane of a measuring site according to tion separator of the TV receiver. The signal is
CISPR Publication 16. The antenna signal from a transmitted via an optical fibre to prevent dis-
calibrated measuring antenna (e.g. a dipole) is fed turbance of the radiated high frequency field of
into a receiver suitable for measurements in the the video unit or terminal during measurements.
range of 30 to 1000 MHz. The IF signal of the
measuring receiver is used as an input to the TV
6. Conclusions
receiver, thus using the former as a frequency
convertor. If the TV receiver is tuned to this IF 1. Video display units or terminals generate
frequency, it is possible to observe whether recon- electromagnetic fields with frequency compo-
struction of information from the received signal is nents up into the UHF region which contain
possible. If the IF frequency is located outside the the harmonics of the video signal.
TV broadcast bonds, it must be converted to an A normal TV receiver made suitable for this
arbitrary VHF channel as shown in Fig. 2. This purpose will in some cases be able to restore the
cascading of two receivers has several advantages: information displayed on a video display unit
measurements can be carried out over the entire or terminal on its own screen, when this field is
frequency range 30-1000 MHz because the mea- picked up. Depending on the type of video
suring receiver determines the measuring display unit or terminal, this reconstruction may
frequency; under optimum conditions be feasible from dis-
the measuring receiver determines the sensitiv- tances of up to 1 km.
ity of the entire set-up; and 3. The information in video display units or termi-
field strength measurements can be carried out nals will not be detectable at such distances if
simultaneously, thus enabling comparison be- an electromagnetic shield is applied. Adequate
tween picture reception quality and ambient shielding of the electromagnetic fields gener-
fieldstrength. ated may double or even triple the price of a
Since the IF signal is filtered at the detection video display unit or terminal.
bandwidth of the receiver, a detection bandwidth 4. If the writing sequence of screen lines of the
of at least 1 MHz should be selected. In order to video display unit or terminal screen is changed
obtain a clear picture on the TV screen, the band- into a random sequence, reconstruction on a
width of the measuring receiver should be at least normal TV receiver is made impossible. The
4 MHz. With a 1 1MH.z bandwidth, a page of text costs of this type of video display unit or termi-
is not clearly readable but will be recognized as nal data protection are estimated to be much
such. At bandwidths smaller than 1 MHz the lower than those of electromagnetic shielding.
picture on the TV screen will hardly be recogniz- 5. The measurement method developed can be
able as a page of text. used to obtain information on the reconstructa-
In contrast to the eavesdropping situation, the bility of the data displayed on a video display
video display unit or terminal is available during unit or terminal, at a predetermined measuring
measurements, thus enabling the pick-up of syn- distance. The set-up is simple and measure-
chronization signals directly. This can be done ments do not take an unreasonably long time to
easily by picking up the magnetic field from the be carried out.
high voltage transformer, close to the unit. This
 W. LWI Eck / electromagnetrc Radiutton from Cideo Dtspluy L’nrts 277

Technical Appendix

Principles of Television rection of flow will change that direction. If a field

is generated in both the horizontal and the vertical
Picture Build- Wp direction by applying voltages between the steer-
ing electrodes, the place of the lighted spot on the
The picture on a television screen is built up screen can be changed. It is easy to see that the
sequentially. The moving picture is the result of 50 application of sawtooth-voltages with different fre-
frames (European standard) being displayed per quencies leads to the movements of the spot over
second. Each picture consists of a number of the screen as shown in Fig. 3. Thus. the picture on
horizontal lines which are so close together that a television screen is built up by modulating the
individual lines cannot be recognized uhen look- light intensity of a spot moving over the screen in
ing at the TV screen from a reasonable distance. the predetermined manner.
These horizontal lines are written on the TV screen
in a predetermined sequence: the first line is writ- Video Signal
ten at the top of the screen and the last line at the
bottom. The individual lines are written from left The signal required to modulate the light intensity
to right on the screen. of the moving spot is called the video signal. A TV
Each TV picture is built up as illustrated in Fig. broadcast receiver receives this video signal from
3. The device used to build up the TV picture is the transmitter. To enable the TV receiver to de-
called a Cathode Ray Tube (CRT) (see Fig. 4). code the received video signal into a readable
If a high voltage is applied between the (heated) picture it has to ‘know’ at what moment in the
electrode (cathode, negative voltage) and the con- received signal the information on each line starts.
ductive layer on the inside of the screen (anode, As the picture is built up according to a prede-
positive voltage), electrons will start flowing from termined scheme, information on the starting mo-
the cathode to the anode. ment of each first line of a frame and all the
As a result of the application of a magnetic following lines has to be transmitted as well. The
field around the ‘foot’ of the CRT the electron signal used to feed this information to the TV
flow is guided into a very narrow beam. If the receiver is called the synchronization signal. For
voltage is high enough, the kinetic energy of the practical reasons the video signal and the synchro-
electrons is so large when reaching the screen that nization signal are combined into one signal, the
the screen will emit photons of visible light. Thus, line feed [LF] signal. The LF signal in a TV
by controlling the voltage applied between the receiver consists of:
cathode and the anode, the light intensity of a - a positive part. the video signal, and
- a negative part, the synchronization signal (pulse
given spot on the screen can be controlled. Since
electrons are charged particles, application of a train).
magnetic or electric field perpendicular to the di- The signals can be combined into one signal,
because the synchronization signals can be trans-
mitted in between separate lines and frames. A
Inior-ac;on ir,ti”g small part of an LF signal in a TV receiver is
__--/--- --
---M---HOTIZ:Pial iiv9aci shown in Fig. 5.
_____----
__----- --- -‘,er:.Zal ?I,,>dCL
-- --
__- ----
___-- ----
___--- -- Video Display Units
___----- ---
__--a--
_---4-- --- Picture Build- Up
1__
The build-up of the picture on a video display unit
Fig. 3. Sequential frame build-up in screen lines. is much the same as in a TV broadcast receiver.
278 W. oan Eck / Eiecrromagnerrc Rodiarron from Video D~splux C‘ntrs

Vertical deflecrior 2lec[ro3es

Porizonial deflec:ion e!ec:roae;

Cd those i-1 /Electron bear

&

Fig. 4. Cathode ray tube.

Line (k-l) Line k Line I Line 2 Line 3

‘Whir? :evel
Video tit Blaci levzl
LJ I u Ll
Svnchronization
L i ne Frame Line Line Sync. = ‘Jl:ra black
sync. sync. sync. sync.
-Time

Fig. 5. Line feed [LF] signal in a TV receiver.

Fig. 6. VDU screen with text. Fig. 7. Detail of Fig. 6.

 W. cm Eck / elecrromugner~c

Fig. 6 shows the image displayed on a video The modulation process is easily performed by
display unit, Fig. 7 is a close-up photograph of the applying both signals to a logical AND. since the
screen which shows that the symbols displayed signals can onlv have the values 0 and 1. Evi-
consist of small dots. These dots (pixels) are dently. a number of adjacent pixels in a horizontal
arranged in horizontal lines. just as in a normal row are written on the screen as individual pixes.
TV receiver. However. the optical size of the pixels is so large
that this is not noticed when looking at the screen
Video Signal from a reasonable distance. The fact that horizon-
tal lines are also displayed as a row of individual
A graphic representation of the shape of the video pixels is important for the detectability of the
signal needed for a picture is shown in Fig. 8. To information by means of a TV receiver. as will be
build up the display in pixels, the current of the shown in the section on detection of information.
electron beam is on-off modulated. Thus, the video
signal in a VDU is a digital signal. a logical “one” Spectral Contents
producing a “white” spot on the screen and a
logical zero preventing this spot from appearing.
If the text displayed on the VDU screen is non-
The initial video signal is shown in Fig. 8b. corre-
repetitive, the signal may on a first approximation
sponding to the shaded screen line in Fig. 8a.
be considered as a random digital signal. The
To obtain the required resolution on the VDU
power spectral density of this signal is given by:
screen, the bit duration in the video signal should
2
be short. If the bit duration were to be as long as
shown in Fig. 8b, the pixels on the screen would s.,,(f) =A y { V’,‘Hz}
! i
be distorted into ovals instead of circles, because
of the scanning speed of the electron beam. There- where T, is the duration of one bit in the final
fore the bit duration is decreased through modula- video signal, and A is a function of the number of
tion of the initial video signal 8b on a square wave pixels displayed on the screen and the signal am-
(the video-dot-clock) with the same period as the plitude in volts [2]. A part of S,,(f) is given in
bit duration in the initial video signal. In this way Fig. 9.
the bit duration is decreased to 50 per cent of the As the video signal in a VDU can only be
original value, as shown in Fig. 8c. realised with finite transition times, 7;. the real

C)
n l-l nnn nnnn
a) Screen build-up in pixels

b) Initial video signal (optical intensity)

c) Final video signal (electron bean intensity)

Fig. 8. Build-up of the video signal in a VDU.

Fig. 9. Power spectral density of the video signal

power spectral density of the video signal is de- spectral lines are far apart. each of the lines may
scribed more appropriately by: be looked upon as an independent narrowband
source for our purposes. The intensity of these
T,.(/) = S,,(f).
(q-y
l +1
{ V’/Hz} narrowband components decreases with increasing
frequency, as is the case with the video signal’s
thus: power spectrum. Because the power spectral den-
sity of the clock signals is concentrated in individ-
ual spectral lines, rather than spread over the
entire frequency axis, the power in each of those
lines can be fairly high compared with the power
In the above expression the last factor denotes a density in the video signal (depending on the
first order low pass filter characteristic with cut-off measuring bandwidth).
frequency fi = l/rT,. It is readily seen that the What is even more important, the clock signals
envelope of the power spectral density of the sig- are often obtained by frequency division of the
nal is fairly constant up to the frequency f, = video-dot-clock. This means that many spectrum
l/rT,,. from which frequency it will be decreasing lines will coincide with the centre of a lobe in the
at a rate of - 20 dB per decade down to f2 = 1/~7;. video signal. This phenomenon has a great impact
At frequencies higher than f1 the spectral density on the detection of information with a TV re-
decreases at a rate of -40 dB per decade. ceiver.
Generally, the frequency f, is in the range 20 to 50
MHz. and fz is in the range 200-500 MHz, de-
pending on the type of components and circuits Electromagnetic Radiation
used.
Principles

Other Signals It can be derived from Maxwell’s equations that

the acceleration of electric charges results in the
The video signal is not the only signal in a VDU. generation of an electromagnetic field. This phe-
It will be shown in the next section that the video nomenon is well known, and it is used to our
signal is the most powerful broadband contribu- advantage in radio communications. A current is
tion to the radiated emission. T:iis is because it is forced to flow through a conductive wire (antenna).
the only signal in a VDU that is amplified to far This current will generate an electromagnetic field
above TTL level. It is therefore unnecessary to go which can be picked up at a large distance from
into the nature of other broadband signals in a the transmitting antenna using another conductive
VDU for this analysis. wire in which the field will generate a current. It
Because clock signals are repetitive, their power can be proved that the current generated is similar
spectrum consists of individual spectral lines at the to that in the transmitting antenna.
add harmonics of the clock frequency. Because the This means that any conductor carrying a cur-
 W. can Eck / electromaqnr~tc Raduttron from Cideo Dlsplyc L’nrts 281

rent with varying strength (alternating current) can Emission Measurements

be regarded as a transmitting antenna. Digital
equipment will therefore generate electromagnetic 1Measurements were made in respect of the inter-
fields containing all frequency components of all ference produced by a VDU. The video-dot-clock
signals inside the equipment. Since the electromag- of the VDU chosen was 11.004 AMHz. The system
netic field is generated by the acceleration of clock frequency was 1.57 MHz, so narrowband
charges, its strength is related to the derivative of components could be expected to occur in the
the current in a circuit. rather than to the current radiated spectrum at 1.57 MHz intervals.
itself. Two types o’f measurement were carried out:
l The maximum available interference power on
Radiation o/the Video Signal the mains power cord was measured using the
CISPR absorbing clamp.
It is obvious that the video signal in a VDU is l The electric field radiated by the VDU in the
radiated into the surroundings of the equipment direction of maximum radiation was measured
by the video processing circuitry. and by the elec- with a biconical antenna according to MIL-
tron beam in the CRT. On a first - rough - STD-461/462 at a distance of 1 metre. In this
approximation, the radiation effectiveness of a set-up the main power cord was shielded.
circuit increases monotonously with frequency at a IMeasurements were made with an HP 8586A spec-
rate of +20 dB per decade, up to a frequency (of trum analyser in the frequency range 30-300 MHz.
several hundred MHz) which is determined by the at a detection bandwidth of 10 kHz and the func-
physical size of the circuits used for video signal tion ‘MAX HOLD’ having been selected.
processing [3]. The results are given in Fig. 11 and Fig. 12. The
If it is assumed that this cut-off frequency is upper portions of Fig. 11 and Fig. 12 illustrate the
higher than fi, the radiated power spectral density measurement results for a full screen of text of a
S,.,.(f) can be estimated by: VDU. The lower portions of Fig. 11 and Fig. 12
illustrate the VDU screen on which only the cursor
S,,(i) : : sin’(7TfTh) ( V’/m’Hz}
is displayed. These results show that:
where f, >f> f2. A part of this spectrum is shown 0 the level of broadband interference is largely
in Fig. 10. dependent on the number of characters dis-
It can be derived from communication fheory played on the screen:
that if the receiver is tuned to one of the lobes of 0 the level of narrowband interference is indepen-
the spectrum, the entire video signal can be re- dent of the contents of the display, and individ-
stored. This can be made plausible if each of the ual narrowband components are determined by
radiated lobes of the power spectral density is the VDU system clock and the video-dot-clock.
regarded as an AM-modulated version of the line It can thus be concluded that the video signal is
feed portion of the video signal up to the Nyqvist the most powerful source of broadband emission.
frequency f,v = 1/2T, [4]. and that the clock signals are the most powerful

S,,(f)
SpeCtraI
intensity
(LOG-axis)

1/T -Frequency (Lin. axis)

b

Fig. 10. Power spectral density .S,,(/) of the radiated field (/ < fi).
 ,Paxi-u---OiSe level

30 43 60 100 :20 140 160 133 200 220 240 260 13: 300
-Freqe-;:. :rcz)

-20 4 I L I I
30 40 60 80 100 120 140 160 I80 200 220 240 260 ?t: 300

-FreqLeTci !YHz)

Fig. 11. Maximum interference power available on the main power cord.

iieldjtrengih
(d’d_V/?i aandwidth IOkHz, VOU screc- ‘_‘I !

i 60 -__-

a-- 30 40 60 80 100 120 140 160 130 200 220

1

240 260 ;:: 300

I

-Freque-:v !Wz)

Fieldstrength
(dBeV/n)

30 40 60 80 100 120 140 I60 rao 200 220 240 260 2:: 300

-Freqe-zy !fiHz)

Fig. 12. Fieldstrength in the direction of maximum radiation at 1 meter distance (horizontal polarization).
sources of narrowband emission. The measure- Fig. 14 shows that the TV receiver will not
ments clearly show that the emitted broadband notice the difference between the radiated signal
spectrum does not follow the assumed sin’( ;;fr,) (solid line) and the video signal which has the
function. Especially in Fig. 11 it is clear that at same spectral density at the reception frequency of
some frequencies (e.g. around 125 and 210 MHz) the TV receiver (dotted line). The band filtered out
resonances occur which cause the emission to in- by the detection filter of the TV receiver is dis-
crease to 15 dB above the emission level at adja- played as a shaded block in the same Figure.
cent frequencies. The response of the TV receiver to the radiated
signal can thus be computed using the entire video
Radio Interference Limits signal as an input signal to a TV receiver. The
Measurements of the electromagnetic fieldstrength amplitude of this signal is chosen at such a value
generated by various types of VDUs were carried that the power spectral intensity of the signal is
out on a measuring site as described in CISPR equal to that of the radiated signal at the reception
Publication f6. Field strength measurements frequency. The signal processing in the TV receiver
according to the forthcoming CISPR recom- is visually represented in the time domain in Fig
mendation on data processing and office 15.
equipment (DP/OE) showed that none of the Fig. 15, shows the input video signal.
VDUs we tested produced electromagnetic inter- Fig. 15b shows the IF signal in the TV receiver
ference beyond the proposed limits. In these mea- in response to this video signal.
surements the observed frequency range was In Fig. 15, the LF signal in the TV receiver is
30-600 MHz. shown; due to the AM detector this is the envelope
of the IF signal.
Fig. 15d displays the video signal in the TV
Reconstruction of Information receiver with optimum adjustment of the brigth-
ness and contrast levels.
Signal Processing in TV Receicer
The amplification of the LF signal around the
In its simplest form a TV receiver can be described threshold, determined by the brightness level, is
according to the block diagram of Fig. 13. As can determined by the contrast level. On a first ap-
be seen in this block diagram, the TV receiver can proximation the contrast level determines the
only see a very small part of the spectrum radiated steepness of the flanks in the final video signal in
by the VDU, with a bandwidth of approximately 8 the TV receiver. In contrast to the screen build-up
MHz 5, at an arbitrary frequency someuhere in in a VDU. the maximum of the video signal
the VHF or the UHF region. determines the black level, whereas the minimum
determines the white level. The picture on the TV
’ Normally a TV receiver is equipped with a VSB demodulator
and a detection bandwidth of approximately 4.5 MHz. This screen is therefore a copy of the picture on the
is effectively equal to an AM detector and 8 MHz detection VDU screen and is composed of a white (or gray)
bandwidth. background with black letters.

HF IF LF Initial Final
input signal signal video video

1
-

Fig. 13. Simplified block diagram of a TV receiver.

 IIIlIIIIIIIf:-lIIIIII’I
-irequency

Fig. 14. Video signal and radiated electromagnetic field. with equal amplitudes at the reception frequency of a TV receiver.

Initial ,:ileo signal (in a VDU)

h
c) LF
Contrast

signal
level

in the TV receiver

IF signal in the TV receiver d) Video signal in the TV receiver

Fig. 15. Signal processing in a TV receiver.

I
Video signal
0
in :he VDU

Due to

Video signal
in the recei,/er

Fig. 16. Reconstruction of line elements.

 b IF signal in
TV receiver

C LF signal

d Video signal

Fig. 17. Signal processing in a TV receiver with coherent narrowband component available.

e(t)= la(t)+BI
Fig. 15 shows that a horizontal line element on
the screen of the VDU is composed of a number Assuming 1a(t) 1 > B, we obtain:
of adjacent pixels which leads to reconstruction of
e(t)=a(r)+B
the video signal for line elements. Since the elec-
tromagnetic field generated is related to the Demodulation of this type of signal is shown in
derivative of the video signal, only the leading Fig. 17. In comparison with reception under ab-
edge and the trailing edge of a long pixel would be sence of the narrowband component. there are two
displayed on the TV receiver as a dot. The effect is advantages:
shown in Fig. 16. 1. The total signal power received by the TV set is
determined by the sum of C(r) and B.
Influence of Narrowband Components 2. The signal can be better reconstructed because
the dynamic range of e(r) has increased rela-
The IF signal in the TV receiver as illustrated in tively.
Fig. 15b can be quantified as:
First Measurements
V(t) = a(t) .cos( y.t),

where W,. is the frequence of modulation, which is First measurements during the experimentation
equal to the centre frequency of the receiving showed that a TV receiver will indeed restore the
filter. If the received narrowband signal is an odd video signal of the VDU, although the image does
harmonic of the video-dot-clock, its frequency un- not appear on the screen because of the lack of
der optimum reception will be u/;.. The received synchronization information in the received signal.
signal can then be described as: This is evident if Fig. 17d and Fig. 15d are com-
pared with Fig. 5.
V’(t)=a(t) cos(W,.t)+Bcos(w.t++),
The synchronization signals are separated from
with B and #I being constants. the video signal in the video separator. When the
Or, assuming Q = 0: TV receiver is tuned to a broadcast station, the
synchronization signals are explicitly transmitted;
V’(t)= [a(t)+B] cos(W,.t).
thus the receiver is able to restore the synchroniza-
Envelope detection in the TV receiver now leads tion of the received information. Normally a VDU
to: does not radiate such a smart signal. The TV
receiver picking up the signal radiated by the [2] A. Papoulis: Probability, random variables and stochastic
VDU is thus mable to synchronize on the signal ~~OCCSS~S. Xtdh~ Hill, 1965.
[3] D. White: ELfI control methodology and procedures. Don
received.
White Consultants ,nc. Gamss\~lle. U.S.A.. 19St.
[4] K. Shanmugam: Digital and analog communication sys-
tems. Wiley and Sons. 1979.
[S] R. Freeman and M. Sachs: Electromagnetic compatibility
References
design guide. Artech House (ISBN O-890006-114-9). 1982.
(61 H.W. Ott: Noise reduction techniques in electronic systems.
[l] W. van Eck. J. Neessen and P. RiJsdijk: On the electromag- Wiley and Sons. 1976.
netic fields generated by video display units. Proc. symp. [7] A. IMauriello: Join a government program to gain Tempest
EMC. Zurich. March 1985. expertise. EMC Technology, Vol. 3. July 1984.