A Discrete Time Oscillator
for a DSP based Radio
Clay S. Turner
TGA Systems, Inc.
3 1 0 0 M ed lo ck B rid g e, S u ite 1 5 0
N o rcro ss, G A 3 0 0 7 1
(7 7 0 ) 4 4 1 -2 1 0 0
1 In tr o d u c tio n
W ith th e p ro lifera tio n o f d ig ita l sig n a l p ro cesso rs,
D S P s, a n d th e ev er in crea sin g n eed fo r h ig h er d a ta
ra tes ov er ra d io ch a n n els, it is n o su rp rise th a t
D S P s a re b eco m in g in teg ra l p a rts o f m o d ern ra -
d io s. S o m e o f th e ea rly u ses w ere fo r p erfo rm in g
sp ectra l a n d d elay eq u a liza tio n ta sk s a n d stra ig h t-
en in g o u t th e n o n -lin ea rities o f th e m ix ers a n d a m -
p li¯ ers. N ow th ey a re a lso b ein g u sed fo r p erfo rm in g
th e m o d u la tio n / d em o d u la tio n ta sk s. N o t o n ly d o es
th is en h a n ce th e ra d io 's p erfo rm a n ce, b u t a lso it
p u ts m o re o f th e ra d io u n d er so ftw a re co n tro l w h ich
sim p li¯ es th e co n ¯ g u ra tio n ta sk o f m u ltifu n ctio n ra -
d io s. In th e co n tex t o f th e p a p er, th e term ra d io
refers to b o th tra n sm itters a n d receiv ers.
H ig h d a ta ra te m o d u la tio n m eth o d s u tilize o p -
tim a l ¯ lters, p recise p h a se a n d freq u en cy sh ifters,
a n d h a rm o n ica lly p u re n u m erica lly co n tro lled o scil-
la to rs. T h ere is a lo t o f in fo rm a tio n in th e litera tu re
a b o u t th e d esig n o f su ch m o d u la to rs a n d d em o d u -
la to rs a lo n g w ith m o st o f th eir co m p o n en ts. B u t
th ere is rela tiv ely little a b o u t d iscrete tim e o scilla -
to rs. T h is p a p er p resen ts a n o scilla to r d esig n th a t
ca n b e u sed in a lm o st a n y m o d u la to r to p o lo g y.
T h e o scilla to r fea tu res a sta b ilized a m p litu d e
w ith q u a d ra tu re o u tp u ts, a n d it m ay b e sm o o th ly
tu n ed . T h e a u th o r h a s im p lem en ted th e a lg o rith m
in a M o to ro la D S P 5 6 0 0 2 . T h e su b ro u tin e co n sists
o f 2 3 in stru ctio n s a n d co n su m es ju st 4 6 p ro cesso r
clo ck cy cles p er o scilla to r o u tp u t. T h is in clu d es
freq u en cy tu n in g a n d b o th th e I a n d Q o u tp u ts.
If tu n in g is n o t req u ired , th en o n ly 2 8 cy cles a re
n eed ed .
2 O sc illa to r s
T h e cla ssica l a p p ro a ch to o scilla to r d esig n u ses a n
a m p li¯ er w h o se o u tp u t is fed b a ck to its in p u t v ia
a g a in red u cin g a n d p h a se sh iftin g n etw o rk . T h e
B a rk h a u sen criterio n fo r a n o scilla to r req u ires th e
fo rw a rd g a in tim es th e rev erse g a in to eq u a l 1 a n d
th a t th e p h a se sh ift m u st b e a m u ltip le o f 2 ¼ . T h is
is ea sily d o n e w ith a D S P v ia m u ltip lies a n d d elay s.
2 .1 T a b le L o o k U p
A sim p le, y et freq u en cy lim ited , w ay to g en era te
sin u so id s is b y ta b le lo o k u p . T h e sin u so id is eva lu -
a ted a n d sto red in a ta b le. T o g en era te th e sin u so id ,
th e ta b le's va lu es a re rea d o ® in su ccessio n w ith ta -
b le w ra p a ro u n d . T h is is ea sily d o n e w ith a p o in ter
u sin g m o d u lo a rith m etic. S in ce a p h a se co n tin u o u s
sin u so id is d esired , th e ta b le n eed s to h o ld a n in te-
g ra l n u m b er o f p erio d s. D esp ite th e sim p licity o f th e
ta b le m eth o d , it d o es p o ssess tw o m a jo r p ro b lem s.
T h e ¯ rst is th e req u irem en t th a t th e freq u en cy m u st
d iv id e ev en ly in to so m e in teg ra l m u ltip le o f th e sa m -
p lin g ra te. T h is ca n resu lt in len g th y ta b les. A lso
if q u a d ra tu re o u tp u ts a re n eed ed , th en th e ta b le's
len g th m u st b e a m u ltip le o f 4 , so a n ex a ct 9 0 ± p h a se
rela tio n is m a in ta in ed . T h e seco n d d raw b a ck is th e
q u estio n o f sm o o th freq u en cy a d ju stm en t.
S o m e d ea l w ith th ese p ro b lem s b y u sin g sh o rter
ta b les a n d a ccep tin g th e h a rm o n ic d isto rtio n ca u sed
b y th e p h a se d isco n tin u ities w h ile n o n u n ifo rm ly
step p in g th ro u g h th e ta b le. H ow ev er, so m e a p p li-
ca tio n s req u ire h a rm o n ica lly p u re w av es a n d ¯ n d
th is m eth o d u n su ita b le. A d d itio n a lly, fo r q u a d ra -
tu re o u tp u ts, th is m eth o d h a s a p h a se, b etw een th e T h is rela tio n say s th a t w ith tw o sta rtin g va lu es
ca rriers, n o ise co m p o n en t. a n d tw o feed b a ck va lu es, o n e ca n recu rsiv ely g en -
T h e ta b le lo o k u p m eth o d 's h a rm o n ic d isto rtio n era te a sin e w av e. O n e o f th e feed b a ck va lu es is
a n d p h a se n o ise ca n b e red u ced b y u sin g in terp o - th e co n sta n t,¡ 1 . S in ce th e va lu e o f a in th e trig
la tio n , b u t th en th e to ta l co m p u ta tio n a l e® o rt ex - th eo rem is a rb itra ry, th e recu rsio n is p h a se in va ri-
ceed s th a t o f th e recu rsiv e o scilla to rs. W h en p u re a n t. It is th e sta rtin g va lu es th a t a llow fo r b o th
sin u so id s a re n eed ed , th e recu rsiv e o scilla to rs b e- a m p litu d e a n d p h a se d eterm in a tio n . T h e freq u en cy
co m e th e m eth o d o f ch o ice. en ters in to b o th th e sta rtin g va lu es a n d o n e o f th e
feed b a ck va lu es.
2 .2 R e c u r r e n c e R e la tio n s F o r ex a m³ p le,
´ th e in itia l va lu es co u ld b e set to 0
2¼ f
a n d A sin f s to m a k e a sin u so id w ith freq u en cy
S in ce cla ssica l o scilla to rs fu n ctio n b y feed in g th eir f a n d a m p litu d³e A .´ N o tice th a t th e ¯ lter's ¯ rst
o u tp u t b a ck in to th eir in p u t w ith a n a p p ro p ria te d e- o u tp u t is A sin 2 2f¼sf .
lay, it is d esira b le to ¯ n d a sim ila r rela tio n sh ip th a t
is o p era b le w ith a d iscrete tim e sy stem . T h is d is-
crete tim e rela tio n is ca lled a \ recu rren ce rela tio n ."
¡ ..................
T h e recu rren ce rela tio n sta tes th a t w h en g iv en n .
.
.
.
.
........................................
... §
..
..................
.
.......................................
. ................................................
.
.
.
cos(a + bn )
.
. .
..... .
.
co n secu tiv e va lu es, th e n ex t va lu e ca n b e ca lcu la ted . .
.
.
.
.
.
.
. ....
..
... +
..
.
. .
.
.
.
.
.
.
.
.
. ....
.
.
.
.
If th e recu rren ce is su ch th a t th e n ex t va lu e ca n b e .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
z ¡1
w ritten a s a lin ea r co m b in a tio n o f th e p a st n va lu es, .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
. .
.
. . .
th e recu rren ce rela tio n rep resen ts a n n th o rd er a ll .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 2 cos(b)
.
.
.
.
.
.
.
.
.
. .
. .
.
.
p o le ¯ lter. A n a ll p o le ¯ lter h a s a feed b a ck o n ly .
.
.
.
.
.
.
.
.
.
.
.
.
.
.....................
. ..
.. .
.
.
.
.
.
.
...
....... .
.
.
..
.
..... .
...
.
.
.
.
.
.
.
.
.
.......................................
.
.
.
. ..... .
. .
.
to p o lo g y. .
.
.
.
.
.
.
.
.
...
... .
.
.
.
.
.
.
.
.
.
. .
.
A n o scilla to r th a t g en era tes a sin g le sin u so id is .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
ca lled a sim p le h a rm o n ic o scilla to r, S H O . F ro m th e .
.
.
.
.
.
.
z ¡1
. .
.
. .
th eo ry o f d i® eren ce eq u a tio n s, it ca n b e sh ow n th a t .
.
.
.
.
.
.
................................................................................................
.
.
.
.
.
.
.
.
a d iscrete tim e sin u so id h a s a tw o term recu rren ce
w h ere th e co e± cien ts a re rea l va lu ed . T h is tw o term
recu rren ce tu rn s o u t to b e a cla ssica l trig o n o m etric F ig u re 1 : S im p le H a rm o n ic O scilla to r
th eo rem . It is:

sin (a + b) = 2 co s(b) sin (a ) ¡ sin (a ¡ b) (1 ) 2 .3 Q u a d r a tu r e O sc illa to r s

T o in terp ret th is th eo rem a s a n o scilla to r recu r- P rev io u sly, a tw o term recu rren ce (2 n d o rd er) rela -
sio n itera tio n , let a = 2f¼sf n a n d b = 2f¼sf , w h ere f is tio n w a s u sed to m a k e a S H O . N ow tw o co u p led
th e o scilla to r freq u en cy, f s is th e sa m p lin g ra te, a n d sin g le term (1 st o rd er) recu rren ce rela tio n s w ill b e
n is th e sa m p le n u m b er. T h en th e th eo rem b eco m es u sed to m a k e a n o scilla to r th a t h a s tw o o u tp u ts
w h ere o n e is 9 0 ± o u t o f p h a se w ith resp ect to th e
à ! à ! à ! o th er. T o co n stru ct su ch a n o scilla to r req u ires tw o
2¼ f 2¼ f 2¼ f
sin (n + 1 ) = 2 co s sin n cla ssica l trig o n o m etric th eo rem s. T h ey a re:
fs fs fs
à !
2¼ f sin (a + b) = sin (a ) co s(b) + co s(a ) sin (b) (4 )
¡ sin (n ¡ 1 ) (2 )
fs co s(a + b) = co s(a ) co s(b) ¡ sin (a ) sin (b) (5 )
W ritten in th is w ay, th e recu rren ce is fo u n d to b e
S u b stitu tin g th e sa m e va lu es fo r a a n d b a s b efo re,
à ! th e p rescrip tio n fo r a q u a d ra tu re o scilla to r is:
2¼ f
N ex tS in = 2 co s P resen tS in à !
fs 2¼ f
N ex tS in = co s C u rren tS in +
¡ L a stS in (3 ) fs
 Ã !
2¼ f 2 .4 A m p litu d e S ta b iliz e d O sc illa -
sin C u rren tC o s (6 )
fs to r s
à !
2¼ f
N ex tC o s = co s C u rren tC o s ¡ S o fa r th e d escrib ed o scilla to rs a re b a llistic in th e
fs
à ! sen se th a t th ey a re lo a d ed w ith so m e p reset va lu es
2¼ f a n d a llow ed to free ru n . S in ce th e ¯ lter's co e± cien ts
sin C u rren tS in (7 )
fs a n d d a ta m u st b e q u a n tized to b e p ra ctica l, a D S P
im p lem en ta tio n m ay y ield a n o scilla to r w h o se a m -
F ig u re 2 sh ow s th e n etw o rk fo rm o f th e q u a d ra - p litu d e w ill ch a n g e w ith tim e. If th e w o rd size o f th e
tu re o scilla to r. D S P is la rg e, i.e., 2 4 b its, th e o scilla to r ca n o p era te
L ik e b efo re, th ere a re tw o d elay elem en ts; h ow - fo r a la rg e n u m b er o f itera tio n s b efo re th e a m p li-
ev er, th e in itia l va lu es a re d i® eren t. W ith th e tu d e ch a n g e b eco m es sig n i¯ ca n t. S o m e n u m erica l
S H O ca se, th ey a re tw o co n secu tiv e sa m p les o f a ex p erim en ts h av e d em o n stra ted ov er a m illio n it-
sin u so id | In th is ca se, th ey a re sa m p les o f a sin u - era tio n s a re n eed ed b efo re th e a m p litu d e ch a n g es
so id 9 0 ± a p a rt. T h is la st fa ct m a k es th e q u a d ra tu re m o re th a n 1 0 p ercen t. S o b a llistic o scilla to rs fu n c-
o scilla to r ea sy to co n tro l. tio n v ery w ell fo r sh o rt d u ra tio n to n e b u rsts. H ow -
T h e u sefu ln ess o f th e q u a d ra tu re o scilla to r h a s its ev er, if o n e n eed s to g en era te a ca rrier, w h ich ca n
ro o ts in F o u rier tra n sfo rm th eo ry. M o st im p o rta n t o p era te in d e¯ n itely, a b a llistic o scilla to r w o n 't d o .
is th e freq u en cy sh iftin g th eo rem w h ich say s fo r th e T o sta b ilize th e a m p litu d e, o n e m ea su res th e o s-
fo u rier tra n sfo rm p a ir: cilla to r's a m p litu d e a n d co m p a res it w ith th e set-
p o in t (d esired ) a m p litu d e a n d a d ju sts th e feed b a ck
x (t) , X (! ) (8 ) a cco rd in g ly. T h e q u a d ra tu re o scilla to r a llow s fo r
triv ia l, n o n -freq u en cy d ep en d en t, a m p litu d e m ea -
th a t su rem en t. B y d en o tinpg th e o u tp u ts I a n d Q . T h e
x (t)e i¢ !t
, X (! ¡ ¢ ! ) (9 ) a m p litu d e, A o is ju st I 2 + Q 2 . If th e setp o in t a m -
p litu d e is d en o ted A s , th en th e sta b iliza tio n g a in G
T h e q u a d ra tu re o scilla to r g en era tes th e co m p o n en ts is
o f e i¢ ! t. T h u s th e q u a d ra tu re o scilla to r is a k ey As
G = p 2 (1 0 )
co m p o n en t o f th e freq u en cy sh iftin g p ro cess. I + Q 2
T h e fa cto r G is u sed in th e feed b a ck lo o p s.
If th e in sta n ta n eo u s erro r co rrectio n req u irem en t
.
......................................................................................................................................................
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
is rela x ed , th e co stly d iv isio n a n d sq u a re ro o t fu n c-
.
. .
.
.

z ¡1
.
.
.
.
.
.
.
.
.
tio n s ca n b e av o id ed . T h is is d o n e b y ¯ n d in g th e 1 st
. .
.
.
.
.
.
.
.
.
.
.
cos(b)
.
.
.
.......
.. .. .. ..
..
.... ........
.
.
.
.
.
.
.
.
o rd er T ay lo r's series ex p a n sio n o f G a b o u t A s . T h is
.
. . .......... .
+......... .. .
.
.
.
.
.
.
.
.
....................
.
.
.
.
..
..
..
.......... .
.........
....................... .................... ...................... ..
. §
...... ....
.
. .
.. . .
..
.
.
.
.....................
......................................... cos(a + bn ) a p p rox im a tio n h a s th e p ro p erty o f b eco m in g m o re
. . ..
.
.
. ..... ¡
. ...
.
.
.
.
.
.
.
. sin(b)
........
.....................
.
.
.
.
.
.
a ccu ra te a s A o a p p ro a ch es A s . T h is a p p rox im a tio n
. ..
. ........ .
.
. .
.....................
.
.
............
.
..
. ................................. .
.
.
.
.
.
.
.
.
is:
..
..
..
. .
.
.
.
.
.
.
. 3 1
¡ A s ¡2 (I 2 + Q 2 )
.
. .
.
.
.
.
.
.
.
.
.
.
.
......................
. G ¼ (1 1 )
.
.............................................................
.
.
.
.
.
.
.
.
.
.
.
.......................................................................
.
.
.
.
.
.
.
.
.
.
2 2
.
.
. .
. .
. .
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
S in ce th e a m p litu d e va ria tio n s, w ith o u t sta b iliza -
z ¡1 .
.
. .
.
.
.
.
.
.
.
.
.
.
.....................
.
.
.
.
......................
. ....
.
.
.
.
. tio n , a re v ery sm a ll, th e ¯ rst o rd er co rrectio n w o rk s
.
.
.
.
. cos(b)
...
.
.
.
.
.
.....
. ..
.......
.
.
.
.
.
.
.
.
.
.
....................
.
..
.
.
.
.
.........
. .........
..........
.. ..................... ...................
.
.
.
.
.......................
+
.........
.
.
.
.
.
.
§
.
.. ..... .
.
.
.
....................
.
.
.
.
......................................... sin(a + bn ) v ery w ell.
.
.
.
.
.........
..
..
..
.
.
. + .......................
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
W ith th e co n sid era tio n o f a ¯ x ed p o in tp D S P in
.
.
.
.
.
.
.
sin(b)
........
..
.
..
. .......
.
.
.
.
.
.
. m in d , it b eco m es co n v en ien t to let A s = 2 2 . A lso ,
.
. .
. .................................................
.
.....................
.
. ..............
.
.
.......
..
..
..
. if G 1 is u sed a n d th e co rrected va lu es a re sca led u p
2
b y 2 , th en a ll o f th e va lu es a re in [¡ 1 ;1 ).
F ig u re 2 : Q u a d ra tu re O scilla to r T h u s,
3 1
G 1 = ¡ (I 2 + Q 2 ) (1 2 )
2 4 2
 T h e n etw o rk fo r th e a m p litu d e sta b ilized o scil- cilla to r is ca lled a n u m erica lly co n tro lled o scilla to r,
la to r is sh ow n in ¯ g u re 3 . T h e circles w ith th e N C O .
¦ s in th em rep resen t m u ltip lica tio n b y a va ria b le, T h is m a k es th e N C O id ea l fo r u se a s a freq u en cy
w h erea s, th e tria n g les rep resen t m u ltip lica tio n b y a sh ift k ey in g , F S K , m o d u la to r. A d d itio n a lly, if th e
co n sta n ts. T h is fo rm o f th e o scilla to r w h en u sed in N C O 's, in p u t is ¯ rst d i® eren tia ted , th en o n e h a s a
co n ju n ctio n w ith a d elay lin e, a n d a H ilb ert tra n s- p h a se sh ift k ey in g , P S K , m o d u la to r.
fo rm er, m ay b e u sed fo r sin g le sid e b a n d a m p litu d e T h e a m p litu d e in d ep en d en ce d u rin g tu n in g a rises
m o d u ltio n , S S B A M . L ik ew ise, th is o scilla to r m ay fro m th e fa ct th a t th e tw o sto red va lu es a re a lw ay s
b e u sed to im p lem en t b o th q u a d ra tu re a m p litu d e 9 0 ± a p a rt. T h is is n o t tru e w ith th e S H O . F o r
m o d u la tio n , Q A M , a n d q u a d ra tu re a m p litu d e d e- a tw o o r fo u r lev el F S K sy stem , th e co s(b) a n d
m o d u la tio n , Q A D . sin (b) term s fo r ea ch freq u en cy a re p reca lcu la ted
a n d sto red in a ta b le.
.
.............................................................................................................................................................................................................................................................................
. .
.
F o r a co n tin u o u sly va ria b le freq u en cy m o d u la -
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
tio n , F M , sy stem , co s(b) a n d sin (b) w ill h av e to
.
.
.
z ¡1 .
.
.
.
. b e eva lu a ted in rea l tim e. S in ce th e a m p litu d e sta -
.
.
.
.
. .
.
.
.
.
.
.
.
.
cos(b)
.......
.
..
. .. .................. .
................. . .
.. 2
.......
..
.
.
.
.
.
.
. b ilized fo rm is u sed fo r lo n g term sta b ility, slig h t
.
. . ..........
.
. + .. ... .. ... . . ..........
. .
.
. . . . . .
.
.
.
.
.
. .........
.....................
..........
.
..
..
..
...
..........................................................................
... §
....... .......
.
.
.....................
. .
.
.
.
.
.............................................................
...¦ .
.
...........
.
................ .
.
............
.
..
..
..
..
.......
.......................
I erro rs in th e sin (b) a n d co s(b) term s a re to lera ted .
.
. .... .... .
. .
.
.
.
. ... ¡ .
. .
.
. .
........................ .
.
.
.
.
.
.
.
.
.
.
.
.
.
sin(b)
.
.
.
.......
.. .......
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
T h is m ea n s m o d era te o rd er series ex p a n sio n s m ay
. . ...... . . .
..................... ................................
. . .
.
.
.
.
..
..
..
...........
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. b e u sed fo r sin (b) a n d co s(b).
.
.
.
.
.
.
.
.
.
.
I .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3 1
.
2 2 .............. 1
.
.
.
. A w id e b a n d F M sy stem w ill n eed a C h eb y sh ev
¡ (I + Q ) G
.
. .
............................................................
.
. .....................
.
.
.
...............................
. .
. .
.
. . . 4 2... .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2
ex p a n sio n w h ere th e m a x im u m erro r is m in im ized
. .
.
.
.
.
.
.
.
.
.
.
. Q .
.
.
.
.
.
.
.
z ¡1 .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .......................................
.
.
.....................................................................
.
.
.
.
.
.
.
...................................................
.
.
.
.
.
.
.
.
ov er a n in terva l th a t is m a tch ed to th e w o rk in g
.
. .....................
......................... .
. .
. .
.
.
.
.
.
.
.
.
.
.
cos(b)
..........
..
.
. .
.
.
.
.
.
.
. +
....
.. ....................
. .
.
.
.
.
.
.
. ..
..
.
.
.
.
............... . 2
..........
..
. .
.
.
.
.
.
.
.
ra n g e o f th e o scilla to r. A n a rrow b a n d F M sy stem
. ...........
..................... .
. . .. .
. . .. . . ........... . ...........
.
.
.
.
.
.
.
.
. .........
.
.
...........
..
..
..
.
.
.........................................
.
.
.
.
.
.
..............................
.
... §
+ .................. ..
.
.
.................... .
............................................................
... ¦
.... .... ..
..........
.
.
...
. .
. ........ .........
.
..
..
.
...........
........... Q ca n u se a T ay lo r's a p p rox im a tio n . T h e ex p a n sio n s
. . .
.
. .
.
.
. .
.
.
.
.
.
.
.
.
.
.
sin(b)
.
...........
..
..
.
.
.
.
.
.
.
.
.
.
a re a b o u t th e cen ter freq u en cy. S in ce th e erro r is
.
. .
. ....... .
.
.....................
.
. ..............
.
.
.......
..
..
..
.
.............................................
zero a t th e cen ter freq u en cy a n d sm a ll freq u en cy
d ev ia tio n s a re co m m o n in F M sy stem s, a low o rd er
a p p rox im a tio n ca n b e u sed . S in ce o rd er o f k ilo h ertz
F ig u re 3 : A m p litu d e S ta b ilized Q u a d ra tu re O scil- d ev ia tio n s a re p erfo rm ed o n o rd er o f 1 0 0 m eg a h ertz
la to r ca rriers, it is ea sy to see th a t th e rela tiv e freq u en cy
sh ift, h , is a tin y n u m b er.
If b is th e cen ter freq u en cy, th e fo llow in g 2 n d o r-
d er T ay lo r's a p p rox im a tio n s a re fo u n d :
2 .5 F r e q u e n c y T u n in g
à !
A s m en tio n ed ea rlier, th e q u a d ra tu re o scilla to r h a s h2
co s(b + h ) ¼ 1¡ co s(b) ¡ h sin (b) (1 3 )
m a n y m erits. N o t o n ly is it ea sy to sta b ilize its a m - 2
à !
p litu d e, b u t it is a lso ea sy to a d ju st its freq u en cy h2
w h ile in o p era tio n . S in ce th e tw o term recu rsiv e sin (b + h ) ¼ 1¡ sin (b) + h co s(b) (1 4 )
2
o scilla to r h a s o n ly o n e freq u en cy d ep en d en t co e± -
cien t, if it is a d ju sted w h ile th e o scilla to r is ru n - w h ere h is a sm a ll freq u en cy d ev ia tio n .
n in g , th e o scilla to r w ill ch a n g e a m p litu d e in a d d i- A sim ila r fu n ctio n a l fo rm a rises fo r th e C h eb y -
tio n to ch a n g in g freq u en cy. H ow ev er, th e q u a d ra - sh ev ex p a n sio n s.
tu re o scilla to r w ith its tw o freq u en cy d ep en d en t co -
e± cien ts w ill n o t ch a n g e a m p litu d e w h en th e fre- co s(b + h ) ¼ C (h ) co s(b) ¡ S (h ) sin (b) (1 5 )
q u en cy is ch a n g ed . T h e resu ltin g sin u so id is co n - sin (b + h ) ¼ C (h ) sin (b) + S (h ) co s(b) (1 6 )
tin u o u s. T h u s, th e tu n a b le o scilla to r fu n ctio n s lik e
a v o lta g e co n tro lled o scilla to r, V C O , in a p h a se H ere C (h ) a n d S (h ) a re low o rd er p o ly n o m ia l ex -
lo ck ed lo o p , P L L . T h is d iscrete tim e tu n a b le o s- p a n sio n s o f co s(h ) a n d sin (h ) a b o u t 0 rep ectiv ely.
T h is fo rm u la tio n fo r b o th th e T ay lo r a n d C h eb y sh ev move b,y:(r\reg)-
ex p a n sio n s u ses a n o rth o g o n a l tra n sfo rm a tio n fro m ENDM
th e cen ter freq u en cy b to th e o rig in . T h e o rth o g o -
n a l tra n sfo rm a tio n 's a d va n ta g e is th a t th e p o ly n o - ;amplitude stabilized quadrature
m ia l a p p rox im a tio n erro rs a re n o t m a g n i¯ ed b y th e ; oscillator iteration (CST)
tra n sfo rm a tio n p ro cess. T h is a llow s fo r low er o rd er ;14 instructions -- 28 cycles
p o ly n o m ia l a p p rox im a tio n s a n d sm a ller m a g n itu d es ;424nSec @ 66MHz
fo r th e p o ly n o m ia l's co e± cien ts.
AQ_OSC MACRO reg
move L:(r\reg)+,x
3 D SP C ode mpy x0,x1,a L:(r\reg)-,y
mac -y0,y1,a
T h is sectio n sh ow s th ree u sefu l o scilla to r m a cro s, mpy x1,y0,b
w ritten in M o to ro la D S P 5 6 0 0 2 , d em o n stra tin g va r- mac x0,y1,b a,x0
io u s fo rm s o f th e a lg o rith m 's im p lem en ta tio n . T h e mpy -x0,x0,a b,y0
¯ rst is th e b a llistic fo rm o f th e o scilla to r. T h e sec- mac -y0,y0,a #0.75,b
o n d is a n a m p litu d e sta b ilized q u a d ra tu re o scilla to r, addr b,a
a n d th e th ird is a sm o o th ly tu n a b le sta b ilized o scil- move a,x1
la to r. E a ch u ses tw o lo n g m em o ry lo ca tio n s. O n en - mpy x1,x0,a
try a n d ex it r \ r e g p o in ts to th e ¯ rst o f th e m em ³ o ry´ asl a
lo ca tio n s. T h e ¯ rst lo n g a d d ress co n ta in s co s 2f¼sf mpy x1,y0,b a,y:(r\reg)+
in x sp a ce a n d h a s th e co sin e o u tp u tp in th e y sp a ce. asl b
T h e co sin e o u tp u t is in itia³lized´ to 22 . T h e seco n d move b,y:(r\reg)-
lo n g a d d ress co n ta in s sin 2f¼sf in x sp a ce a n d h a s ENDM
th e sin e o u tp u t in th e y sp a ce. T h e sin e o u tp u t is
in itia lized to 0 :0 . S in ce th e m a cro s m a k e ex ten siv e ;Continuously tunable,
u se o f th e p ro cesso r's p a ra llelism , th e lo n g m em o ry ; stabilized quadrature oscillator (CST)
d a ta sh o u ld b e p la ced in in tern a l D S P ra m to av o id ;Uses 2nd order Taylor's series.
w a it sta tes. ;23 instructions -- 46 cycles
T h e m a cro s m ay ea sily b e m o d i¯ ed to p rov id e ;697nSec @ 66MHz
sta cka b le o scilla to rs b y ch a n g in g th e a u to d ecrem en t
p a rt o f th e ea ch m a cro 's la st in stru ctio n to a n a u - AFQ_OSC MACRO reg
to in crem en t. T h e p o in ter w o u ld a lso n eed to b e move a,y0 #0.5,b ;a has delta freq.
co n ¯ g u red to b e m o d u lo 2 tim es th e n u m b er o f o s- asl b L:(r\reg)+,x
cilla to rs. L ik ew ise, th e m em o ry lo ca tio n s w ill n eed asl b
to b e a t th e p ro p er ty p e o f m o d u lo a d d ress. mac -y0,y0,b L:(r\reg),y
asr b a,y0
;ballistic quadrature mpy -y1,y0,a b,x0
; oscillator iteration (CST) mpy x1,y0,b
;6 instructions -- 12 cycles mac y1,x0,b L:(r\reg)-,y
;182 nSec @ 66MHz mac x1,x0,a L:(r\reg),x
move a,x1
Q_OSC MACRO reg mpy x0,x1,a b,y1
move L:(r\reg)+,x mac -y0,y1,a
mpy x0,x1,a L:(r\reg)-,y mpy x1,y0,b
mac -y0,y1,a mac x0,y1,b a,x0
mpy x1,y0,b mpy -x0,x0,a b,y0
mac x0,y1,b a,y:(r\reg)+ mac -y0,y0,a #0.75,b
addr b,a
move a,x1
mpy x1,x0,a
asl a
mpy x1,y0,b a,y:(r\reg)+
asl b
move b,y:(r\reg)-
ENDM

4 C o n c lu sio n
S ev era l o scilla to r fo rm s h av e b een d escrib ed a lo n g
w ith sa m p le co d e a n d a co n tin u o u sly va ria b le fre-
q u en cy, a m p litu d e sta b ilized q u a d ra tu re o scilla to r
is d ev elo p ed . W h ile th is la st fo rm seem s to b e a
b it co n v o lu ted , its ex ecu tio n is stra ig h t fo rw a rd . It
n eed s o n ly fo u r m em o ry lo ca tio n s| tw o fo r th e o u t-
p u ts a n d tw o fo r th e feed b a ck va lu es. T h ese fo u r
lo ca tio n s m ay b e co m b in ed in to tw o lo n g m em o ry
lo ca tio n s a s is d o n e in th e sa m p le m a cro s.
T h e q u a d ra tu re o scilla to r is n o t o n ly ex trem ely
° ex ib le in th a t it ¯ n d s n u m ero u s u ses in m o d u la to rs
a n d d em o d u la to rs, b u t th is im p lem en ta tio n a lso a l-
low s fo r ea sy co n tro l. A lso , T h e e± cien cy o f th e
o scilla to rs a llow s o n e to d esig n a ra d io w ith a g en -
era l p u rp o se D S P w ith a n IF in th e 1 0 0 k H z ra n g e.
A n A S IC co u ld ex ecu te th e a lg o rith m a t a m u ch
h ig h er ra te.