See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/237764745

Experiments Towards an Erasable Compact Disc Digital Audio System

Article in Journal of the Audio Engineering Society. Audio Engineering Society · July 1984

CITATIONS READS

5 984

2 authors:

Kees Schouhamer Immink Joseph Braat

Turing Machines Inc Delft University of Technology
301 PUBLICATIONS 3,487 CITATIONS 229 PUBLICATIONS 2,448 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Kees Schouhamer Immink on 28 October 2019.

The user has requested enhancement of the downloaded file.

ENGINEERING REPORTS

Experiments Towardan ErasableCompact Disc Digital

Audio System*

K. A. SCHOUHAMER IMMINK AND J. J. M. BRAAT

Philips Research Laboratories, 5600 JA, Eindhoven, The Netherlands

Recording experiments of erasable information in magnetooptical amorphous layers

on pregrooved disks are described. The recording and reading of information is done
on a recorder provided with an AIGaAs laser and equipped for polarization-sensitive
readout of the Kerr effect. The signal-to-noise ratio and the bit error rate of the recorded
information have proved to be sufficient for half an hour of digital music according to
the Compact Disc digital audio standard on a 120-mm-diameter disk (equivalent to
400 00 bit/mm2). With only slight modifications the recorder can be used to read out
normal Compact Disc records.

1 INTRODUCTION ercive force of the layer is decreased by a temperature

rise, and with the aid of a small external magnetic field
Many research groups in the world are now inves-
the magnetization of the layer is locally reversed. The
tigating magnetooptical (MO) thin films for erasable
influence of the internal perpendicular magnetization
storage applications. Early experiments concentrated
of the layer is to rotate the plane of polarization of
on ferrimagnetic alloys of GdFe and GdCo [1]. Later
plane-polarized light reflected by this layer. The rotation
ternary alloys have been investigated in order to optimize is positive or negative, depending on the polarity of
the magnetic and magnetooptical properties [2], [3].
the internal perpendicular magnetization of the layer.
System applications can be divided into two branches:
The top-top rotation is typically 0.7 °. Fig. 1 shows a
low-density storage for computer applications and high-
density storage for applications in the professional and cross section of the actual magnetooptical disk. On a
consumer fields (such as digital audio), glass substrate a 2P (photo polymer) layer has been
At Philips Research Laboratories we have constructed deposited with a groove structure in its top surface. A
thin dielectric layer separates the 2P layer from the
a magnetooptical recorder to study the limits and char-
magnetooptical storage layer. It consists of an amor-
acteristics of these storage materials from a system phous alloy of Fe, Gd, and Tb and is approximately
point of view. So far we have concentrated on alloys 100 nm thick. This layer has been vapor deposited and
of GdFeTbLaboratories
Research that have been vapor deposited
in Hamburg. In this at Philips
paper we shows an internal magnetization that is perpendicular
describe the experiments we have done to establish the to the surface. On top of this storage layer a protective
layer is deposited. The disk is illuminated through the
maximum achievable information density. Important glass substrate. The entire disk can be sandwiched with
parameters determining this figure, such as signal-to-
noise ratio and bit error rate, are dealt with. a second glass or plastic substrate for protection against
handling errors.
1 DESCRIPTION OF THE MAGNETOOPTICAL
1.2 The Optical Recorder
DISK AND THE OPTICAL RECORDER
The optical recorder has to perform two functions:
1.1 The Magnetooptical Disk 1) Local heating of the bit locations
The recording of information is done by locally 2) Polarization-sensitive detection of the created
heating the amorphous magnetooptical layer. The co- magnetooptical domains.
A schematic drawing of the optics is shown in Fig.
* Presented at the 73rd Convention of the Audio Engineering 2. The light source is a high-power A1GaAs laser with
Society, Eindhoven, The Netherlands, 1983 March 15-18. a wavelength of 850 nm. Approximately 40% of the
J. AudioEng.Soc.,Vol.32,No.7/8,1984July/August 531
IMMINK
ANDBRAAT ENGINEERING
REPORTS

light output is usefully captured by an objective lens modulation by the analyzer and detected by an avalanche
with a numerical aperture of 0.3. The parallel beam detector. During recording and reading a small part of
traverses a "sun-glass" sheet polarizer, a 10% and a the light (10%) is coupled out to the tracking (push-
25% mirror, and is focused on the disk by means of an pull) and focusing optics (Foucault double-wedge
objective with a numerical aperture of 0.6. The half- method). An automatic gain control is incorporated in
width of the light spot is slightly smaller than 1 txm. order to compensate for the large difference in light
During information recording the laser is pulsed with power during recording and reading. An oscilloscope
a pulse duration of 50 ns at intervals of 250 ns. The trace of the digital signal delivered by the avalanche
peak power is 60 mW; due to losses in the light path, detector is shown in Fig. 4. Local erasure of the in-
10 mW is available in the focused light spot. formation is done in two steps:
A magnetic domain generated by a single light pulse 1) While the laser is delivering pulses every 250 ns,
has a circular shape with a diameter of typically 1 Ixm. the external magnetic field is reversed. The whole track
Oblong domains are generated by several light pulses surface is now magnetized in the original direction.
being delivered at 250-ns intervals. The size of the 2) New information is recorded in the normal way.
domains is slightly influenced by the strength of the It has been measured so that the information in neigh-
magnetic field generated by the coil behind the disk. boring tracks is unaffected by the erasure of a track.
Depending on the coercive force of the storage layer The signal-to-noise ratio remains unaltered.
at room temperature (typically 80 kA/m), this auxiliary
field varies between 10 and 20 kA/m. 2 SIGNAL-TO-NOISE RATIO CONSIDERATIONS

Fig. 3 shows a polarization-microscope photograph Since the magnetooptical effect is small, the signal-
of written domains in the pregrooved spiral on a disk. to-noise ratio of the detected signal is limited. Fig. 5
The domain sequence represents a typical Compact- shows the detection arrangement consisting of an and-
Disc-like signal with oblong domains of discrete lengths, lyzer (detection angle 13) that is almost crossed with
The minimum domain length (not the bit length) is 2
respect to the incident plane of polarization. The azimuth
i_m; the spacing between tracks is 1.7 i_m. of the plane of polarization is modulated by the magnetic
When detecting written domains the laser is pulsed domains (top-top excursion 20 is 0.7°).
at a high frequency (13 MHz) with a duty cycle of 0.25
When the instantaneous azimuth of the plane of po-
and a peak power of 8 mW. On the disk a quasi-con- larization is represented by an angle a, the light power
tinuous power of 0.3 mW is available, which is low
after passage through the analyzer is given by
enough not to perturb the recorded domains. The mirror
with a 25% reflectivity throws reflected light onto an
analyzer whose transmitting direction is almost at 90 ° P(13, a) = Po sin 2 (13 - a)
with respect to the entrance polarizer. The rotation of
where Po is the light power incident on the analyzer.
the plane of polarization of the reflected light due to Supposing a and 13to be small and 13not equal to 0,
the magnetic domains is converted into an intensity we approximate this expression by

Si02 / protectivelayer P(13, a) = Po sin2 13( 1 2ainu)

t_-n _ '
GOTb
Fe_/ dielectric layer
_2P(-10JJm) In the case of harmonic azimuth modulation (a =
glass substrata
0 cos tot) we obtain the following expression for the
detector current:

Fig. 1. Cross section of a magnetooptical disk. i(13, t) = i0 sin 2 13 ( 1 tan2013II,cost tot )

aisc where i0 is proportional to the light power incident on

to trackingand C_ .T, -- the analyzer and i· (< 1) accounts for the decrease in
focusing optics avalanche H
detector r.J
signal amplitude due to the finite optical bandwidth.

AIGoAs /
' / LV_
Il-
Pi
modu,a,on
given by
d pth
theetecto'sr,_ow
laser _ ' '( J_ g"-kJ
'_ lin I \ .
NA=0.3/ T=90% T=75 Yo NA=0.6 J',lmagnetic m -- JJ, .
polarizer H co,I tan13
The noise current consists of several contributions:
Fig. 2. Optical light path for recording and polarization- 1) Thermal noise
sensitive readout. optical
from a standard When the
diskanalyzer
(such asisaremoved,
Compact the signals
Disc) can 2) Amplifier noise
bedetected. 3) Darkcurrentnoise

532 J. Audio Eng. Soc., Vol. 32, No. 7/8, 1984 July/August
ENGINEERING REPORTS ERASABLE COMPACT DISC DIGITAL AUDIO SYSTEM

4) Photon shot noise iav denotes the average detector current, N 2 is the var-
5) Surface noise of the disk. iance of the detector current, and B is the measuring
When an avalanche detector is used, the photon noise bandwidth.
and the surface noise become predominant. Assuming the noise to be the superposition of photon
A general expression for the carrier-to-noise ratio and surface noise, we obtain
CNR is
N 2 = 2qiav + Clx2iav 2
CNR = 1/2S2/BN 2
where q is the elementary charge andC a constant that
where S is the amplitude of the signal given by characterizes the surface noise power. The photon noise
contribution has a white spectrum, but the surface noise,
S = miav , as the signal itself, is affected by the optical modulation

Fig. 3. The magnetic domain structure as seen in a polarization microscope with almost crossed polarizers.

Fig. 4. Eye pattern of detected magnetic domain structure. The tangential velocity approximately equals 3 m/s.

J. Audio Eng. Soc., Vol. 32, No. 7/8, 1984 July/August 533
IMMINK
AND
BRAAT ENGINEERING
REPORTS

depth ix. teristics of the recording medium. The main target is

The expression for the carrier-to-noise ratio becomes space efficiency, that is, to obtain the highest infor-
mation density permitted by the limiting characteristics
1/2m2iav
2 of the recording channel. Modulation systems, some-
CNR =
B (2qiav + Cix2iav 2) ' times called channel codes, can be designed to match
a wide scale of specific requirements· For example, a
In order to have an estimate of the surface noise channel code can be so designed that only domains of
contribution we can rely on data measured on VLP unit diameter are recorded (domain position modula-
disks, for example [4]. In this situation photon noise tion). Other channel codes transform the data stream
can be neglected, and putting m = Ix, the expression in such a way that the information is contained in the
for the carrier-to-noise ratio equals lengths and the position of the magnetic domains. All
these channel codes have their particular pros and cons,
1 which can only be determined by many experiments.
CNRvLp - 2BC During our experiments with the magnetooptical storage
medium we studied the behavior of a large class of
A typical value is 57 dB measured over a bandwidth modulation systems. As an example of domain-length
of 30 kHz at a scanning velocity of 3 m/s, and the value modulation we studied the class of so-called run-length-
of C can be derived from these figures, limited sequences.
In the case of detection of magnetic domains we A binary string of bits is defined as run-length limited
obtain the following expression for the carrier-to-noise if the number of consecutive ls (or Os) is bounded
ratio: between certain minimumand maximumvalues. For
example, the Compact Disc code EFM is so designed
1 (202ix2/tan 2 13) iav that only strings of at least three and at most eleven
CNR - B 2q + Cix2iav consecutive bits of the same polarity are allowed as a

The average detector current depends on the light 2&e

_

power incident on the detector and equals l_._polarizer

· PL
tar - hv _q sin2 13

where hv is the photon energy and 'q the quantum ef-

ficiency of the dectector (_ 1). Fig. 5. Orientation ofpolarizer and analyzer for the detection
The light power incident on the analyzer equals of magnetic domains. The domains induce an azimuth mod-
ulation of the plane of polarization with an amplitude of 20.
P[ = RTPo
Numerical data on surface noise
where Po is the average power (0.3 mW) incident on curve a: CNR infinite
the disk, T the transmittance of the light path from the curve b: CNR = 60 dB
disk to the analyzer (typically 0.15), and R the reflec- curve c: CNR
measured over =3057 dBat tangential speed of 3 m/s
kHz
tivity of the storage layer (50%).
With a measuring bandwidth of 10 kHz the final ICNRtdB)
expression for the carrier-to-noise ratio becomes (ix _ 60
_
1) o

CNR = 1 + 4.5 x lO13Csin2[_ 40

30
In Fig. 6 this theoretical carrier-to-noise ratio is shown
as a function of the analyzer angle [3. The carrier-to- 20
3.3 x 105 severalvalues
noise ratio is depictedfor cos 2 [3 of the disk 50
10NN_,_
_ __,_, ,
surface noise. The optimum for a standard disk (curve o* 10° 200 300 to* 500 600 700 800 90°
c) is seento be 50 dB and is reached at analyzerangles ' @
close to 0°. Fig. 6. Theoreticalcarrier-to-noiseratio of signaldue to
magnetic domains. Curve a applies to a perfect disk without
3 DESIGN CONSIDERATIONS AND surface noise. The noise is due only to the shot noise of the
photons. Curves b and c show the influence of the surface
MODULATION SYSTEMS noise of a disk. The optimum detection angle is seen to be
close, to 0° (complete extinction). A practical limit is 2-3 °
Modulation systems are normally used to adapt the because otherwise the detection becomes nonlinear (second-
incoming binary data stream to the particular charac- harmonic distortion)·

534 J. Audio Eng. Soc., Vol. 32, No. 7/8, 1984 July/August
ENGINEERING
REPORTS ERASABLE
COMPACT
DISCDIGITAL
AUDIOSYSTEM

modulator output. The backgrounds to the choice of to measure characteristic properties of the magneto-
certain maximum and minimum run lengths are briefly optical channel is the pulse-length distribution mea-
outlined in the following. If the clock regeneration is surement described in Sec. 4.3.
derived from the actual readout data, which is normally
the case, then level transitions yield the synchronizing 4.1 Measurementof Carrier-to-Noise Ratio
information. A small maximum distance between tran-
We recorded a pure frequency of 500 kHz on the
sitions benefits the worst-case clock regeneration. A
large minimum run length decreases the maximum fre- disk at different linear velocities. The frequency was
quency of the modulation stream, so that the bandwidth chosen equal to 500 kHz because the signal power of
requirements of the channel may be reduced, the EFM modulation system is maximum in this fre-
State-of-the-art high-power solid-state lasers have a quency region. We measured the different noise levels
limited time during which they can emit a light pulse, relative to the detected
over a bandwidth of 10signal
kHz. power, always
The noise integrating
level that de-
and they can only be used in the pulsed mode, which
naturally leads to modulation systems using (unity) termines the final carrier-to-noise ratio is due to surface
roughness of the disk. In nonrecorded regions and
domain position modulation. We can, however, use equally on an aluminum-coated disk this noise level is
domain-length modulation, with its much higher space identical. A characteristic Value is - 44 dB when the
efficiency, if we record the domain lengths as overlap- scanning velocity is 3.25 m/s (see Fig. 8). As should
ping unit domains, be expected in the case of surface noise, the noise level
In Fig. 7(a) we have depicted as an example a binary
modulation stream that has to be recorded on the disk. increases when the tangential velocity is decreased.
T is a channel bit time. The domain lengths are integral Other noise sources are photon noise, amplifier noise,
multiples i of T, with certain minimum and maximum and dark current noises. The photon noise can be mca-
bounds of i. By electronic means we derive a pulse sured separately from disk surface noise when the disk
sequence [Fig. 7(b)]. If the signal in Fig. 7(a) is high, is not rotating. This noise level is - 48 dB with respect
then a domain has to be recorded, resulting in the vis- to the signal. Dark current noise and amplifier noise
are at a level of at least -56 dB.
ualized domain pattern plotted in Fig. 7(c).
Designing and building real-time (de)modulators is 4.2. Measurement of Bit Error Rates
a time-consuming activity, especially when a wide range
of codes has to be tested. We proceeded therefore in a Following the procedure described in Sec. 4, we have
way used before in experiments with a Te-based storage detected bit error rates by comparing a detected bit
medium. We programmed a number of PROMs so that sequence with the originally recorded sequence. An
they contained short binary sequences satisfying acer- important parameter is the linear scan velocity that
tain modulation rule. All PROMs contained a frame influences the carrier-to-noise ratio and consequently
sync pattern of 27 bits and a sequence of 561 bits sat- the bit error rate. The measured 44-dB carrier-to-noise
isfying the modulation rule. The frame sync pattern ratio yields a bit error rate of 10 -5 or less, and the few
was identical in all experiments. The PROMs were
periodically read out with a clock frequency of 4 MHz. T
The angular velocity of the disk was adjusted to the :__
desired tangential information density. After writing .__ _ [_1 o
one or more tracks, we read out using the repeat track [ IIll Ill I Il b
feature. The signal from the avalanche detector was
further electronically processed (digitized, clock re- G _ _ ¢
covery, frame sync, etc.). The binary processed signal
was now compared with the output of a PROM con- Fig. 7. Electronic translation of a pulse-width modulation
into a signal usable for a pulsed mode laser.
taining the original sequence. The PROM was syn-
chronized by the frame sync pulses derived from the
recovered information. In this way we were able to
generate a sequence of bit errors that could be inves-
tigated for its statistics, such as number of single-bit CUR _10d8

errors, average error burst length, and so on.

4 EXPERIMENTAL RESULTS (dB)m

------_.__ /A
B
In this section we describe some of the experimental _
results obtained during our investigations. The main c
goal of all experiments was to characterize the mag-
netooptical channel establishing the maximum reliable 0 ' 2_0 ' ,-_0 ' 6_0 ' 8_0 ' lobo
information capacity. First we studied the noise levels ,. freq.(kHz)
and the main noise sources. Further we studied the bit
Fig. 8. Measured carrier-to-noise ratio (bandwidth 10 kHz)
error rates, that is, the ratio of bits that are in error to of signal due to magnetic domains. The upper noise level is
the total number of bits that are read out. A new tool due to medium noise.

J. Audio Eng. Soc., Vol. 32, No. 7/8, 1984 July/August 535
IMMINK
ANDBRAAT ENGINEERING
REPORTS

detected errors are nearly all fixed in position on the

disk and are thus due to disk imperfections. At lower /

scan velocities the carrier-to-noise ratio slightly de- [ 1

creases. A more serious consequence of the lower ve- ]1 l/ _ _ A , A
locity is the asymmetry in the recorded signal due to
the finite minimum length of a recorded domain. By
deleting the first laser pulse during recording from each
sequence of 3, 4 ..... 11 pulses we can establish a , .._ /_ ,1_ A · ,
new recording optimum at a reduced linear velocity. 5o0 _000 _500 2000 2500
The experiments showed that at a tangential speed of , tins]
2.4 m/s a bit error rate of 10 -4 is still possible. (a)

As stated in Sec. 4.2, the bit error rate is a measure

4.3 Measurement of Pulse-Length Distribution J J_
of the quality of the system, the disk included. The bit _,__-.A

system performance. Measuring the electrical pulse Lmls

widths in the disk playback channel and displaying
error rate is not always an adequatemeasure of the I _ I
their distribution can give insight into characteristics _
500 1000 1500 Ix2000
_ A2500
A
of the system. , t[nsl
The pulse-length distribution system measures, as (b)
its name indicates, the distribution of the lengths of

generates plots as depicted in Fig. 9. Each plot is divided

into an upper
the pulses and a between
(distances lower part depicting The
transitions). the system
length _f_f_ fk/,. A _/-,
distributions of the domains depending on polarity. To I __

distribution of an EFM sequence was measured directly,

without the disk channel [see Fig. 9(a)].
calibrate the horizontal axis of the plots, a pulse-length I _ 325rots
Some interesting properties appear in those plots. ,
500 1000 1500 /x._
2000 _ 2500
A
As mentioned, a serious consequence of a deviation ,. t[nsl
from thecorrect speed is that the symmetryis disturbed. (c)
The plot of Fig. 9(c) was recorded at approximately I
the nominal tangential velocity (3.25 m/s) and the plots I /_

1
of fig. 9(b) and (d) with velocities of 4.0 and 2.5 m/s,
respectively. To correct for the asymmetry, one (or j /xJ-,__,_
more) of the successive laser pulses can be deleted.

without and with this asymmetry correction. The upper 2.5m/s

and lower parts of the graphs shift with respect to each
Fig.
other.9(h)
As and(i)depictsthepulse-lengthdistribution
a consequence of an increasing density the I _ AJ.._,2500
soo 1000 1500 2000
intersymbol interference (or tangential crosstalk) in- = t[nsl
creases. Due to this phenonemon the bell-shaped area (d)
centered around q, t3, ts, etc. [Fig. 9(b) and (h)] spreads
out and the area centered around t2, t4, etc., is filled
up [see Fig. 9(d), (f), and (h) for tangential velocities
of 2.5, 2.25, and 2 m/s, respectively]. This results in
an unreliable bit length decision, and hence the bit _/k
error rate increases.
OS corr
4.4 Compact Disc Digital Audio System
2.5 mis
Initially bit sequences generated by a PROM were 500 1000 1500 2o00 2500
used as a data source. To explore the possibilities of _ tins]
the magnetooptical system channel we also investigated (e)
the use of Compact Disc music as a data source [5].
The preceding experiments showed that the CD channel Fig. 9. Pulse-length distribution as a function of linear density
code EFM, as described in Sec. 4, was quite feasible. (arbitrary vertical
Calibration axis) for
of horizontal different
scale. tangentialbetween
The distance speeds. tops
(a)
A standard Compact Disc consumer player was used corresponds to 231 ns. (b) 4 m/s. (c) 3.25 m/s. (d) 2.5 m/s.
to deliver the digitally encoded music. (e) 2.5 m/s, one pulse deleted for asymmetry correction.

536 d. Audio Eng. Soc., Vol. 32, No. 7/8, 1984 July/August
ENGINEERING
REPORTS ERASABLE
COMPACT
DISCDIGITAL
AUDIOSYSTEM

500 1000 1500 2000 2500 500 1000 1500 2000 2500
tins] = tins]

(f) (g)

500 1000 1500 2000 2500 500 1000 1500 2000 2500
.- t[nsl -. tins]

(h) (i)

Fig. 9. Pulse-length distribution as a function of linear density (arbitrary vertical axis) for different tangential speeds. (f)
2.25 m/s. (g) 2.25 m/s, one pulse deleted for asymmetry correction. (h) 2 m/s. (i) 2 m/s, one pulse deleted.

The high-frequency signal from the Compact Disc's music signal (up to 2 MHz) look promising.
photodiode was electronically processed and fed to the
magnetooptical recorder. 6 ACKNOWLEDGMENT
With a linear bit density of 40% of the Compact Disc The authors are indebted to many researchers at the
density the music was recorded and reproduced with Philips Research Laboratories. They particularly wish
the same performance as obtained from a Compact to thank M. Urner-Wille, K. Witter, and J. Reck of
Disc player. PFH Hamburgwho provided the magnetoopticallayers.
The experimental setup for recording the music is P. Vromans and J. Ramaker supplied the pregrooved
equally suited for playing back standard compact disks disks. R. Aarts and J. Kahlman designed the electronics
manufactured for a consumer Compact Disc player, of the experimental setup.
only the analyzer sheet must be removed.
7 REFERENCES
5 CONCLUSIONS
[l] P. Chaudhari, J. J. Cuomo, and R. J. Gambino,
Thermomagnetic recording on pregrooved disks (track IBM J. Res. Dev., vol. 17, p. 66 (1973).
spacing 1.7 gm) has been demonstrated with domain [2] M. Urner-Wille,J. Mag. Mat., pp. 15-18 (1980).
dimensions in the micrometer range. The signal-to- 1339-1340.
noise ratio of the detected signal is sufficient for an [3] M. Urner-Wille, P. Hansen, and K. Witter, IEEE
error-free recovery of a digital music signal according Trans. Magn., vol. MAG-16, no. 5 (1980).
to the Compact Disc standard. The information density [4] J. P. J. Heemskerk, Appl. Opt., vol. 17, p. 2007
in the track direction is 40% of the density on a Compact (1978).
Disc. At the present time the density is limited by disk [5] L. B. Vries, K. A. Immink, J. G. Nijboer, H.
Hoeve, J. Timmermans, L. M. Driessen, T. T. Doi,
surface roughness and not by random variations in do- K. Odaka, S. Furukawa, I. K. Iwamoto, Y. Sako, H.
main dimensions. Ogawa, and T. Itoh, "The Digital Compact Disk:
Erasure of recorded domains has been accomplished Modulation and Error-Correction Schemes," presented
in two stages (one erase step and one writing step), at the 67th Convention of the Audio Engineering So-
Preliminary experiments with real-time overwriting by ciety, J. Audio Eng. Soc. (Abstracts), vol. 28, p. 931
modulating the external magnetic field with a digital (1980 Dec.), preprint 1674.

d. Audio Eng. Soc., Vol. 32, No. 7/8, 1984 July/August 537
 IMMINKANDBRAAT ENGINEERING
REPORTS

THE AUTHORS
x

K.Immink J. Braat

Kees Schouhamer immlnk joined the the Philips Re- servo, acoustical, and modulation systems.
search Laboratories, Eindhoven, The Netherlands, in
1968. After part-time studies, he obtained a master's ·
degree in electronic engineering at the Eindhoven Uni-
versity of Technology. Joseph Braat graduated from Delft Technical Uni-
He then joined the measurement and control group versity in 1970 and wrote his thesis during a three-
of the Philips Laboratories to do research on servo year stay at the French Institut d'Optique in Paris. The
systems for the laser video system, subject was holography using spatially incoherent light.
In 1979 he joined the optical recording group working In 1973 he joined the optics group of the Philips Re-
on channel codes for optical disk systems, in particular search Laboratories in Eindhoven and became active
the digital audio Compact Disc system, in the field of optical recording, scanning, microscopy,
Mr. Immink holds several patents in the fields of and lens design.

538 J. AudioEng.Soc.,Vol,32, No.7/8, 1984July/August

View publication stats