Design of an Advanced Signal Conditioning Unit for

Design of an Advanced Signal Conditioning Unit for Sensor with Reduced off-the-Shelf Components Arun Kumar Sinha§, Daniele D. Caviglia§ and Pankaj Gaur-

§ Department of Biophysical and Electronic Engineering, University of Genova, Genova, 16145, Italy Atrenta India Pvt. Ltd., N oida R&D center, Noida, 201301, India § Emails: (arun.kumar.sinha; daniele.caviglia)@unige.it. [email protected] Abstract-In this paper we presents a signal conditioning unit working at single supply. The unit consists of high pass filter, buffer and a cascaded stage of low pass filter with amplifier. The order of high pass filter and low pass filter was first and second respectively. The unit was realized by using minimum off-the-shelf components through effective sizing rules proposed in this paper. Software

simulation

was

done

to

ascertain

the

working

of

conditioning unit according to thc specifications. A hardware prototype has been made to demonstrate the working of the unit.

Keywords- Amplifiers; High-pass filters; Low-pass filters; Operational amplifiers

I.

INTRODUCTION

There has been lot of developments in the field of sensors focused on their applications. The main challenge in this regard is their interfacing and in data processing. In the interface circuitry the main role is played by the signal conditioning unit, in optimizing the sensed signal for better processing at higher level [ I]. In recent years proposals have been made for single supply from conditioning electronics to data processing [2]. Such an arrangement can reduce the size of Printed Circuit Board (PCB), by reducing the number of off-the-shelf compon­ ents. It can also solve the problem of generating different voltage source for the system. Barboni et al. [3] have proposed such system for the conditioning of dynamic signal from Piezoelectric Oxide Semi-conductor Field Effect Transistor (PO SFET) sensors. Though such method was efficient, there was further scope for improvements by reducing the number of components. Based on such an idea one of the improvement, can be proposed in the reference generation at virtual ground. In generating the Voo/2 for single supply circuits for level shifting [4]. Extra set of op-amps and resistors were required, which can make the PCB size bulkier even if readout is smart [3]. The next improvement can be done in the filters itself; because of the fact that lot of approximations for sizing the resistors and capacitors, based on tables can vary the response of the filters [5]. The authors agree that for higher order filters, tabular methods are easy and worth appealing. But for the lower order system (i.e., first, second and third), an indigenous method for sizing the passive components can be developed. In this paper we propose a signal conditioning unit, with reduced off-the-shelf components for single supply system. The unit consists of a first order high pass filter (HPF), a buffer, a second order low pass filter (LPF) and an amplifier. The novelty of this work consists in minimizing the circuits, using the proposed sizing rules for filters and amplifier. This method

has potential to reduce the size of hardware, and can offer the precise control of the cut-off frequency. The layout of this paper is as follows: in Section II we will briefly discuss the specification, for the signal conditioning system. Section III will deal with the design of signal conditioning system. In Section IV we first presented the results in form of simulation, and then by implementing hardware prototype, made from the off-the-shelf components. Finally in Section V, the conclusion of this paper is drawn in brief. II.

SYSTEM SPECIFICATIONS OF ANALOG STAGE

The reference system consists in a sensor array of m x n POSFET cells, a corresponding number of signal channels, each with a front-end circuit able to bias and read the sensor output. In particular the specifications of each analog front end circuit can be summarized as follows: •

Single supply: O-Voo (=3.3 V).

•

Peak-to-Peak maximum input voltage: Vpp =1 V.

•

Amplifier gain: Av =VoolVppo

•

HPF order and cut-off: 1st and 2 Hz respectively.

LPF order and cut-off: 2nd and I kHz respectively. As reported in [6] for the dynamic (i.e., ac) signal of peak amplitude from the POSFET sensor is ± 0.5 V. The gain of the amplifier should be fixed to VoolVpp, so that the signal after amplification lies within the supply rails as shown in Fig. 1. •

I-'---+--.,.VDD/2

-------------------------------------------------

0

Figure 1. Amplification of dynamic signal within rail to rail supply.

The sensor was meant for tactile sensing, therefore the band of sensing frequencies was between 2 - Ik Hz [7]. To avoid aliasing during signal sampling, the order of LPF should be higher than HPF. This is an open point as we assumed in this paper that the slope of a 2nd order LPF, will be sufficient in achieving the goal of antialiasing. This will involve a smaller

number of passive components; on the other hand SNR can be improved by oversampling technique at digital stage [3]. On the basis of the above specifications, we propose the arrange­ ment of the blocks in the front-end analog stage as shown in Fig. 2.

V5(S) V4(s)

=Av =1 +

RF Ra+(1/ sCa)

1 + Av1'

(3)

The Avi can be written in term of magnitude and phase as, 1 AVI =I ffi RFCa / I + ffi 2 R �C � I L900- tan- ffi RaCa .

(4)

Based on (4) the ratio of resistance can be specified, in term of capacitance Ca and cut-off ffiL as by, Figure 2. Blocks in the proposed analog stage for signal conditioning and overall interface system.

In the first block the coupling capacitor Cp, removes the dc bias voltage from the output of POSFET sensor, and sets the ac signal on V0012 level through the resistive network. This HPF block also set the lower cut-off to 2 Hz. The unity gain buffer stage separates the two filter blocks, thereby keeping the filter calculation simple. The cascaded block consists of a second order LPF and an op-amp based amplifier. This stage provides a -40 dB/dec roll-off and dynamic signal amplification. III.

If we equate the expression CaffiL Ra in (5) to I, then Ra can be expressed in terms of Ca and ffiL by the new resistance value, R� =I/CCa x ffid.

(6)

The feedback resistance RF can be expressed in term of the new value as, (7) RF =..fi x R� x Av1' The value of AVI in (7) should be equal to RF/Ra; this can be achieved after dividing the value of R� by --J2,

)

(

(8)

AVI =R d Ra e=R�/ ..fi) .

DESIGN OF FILTERS AND AMPLIFIER IN ANALOG STAGE

In this section we will express each block in analog stage by Laplace transfer function. The HPF and buffer blocks have not been detailed as they are widely understood. The overall transfer function of HPF can be given by,

In (8), AVI can be substitute by (3), in order to get the value of RF directly in term of gain Av as, (9) RF =Ra xC Av - O.

The condition ffi =ffiL will relate the lower cut-off frequency ffiL, with the values of Rp and Cp as by,

Fig. 4 shows the circuit configuration of the proposed amplifier cascaded with the 2nd order LPF. The 2nd order filter is based on the Sailen-Key topology. The sizing of passive components was done according to the proposed rules, to have the Tschebyscheff filter response. The 2nd order transfer function was derived in term of gain of the non-inverting amplifier. Then by using (3) we derived the 3rd order overall transfer function of the cascade circuit.

(2) Rp x Cp =2/ffiL ' Before ffiL the filter will have a +20 dB/dec roll-off. The buffer has unity gain transfer function given by IV3Uffi )IVzUffi )1 =1. We will now focus on the proposed amplifier block as shown in Fig. 3, used in this signal conditioning unit.

.... ,..- - ---- ------- ---- - -- -

Vis)

Eq. (6) and (9) will give the value of resistance Ra and RF, which should be present in the feedback loop, to give the chosen value of gain (Av).

IIsC,

--1

--

---- --

-

--

I �T_R'� ! lIsC L----'

+

C

C1

-- ---

R

V,(s) �----

!

Iv/s I I

Avi

!

--�-----------------------

!

L-

This proposed amplifier configuration should provide gain (Av) to the input ac signal without adding an offset. The signal entering in amplifier block will have a dc value of V00/2 plus an ac signal. Therefore this amplifier block should amplifY only the dynamic content. The addition of an offset can shift the ac signal towards the supply rail, thereby clipping the signal content. Therefore we proposed the overall transfer function of the amplifier as given by,

R

I -�-f

I C

Figure 3. Non-inverting amplifier with capacitor in the feedback loop.

-----------------

•

RF

-

,

! I I I

., ! ----�---------------j a

A

IIsC

i

I

Figure 4. Cascade circuit consisting of a second order LPF with an op-amp based amplifier.

The 2nd order transfer function of the configuration shown in Fig. 4 can be given by, V5(S) Av/R2 R1C1C2 (10) = I V3(S) 2 (1-Av)R1Cl+R2 C2 +R1C2 s+ S+ R2 R,C,C2 R2 R1C,C2

(

)

On assuming the coefficients "a" and "b" as by, a =( I-Av)R,C,+R2C2+R,C2.

•

Calculate the value of Ra (=R�/ .,fi) where R� is given by (6), and RF is given by (9).

•

After substituting m x n =I in (18), the value of n will be, (23) n =Av + 2s - 2.

•

The values of RI will be nR2 and CI will be ( I/n) x C2•

(11)

b =I/R2R,C,C2 .

(12)

Eq. (10) will reduce to, Vs(s)

Avxb V3(s) s2+abs+b '

(13)

The second order transfer function given by (13), can be compared with the standard second order equation (14) [8].

Km�

Xo(s)

(14)

s2+2smns+m�'

Xj(s)

where K is the gain, mn is the natural frequency and zeta (S) is the damping factor. Eq. (14) has a pair of complex conjugate poles PI and P2, therefore for stable filter its zeta must be in range 0 < s:S 1. Substituting s =jm in (14) we will get,

XoUm) K L- tan" = XjUm) 1-;===1==

( (:i)r e��/ 1-

+

(��mn ) UJ

( (:i))

Resonance condition will exist at vicinity of frequency m = mn, and the magnitude will be M2 = Av / 2s . The maximum peak will be Mm =Av/(2s--J(l-S )), at frequency m =mm (= 2 mn--J(1_2S )). For the peak S should be less than 0.7071. Coefficients a and b can be expressed in term of S and mn as, •

a = 2S!mn,

(16)

b =m�.

(17)

If RI =nR2 and CI =mC2 then from (11) and (16) we will get,

((

(

l/ vlmxn )x 2s- ((n+ l)Nmxn )

))

,

(18)

and from (12) and (17) we will get, R2xC2 = 1/ (mnvlmxn ) .

•

al =(R,+R2)C2 - R,Clao'

(19)

For the maximum peak value Mm =VoolVpp and with the given value of S « 0.7071), the value of Av can be given by, (20)

(25) (26)

)

(

(27)

b, =boC2/ C2 - aa(R,CI/(RI+R2)) . IV.

RESULTS AND DISCUSSION

In this section we will present the frequency and transient response of the signal conditioning unit (analog stage). The simulation of the circuit was done on Orcad® Capture using Spice off-the-shelf components. We made a hardware proto­ type of the unit to demonstrate the transient response. The cut­ off frequency for HPF was mL =12.56 rad/sec; for LPF was mH =6280 rad/sec. The op-amp was TLC2272CP (0 - 5 V), Ca = Cp =1 IlF; Ra =56.3 ill, Rp =160 kQ, RL =1.5 MQ and V00 =3.3 V. Table I shows values of the parameters and compon­ ents used in simulating the circuit configuration. TABLE I

COMPONENTS VALVES FOR DIFFERENT lETA C, R, R2 (kil) (kil) (nF)

C, (JlF)

1.22

I

180

0.219

80

1.85

1

109

0.202

104

77

0.183

122

�

Av

0.4

2.42

4565.56

1.22 0.82

0.5

2.85

4937

1.85 0.54

5469.2

2.37 0.42

2.37

0.6 3.168

In (18) and (19), m and n are constants, to be determined. The second order filter can be tuned based on the following rules:

b(l+ao)x (s+(bo/(l+aa)) ) = 3 s +(bo+alb)s2+b(l+albl)s+bob '

In (25), coefficients al and bl are given by,

1-

When m/mn » I then IXoUm)/XiUm)ldB =- 4010glO(m/mn) (roll-off is -40 dB/dec).

Av =1-

Vs(s) V3(S)

. (15)

On comparing (13) and (15), K = Av, mn = --Jb and S = a--Jb/2. From (15) following interpretation can be made: •

As given by (3), the Av can be rearranged as, bo+ s(ao+l) (24) Av = ' b 0+ s where a a =RFlRa and b o =I/RaCa. Substituting the value of Av given by (24) in (13), the 3rd order overall transfer function of the cascade circuit can be given by,

Ill.

(rad/sec)

n

m

I

R. (kil)

The components in Table I were sized for the simulation purpose. For three values of zeta, Fig. 5 shows the frequency response of the signal conditioning unit at the output of cascade circuit.

25

[

---

2

-iij 15 (!) c

•

Substitute the particular value of S in equation given by,

05

(21) where mH is the higher cut-off frequency. •

By substituting the value of m x n = I in (19), determine the value of R2 and C2 as given by relation, (22)

?o-'

10'

10'

freq (Hz)

10'

Figure 5. Frequency response of the unit,on Y-axis is the linear scale of gain instead of dB,in order to show the gain is within supply rails.

For the same values of zeta, Fig. 6 shows the transient resp­ onse at the output of the unit.

shows transient response as given on CRO. The response shows a smooth output waveform with the dc level at V 00/2. � - ,-,

1 4 1----+-\"'·1+- ---+\"1 +----+-- \'''I-'---t- \''I--:f-1

30

001

002

003

004

005

time (sec)

006

007

008

009

Figure 6. Transient response for the input sine wave of amplitude 0.2 V (p-p), freq = 50 Hz and offset = 0.5 V.

TABLE II OBTAINED VALUE OF THE SPECIFICATIONS

0.4 0.5 0.6

Mm (VN) 3.28 3.23 3.25

Av (VN) 2.41 2.81 3.12

OlH

(rad/sec) 6220 6314 6233

,..

�

! t.1I! I�

I�

I�

OlL

(rad/sec) 12 12 12

In Table II we can observe the excellent matching with specif­ ications of signal conditioning unit, achieved by our design approach. As shown in Fig. 7 the prototype of the hardware with off-the-shelf components was made, in order to test our design through physical components. The value of components for zeta 0.5 was chosen in this prototype, because of the tradeoff between gain and damping. [f the value of zeta is large then the system response will be sluggish. If zeta is small then we need to decrease the value of gain according to ( 20). =

�

Ill! �

;;,d

=

Table II shows the achieved values of specifications,

�

I .I!.i

01

riI

Ii �

p;jI� I

,..4

Figure 8. Transient response for input sine wave,amplitude = 0.2 (p-p) V, freq = 50 Hz and offset = 0.5 V; note the average value of 1.64 V.

It can be concluded from this section that our approach, for making a single supply signal conditioning unit, with reduced number of off-the-shelf components was very effective. Authors verified the stability of the cascaded stage from (25), using Bode plot and the circuit was indeed stable. V.

CONCLUSION

We presented in this paper a single supply signal conditioning unit for POSFET sensors. The innovation of the work was enhanced by the approach given by the authors, for sizing the components present in the unit. Software simulation and hardware prototype proved the working of this design based on the given approach. ACKNOWLEDGMENT

=

39 nF

..---" •. -220 kfl

=W+--- RI.'=

Figure 7. Prototype of the hardware (dimension: 3.5 cm

This work was performed in Microelectronics laboratory at D1BE-UNIGE. Authors would like to thanks Giorgio Carlini technical staff of DIBE. Authors would also like to thanks Luigi Pinna (research fellow) for his valuable suggestions. REFERENCES

1.5 Mfl

x

[I]

B. J. Hosticka, "Analog circuits for sensors", in Proc. of the 37th European Solid-State Device Research Conf, Munich, Germany, 2007, vol. 7,pp. 97-102.

[2]

L. Barboni, R.S. Dahiya, G. Metta and M. Valle, "Interface electronics design for POSFET devices based tactile sensing systems", in Proc. of 19th IEEE Int. Symp. in Robot and Human Interactive Comm., Viareggio, Italy,2010,vol. 10,pp. 686-690.

[3]

Presented - L. Barboni, M. Valle and G. Carlini, "Smart readout design for tactile sensing devices",in Proc. of 18th IEEE Int. Conf on Electron., Circuits and Systems, 2011,Beirut,Lebanon.

[4]

A Single Supply Op-Amp Circuit Collection, Application report, [Online]. Available: http://www.ti.comJIitlanlsloa058/sloa058.pdf

3.5 cm).

The values of the resistance and capacitance for this proto­ type, were adjusted according to the availability. Therefore accurate matching with specifications cannot be expected, because our goal was to verify the transient response of this prototype. From the transient response we recorded the gains at different frequencies after observing on Cathode Ray Oscillo­ scope (CRO). The value of gains at different frequencies is shown in Table III.

[5]

S. Franco. (2001, Aug. 8). Design with Operational Amplifiers and Analog Integrated Circuits. McGraw-Hili Higher Education,pp. 160-205.

[6]

R.S. Dahiya, G. Metta, M. Valle, A. Adami and L. Lorenzelli, "Piezoelectric oxide semiconductor field effect transistor touch sensing devices ", Appl. Phys. Lett., vol. 95, pp. 034 105-1--D34 105-3,2009.

[7]

R. S. Dahiya, D. Cattin, A. Adami, C. Collini, L. Barboni, M. Valle, L. Lorenzelli, R. Oboe, G. Metta and F. Brunetti. (2011, Dec.). Towards tactile sensing system on chip for robotic applications,IEEE Sensors, vol. II, no. 12, pp. 3216-3226.

TABLE III

[8]

J. J. D'Azzo and C. H. Houpis. (2003, Aug. 14).

Linear Control System

analysis and Design with Mat/ab. Marcel Dekker Inc.