FACULDADE DE ENGENHARIA DA UNIVERSIDADE DO PORTO
Unde wa e Op ical Communica ion
an App oach Based on LED
Be na do Miguel Ca alho Sil a
FINAL VERSION
Mas e s in Elec ical and Compu e s Enginee ing
Supe iso : Nuno Alexand e Lopes Mo ei a da C uz
Co-Supe iso s: José Ca los dos San os Al es, Luís Manuel de Sousa Pessoa
Oc obe 30, 2015
© Be na do Miguel Ca alho Sil a, 2015
ii
Abs ac
One o he majo obs acles in he use o unde wa e obo s on a la ge scale is he di icul y in
communica ing wi h he pla o ms du ing missions, since he mos common me hods o commu-
nica ion ( ia adio o o he elec omagne ic wa es) used in comme cial sys ems use signals ha
a e s ongly a enua ed in wa e . Al e na i ely, acous ic-based solu ions ha e long been he de aul
wi eless communica ion me hod o unde wa e applica ions, since hey allow easonable anges.
Howe e , due o he se e e limi a ions in bandwid h and he slow da a a es wi h high la ency, hese
a e no he mos e icien solu ions. Summing up, using acous ics, applica ions such as moni o ing
and con olling emo e ope a ions a e no p ac ical.
In his p ojec , i is in ended o ake ad an age o he ecen de elopmen s o he ligh emi ing
diode (LED), (specially wi h he highe ligh ou pu and mo e p ecise uning o he wa eleng h)
and he pho o-senso echnology. These will be used o de elop an imp o ed communica ion
module, in esponse o he g owing demand o obo ic solu ions o he ma ine en i onmen ha
ea u es a high speed communica ion sys em a sho anges, whe e low powe , low complexi y
and small dimensions a e p e ended.
Using high b igh ness blue, cyan and g een LED based ansmi e s and a blue/g een enhanced
pho o-diode based ecei e s, he main goal is o achie e, wi h an abundan numbe o es s in all
so o condi ions and en i onmen s, da a ansmission a es up o 1 Mbps o e 5 me e s wi h he
capabili y o ansmi ing in sea and i e wa e s.
The inal s ep consis s o he c ea ion o an ope a ional pla o m, o es mul iple combina ions
o Tx/Rx con igu a ions and ela i e o ien a ions, and, using he in o ma ion ob ained h ough he
es s, une up he sys em, so i s maximum e iciency can be eached, ma ching up all he delinea ed
objec i es.
iii

i
Acknowledgmen s
Começo es e capí ulo de Ag adecimen os po dize que apesa de e ei o ques ão de esc e e es a
disse ação em Inglês, po que achei que se ia mui o mais in e essan e um abalho cien í ico que
dema ca o inal do pe cu so académico na aculdade, se legí el e e isibilidade a um público
mais ab angen e nacional e in e nacional, esc e o es as pala as de ag adecimen o e de econhec-
imen o no bom e elho Po uguês, pois es as pala as são dedicadas a algumas pessoas que me
ma ca am mais e que me ajuda am ao longo des e pe cu so. Apesa de apa en emen e mais leigas,
es as pala as al como es as pessoas, pa a mim são as e e ências mais especiais.
Após es a in odução, começa p op iamen e os di os ag adecimen os, e que ia começa po
ag adece a quem o iginou es e ema e es e abalho e ambém quem o apoiou, po isso aos meus
o ien ado es P o esso Nuno C uz e D . Luís Pessoa. Ob igado po e em colocado es e desa io,
e ob igado po oda a disponibilidade, conselhos e oca de expe iências. Não e ia conseguido
assimila odo o abalho sem a ossa ajuda. Ao P o esso Nuno C uz que desde o início, oi
um dos a o es mais p eponde an es na escolha des a disse ação, po que pela minha expe iência
no meu pe cu so académico, assumi de imedia o que com o P o esso i ia conseguia abalha de
o ma mais au ónoma, mas se p ecisasse de qualque ipo de apoio, sabia que podia con a semp e
com o P o esso . Ao Luís Pessoa, mui o ob igado não só po e mui as ezes “iluminado” as
dú idas que pe sis iam, mas ambém po da uma “luzes” ace ca da componen e mais cien í ica
do abalho e po odo o apoio dado semp e que oi p eciso a anja algum ipo de ma e ial de
labo a ó io, que nunca conseguia a anja no depa amen o da aculdade.
De seguida, que o ag adece a oda a minha amília, em especial aos meus a ós, ao meu
pai Al edo, à minha mãe Flo bela, e às minha duas i mãs, po o dem hie á quica, a Ma iana e a
Camila. Toda ela oi capaz de me da o ça semp e que p ecisei, educação po que semp e p ecisei,
mo i ação, inspi ação, ca inho,... basicamen e um ho o de coisas que nunca cabe iam numa
disse ação só. Ag adeço só em especial a ambos os meus pais, po nunca du ida em de mim,
quando eu p óp io o du idei, e de me e em dado oda a libe dade pa a aze udo o que que ia e
quando que ia, pois con ia am (e con iam) cegamen e que a ei oma ei semp e a escolha co e a.
É g aças a eles que i i udo nes e cinco anos. De es o, nes e cu o empo pa a a disse ação,
o am só boleias incon á eis a é a aculdade a odas as ho as, i busca a aculdade bem cedo
(3h, 4h da manhã) pa a ainda pode do mi umas ho i as, aze di e as a ajuda a aze es es num
anque de água na aculdade. Depois de cinco anos a i e ao máximo a ida académica, es a
disse ação oi um passeio pa a eles... Às minhas i mãs, à Ma iana que o ag adece po me e
mos ado o que é ida académica. Fos e um exemplo e uma asquia pa a mim, sendo a p imei a
a en a na aculdade e a en a i e udo a que inhas di ei o. Foi o eu amo ao eu cu so e udo
que ele ep esen a, que me ez que e aquilo que u inhas, e ama com an a o ça como u. À
Camila ob igada po es a es semp e p esen e quando p ecisa a de desca ega o meu s ess em
alguém. A pa i de ago a, já sabes quem é que ens de supe a . Boa so e. Fico à espe a dos meus
ag adecimen os. Ahhh... ambém deixo uma pala a à minha p ima Ca a ina, po semp e ica
com aquele b ilho nos olhos quando ala a do p imo que e a Engenhei o.
i
À minha namo ada Tina, ambém é ou a pessoa que pode ia da numa disse ação. Desde de
me liga mais de 20 ezes de manhã pa a me aco da , a le a a casa quando a es ada es a a com-
plicada, a ob iga a descansa quando não que ia mas p ecisa a,... não sei como consegues, mas
ens uma capacidade de me a u a que não acho no mal. Não há ninguém como u... Ob igado
po oda a mo i ação, amo , ca inho, paciência que me des e ao longo des es anos, e ob igado po
nunca desis i es de mim, mesmo quando nada unciona a. Mesmo NADA! G aças a i que ago a
sou capaz de pensa em coisas de uma o ma di e en e, e de i ê-las de uma o ma di e en e. Po
mais pala as que esc e a, es as nunca a ão jus a quem és nem ao que ep esen as pa a mim.
Des e-me um no o signi icado di e en e ao núme o 13.
Es a egicamen e, coloquei es e úl imo pon o nos ag adecimen os. Deixo es e úl imo pa a
odas as amizades que c iei nes e cu os cinco anos. Todas elas não ize am pa e da minha ida
académica. Todas elas FORAM a minha ida académica. Não é possí el esc e e uma disse ação
ou ese sob e elas, é p eciso i ê-las pa a pe cebê-las. Foi só g aças a elas, que consegui ap ende
an o, num ão cu o espaço de empo. Com essas amizades, é que ui capaz de c ia ainda mais
amizades, e sen i como uma pessoa que já i eu mil e uma coisas e que conheceu udo que ha ia
pa a conhece , pa a no memen o a segui , ol a a po numa si uação o almen e no a e di e en e.
Foi a melho expe iência de ida, que alguma ez ambicionei e . Não que o se injus o pa a
ninguém, mas deixo um ab aço especial à F.A.N.F.A.R.R.A. de Elec o, po me e dado algo que
nunca pedi, mas nunca i ei esquece , a pessoal mais elho que me moldou e ez pensa , pessoal
mais no o com quem me di e i a goza , e a pessoal do meu ano. A jun a a es a legião, aquele
ab aço especial ao Chuck, Co ec o , Rod iguez, Locomo i a e Pin o . Ao Chuck po se uma das
pessoas que me ensinou mais, e um dos gajos mais ilho da mãe que eu conheço. Ao Co ec o po
se uma das pessoas que semp e ac edi ou em mim desde no o e com mais espí i o, e um dos gajos
mais ilho da mãe que eu conheço. Ao Rod iguez po es a semp e abe o a ala comigo semp e
que p ecisa a de um omb o amigo, e um dos gajos mais ilho da mãe que eu conheço. Qualque
um des es ês senho es em uma das ca a e ís icas di as. Ao Locomo i a não ag adeço só eu, mas
ambém es a disse ação, pois semp e que i e um p oblema no pe cu so académico, mui o mais
du an e a ealização des a disse ação, e a semp e a i que di igia e i a a dú idas. Es e abalho
oi possí el ambém g aças a i. E ao Pin o , po e sido p o a elmen e quem so eu mais comigo
em cinco anos, mas que ambém quem pude semp e con ia pa a o que de , e ie . E ambém
po se um ilho da mãe. Me eces udo que ganhas e e conquis as e. Se ão semp e es as g andes
amizades que nunca esquece ei.
Dedico es a disse ação a odos ós.
Ab aço,
Be na do "Maque is a" Sil a
“Não cho es po que acabou.
Ri po que acon eceu.”
Chuck No is
ii
xi LIST OF TABLES

Abb e ia ions
APD A alanche Pho o-Diode
AUV Au onomous Unde wa e Vehicle
BER Bi E o Ra e
BPF Band-Pass Fil e
bps bi s-pe -second
BPSK Bina y Phase-Shi Keying
DEEC Depa men o Elec ical and Compu e Enginee ing
EM Elec omagne ic
ESR Equi alen Se ies Resis ance
IC In eg a ed Ci cui
IR In a ed
LED Ligh Emi ing Diode
LD Lase Diode
LOS Line o Sigh
MCPCB Me al-Co e P in ed Ci cui Boa d
MOSFET Me al–Oxide–Semiconduc o Field-E ec T ansis o
OP-AMP Ope a ional Ampli ie
OTS Op ical Teleme y Sys em
PC Pe sonal Compu e
PCB P in ed Ci cui Boa d
PD Pho o-Diode
PMT Pho omul iplie
PSK Phase-Shi Keying
QAM Quad a u e Ampli ude Modula ion
QPSK Quad a u e Phase-Shi Keying
RF Radio-F equency
RGB Red-G een-Blue
ROV Remo ely Ope a ed Vehicle
Rx Recei e
SMD Su ace-Moun De ice
TIA T ansimpedance Ampli ie
TTL T ansis o -T ansis o Logic
Tx T ansmi e
VGA Va iable Gain Ampli ie
VSF Volume Sca e ing Phase Func ion
x
Chap e 1
In oduc ion
1.1 Con ex
Technology has eached a poin whe e much impo ance has been gi en o he c ea ion o e-
mo e and au onomous obo s, in o de o enhance he p ese a ion and com o o human li e,
o con ibu e o he in es iga ion and e olu ion o echnology, among o he s. Wi h such ecen
de elopmen s, he asks a hand ha e been inc easing in complexi y and p ecision. In hese case
scena ios, when i is oo dange ous, isky o complica ed o a human o pe o m a ask, hese a e
mainly en us ed o a machine. To ins uc his machine y and obo s, i is necessa y o es ablish
communica ion be ween he ope a o and he machine. This communica ion is also aluable since
i can ansmi in o ma ion ha has been ex ac ed om senso s.
Besides es ablishing communica ion ia a physical cable, which has some limi a ions since
i is a wi e communica ion, wi eless adio wa es ha e been one o he mos commonly used so-
lu ions in he ae ial and e es ial en i onmen s. Howe e , ying o implemen communica ions
in unde wa e condi ions aises comple ely new p oblems, since adio and o he elec omagne ic
(EM) wa es a e highly a enua ed unde wa e . Since wi eless communica ions a e a big ad an age
o his kind o en i onmen s, al e na i es such as he use o acous ic modules became he s an-
da d op ion o communica e unde wa e . The inhe en p oblems o his solu ion, despi e he low
a enua ion and long dis ance ange, e eals limi a ions in bandwid h and p esen s a high la ency
and high powe consump ion.
These p oblems ha e been he main ba ie ha es ains he use o unde wa e obo s in a
la ge scale, making unde wa e explo a ions an e en mo e challenging and di icul ask. Due o
such limi a ions, al e na i e me hods ha allow unde wa e obo s (like au onomous unde wa e
ehicles (AUVs) and emo ely ope a ed ehicles (ROVs)) o ansmi in o ma ion ha e been an
eme ging hema ic in he pas ew yea s and s ill ep esen s many challenges, in o de o achie e a
mo e eliable solu ion.
1
2In oduc ion
1.2 Mo i a ion
The necessi y o unde wa e wi eless communica ions esul s om he mobili y equi emen s o
obo ic-based ac i i ies [12]. This is he main eason why physical connec o s a e no a iable
solu ion o es ablish communica ion unde wa e [13]. In subma ine ope a ions, whe e a obo ic
ehicle has o pe o m complex mo emen s, being dependen on a physical connec o could limi
he ange and he p ecision o i s mo es. Besides, i ep esen s an addi ional isk i he cables,
o some eason, end up en angled, which could esul in loss o aluable equipmen and subjec s
o s udy. E en hough acous ical-based sys ems a e he de aul choice and a e capable o es ab-
lishing wi eless communica ions unde wa e a a long dis ance ange, he slow da a- a es and high
la ency does no allow a eal- ime con ol and da a ha es ing. Fo hese easons, main enance and
su eillance ope a ions based on acous ics a e e y limi ed.
The e a e many a eas whe e an imp o ed pe o mance o he unde wa e wi eless commu-
nica ions may ha e a big impac , such as in oil, gas and mining explo a ions, unde he sea
cons uc ions, he s udy on he impac o deep-sea aw ma e ial esou ces exploi a ion, and he
measu emen and moni o ing o physical, biological and bio and geochemical pa ame e s (e.g.
clima e change) [12,14,15,16,2].
In o de o su pass many o he obs acles p esen ed in he unde wa e wi eless communica ion,
an op ical ligh emi ing diode (LED) based solu ion is sugges ed, since i p esen s he po en ial o
o e come many limi a ions shown by o he me hods.
1.3 Objec i es
The p oposed solu ion is o c ea e a sho ange unde wa e op ical wi eless communica ion sys-
em based on LED echnology, in o de o inc ease he e iciency in unde wa e communica ions.
Ideally, he solu ion has o be a iable op ion o be implemen ed in an ope a ional pla o m, such
as an AUV. The expe imen s ca ied ou along he whole pe iod o wo k will be es ed in wide
and a ied es scena ios, such as in a con olled labo a o y scena io, simula ing communica ions
inside a es pool, and in eal li e scena ios, ying o communica e in wo dis inguished en i on-
men s: he ocean and he i e . To each his goal he ollowing objec i es will be conside ed:
• Assembly o a LED and pho o-senso based es sys em;
• C ea ion o an unde wa e wi eless link ha enables he ansmission o in o ma ion;
• C ea e a sys em ha is able o es di e en ansmission pa ame e s;
• C ea e a eal-li e solu ion, ha could be implemen ed in a obo ic pla o m;
• Tes o he sys em in he labo a o y;
• Tes o he sys em in he labo a o y pool; and
• Tes o he sys em in eal li e en i onmen s (Ocean and Ri e ).
Chap e 2
S a e o he A
In his chap e , i is p esen ed all he majo scien i ic wo k de eloped abou unde wa e communi-
ca ion. I s a s by analyzing he many me hods o unde wa e communica ion, conside ing all he
p os and cons o each p esen ed app oach.
Subsequen ly, i will ocus on de eloping he hema ic o unde wa e op ical wi eless commu-
nica ion, aking in special conside a ion he op ical p ope ies o he a ious wa e ypes ha a e
used as ansmission channels, and i will be s udied all he ela ed wo ks and solu ions ha ha e
been de eloped un il his momen , academic and comme cial.
In he end, he e will be a sum up abou all he chap e .
2.1 Unde wa e Communica ion Me hods
As analyzed by [12,17], unde wa e communica ions a e adi ionally made wi h a physical con-
nec o , made by cables o ibe -op ics. These ep esen some ad an ages, since hey a e capable o
anspo elec ical powe and o e a high speed and eliable communica ion. Howe e , hei use
can limi he ange and maneu e abili y o unde wa e ope a ions, and he p esence o a long and
hea y cable and he associa ed hyd odynamic d ag, inc eases he isks o an ope a ion.
Fo hese easons, he e is a high in e es on he in es iga ion o al e na i e ways o ansmi
in o ma ion unde wa e , wi hou being dependen on a cable connec ion. This is why i has been
gi en a big impo ance o he wi eless communica ion echniques. Nex i is p esen ed he mos
common me hods o unde wa e wi eless communica ion, e e ing hei ad an ages bu also he
p oblems hey ace [16].
2.1.1 Acous ics
Acous ics has o a long ime been he s anda d solu ion o subma ine missions because o he ac
ha sound p opaga es well unde wa e , i enables he ansmission o he in o ma ion wi hou been
dependen on an "umbilical" cable, which gi es o he unde wa e ehicles he necessa y mobili y
o pe o m complex mo emen s eely. Adding o his, i s ela i e low signal a enua ion and long
ange allows he ansmission o in o ma ion in he o de o he kilome e s. Despi e all hese
3

4S a e o he A
ad an ages, his solu ion also ca ies some p oblems ha begin when he speed o ansmission is
impo an [16].
As e e enced by [18], his me hod has i s disad an ages, because i p esen s a low p opaga ion
speed (high la ency). Mo eo e , i has a low bandwid h wi h a maximum da a- a e in he o de o
he ens o housands bi s-pe -second (bps), making he acous ics communica ion a slow op ion
o ansmission. This me hod also p esen s o he p oblems, such as i s suscep ibili y o mul ipa h
p opaga ion, dispe sion, equency ading and ambien noise (specially in shallow wa e s).
The nex able 2.1 con ain some a ailable acous ical solu ions and hei main cha ac e is ics.
Table 2.1: Compa ison able o some con en ional unde wa e acous ic solu ions [6,7,8,9,10,
11].
Model Dis ance (m) Da a-Ra e
(kbps) Powe (W) Weigh in
Ai (kg)
HERMES 120 150 32 —
Sona dyne A T ak6 Type 8220 3000 9.0 50 5.1
AquaSeNT - AM-OFDM-P1 4000 9.0 20 7.6
AquaSeNT - AM-OFDM-S1 4000 1.6 20 2.4
Ben hos - ATM-916 6000 15.36 15.75 4.5
Ben hos - ATM-926 6000 15.36 15.75 4.1
Ben hos - ATM-966 6000 15.36 24.5 6.8
E oLogics - S2CR 48/78 1000 31.2 60 6.5
E oLogics - S2CR 18/34 3500 13.9 80 6.5
E oLogics - S2CR 12/24 6000 9.2 40 7.78
E oLogics - S2CR 7/17 8000 6.9 80 7.78
LinkQues Inc. - UWM1000 350 19.2 2 4.2
LinkQues Inc. - UWM2200 2000 38.4 6 3.0
LinkQues Inc. - UWM3000 5000 5 12 4.1
LinkQues Inc. - UWM4000 4000 9.6 7 7.6
LinkQues Inc. - UWM10000 10000 5 40 21.0
2.1.2 Radio-F equency
On he o he hand, he e a e he EM wa es such as adio, a e y popula choice o ansmi in-
o ma ion ia wi eless in e es ial communica ions. As s a ed by [15,16], EM wa es such as
adio- equency (RF) compa ed o acous ic wa es, a e much as e and ha e a highe bandwid h,
in his e es ial ansmissions.
In addi ion, adio p oduces a high h oughpu compa ed o sound wa es, wi h speed le els
su ounding he hund eds o kbps up o en me e s, and achie ing ens o Mbps a he ange o
one me e . Again, compa ing bo h powe consump ions, he adio e eals o be mo e ene ge ically
e icien han acous ic echnology, needing less powe .
E en wi h hese cha ac e is ics, he e a e many limi ing ac o s when using RF in wa e . The
EM wa es ha e di e en beha io in eshwa e and seawa e .
2.1 Unde wa e Communica ion Me hods 5
Acco ding o [16], eshwa e is conside ed o he EM as a low-loss medium, and i s p opaga-
ion speed c, and he abso p ion coe icien αcan be ep esen ed as:
c≈1
√εµ (2.1)
α≈σ
2 µ
ε(2.2)
In he equa ions εis conside ed he dielec ic pe mi i i y, µis he magne ic pe meabili y
and σis he elec ic conduc i i y. By obse ing his exp essions, i can be concluded ha he
abso p ion o eshwa e is independen o he equency o he signal, and p opaga ion speeds o
EM a e smalle han ligh , bu s ill as e han sound [16]. The only incon enien o his solu ion,
in eshwa e , is he an enna size ha is equi ed in o de o ob ain a la ge bandwid h.
To EM wa es, he seawa e is conside a e as a high-loss medium, and i s p opaga ion speed c,
and he abso p ion coe icien αcan be ep esen ed as:
c≈s4π
µσ (2.3)
α≈pπ µσ (2.4)
As i can be concluded by he equa ions 2.3 and 2.4 bo h a e dependen o he equency o he
signal , meaning ha i a high equency is used, a high speed is ob ainable, bu he abso p ion
o EM wa es is also high.
I is possible o conclude ha communica ion h ough EM wa es is limi ed in eshwa e by
he size o he an enna needed, and has a c i ical p oblem because o he high a enua ion in
seawa e . In igu e 2.1 i is shown he di e en RF wa es in di e en ypes o wa e , compa ing
hei pe o mance wi h he ligh and sound wa es.
Al hough i is known ha RF is no he ideal me hod o communica e, he e a e some com-
me cial possibili ies. WFS Technologies L d comme cializes a unde wa e wi eless RF sys em o
sub-sea explo a ion, he Sea oo h® , a sys em se ies claimed o ha e a model capable o achie ing
156 kbps o e 7 me e s [19]. Al hough he e a e some si ua ions whe e his sys em could be used,
i s ill equi es a high powe consump ion, a ound 15 Wa s, and an an enna wi h la ge dimensions
(1 me e squa ial an enna).
2.1.3 Op ical
An op ical solu ion is ano he ansmission me hod s ill in de elopmen ha sha es some ad an-
ages o he RF, bu no all o i s disad an ages. The majo ad an age compa ing o RF commu-
nica ions is he possibili y o main aining he same condi ions unde wa e wi hou conside able
losses. As RF, his solu ion can ob ain in a unde wa e en i onmen , a high h oughpu in he
6S a e o he A
103104105106107108109
103
104
105
106
107
108
109
F equency (Hz)
P opaga ion speed (m/s)
ligh
σ=.001
σ=.05
σ=4
acous ic
Ligh
Sea
F esh wa e
Dis illed wa e
Acous ic
Figu e 2.1: RF p opaga ion speeds in di e en ypes o wa e .
o de o se e al Mbps o Gbps, a high p opaga ion speed (low la ency) and achie e sho and mid-
ange link dis ances in he o de o 100-200m [14,1,17]. In [14] also e e ed ha unde wa e
op ical communica ion can bene i meaning ully om he p og ess made in he e es ial op ical
wi eless communica ion, becoming an a ea in cons an e olu ion.
Like any sys em, op ical solu ions ha e some inhe en p oblems. Al hough op ical signal ha e
a low a enua ion, i becomes ele an in long dis ances [14]. Fo ha eason, acous ic as ly
ou pe o m he op ical channel when comes up o dis ance ange. O he ele an disad an ages, in
addi ion o he abso p ion, op ical sca e ing is signi ican and he ambien ligh can also in e e e
wi h communica ions [16].
Ano he complex p oblem associa ed o his me hod is he dependence on a clea way o he
ligh o a el, a line o sigh (LOS) [14]. In he beginning, wi hou a clea LOS i was impos-
sible o es ablish communica ion, since he ligh canno each he des ina ion. Fo ha eason,
some al e na i e me hods we e c ea ed o p e en he loss o he LOS, and o c ea e op ical links.
Figu e 2.2 shows some o hese me hods.
Wi h hese a ian high bandwid h links op ions, i is possible o c ea e sophis ica ed and
collabo a i e pa hs o planning and obse a ion [14,1].
In igu es 2.2a and 2.2b is ep esen ed a communica ion channel based on a clea and sim-
ple LOS, whe e communica ion exis s by emi ing ligh in a na ow beam (simila o lase ) and
in a di use beam, espec i ely. Figu e 2.2c shows a e o- e lec o link ype, used when limi ed
dual communica ion is necessa y, and he ecei e is aw ully low powe ed o pe o m as a ull
anscei e (such as unde wa e senso nodes). In his con igu a ion he ligh o he ansmi e is
2.2 Unde wa e Op ical Wi eless Communica ion 7
(a) LOS (b) non-di ec LOS
(c) e o- e lec o (d) non-LOS be ween a ansmi e and a ecei e
Figu e 2.2: Di e en unde wa e op ical wi eless link con igu a ion [1].
e lec ed back om a modula ed e o- e lec o on he ecei e ha , in he mean ime, is encoded
wi h in o ma ion. Ano he link ep esen ed in 2.2d is a non-LOS ansmission whe e he ans-
mi e and he ecei e posi ion hemsel es diagonally upwa ds o ansmi and ecei e he ligh
e lec ed in he su ace o he wa e .
2.2 Unde wa e Op ical Wi eless Communica ion
Seeing he po en ial o he op ic channel in he unde wa e en i onmen , he a en ion o his wo k
is ocused on s udy, wi h some de ail, he op ical communica ions. I s a s by analyzing he op ical
p ope ies o he wa e channel o ob ain a de ailed o e iew o he beha io o ligh in wa e , and
will end in e iewing some majo achie emen s in he ield o knowledge.
2.2.1 Op ical P ope ies o Wa e
As concluded by [2], he beha io o ligh in wa e is a e y complex issue. The e a e ex ensi e
wo ks ega ding his hema ic, since i has many in e es s in many ields o science. Since his
wo k is mos ly ocused in he concep ion and implemen a ion o a p ac ical solu ion, his hema ic
will be b ie ly e e enced, aking in o accoun he wo k done by [2].
Assuming a di e en ial po ion o wa e ( ep esen ed by a olume wi h h ee dimensions), wi h
a wid h o ∆ and wi h an small heigh and leng h, i is assumed ha he amoun o ligh en e ing
he olume equals he amoun o ligh lea ing i , less any loss ha occu ed on he olume in he
o m o abso p ion and/o sca e ing. This phenomenon is ep esen ed in igu e 2.3.
The inciden powe is ep esen ed by Φi, he abso bed powe by Φa, he sca e ed powe by
Φsand he ansmi ed powe is ep esen ed by Φ . These alues a e dependen o he wa eleng h
λ.
14 S a e o he A
ex e nal 10 Mbps E he ne in e ace. The communica ion has a da a a e o 4Mbps a a dis ance o
up o 50 m in a pool. In a labo a o y se ing ansmi ing h ough ai i is 32 Mbps.
In bo h [35,36] is desc ibed an op ical eleme y sys em (OTS) wi h which i downloaded 20
MB o da a o e a 5 Mbps link a a ange o 80 me e s. I epo s communica ions a a da a a e
o 10 Mpps a a dis ance o 108 me e s, 5 Mbps a dis ances o up o 128 me e s, and 1 Mbps a
dis ances o up o 138 me e s.
In [37] he au ho ies a di e en app oach by designing a mo e e icien ansmi e and
ecei e . Wi h an a ay o 7 LEDs o he ansmi e , and 7 PD o he ecei e , displayed in he
o m o a unca ed hexagonal py amid s uc u e wi h a special lens, he ansmi ed ligh ge s mo e
ocused and he ecei ed ligh can be cap u ed mo e e icien ly, ob aining a quasi-omnidi ec ional
ansmi e and ecei e .
Documen [38] gi es a con inua ion wo k on he AquaOp ical p ojec [31,34], which uses he
de eloped ha dwa e, and i s upg ades esul ing in he AquaOp ical II. Also in his pape , by using
he AquaOp ical II, a base signal s eng h model o end- o-end communica ions is p ojec ed. I is
a g ea li e a u e ha e iews all he ecen wo ks done in he a ea.
In [39], using a labo a o y se -up, i was in es iga ed he impac o backg ound noise on he
ecei e in shallow wa e s. I was used a 2.5 me e s es pool o s udy he e ec . The wo ansmi -
e s used LEDs, bu he ecei e was cons i u ed by an APD and a PD. Fo his expe imen i was
ob ained, using Manches e Coding 6.25 Mbps, 12.5 Mbps using NRZ 8b/10b code and 58 Mbps
wi h DMT. I was es ima ed ha maximum dis ance was a ound he 60 me e s in clea wa e and
10 me e s in mu ky.
Now, i will be shown some o he comme cial solu ion a ailable, o unde wa e op ical wi e-
less communica ion.
As e e ed be o e, he Ambalux Co po a ion [28] manu ac u es modems ha es ablish com-
munica ion poin - o-poin , capable o achie ing 10 Mbps and up o 40 me e s ange. The in e ace
is 10BaseT E he ne compa ible allowing TPC/IP and UDP connec ions. The ansmi e con-
sumes up o 36 W and he ecei e 7.2 W.
As a p oduc om he in es iga ion in [26,27], he de ices c ea ed by he au ho we e pos e i-
o ly comme cialized by he company Penguin Au oma ed Sys ems, Inc. [40]. Same cha ac e is ics
o he de eloped de ice a e conside a e.
Ano he comme cialized p oduc is he Nep une, p oduced by SA Pho onics [41], ha is
claimed ha i can achie e 250 Mbps o e 200 me e s, wi h a powe consump ion la ge han
5 W.
F om he wo k o [30], he company Sona dyne In e na ional L d. adop ed he sys em along
wi h an LED-based model o nea -su ace communica ions. The p oduc is he BlueComm [42]
and i claims ha is capable o achie e up o 20 Mbps o da a a e and up o 200 me e s o ange.

2.3 Summa y 15
2.3 Summa y
This chap e p esen ed he esea ch wo k ha has been done in unde wa e communica ions, be-
ginning wi h a e e ence on he unde wa e op ical wi eless communica ions. I was possible
o conclude ha , in all he unde wa e communica ion me hods, wi eless communica ions a e a
mus -use solu ion and he op ical communica ions a e a new and p omising me hod o make com-
munica ion links unde wa e . A e all, his is an a ea ha is s ill in in es iga ion, o i s inc edible
bene i s, bu also i s complex p oblems.
When s udying he op ical p op ie ies o wa e , i was clea ha di e en en i onmen s, whe e
wa e isibili y is no he ideal, will ob iously limi he p opaga ion o ligh , no only a ec ing he
speed o he connec ion, bu also in luencing he wa eleng h needed o he a enua ion o emain
low.
In he e iew o o he wo ks, some knowledge was ob ained abou he a chi ec u e o he
sys em, hei main componen s and wha a e he bes ma e ials o ob ain he bes , expec ed esul s,
and common e o s and di icul ies ound in his a ea o wo k. Wi h his acqui ed in o ma ion, a
solid and s udied design sys em can now be done.
16 S a e o he A
Chap e 3
Implemen a ion
In his chap e i is p esen ed he implemen a ion and de elopmen o he unde wa e communica-
ion module p o o ype.
Ini ially, i is shown he p oposed me hodology o he implemen a ion o he sys em, show-
ing he design and i s cons ain s, he di e en laye s ha o m he p oposed module, speci ying
he componen selec ion, dimensioning and i s design, ending wi h he display o he p oduced
p o o ype.
3.1 Sys em Design
In he p ima y sys em design, shown in igu e 3.1, i was in ended ha a pack o da a would
be p esen in a compu e and, by modula ing his da a in o a ligh signal, i would be sen by a
ansmi e (Tx), hen his ligh signal would be cap u ed by he ecei e (Rx), and he end e minal
would demodula e and show he da a.
Figu e 3.1: P ima y sys em design o e iew.
A e some de elopmen , he mo e de ailed edesign solu ion in igu e 3.2 was achie ed. I
is simila o he ini ial one, whe e a compu e e minal ies o communica e wi h a pee , using
he op ical channel, bu i is in ended ha he compu e communica es wi h a mic o-con olle
ha connec s o he compu e (PC) and he Tx/Rx ia se ial communica ion. Also, a e some
conside a ions, he design would be physically able o es ablish bidi ec ional communica ions
wi h wo dis inc colo s.
17
18 Implemen a ion
Figu e 3.2: De ailed sys em design.
3.1.1 Design Cons ain s
The e iews ha we e made show ha unde wa e op ical wi eless communica ion is a iable solu-
ion o be implemen ed in an ope a ional pla o m, such as an AUV, o communica ion pu poses.
So, i i is p e ended o implemen a module in an AUV, some s anda d necessi ies mus be aken
in o accoun . Because an AUV is an au onomous ehicle ha ope a es in an unde wa e en i on-
men , he de ice ha will be a ached o i , in addi ion o he ull p o ec ion o he elec onics om
wa e , which has o be assu ed, mus be as e icien as possible, mainly in e ms o ene gy, and
also ha e a small physical size. Fo ha eason, hese ini ial simple equi emen s a e no ed:
1. Maximum ange - up o 5 me e s;
2. High da a- a e - a 1Mbps;
3. Low powe - below he 5W;
4. Small size - ela i ely smalle han 15 cen ime e long;
5. Low weigh - made wi h ligh ma e ials; and
6. Low complexi y - simple and modula .
The e a e o he equi emen s, such as a low-cos solu ion, ha could be conside ed, bu o
he pu pose o his disse a ion, hese will be somewha o e looked, making he choice based on
common sense.
3.2 Op ical T ansmi e - Tx
Wi h he op ical ansmi e , i is p e ended o design a ci cui ha is capable o ou pu as much
ligh as possible. This ope a ion will be done h ough a d i e ha eeds he chosen ligh sou ce
o ob ain he maximum pe o mance o i , o be able o achie e a wide ange. The da a a e will
also be limi ed by he ime esponse o he componen s ha he ansmi e is made o , so a s udied
3.2 Op ical T ansmi e - Tx 19
selec ion o he module componen s is ad ised. Figu e 3.3 gi es a simple iew o he op imal
solu ion.
Figu e 3.3: Simple Tx sys em design.
In he ollowing subsec ions, a s udy and some unc ional es s will be pe o med o be e
choose he bes ligh sou ce and adequa e d i e o i .
3.2.1 Ligh Sou ce
Since one o he objec i es o his disse a ion was he c ea ion o a LED-based sys em, p ima ily
i was necessa y o decide a ligh sou ce o he ansmi e . Two possibili ies o a gene al LED-
based ligh engine we e conside ed: a di use ligh sou ce, a ailable as a LED o a concen a ed
beam, as a lase , in his case a lase diode (LD).
Be ween he wo choices, he di use LED ligh sou ce was conside ed as a mo e appealing
solu ion. Al hough using a LD would be e icien since i has a na owe EM spec um, highe
ligh in ensi y a long dis ances and as e swi ching speed, he LD p esen s a complex p oblem o
misalignmen a long dis ances, and a di use ligh engine is able o adjus i s ligh beam using an
op ic, making i a much mo e e sa ile solu ion. O he issues ega ding he LD we e also aken
in o accoun , such as a highe powe consump ion and a much g ea e cos .
The ligh sou ce o he Tx ended up o be high-powe LEDs, whe e cha ac e is ics such as
small size, high-b igh ness, na ow EM spec um and o he op ical ea u es we e desi able. Also,
he exis ence o he h ee colo s (blue a 470 nm, cyan a 505 nm and g een a 530 nm) o iginally
p e ended would be an ad an age.
Figu e 3.4: A MCPCB
wi h g een LUXEON Rebel
LEDs.
Fo hese easons, o he ini ial es ing and unde s anding o
he sys em, i was only used LUXEON Rebel g een LEDs, shown
in igu e 3.4, since i was a ailable an example o hese LEDs (p o-
ided by INESC TEC) and hese me up he p incipal necessi ies
desc ibed be o e.
Table 3.1 shows some o he mos ele an in o ma ion abou
he LUXEON Rebel LEDs, dis ibu ed acco ding o hei ligh
colo . Since he se en LEDs a e connec ed in se ies, o ob ain
he maximum ou pu o ligh , bu s ill espec ing he ypical da a-
shee alues, i is necessa y o supply abou 20.3V a 700mA. Bu
since one o he objec i es is o be implemen ed in an ope a ional

20 Implemen a ion
base o an AUV, he LEDs supply will be se o mo e ypical alues o 24V a 700mA (which co -
esponds o wo ba e ies o 12V), wi h he emaining ol age been dissipa ed by a dimensioned
esis o , indica ed o his ope a ion.
Because he sys em is supposed o be con olled by a mic o-con olle , he swi ching signal
espec s he ansis o - ansis o logic (TTL) le els. So i was necessa y o dimension a LED d i e
ha was able o swi ch he LEDs a he necessa y equencies and, a he same ime, supplying he
necessa y cu en and ol age hey equi e.
Table 3.1: Elec ical and spec al cha ac e is ics o he LUXEON Rebel LEDs.
Componen Typical
Vol age (V)
Typical D i e
Cu en (mA)
Typical
Wa eleng h (nm)
G een: LXML - PM01 2.9 700 530
Cyan: LXML - PE01 2.9 700 505
Blue: LXML - PB01 2.95 700 470
As e e ed, al hough wi h he sample he e was no op ical lens a ailable o es ing, o he
inal p o o ype i is in ended o use some op ical de ice o inc ease and op imize he ligh ou pu
o he 7 LEDs.
3.2.2 Ligh Sou ce D i e
As explained p e iously, i is expec ed ha he signal espec s he TTL, meaning ha i will a y
be ween 0V and 5V (depending i he bi alue is ’0’ o ’1’, espec i ely).
Figu e 3.5: P ima y d i e
design.
The e a e some d i e solu ions a ailable in he ma ke ha
can d i e hese kind o high-b igh ness LEDs, jus like he p e i-
ously used, bu o his applica ion in his disse a ion, he LEDs
mus be swi ched in he highes equency a e as possible, main-
aining a maximum h oughpu powe a he same ime. Al hough
comme cially he e a e some solu ions o d i e his so o LEDs
wi h high powe ou pu s p e ended o wi h ela i e high equency
swi ches, inding a solu ion ha can pe o m bo h ope a ions a he
same ime is ano he issue. The solu ions p esen ed in he ma ke
a e oo complex, la ge in size o ine icien . Fo hose easons,
i was conside ed mo e bene icial o design an app op ia e LED
d i e o his applica ion, which is capable o mee ing he equi-
si es, while main aining a simple complexi y.
One o he ini ial ideas was o con ol he high cu en supply
o he LEDs h ough a ansis o . The e a e many ypes o ansis-
o s ha could pe o m he p e ended unc ion (BJT, JFET, IGBT
and mo e), bu o his kind o applica ion, p ecisely because o he high cu en and ela i ely
high ol age o con ol, a MOSFET was conside ed a mo e logical choice, no only because o i s
low cos , wi h a small size and a as esponse ime bu also because a mic o-con olle , by de aul ,
3.2 Op ical T ansmi e - Tx 21
is no able o p oduce a high cu en , and a MOSFET is able o ope a e using a signal wi h low
cu en , needing jus o be con olled mainly by he ol age applied o he ga e.
The idea is ha a command signal is applied o he ga e o he MOSFET. I he signal is a
+5V, he LEDs u n on and when he signal is a 0V he LEDs u n o . The p oposed LED d i e ,
as shown in he ci cui 3.5, is consis ed o a MOSFET, whe e he d i e signal is applied o he ga e,
he 24V and he LEDs a e connec ed o he d ain and he g ound o he sou ce.
A e some p ima y conside a ions, an addi ional esis o was placed be ween he 24V and he
LEDs o p o ec he LEDs and o dissipa e he esidual powe ha ypically emain om he LEDs.
The alue o he LEDs load esis ance RLis ob ained:
VL=RLI ⇔RL=VL
I ⇔RL=Vsupply −V
I ⇒RL=24−(2.9×7)
700×10−3≈5.3Ω(3.1)
And RLmus be able o dissipa e a powe o :
PL=I 2RL⇒PL= (700×10−3)2×5.3≈2.6W(3.2)
The alue o he cu en I , used in he p e ious calcula ion, was es ablished by he da ashee
in o de o ensu e ha he de ice could sus ain he wo s case scena io: a con inuous ansmission
a ull powe .
3.2.3 T ansmi e Tes s
Fo hese i s es s, o analyze he beha io o he LEDs and check i he concep would wo k, i
was used some powe and swi ch MOSFETs a ailable in he DEEC.
Fo his i s s age es s, he inpu signal is c ea ed by a unc ion gene a o ha ou pu s a squa e
wa e ha ope a es be ween he +5V and 0V. The objec i e was o see he delays gene a ed by he
ansis o s and o s udy hei beha io a di e en signal equencies. To achie e his goal, a es
se up, as shown in igu e 3.6, was assembled.
Figu e 3.6: The es se up o he MOSFET es ing.
Ini ially, i was assumed ha MOSFETs o powe applica ions, al hough able o swi ch g ea e
powe loads, hey would only be capable o p oducing slowe esponse imes.
By using a sample ounded in he labo a o y, in his case he P30NF10, i is isible in he p in
3.7, he in luence o he slow esponse ime in a ansmission.
22 Implemen a ion
Figu e 3.7: P30NF10 esponse ime es : F = 1MHz, T = 250ns.
So, since he p e ended ope a ion equencies demand a as swi ching speed, a mo e app op i-
a e MOSFET o hese ope a ions is needed. The challenge behind his ype o swi ching MOS-
FET, is o ind a sample ha can bea he supply ol age and d i e cu en .
In o de o es he concep he BS170, which is a mo e app op ia e MOSFET o his applica-
ion, was used. Howe e in he u u e, a mo e conside ed choice will be made, since his MOSFET
is no capable o suppo ing he necessa y 24V o ol age supply and 1A o d i e cu en .
In igu e 3.8, i is possible o see ha he BS170 is capable o ollowing he ansmission signal,
wi h a conside able small delay along he b oadcas .
Figu e 3.8: BS170 esponse ime es : F = 1MHz, T = 250ns.
Based in he wo k done by [15], i was also he possibili y o using a MOSFET d i e was also
explo ed, in o de o inc ease he cu en o he ansmission signal inse ed in he MOSFET’s ga e.
This is a desi able ampli ica ion because i a highe cu en is injec ed in he MOSFET’s ga e, he
in insic capaci ies o he MOSFET would be sa u a ed, which could dec ease he commu a ion
delay p oduced by i .
In igu e 3.9, i is shown a possible ci cui , ha would wo k based on a MOSFET d i e . The
complexi y o his layou is highe , compa ing wi h he p e ious scheme, bu s ill ela i ely simple.
3.3 Op ical Recei e - Rx 23
Figu e 3.9: Se up scheme o Tx wi h MOSFET d i e .
Despi e o i s ad an ages his op ion was o e looked, since he delay p oduced by his d i e ,
in he ou pu , was supe io in compa ison o ime gain by cha ging he MOSFET capaci ance.
3.3 Op ical Recei e - Rx
Jus like he op ical ansmi e , i is also necessa y o design ano he impo an componen o he
sys em, he op ical ecei e . Basically, as shown in 3.10, his pa has o be capable o cap u ing
he emi ed ligh signal, ansmi ed by he Tx, h ough a pho o- ecei e . Then he ligh signal
sen , passes h ough some p ocessing o be able o e ie e he o iginal message. This block is one
o he mos complex and impo an pa s o he sys em, because no only he esponse ime o he
laye is impo an , again, no o limi he da a a e, bu also he ecei e mus be able o ecei e he
ligh o he signal igno ing he ambiance ligh ha adds noise o he ansmi ed signal.
Figu e 3.10: Simple Rx sys em design.
In his sec ion i will be desc ibed he selec ion o he pho o- ecei e , he many s ages o he
acquisi ion and p ocessing o he ecei ed signal, and es ing each elemen o i , inalizing wi h a
es o he all ecei e laye .
30 Implemen a ion
3.3.3.2 T ansimpedance Ampli ie
So a mo e simple and di ec se up was c ea ed o be able o acqui e he ansmission as shown in
igu e 3.17, by using he al eady es ed Tx. Al hough his se up is mo e limi ed, since i canno
es he ange wi h p ecision, i pe mi s o be e es ing he TIA ime esponse.
Figu e 3.17: The inal es se up o he TIA es ing.
This simple es is pe o med by placing he Tx and he "Rx" in a ixed posi ion, whe e a
plausible ansmission migh occu , and es he ou pu o he TIA wi h some ad isable alues o
he esis o s and capaci o s, a di e en ansmission equencies o 100kHz, 500kHz and 1MHz.
The alues o he esis o s used we e 10kΩand 100kΩ, and he capaci o s alues we e 1.8pF,
2.2pF, 2.7pF and 3.3pF.
In a i s app oach, he ou pu gain was s udied by es ing he TIA using only he eedback
esis o . As i can be seen in he p in s 3.18, he gain o he TIA, is en imes g ea e wi h he
100kΩ esis o , as i could be p edic ed.
(a) R =10kΩ. (b) R =100kΩ.
Figu e 3.18: Feedback esis o compa ison: F=100kHz, wi hou CpF.
Al hough a highe gain can be seen as an asse , wi h a bigge eedback esis o , he g ea e
is he delay gene a ed by i , since he RC ime cons an τ(exp essed in seconds) is ela ed o he
esis ance o a ci cui :
τ=RC (3.10)

3.3 Op ical Recei e - Rx 31
In addi ion, as i can be seen in he p in s 3.19a, a a equency o 1MHz, he ou pu signal,
e en wi hou a capaci o ha could ex end he delay, he signal is highly a ec ed. Al hough ha
a a i s sigh a 3.18b, seems ha i does no need o be il e ed, i a capaci o is placed, as igu e
3.19b shows, he signal becomes almos impossible o be econs uc ed, making i imp ac icable.
(a) Wi hou C . (b) C =1.8pF.
Figu e 3.19: Time esponse o he TIA wi h high gain: F=1MHz,R =100kΩ.
Because o his isible delay, i was assumed ha using a eedback esis o o 10kΩwould
end up being he bes op ion. E en hough , he ou pu signal o he TIA a 1MHz equi es some
il e ing, as i can be concluded wi h igu e 3.20.
Figu e 3.20: Ou pu signal o TIA: F=1MHz,R =10kΩ, wi hou C .
The nex s ep o he TIA es ing only he eedback esis o o alue 10kΩwas used, a ying
only he alues o he eedback capaci o s. Ne e o ge ing ha he TIA delay is always depen-
den o equa ion 3.10, which means he la ge he capaci o , he highe he delay. In he lowe
equencies es ed, he delay gene a ed by he capaci o s is no en i ely pe cep ible, he only sig-
ni ican di e ence is in he ou pu noise o he signal. The la ge he capaci ance, he less noise
emains in he ou pu signal, as i can be assumed by he p in s 3.21.
32 Implemen a ion
(a) C =1.8p . (b) C =2.2pF.
(c) C =2.7pF. (d) C =3.3pF.
Figu e 3.21: Time esponse o he TIA: F=100kHz,R =10kΩ.
By c i ically analyze and compa e wi h 3.18a, he esul s in 3.21, a 100kHz, i was conside ed
he bes esul we e he 2.2pF and 2.7pF capaci ances, he igu es 3.21b and 3.21c espec i ely.
Figu e 3.22: Ou pu signal o TIA: F=500kHz,R =10kΩ, wi hou C .
A he ansmission equency o 500kHz in igu e 3.22, i is no iceable he signal slowing
down and he noise s ill emaining, and wi h he capaci o s he delay begins o inc ease, al hough
3.3 Op ical Recei e - Rx 33
he e a e no e y signi ican a ia ions be ween capaci ances, being he noise emo e ac o mo e
p eponde an , as shown in igu e 3.23.
(a) C =1.8p . (b) C =2.2pF.
(c) C =2.7pF. (d) C =3.3pF.
Figu e 3.23: Time esponse o he TIA: F=500kHz,R =10kΩ.
Again by analyzing he igu e 3.23, he esponses o 3.23b and 3.23c seem o be he be e
esul s.
Finally, he TIA is es ed a 1MHz, whe e an imp o emen mus be done o he ou pu signal
ep esen ed in igu e 3.20. In he p in s p esen ed in 3.24 a e displayed he esul s om he a ious
capaci o s.
34 Implemen a ion
(a) C =1.8p . (b) C =2.2pF.
(c) C =2.7pF. (d) C =3.3pF.
Figu e 3.24: Time esponse o he TIA: F=1MHz,R =10kΩ.
A his equency, he esponse imes a e a hei lowes ob aining he highes delay. In he
p in 3.24d, al hough i migh show a mo e il e ed ou pu , he delay displayed in he all ime is
oo high o be conside a e a iable choice. Taking in o accoun all he p e ious es s and his las
one, i seems ha he p in 3.24b, wi h a capaci o alue o 2.2pF is he mos iable choice o
he eedback capaci o o he TIA. Wi h hese es s, i is possible now o p esen he ecommended
ea u es o he TIA, as shown in able 3.2.
Table 3.2: TIA block speci ica ions.
T ansimpedance Ampli ie
Componen THS4631
Powe Supply (VS+,VS−)+12V, -12V
Feedback Resis o (R )10kΩ
Feedback Capaci o (C )2.2pF
3.3.3.3 Bandpass Fil e
The nex block o he p ocessing is he BPF, and as explained p e iously he objec i e o his es
is o analyze he beha io o he ou pu signal by changing he equency o he inpu ansmission
3.3 Op ical Recei e - Rx 35
and de ec unc ional esul s, as when he magni ude o he signal declines and i he phase shi s.
Jus as a gued in he pas subsec ion 3.3.2.2, i was belie ed ha using a single supplied ampli-
ie o he il e , would be mo e bene icial, since i simpli ies he supply necessi ies and limi he
ou pu o he desi ed le els, bu in e y i s es s, he ou pu signal was no emo ely he expec ed,
since he single supply dis o ed he il e ed ou pu , whe e an inpu sinusoidal wa e did no ha e
he desi ed magni ude ou pu , and phase shi s we e no isible. Luckily, he chosen ampli ie was
able o wi hs and single a dual supplies o +5V and -5V, which would allow o ob ain he desi e
ou pu .
Fo ha eason, a se up simila o he igu e 3.25 was a anged.
Figu e 3.25: Tes se up o he BPF es ing.
Due o he absence o componen s wi h he p oposed alue o ob ain he p ecise cu o e-
quencies, o hese es s i was used componen s wi h simila alue o app oxima e as close as
possible o he ideal alues. The alues o he componen s o hese es s and espec i e cu o
equencies a e:















R1=100Ω
R2=1kΩ
C1=1µF
C2=18pF
⇒


c1≈1.6kHz
c2≈8.8MHz
(3.11)
The i s equency es ed was a 1MHz, because i is one o he p e ended ansmission e-
quency a es and also has been used as a base es equency all along he wo k. The esul ing
ou pu wa e is shown in igu e 3.26

36 Implemen a ion
Figu e 3.26: BPF ou pu a 1MHz
As i may be assumed, a a i s sigh he ou pu signal has he p e ended magni ude and igh
phase, since i is an in e ing il e , bu is di icul o see he delay caused by i . Because o ha
p oblem, he p in s o he oscilloscope we e in e ed o enable a be e ime esponse pe cep ion.
Figu e 3.27: BPF a 1MHz wi h in e ed ou pu .
Now in igu e 3.27 i is possible o clea ly see ha he ou pu is su e ing om a na u al delay,
since his equency is close o he egion o high cu o equency. I is con i med ha he close
i ge s om his egion he g ea e he delay om he il e , as shown in igu e 3.28, whe e he
inpu is a 5MHz signal. As also can be seen he magni ude o he signal is a ec ed as well.
3.3 Op ical Recei e - Rx 37
Figu e 3.28: BPF in e ed ou pu a 5MHz
Bu e en ually, i he signal eaches o su passes he cu o limi egion, he magni ude o he
signal is highly a enua ed and he phase is comple ely shi ed, as i is possible o see in igu e
3.29, whe e he inpu signal is il a ed.
Figu e 3.29: BPF in e ed ou pu a 10MHz
Then, i a signal app oxima es he high equency cu o egion su e s om a delay, con a y,
i a signal disloca es o he low equency cu o egion he ou pu is in ad anced. The igu e 3.30
shows exac ly his phenomenon, whe e he ou pu ace is ahead o he inpu ace.
38 Implemen a ion
Figu e 3.30: BPF in e ed ou pu a 10kHz
And jus as a he high equency cu o egion, a 1.6kHz he ou pu signal su e s om high
a enua ion in he magni ude, and ge s in ad ance o he inpu signal, as can be concluded in igu e
3.31.
Figu e 3.31: BPF in e ed ou pu a 1.6kHz
Wi h hese es s i was possible o con i m ha i wo ks like i should and i was able o analyze
and p edic he beha io o he il e in he ange o equencies. The possible speci ica ions o he
BPF a e shown in able 3.3.
Table 3.3: BPF block speci ica ions.
Bandpass Fil e
Componen AD8041
Powe Supply (VS+,VS−)+5V, -5V
High Pass Componen s (R1,C1)100Ω,1.2µF
Low Pass Componen s (R2,C2)1kΩ,2.2pF
3.3.3.4 Compa a o
Fo his inal block, o es he compa a o , a e y simple se up, as shown in igu e 3.32, was
assembled o es i s esponse.
3.4 Physical Casing 39
Figu e 3.32: Tes se up o he compa a o es ing.
Because he h eshold was ixed as hal o he supplied ol age, he esis o s Rd1and Rd2mus
be equal. Also he esis o Rpwo ks as a pull-down esis o ha is esponsible o speed up he
compa a o ime esponse.
(a) T = 250ns. (b) T = 100ns.
Figu e 3.33: Time esponse o he compa a o a 1MHz.
As shown in he images o 3.33, he ime esponse o he compa a o is mo e han sa is ac o y,
he only de ail o une in he p o o ype is he h eshold.
Table 3.4: Compa a o block speci ica ions.
Compa a o
Componen AD8561
Powe Supply (VS+,VS−)+5V, 0V
Pull-Down Resis o (Rp)10kΩ
Vol age Di ide (Rd1,Rd2)1kΩ,1kΩ
3.4 Physical Casing
As e e ed in he beginning o his disse a ion, he main goal is o c ea e a communica ion module
ha is able o communica e in an unde wa e en i onmen . In o de o achie e his i was necessa y
o c ea e a wa e p oo casing o place all he elec onic componen s inside. Addi ionally, i mus
46 Implemen a ion
Also some addi ional decoupling capaci o we e included in all he in eg a ed ci cui s (ICs),
such as in he egula o s, acco ding o hei da ashee , in o de o educe hei noise. Fo each i
was used a pai o ce amic and an alum capaci o s wi h a lowe equi alen se ies esis ance (ESR)
han he ypical elec oly ic capaci o , o boos he decoupling pe o mance, and o ease he o e all
PCB layou .
Finally he las change made o he ci cui y was o include a MOSFET d i e . Despi e p e-
iously his was a disca ded op ion, i was hough ha wi h his d i e a be e isola ion o he
g ound om he MOSFET, and a smalle dis o ion o he Tx signal was possible. Wi h a new
ma ke sea ch i was ound he UCC27517 ha looked like a iable op ion, since i is sui able o
he use wi h mic o-con olle s and is compa ible wi h he TTL, wo king wi h a logic le el signal o
3.3V o 5V, and as a as esponse ime ( ounding abou he 20ns) ha does no induce a signi ican
delay o he ansmi ed signal. All he alues o he ga e esis o ( ep esen ed by Rga e) whe e he
ecommended om he da ashee .
In igu e 3.41 i is shown he e olu ion and he di e ences o he i s p o o ype and he inal
e sion.
Figu e 3.41: The p ima y Tx p o o ype on he le and he inal Tx PCB on he igh .
As i will be e e eed la e on his wo k, i was possible o achie e he possibili y o ha e
only he ci cui y wo king wi h a supply o -5V and +5V, which esul ed in emo ing he +12V
egula o om he Tx laye , imp o ing he sys em e iciency and simplici y. The nega i e ol age
con e e was eloca ed o he Tx laye .
In he appendix Ai is possible o examine mo e p ecisely he PCB schema ic and i s layou .
3.5.4 Rx Laye
Finally, in he Rx laye is whe e he mos complex ope a ion o ecei ing he ligh message, is p o-
cessed. Because o ha , his laye has mo e componen s in i , o be able o pe o m he necessa y
ope a ions.
Jus like he Tx laye , he Tx equi ed mo e han one e sion o achie e an accep able p o o ype
esul . In he i s app oach, beginning wi h he TIA, he choice o componen s was main ained
he same ( he THS4631 wi h he R =10kΩand he C =2.2pF) as he p e ious es s, aking in o

3.5 P oposed P o o ype 47
accoun his same es s. In he BPF he ampli ie used was changed, in o de o a oid he necessi y
o ha ing an addi ional egula o p oducing -5V ha would only be used by his componen . So
he LM7171 was chosen o pe o m he il a ion, since is mo e han able o handle he equencies
needed and as a simila ol age supply like he TIA, he es o he componen s s ayed he same
as es ed. Fo he compa a o , he only change is he h eshold alue V e will be egula ed wi h a
po en iome e . Also because o he necessi y, a ol age egula o was placed in he PCB o ou pu
he necessa y -12V o he componen s.
The only p oblem in he assembly o his i s e sion was only he ac ha he p e ended
elec oly ic capaci o s o be used in he nega i e con e e we e no deli e ed un il he end o
June, making i impossible o be used acco ding o he ini ial planning, so some de aul ce amic
capaci o s we e used. Figu e 3.42 shows he inal p oduc o he i s e sion o he Rx PCB.
Figu e 3.42: Fi s assembly o he Rx laye .
S ill sho a e he i s es s on he PCB, some de ec s we e ound in he design. In addi ion
o he inabili y o acqui ing he signal while ansmi ing due he noise gene a ed by he MOSFET,
he no mal acquisi ion was no possible by mo i es unknown. As well an audible sound noise was
gene a ed by he PCB om his Rx laye . I was p esumed ha a componen was damaged o
mal unc ioning.
The i s debug es was o cu and sepa a e he unc ion blocks om each o he and analyze
we e he p oblem was. Meanwhile, i was iden i ied he necessi y o a bu e be ween he TIA and
he BPF in o de o sepa a e he ope a ions in cu en o he TIA, and he ol age ampli ica ion
ope a ion o he BPF, making i a good p ac ice in his ci cui .
In one o he Rx laye s he e was also an issue ega ding he nega i e ol age con e e , because
one o hen was wo king and he o he was no . La e his complica ion was iden i ied due o an
unp edic able p oblem wi h his de ice, whe e in one o hem a la ch occu ed because o esidual
cha ges in he capaci o s. This p oblem can be ixed by ollowing he da ashee s eps, and use a
de aul diode (LS4148) o his andom issue, ha can occu .
48 Implemen a ion
I was checked ha he TIA was wo king p ope ly, bu a e cu ing he blocks connec ions, i
was diagnosed a p oblem in he BPF we e no signal was been il e ed.
So a na owe debug es was applied o he BPF, by i s ly using only he esis o s as a sim-
ple In e ing Ope a ional Ampli ie and hen adding he capaci o s o check i he ou pu is he
expec ed. By edoing and es ing s ep-by-s ep he BPF, i was possible o make i wo k like i
was ini ially p e ended. Subsequen ly an imp o ised bu e was included be ween he TIA and he
BPF (using a BUF634, acqui ed in INESC TEC) and i was possible o see he ecei e signal wi h
he co ec il a ion and ampli ica ion.
Bu e en wi h his co ec ions when a emp ing o connec he BPF ou pu o he compa a o
he audible noisy sound eappea ed, so i was assumed ha he p oblem esided in he compa a o
o e en in i s connec ion. Ini ially be ween he BPF and he compa a o i was placed a de aul
diode (again he LS4148) in o de o a oid a nega i e inpu in he compa a o , making i basically
a sa e y measu e and no hing mo e. Bu i was p esumed ha his diode could ep esen a compli-
ca ion, and ha he po en iome e ha con ol he h eshold V e o he compa a o was imp ope
o his p ecision applica ion necessi y.
So e en mo e changes we e made o his i s e sion o he PCB, by elimina ing he connec ion
diode, connec ing a p ope imme , o ge a mo e p ecise V e and eplacing he compa a o o
exclude any e o ega ding damaged ma e ial o o he possible causes. A e applying all his
changes, o he i s ime, a success ul ou pu signal was possible o be ex ac ed.
In he end o he p elimina y es ing o he PCB, in igu e 3.43, shows he necessa y modi ica-
ions made.
Figu e 3.43: Visible imp o ised modi ica ion in he i s e sion o he Rx laye .
On accoun o he same p oblem wi h he PCB aces such as he Tx ini ial laye , and all he
o he mis ake men ioned abo e and some mo e conside a ions, i was delibe a ed ha an enhanced
e sion o his PCB would be an ad an age.
3.5 P oposed P o o ype 49
So o new inal e sion be ween he TIA and he BPF a Bu e was included, he only aspec
o wa ch o is i his bu e is able o ope a e in he equi ed equencies and ha would delay he
signal as less as possible, and he p e iously used bu e BUF634 was mo e han able o ope a e
in he equi ed equencies wi hou delaying he signal signi ican ly. I s simple connec ions and
ci cui layou we e also a ac i e ai s o i s choice.
Like was men ioned ea lie he imme was also an impo an change, so o his e sion
a SMD 10kΩ imme was used o con ol he h eshold wi h mo e p ecision, in o de o make
ansmissions mo e e ec i e.
Since i was gi en a new oppo uni y o c ea e a new e sion o he Rx, a new hypo hesis was
explo ed o be able o ope a e he Rx laye wi hou eso ing o he +12V and -12V. To make his
hypo hesis o be ue, i was sea ched an AMP-OP ha would manage he p e ended equencies o
ope a ion, wi h a low noise inpu ol age and ha can wo k a only +/-5V supply. A e an ex end
ma ke sea ch, i was ound an good op ion, he LMH6624, which has a ema kable pe o mance
a a low supply ol age o +5V and -5V, wi h a low noise inpu ol age and is able o ope a e in
he in ended equencies.
Wi h his modi ica ion, a new imp o emen was made o all he ci cui y, making unnecessa y
he exis ence o he +12V egula o , simpli ying he ci cui . Wi h his, as men ioned p e iously
he nega i e ol age con e e was placed in he Tx laye , changing sligh ly he module pin-ou .
The emaining modi ica ion we e simila o he Tx laye , by c ea ing a ecommended g ound-
plane in he PCBs and ein o cing he decoupling capaci o in each unc ion block o he laye .
Fo his laye , due o he complex layou i was addi ionally necessa y o make his laye a dual
plane laye . This made ac ually easie o es sequen ially he laye blocks one-by-one.
In
(a) Ini ial P o o ype (b) Final P o o ype
Figu e 3.44: The change om he ini ial p o o ype o he inal.
50 Implemen a ion
To inalize his laye , in igu e 3.44 i is shown he e iden changes o he ini ial model and he
end p o o ype.
3.5.5 Assemble and Connec ion
To assemble all he elec onic componen s o he Tx and Rx i was necessa y o c ea e di e en
PCBs o each laye . To be able o do such a hing, each PCB would ha e o ma ch up he
dimension equi emen s o he physical casing, which means ha he ideal solu ion is o c ea e
ound PCBs o each laye wi h a maximum diame e o 40mm.
Figu e 3.45: P e iew o he o de o he laye s.
The module as shown in igu e 3.35 begins wi h a physical cable whe e he necessa y 24V and
g ound a e supplied alongside wi h he ansmi e and ecei e cables. They connec o he i s
PCB, he Tx d i e , by solde ing he cables di ec ly o he PCB. This one is ixed by sc ews o he
casing, and hen his laye connec s o he Rx PCB h ough pin head ( esembling he connec ion
be ween an A duino and a shield). Some o his pin heads a e jus some supply ol ages o he
Rx PCB, bu also as some pin heads ha ha e o connec o he LED PCB and o he ou pu pin
heads o connec o he PD PCB.
Finally, by placing he PD laye in ing shape be ween he op ical lens and he LED laye and
a ach he lens o he LED PCB, he module is comple e. The op ical lens ocus each LED o he
PCB and he PDs ha e some clea space o de ec ligh h ough he hole p esen in he lens.
The igu e 3.46 shows he inal aspec o he module, showing he connec ions be ween all he
laye s o he sys em ha a e inside he casing.
3.6 Summa y
In his chap e i was p esen ed all he main componen s o he communica ion sys em sepa a ely,
showing he p elimina y es s made o alida e he concep and o ake some p io conclusions on
how i will pe o m in di e en es condi ions.
I was ob ious ha his eal implemen a ion o he p o o ype was way mo e elabo a ed han
he expec ed, and a subs an ial amoun o ime was in es ed in i , in o de o achie e a wo king
3.6 Summa y 51
Figu e 3.46: The nucleus o he communica ion module, assembled all oge he .
and es able p o o ype. Wi h all his wo k is accu a e o say ha he physical concep ion o his
p o o ype, is he solid co e o all his wo k.
I goes wi hou saying ha a lo o imp o emen s can be done o all he p o o ype, some which
will ha e e e ence in he chap e o conclusions and u u e wo k.
Now ha he implemen a ion o he p o o ype is concluded some es s on bench and unde -
wa e mus be done o alida e he concep , and o see i he e a e any adjus s o be made o he
sys em.

52 Implemen a ion
Chap e 4
Tes Resul s
In his chap e , i is p esen ed all he es s ha ha e been done o alida e he implemen ed solu ion.
The es me hodology s a ed wi h he simples es s in a wo kbench wi h di e en physical
ac o s and when he p o o ype is all es ed in ai -bo ne, i s ansmission is subjec ed unde wa e
in he labo a o y pool p esen ed in he depa men .
Fi s he es p ocedu e is p esen ed, explaining wha impo an da a is collec ed, how he es
is done and how a esul is egis e ed. Then i is p esen ed he esul s o he es s pe o med is
di e en en i onmen s.
4.1 Tes P ocedu e
Fo e e y es ini ially, he pa ame e s which a e eco ded and a ied, in each di e en es , a e he
ollowing:
•Da e and Hou o he es ;
• The ype o Tes Scena io;
• Type o Casing o be es ed;
• The Ligh Colo o he ansmission;
• Which F equency is ansmi ing;
• The De ia ion o he modules be ween each o he ;
• The ambiance Ligh Le el;
• The Time o ansmission;
• And he alue o he compa a o Th eshold;
The e a e wo es scena ios, i s ly es ing he p o o ype ou o wa e in he wo kbench an
in he es pool. In he casing pa ame e i is possible o see he p o o ype pe o ming wi hou
53
54 Tes Resul s
casing no he op ical lens, only wi h he ocusing c ys al and wi h no casing, o wi h wo di e en
ypes o casing. One wi h a long end, able o be e ejec he in luence o he ambiance ligh , bu
ob iously mo e sensi i e o de ia ions, and a sho end wi hou he de ia ion p oblem bu mo e
a ec ed by ambiance ligh . The es ing end pa s o he casing can be seen in igu e 4.1.
Figu e 4.1: The wo es ed casings. On he le , he long one, and on he igh , he sho one.
Also, he wo s udied high b igh ness LEDs ligh colo s we e he g een ( igu e 4.2c) and blue
( igu e 4.2a), es ing a wo di e en equencies o 1MHz and he 100kHz. In e ms o he de i-
a ion, he holde had some ma ks ma king he de ia ions o 0, 15, 30 and 45 deg ees, being his
de ia ions subjec o es ing ( igu e 4.2b).
Since he es s we e pe o med in a labo a o y, he ambiance ligh le el was also somewha
con ollable, so h ee cases we e s udied, wi h medium ligh s, maximum ambiance ligh and wi h
no ligh a all. The emaining pa ame e s we e only moni o ing a iables, so no es s we e pe -
o med by a ia ing hem.
(a) (b) (c)
Figu e 4.2: The blue Tx, he module holde and g een Tx.
4.1 Tes P ocedu e 55
In addi ion, o connec he e minal o he modules cables o he espec i e sou ce o ou pu , i
was necessa y o implemen a "de i a ion" boa d, in o de o pe o m his ask. So a e y simple
ci cui boa d was c ea ed o ul ill his necessi y. I ea u es a connec o o each cable o he
module, some es poin s, a command swi ch and a use in o de o p o ec he ba e y om a
sho -ci cui i needed. The boa d is shown in igu e 4.3.
Figu e 4.3: The de i a ion ci cui boa d.
In e e y es i is ollowed a ce ain p ocedu e, ha can be in e p e ed like a p o ocol, ha is
oughly he same in each es scena io.
The pai Tx and Rx a e placed in a ce ain ange o be es ed, ixed and hen hey a e bo h
supplied wi h hei ol age supplie s. A his ime, bo h o he modules a e in ac i e Rx mode,
which means ha bo h a e able o ecei ing a ligh signal.
Then a signal, gene a ed by a unc ion gene a o o a mic ocon olle , is placed in he p e ended
ansmission cable and he Tx u ns on. By now, he ecei e s o he Tx a e ine ec i e because
hey ge absolu ely dazzled by he ligh o he Tx.
Then he ol age is egula ed un il a plausible communica ion is es ablished. When a ans-
mission is se led, he ini ial consumed cu en is w i en down and his ope a ion is main ained
du ing a p ede ined ime pe iod. When his pe iod ends he inal cu en is eco ded.
The es ing case ends when he maximum supply, a ound 24V, is eached. A his poin , i he
de ice does no ansmi a a ce ain ange, his dis ance is known as i s ange limi .
4.1.1 De e mining a Resul and Rela ed P oblems
Righ a e he implemen a ion s a ed o show some p omising pe o mances, a ques ion occu ed
in e ms o es ing and e ie ing esul s: How can a ansmission be classi ied as success ul o
e en plausible, and how o measu e e o a es o de i a i es?
62 Tes Resul s
(a) G een Tx
(b) Blue Tx
Figu e 4.9: Tes esul s using di e en ansmission equencies o 100kHz and 1MHz.
Summing up, i is concluded ha in en i onmen ou o wa e , in e ms o he wa eleng h o
he emi ing ligh , he ela ion o he PD and he blue ansmission LEDs seems o be he s onges
combina ion, a he han he g een ligh , which seems o be way less b igh han he blue Tx.
I was con i med ha he ac ylic lens does enhance he deice ange signi ican ly, by pe o ming
i s unc ion o ocusing he ou pu ligh o he Tx, and s ill enabling he ecep ion o ligh by he
Rx.
In he o he expe imen al cases, he ligh le el in e e ence es s we e p e y much incon-
clusi e, since he high in luence o he de ia ion due o he using o he ac ylic lens, p oduced

4.3 Wo kbench Tes s 63
inconsis en esul s ha did no allow a mo e c i ical e dic .
Meanwhile, he impac o he h eshold ha con ols he compa a o ou pu , seemed o be
ano he p eponde an ac o o achie e a be e ansmission. S ill his is a delica e ma e since i
a oo low h eshold is de ined, highe he p obabili y o e o cap ions and e a ic ou pu s due o
he ambiance ligh in e e ence.
Finally, in he wo di e en ansmission equencies, no majo in luence was de ec ed, which
means ha in bo h equencies he ansmission is much alike.
4.3 Wo kbench Tes s
Since he i s es s a e concluded, i is necessa y o es he module wi h he wa e p oo casing, o
check i i can ansmi e en wi h he closed casing. This casing capabili y es will be con i med
i s ou o wa e and a e in he unde wa e en i onmen .
Now ha he casing is going o be es ed, all he s uc u e ha will be used in he unde wa e
es s will also be added. So o his simple se up, as i is shown in 4.10, he casings o he Tx and
Rx we e ixed o a holde wi h ma ked posi ions o he de ia ion angle o be es ed. This holde
i is se led in a so o unway wi h ma ks on i , in o de o ma k he a ious ange posi ions.
Figu e 4.10: Wo kbench es se up.
The es ing pa ame e s a e he same as he p e ious es s, bu now he de ia ion angle is con-
side ed in ou di e en posi ions: no de ia ion, 15 deg ees, 30 deg ees and 45 deg ees. Also i
was possible o es his casing wi h mo e han one end e minal, and wi h a long head and a sho
head.
In a mo e impo an no e, since he con ol o he V e is no accessible wi h he casing closed,
i was needed o calib a e he h eshold in a ixed alue. Since, o he Rx, a lowe alue would
enhance i s pe o mance, bu a he same ime, makes i mo e suscep ible o he noise, a ponde ed
middle e m needed o be eached. So bo h h esholds we e calib a ed o 740mV, in o de o
imp o e he communica ion, wi hou inducing so much noise.
So o he i s es , i will be es ed he pe o mance o he g een and he blue ligh a 1MHz,
by applying no de ia ion, and wi h an a e age in e e ence o ligh .
64 Tes Resul s
Figu e 4.11: Resul s o he g een and blue ansmissions, using he wo casing ypes.
As i is shown in he g aph 4.11, he use o he casing, independen ly i i is he long o sho
ype, he ange o he ansmission, isually compa ing o he con ol subjec , is mo e limi ed. This
is p obably because o he design o he casing and he posi ioning o he PD ing. E en hough i
is able o cap ioning he signal, a a sa is ac o y ange, i always limi s he cap u e o he signal.
In e ms o which casing in his es was he e ec i e, hei di e ence is almos i ele an . In
he g een ligh pe o mance, he aces o he long and sho casing a e almos coinciden , so he
di e en casings did no in luence he ansmission in e ms o he ange o he powe consumed.
Only in he blue ligh case, when no de ia ion is p esen ed in he ansmission, he long casing
seems o pe o m a li le be e han he sho one.
In a i s imp ession, he long casing compa ing o a sho one, in a ansmission, such as
expe imen ed in he es 4.11, indeed does no ep esen a e y ele an di e ence, since he en y
and exi poin a e supposedly well aligned.
The mos ele an di e ence be ween his wo casings migh occu in wo di e en scena -
ios: when he in e e ence o he ambiance ligh is highe , whe e in his si ua ion, he long casing
po en ially has an ad an age, since i p ope ly isola es he Rx om his e ec . And when a de-
ia ion be ween he wo modules exis , whe e now in his si ua ion he long casing migh ha e a
disad an age compa ing o he sho , since i can limi he Tx ligh ou pu .
So o he ollowing expe imen s, i s ly, he ambiance ligh in luence will be es ed by ap-
plying he h ee di e en ligh in ensi ies. Then a e some conclusions a e made, he de ia ion
beha io will be es ed, in o de o con i m o no he p esumed conclusions.
4.3 Wo kbench Tes s 65
(a) Long Casing
(b) Sho Casing
Figu e 4.12: Resul s o he in e e ence o di e en le els o ambien ligh .
The esul s displayed in he g aphs 4.12, shows he p edic ed beha io . By analyzing he
esul s on he g een ligh , al hough i is a e y small di e ence, in he long casing end, shown in
g aph 4.12a, he ecep ion o he h ee di e en in e e ence main ained p ac ically he same, and
in he sho end, shown in 4.12b a small di e ence be ween he ecep ions can be de ec ed. As o
he blue ligh , i has hinne ace all along i s domain, in he long end, meaning ha has a smalle
a ia ion, and in he sho end shows a mo e uns able ou pu .
Despi e all his conclusions, e en wi h he di e en ligh in e e ence all he ou pu s a e sa is-
ac o y.
66 Tes Resul s
Since he impac o a de ia ion, on he p e ious con ol es s, was e y isible, i is possible o
assume, ha by de ia ing he modules be ween each o he , he ou pu will espond o he s imula-
ion. I is only necessa y o analyze how big he casing will limi he ecep ion and ansmission.
(a) Long Casing
(b) Sho Casing
Figu e 4.13: Resul s o he di e en module de ia ions.
As expec ed, in he g aph 4.13 he de ia ion add a huge impac on he ligh ou pu , bo h in
he g een and he blue Tx. The de ia ion ha achie ed he maximum ange was he alue o 15
deg ees. The o he de ia ion alues g ea e han his had a wo se pe o mance han he a e age
signal.
4.3 Wo kbench Tes s 67
I is impo an o s a e ha by compa ing he long and sho casing ou pu s, espec i ely g aph
4.13a and 4.13b, is clea ly isible he limi ed ange o he Tx when he de ia ion is supe io o
15 deg ees, which means ha he long casing limi s he ansmission o he signal o de ia ions
supe io o 15 deg ees.
The only expe imen ha is le o do is he ansmission a 100kHz, in o de o s ipula e
he beha io o he module wo king a a lowe equency, capable o being se ial communica ion
suppo ed. Fo his es , i was only pe o med using he long casing, since no ad an age was
p edic ed in es ing in he sho casing.
Figu e 4.14: Tes esul s using di e en ansmission equencies o 100kHz and 1MHz.
In his ansmission a 4.14 he only pa icula occu ence, only applicable o he blue ligh , is
he mo e limi ed ange, which s a ed o be insigni ican bu in he end i go conside ably la ge.
Also a 100kHz he powe consump ion is highe .
Now ha he beha io ou o wa e is known, he concep will be es ed unde wa e .

68 Tes Resul s
4.4 Labo a o y Pool Tes s
Figu e 4.15: Labo a o y pool es se up.
Be o e beginning he ansmission unde wa e , some a en ion like a oiding he cables being we
we e also necessa y.
Figu e 4.16: Resul s o he g een and blue ansmissions, using he wo casing ypes.
The esul s o g aph 4.16 show a esul o he han he p edic ed. Due he communica ion
unde wa e , he signal ou pu is sligh ly a enua ed, being he maximum ange o 1.80 me e s, no
as good as o he cases.
The es o he g een ligh wi h he long casing, was no e y success ul, inducing some e o
o he measu emen s.
4.4 Labo a o y Pool Tes s 69
Now he sys em will be es ed o he di e en ambien ligh le el in e e ence. Since his
happened in he p e ious es , some a enua ion is p edic able o his case.
(a) Long Casing
(b) Sho Casing
Figu e 4.17: Resul s o he in e e ence o di e en le els o ambien ligh
As i can be seen in he g aphs 4.17, i main ained he simila pe o mance, being he long
casing ace hinne han he ace o he sho case, due o he in e e ence o ligh . The e o is
he g een measu emen s a he de aul ligh le el.
Fo nex es s he pe o mance o he modules will be es ed by applying he de ia ions ma ked
in he holde . Since he p e ious wo kbench es s, he bes ange esul s we e ob ained when he
70 Tes Resul s
de ia ion was applied o he modules, and al hough he gene al pe o mance o he sys em declines
unde wa e , i is p edic ed ha he bes ange esul s unde wa e will be ob ained also by applying
he de ia ion.
(a) Long Tx
(b) Casing Tx
Figu e 4.18: Resul s o he di e en module de ia ions.
In he g aphs 4.18, mo e p ecisely in he 4.18a, he signal o he g een ligh wi h no de ia ion,
ha was assumed like an e o measu emen , looks like an ou pu esponse o he al eady de ia ed
ligh . In 4.18b, also he g een does no ep esen a big de ia ion like i should show, which may
indica e ha he g een ligh is mo e in luenced by he wa e han he blue ligh , which is a bi
a enua ed, bu no in a such signi ican di e ence.
4.5 Summa y 71
Figu e 4.19: Tes esul s using di e en ansmission equencies o 100kHz and 1MHz.
In his las case, as shown in he g aph 4.19 he ansmission in 100kHz using blue ligh ,
agains he no mal endency shown by he pas expe imen s, is eaching highe anges wi h lowe
consump ion. The g een ligh p esen s a e y p edic able beha io , bu does no shown de ia ion
laws.
4.5 Summa y
In his chap e , i was possible o es and analyze, e en i b ie ly, he cha ac e is ics o he com-
munica ion module. Al hough a om pe ec ion some basic concep s we e also alida ed, which
con ibu ed o he alue o he wo k.
As some pe o mance speci ica ions, he p o o ype was able o ansmi ing up o 2.80 me e s
wi h a de ia ion o 15 deg ees ou o wa e , and was able o ansmi ing up o 2.00 me e s wi h
he same de ia ion o 15 deg ees in unde wa e .
This es s enable he possibili y o alida ing and ul ill he p ima y objec i e ha is c ea ing a
p o o ype able o ansmi ing in o ma ion unde wa e .
The ecommended imp o emen s and new ideas o inc ease he po en ial o all his p ojec will
be desc ibed in he u u e wo k sec ion.
78 PCB Foo p in s and Schema ics

A.1 Ini ial Ve sion 79
A.1 Ini ial Ve sion
A.1.1 Tx Laye
Figu e A.1: Ini ial PCB Schema ic - Tx laye .
80 PCB Foo p in s and Schema ics
Figu e A.2: Ini ial PCB Foo p in - Tx laye (only op laye ).
A.1 Ini ial Ve sion 81
82 PCB Foo p in s and Schema ics
A.1.2 Rx Laye
Figu e A.3: Ini ial PCB Schema ic - Rx laye .
A.1 Ini ial Ve sion 83
Figu e A.4: Ini ial PCB Foo p in - Rx laye (only op laye ).

84 PCB Foo p in s and Schema ics
A.2 Final Ve sion 85
A.2 Final Ve sion
A.2.1 Tx Laye
Figu e A.5: Final PCB Schema ic - Tx laye .
86 PCB Foo p in s and Schema ics
Figu e A.6: Final PCB Foo p in - Tx laye (only op laye ).
A.2 Final Ve sion 87
94 REFERENCES
[42] Sona dyne. BlueComm Unde wa e Op ical Modem. h p://
www.sona dyne.com/p oduc s/all-p oduc s/ins umen s/
1148-bluecomm-unde wa e -op ical-modem.h ml. [Online; accessed 13/Feb u-
a y/2015].