A very compact KHN filter with multidecade tuning
Abstract
A very compact implementation of a multifunction Kerwin Huelsman Newcomb (KHN) filter that can be frequency tuned almost seven decades, from 0.2 Hz to 1 Mhz, is presented. Tuning is achieved by means of high-value, programmable active resistors biased using the quasi floating gate (QFG) technique and linearisated through capacitive gate voltage averaging. The circuit, realized in a 0.5 μm standard CMOS technology using only four CMOS inverters, six small capacitors, six small resistors and two programmable active resistors, occupies a total area of 0.02 mm2, dissipates 3.45 mW and presents a dynamic range at 0.1 % THD of 55.86.
Full text
A Ve y Compac KHN Fil e wi h Mul idecade Tuning
Ca los Muñiz-Mon e o1, Alejand o Díaz-Sánchez1,2
Elec onics depa men s
1. Na ional Ins i u e o As ophysics Op ics and Elec onics
2. Poli echnical Ins i u e o Puebla.
Puebla, MEXICO
cmuniz, [email protected]
Ramón González-Ca ajal
Depa men o Elec onic Enginee ing
School o Enginee ing, Uni e si y o Se illa
Se illa, SPAIN
ca ajal@g e.esi.us.es
Abs ac —A e y compac implemen a ion o a mul i unc ion
Ke win Huelsman Newcomb (KHN) il e ha can be equency
uned almos se en decades, om 0.2 Hz o 1 Mhz, is p esen ed.
Tuning is achie ed by means o high- alue, p og ammable
ac i e esis o s biased using he quasi loa ing ga e (QFG)
echnique and linea isa ed h ough capaci i e ga e ol age
a e aging. The ci cui , ealized in a 0.5 µm s anda d CMOS
echnology using only ou CMOS in e e s, six small
capaci o s, six small esis o s and wo p og ammable ac i e
esis o s, occupies a o al a ea o 0.02 mm2, dissipa es 3.45 mW
and p esen s a dynamic ange a 0.1 % THD o 55.86.
I. INTRODUCTION
In low equency applica ions, such as analog p ocessing
o biomedical signals, analog il e s a e used o ejec
unwan ed signals, such ha la ge DC o se s gene a ed by
senso s and licke o ou band noise. Addi ionally, he
bandwid h o mos o he biomedical signals is in he ange o
0.1 Hz − 10 kHz and, because o he ypical equi emen s o
biomedical applica ions, p og ammabili y is manda o y [1].
Un o una ely, he design o p og ammable in eg a ed il e s
wi h bandwid hs below 1 kHz is no i ial, especially i o he
design speci ica ions such as low noise, low dis o ion, small
a ea and educed powe consump ion, mus be sa is ied. The
main challenge comes om he lack o la ge ime cons an s
a ailable in in eg a ed ci cui s. To o e come his p oblem,
imp ac ical echniques ha e been used in he li e a u e, among
hem: use o ex e nal capaci o s (addi ional ou pu pads),
capaci ance mul iplica ion (coa se p og ammabili y ha is
only possible wi h disc e e con ol) and echniques o cu en
di ision and cu en cancella ion (p one o p esen misma ch
and la ge o se componen s) [2-3]. A mo e success ul
al e na i e ha eaches un il i e decades o equency uning
is he companding app oach in log-domain p ocessing, whe e
he inpu is comp essed, hen nonlinea ly p ocessed, and
inally expanded a he ou pu , p ese ing dynamic ange (a
low supply ol ages) and he o e all linea i y [4]. This
app oach exploi s he exponen ial law o MOS ansis o s
biased in weak in e sion and equen ly equi es de use o
Mul iple Inpu T anslinea Elemen s (MITE) ealized wi h
loa ing ga e ansis o s (FGT) [4, 6].
The main p oblem o all he p e iously men ioned
app oaches is he huge die a ea equi ed, and he ac ha only
he log-domain il e s each mul idecade uning. This wo k
p oposes a e y compac and ex emely simple ealiza ion o a
KHN il e (sui able o biomedical applica ions) wi h
mul idecade uning ha does no equi e o log-domain
p ocessing. The p oposal is based on he use o e y la ge-
alued, p og ammable ac i e esis o s biased using he quasi
loa ing ga e (QFG) echnique and linea isa ed h ough
capaci i e ga e ol age a e aging. A cu en mode e sion o
his s a egy was ecen ly epo ed in [7]. The pape is
o ganized as ollows. Sec ion II in oduces he quasi ideal
in eg a o used in he KHN il e syn hesis o Sec ion III.
Simula ion esul s a e p esen ed in Sec ion IV. Finally, in
Sec ion V some conclusions a e d awn.
II. QUASI-IDEAL INTEGRATOR DESIGN
Le ’s conside he quasi-ideal in eg a o shown in Fig. 1a.
The CMOS in e e is conside ed o analysis pu poses as an
OPAMP o gain A wi h he non-in e ing inpu connec ed o
VDD/2. Rg is a p og ammable ac i e esis o con olled by he
ol age Vcp. The co esponding ans e unc ion is gi en by
∞→−≈
+
+
−== A
CsR
AA
CsR
V
V
sH
g
g
in
ou ,
1
11
1
1
)( (1)
The ci cui implemen a ions o he CMOS in e e and he
ac i e esis o Rg a e illus a ed in igu es 1b and 1c,
espec i ely. The in e e is ealized by using la ge ansis o s
o educe noise and misma ch con ibu ions. Ne e heless,
mo e elabo a ed ampli ie s can be employed o ob ain low
ol age and/o low powe e sions o his in eg a o . Rg is a
h ee ansis o e sion o he linea ised quasi- loa ing ga e
esis o p oposed in [8], bu wi h he ansis o s Mp2 biased in
weak, mode a e o s ong in e sion egion, always in iode
mode, by he ol age Vcp. Consequen ly, he e ec i e
esis ance can be uned om hund eds o kilo-ohms o he ange
Wo k suppo ed by CONACyT, Mexico, unde G an I-37470-A
1-4244-1342-7/07/$25.00 ©2007 IEEE 44
Figu e 1. (a) Quasi-ideal in eg a o ealized wi h a unable ac i e esis o .
(b) CMOS in e e ealiza ion. (c) QFG p og ammable esis o linea ised
wi h he capaci i e ga e ol age a e aging scheme p oposed in [8].
o Gigaohms, while a capaci o o a ew pico a ads is enough
o ob ain cu o equencies in he o de o he z. Th ee
ansis o s a e necessa y o a oid di ec pola iza ion o he
pa asi ic ac i e-subs a e diodes due o signal luc ua ions
be ween he e minals o Rg, inc easing he dynamic ange [9,
10]. In addi ion, he weak-in e sion ope a ion o ansis o s
Mp2 o ces Rg<<RB (whe e RB is he di usion esis ance), in
o de o a oid a pa asi ic esis i e di ide [11]. The subs a e
connec ion o he i s ansis o Mp2 (ma ked wi h a black
iangle in Figu e 1a) should be connec ed o a d ain ins ead o
a ga e o a oid unp edic able o se s due o subs a e leakage
cu en s [10].
In single-ended ci cui s wi h loa ing MOS esis o s,
linea isa ion can be pe o med by including in he ga e ol age
a componen co esponding o he a e age o he ol ages in
he e minals o he esis o [8]. This a e aging is achie ed by
he small capaci o s Ca. Consequen ly, he ga e ol age o
ansis o s Mp2 becomes (VA+VB)/2+Vcp.
III. KHN FILTER WITH MULTIDECADE TUNING
The Ke win Huelsman Newcomb (KHN) il e is a
opology wi h ex eme lexibili y, good pe o mance and low
sensi i i ies [12]. A ci cui app oxima ion o his il e , using
he quasi-ideal in eg a o desc ibed in he p e ious sec ion, is
illus a ed in Figu e 2. Wi h a s aigh o wa d analysis, he
high-pass, band-pass and low-pass esponses becomes
+++
++
+
=
2
4
6
5
6
2
4
6
2
2
3
6
12
)(
A
AR
R
R
R
s
AR
R
s
A
s
R
R
sH HP
ψψ
ψ
(2)
Figu e 2. Tunable KHN il e ealized wi h he quasi-ideal in eg a o o
Figu e 1a and he QFG p og ammable ac i e esis o wi h capaci i e ga e
ol age a e aging linea isa ion o Figu e 1c.
+++
++
+−
=
2
4
6
5
62
4
62
3
6
12
)(
A
AR
R
R
R
s
AR
R
s
A
s
R
R
sH BP
ψψ
ψ
ψ
(3)
+++
++
=
2
4
6
5
6
2
4
6
2
3
6
2
12
)(
A
AR
R
R
R
s
AR
R
s
R
R
sH LP
ψψ
ψ
(4)
CR g
1
=
ψ
(5)
while he na u al equency
ω
n and quali y ac o Q a e:
2
4
6
5
61
A
AR
R
R
R
n++=
ψω
(6)
AR
R
A
AR
R
R
R
Q2
1
4
6
2
4
6
5
6
+
++
= (7)
F om (6) and (7) is obse ed ha he na u al equency can
be uned modi ying he pa ame e
ψ
=1/(RGC) h ough he
ol age Vcp. Mo eo e , his uning does no a ec Q.
45
Figu e 3. Layou o he p og ammable KHN il e o Figu e 2 designed in a
0.5-µm CMOS echnology.
TABLE I. DESIGN DETAILS OF THE CIRCUIT OF FIGURE 2
Mn1, Mp1, Mp2 W/L (µm/µm) =
12/1.2, 71.1/1.2, 1.5/0.6
R3, R4, R5, R6 10 KΩ
C, Ca, CL 1 pF, 0.1 pF, 1 pF
VDD, Vcp, 3 V, Vcp ∈ (0 V, 1 V)
IV. RESULTS
The KHN il e o Figu e 2 was designed using BSIM3.1
models o a 0.5 µm CMOS AMI p ocess (VTHn=0.65 V,
|VTHp|=0.95 V). Figu e 3 shows he layou o his il e using
he design de ails summa ized in Table I. All he capaci o s
a e e y small, consequen ly, he o e all ci cui occupies and
a ea o only 0.02 mm2. The equency esponses o he low-
pass, band-pass and high-pass signals a e illus a ed in Figu e
4. Whi a sweep o Vcp om 0 o 1 V, he cu o and cen e
equencies ange goes om 0.2 Hz o 1 MHz, he e o e
ea u ing a uning ange o se en decades.
The simula ed To al Ha monic Dis o ions (THD) o he
low-pass and high-pass signals a e shown in Figu e 5 as a
unc ion o he ampli ude o he inpu signal and he uned
equency. The maximum le els o dis o ion appea wi h
signals o equency 0.1 Hz and ampli ude abo e 0.4 V.
Howe e , he THD emains in mos o he cases below 1 % (-
40 dB) wi h signals o ampli ude below o 0.3 V.
The inpu - e e ed noise spec al densi y o he low-pass
signal is illus a ed in Figu e 6. The co esponding dynamic
ange, o a 100 Hz-bandwid h ( uned equency) and a signal
o ampli ude 100 mV (co esponding o a THD o 0.1 %) and
equency 10 Hz is equal o 55.86 dB.
In Table II some o he cha ac e is ics o unable il e s
p oposed in he li e a u e a e summa ized and compa ed wi h
he il e p oposed in his wo k. As can be obse ed, he a ea
Figu e 4. Low-pass, band-pass and high-pass esponses o he KHN il e
wi h a sweep o Vcp om 0 V o 1 V in inc eases o 25 mV.
o he p oposed il e is only he 2 % o he a ea epo ed in
e e ences [2] and [5], and he unable ange is wo decades
bigge han he ange o he log-domain il e p oposed in [4].
Finally, an a e age powe consump ion o 3.45 mW was
ob ained using he echnique epo ed in [13]. Ne e heless,
mo e elabo a ed ampli ie s can be employed ins ead o he
in e e o ob ain a be e powe consump ion.
V. CONCLUSIONS
A new s a egy o design low- equency, unable in eg a ed
il e s has been p esen ed and e i ied by simula ion. This
echnique uses la ge- alued ac i e esis o s o o e come he
p incipal challenges o he p e iously epo ed s a egies in he
li e a u e, allowing an ex emely compac and easy ci cui
ealiza ion, wi h low ha dwa e complexi y, ully in eg a ed
capaci o s and mul idecade uning capabili y wi hou using he
log-domain p ocessing app oach.
The s a egy can be used in o he il e opologies and
scales wi h he CMOS p ocesses. Also, he powe
consump ion and ol age equi emen s can be easily educed
i mo e elabo a ed ampli ie s a e used ins ead o CMOS
in e e s.
46
Figu e 5. Simula ed THD o he low-pass and high-pass signals as a
unc ion o he ampli ude o he inpu signal and he uned equency. The
equency o he inpu is one decade below o abo e o he uned equency,
o he case o he low-pass and high-pass il e s, espec i ely.
Figu e 6. Inpu e e ed noise o he low-pass esponse uned a 100 Hz.
TABLE II. COMPARISON OF TUNABLE FILTERS IN THE LITERATURE
Re e ence
and
CMOS
p ocess
Tunable ange O de Die a ea
[2]
(1.2 µm) 2 Hz 6 1 mm2
[4]
(1.2 µm) (1 Hz – 100 kHz) 2 No epo ed
[5]
(0.8 µm) (50 kHz – 2.1 MHz) 3 1 mm2
[6]
(0.8 µm) (25 Hz – 35 kHz) 1 0.1 mm2 *
This wo k
(0.5 µm) (0.2 Hz – 1 MHz) 2 0.02 mm2
* Wi hou conside wo ex e nal capaci o s o 150 pF.
REFERENCES
[1] J. D. B onzino, The Biomedical Enginee ing Handbook, CRC P ess
Inc. and IEEE P ess, 1995.
[2] S. Solis-Bus os, J. Sil a-Ma ínez, F. Mallobe i and E. Sánchez-
Sinencio, “A 60 dB Dynamic Range Six h O de 2.4 Hz Lowpass Fil e
o Medical Applica ions,” IEEE T ansac ions on Ci cui s and Sys ems
Pa II, Vol. 47, No. 12, pp. 1391-1398, Decembe 2000.
[3] J. Sil a-Ma ínez and S. Solis-Bus os, “Designs conside a ions o high
pe o mance e y low equency il e s,” P oc. o he 50 h In e na ional
Symposium on Ci cui s and Sys ems, ISCAS 99, Vol II, pp. 648-651,
O lando, USA, June 1999.
[4] B. A. Minch, “Mul iple-Inpu T anslinea Elemen Log-Domain
Fil e s,” IEEE T ansac ions on Ci cui s and Sys ems—II: Analog and
Digi al Signal P ocessing, Vol. 48, No. 1, janua y 2001.
[5] J. A. De Lima and C. Dualibe, “A Linea ly Tunable Low-Vol age
CMOS T ansconduc o Wi h Imp o ed Common-Mode S abili y and
I s Applica ion o gm–C Fil e s” IEEE T ansac ions on Ci cui s and
Sys ems Pa II, Vol. 48, No. 7, pp. 649-660, July 2001.
[6] A. J. López-Ma ín, C. A. De La C uz-Blas, and A. Ca losena,’’1.2 V 5
-
µ
W Class-AB CMOS Log-Domain In eg a o Wi h Mul idecade
Tuning’’ IEEE T ansac ions on Ci cui s and Sys ems Pa II, Exp ess
B ie s, Vol. 52, No. 10, pp. 665-668, Oc obe 2005.
[7] C. Muñiz-Mon e o, R. González-Ca ajal, A. Díaz-Sánchez and J. M.
Rocha, “Low equency, cu en mode p og ammable KHN il e s
using la ge- alued ac i e esis o s,” In p in , 58 h In e na ional
Symposium on Ci cui s and Sys ems, ISCAS 2007.
[8] J. Rami ez-Angulo, M. S. Sawan , R. G. Ca ajal and A. López-Ma ín,
“Linea iza ion o MOS esis o s using capaci i e ga e a e aging,”
Elec onics Le e s, Vol. 41, No. 9, Ap il 2005.
[9] C. U quidi, J. Rami ez-Angulo, R. Gonzalez-Ca ajal and A. To alba,
“A new amily o low- ol age ci cui s based on quasi- loa ing ga e
ansis o s,” in P oc. MWSCAS, Tulsa, OK, Augus 2002.
[10] I. Seo and R. M Fox, “Compa ison o Quasi-/Pseudo-Floa ing Ga e
Techniques and Low-Vol age Applica ions, ” in Analog In eg a ed
Ci cui s and Signal P ocessing, V. 47, pp. 183-192. 2006.
[11] M. Bikumandla, J. Ramí ez-Angulo, C. U quidi, R. G. Ca ajal and A.
J. Lopez-Ma in. “Biasing CMOS ampli ie s using MOS ansis o s in
sub h eshold egion,” in IEICE Elec onics Exp ess, ol. 1 no. 12,
Sep embe 25, 2004.
[12] L.P. Huelsman and P.E. Allen, In oduc ion o he heo y and design o
ac i e il e s, McG aw-Hill Book Co. USA, 1980.
[13] S.M. Kang, “Accu a e Simula ion o Powe Dissipa ion in VLSI
Ci cui s,” IEEE Jou nal o Solid S a e Ci cui s, Vol. Sc-21, No. 5, pp.
889-891, Oc obe 1986.
47