• Nem Talált Eredményt

[lJ se

N/A
N/A
Protected

Academic year: 2022

Ossza meg "[lJ se"

Copied!
5
0
0

Teljes szövegt

(1)

A STRAY-INSENSITIVE SWITCHED-CAPACITOR DELAY CIRCUIT

Gy. SIMON

Institute for Telecommunication Electronics, Technical University, H-1521 Budapest

Received: December 15, 1985 Presented by Prof. S. Csibi

Summary

A circuit configuration is introduced for two-phase clocked delay and summing units with reduced sensitivity to stray capacitance spread. An example of a three-point running DFT processor circuit is given.

Introduction

Most of the known

se

delay circuits seem to be very complicated for practical purposes requiring either two operational amplifiers for one stage [lJ or a multi-phase clocking scheme [2]. Another two-phase clocked solution with one operational amplifier [3J suffers from being sensitive to top-plate parasitic capacitance tolerances. Starting from the latter configuration a modified structure is introduced. The main goal of the present contribution is to get a circuitry less sensitive to parasitic capacitance spread for applications in delay and weighted summing stages.

Improving the sensitivity properties

The first version of a two-phase SC subcircuit is illustrated in Fig. 1. Cl is the integrating capacitor. C2 is charged to a voltage equal to Vout in phase 2. For C 2 = Cl (i.e. the relative error value Lt is zero) the charge sample on Cl preceding the last one is compensated in the next phase 1 resulting in a zero output component by recharging C2. On the other hand the new charges qin(Z) entering the system in phase 1 and 2 are summed and disappear in the next cycle from the stage involved. The last sample should be saved or processed before the charge representing that sample is compensated. From a theoretical point of view the circuit in Fig. 1 seems to be a suitable unit delay element.

(2)

88 GY. SHWS

ICt'(1+t»C:~

1 I'

!

C,

~---4.--oVou" (z)

Fig. I. Basic two-phase delay circuit

However the relative top-plate parasitic capacitance spread of C 2 causes an error represented by the relative error parameter ,1. With proper dimensioning of the nominal capacitance values the sign of ,1 in a realized stage may be positive as well as negative.

1

vaul(z)= Cqin(Z)·Z-l. (1)

1

As qin(Z) enters the st,!:ge by coupling capacitors, the properties of the stage depend on ratios of capacitances. However, for a finite value of ,1

Vaul1 (z)· (1-,1 . Z-l) =

Cr

1 Qin(Z)z-l

(2)

i.e. the intended finite memory delay element changes into an infinite impulse response unit. As the top-plate parasitic capacitance has a wide spreading range the structure of Fig. 1 is not good for practical applications. If the value of

,1 for the different stages on a given chip has a wide spread the method of determining the pole and zero shifts in the z-plane for a given constant value of

,1 [4J is not feasible. On the other hand, even if the value of ,1 on a given chip can be considered essentially constant (i.e. correlated tolerances) the method of [4J can be applied for analysis purposes only, because the value of ,1 for a chosen chip sample is stochastic in nature. As the top-plate capacitance spread (relative to the "main" capacitance level) is determined by the technology applied we try to find a solution to reduce the effect caused by it. Reducing the coefficient of the error term (corresponding toLl in Eq. (2)) results in a response closer to the parasitic-free ideal case.

Let the structure be extended according to Fig. 2. In this case the new information sample (charge) should enter the stage at the beginning of phase 1.

The error is represented by the relative error parameter of C 3' ,1. The top- plate capacitance spread of C 3 is supposed to be the dominant source of error.

Equation (2) should be modified yielding

( ,1

-1)

1

-1

VauI2(Z)· 1 - ; z = Cl Qin(Z)Z . (3)

(3)

SWITCHED-CAPACITOR DELA Y CIRCUIT 89

Increasing the value of a results in decreasing error coefficients in the left-hand side of Equation (3). Supposing that the (nominally equal) capacitors C

d':l.

and C 3 are chosen to be unit capacitors the total capacitance involved is (1

+

':I.); e.g.

for ':I.

=

10 the summed value is 11. If eland C 2 in Fig. 1 are of unit values the

total capacitance is 2. The circuitry presented in Fig. 2 compared to that of Fig.

2

~

C, /""

Fig. 2. Improved delay circuit schematic

is more complex (four switching transistors and one capacitor should be added) and the summed capacitor value is higher as well. The price paid is, however, converted into the reduced influence of the top-plate capacitance value spread. The relative stray capacitance uncertainty error Lt and the application dependent allowable error coefficient Lt/a are to be considered for choosing the value of ':I..

An example

The charge input of each delay unit of a circuit may be used as a weighted inverting and/or non-inverting summing point. Recursive and non-recursive circuits may also be implemented. Figure 3 shows a unit delay or summing element with IX = 10. As a system illustration a three-point running DFT processor is given in Fig. 4. The unit elements are as in Fig. 3 and are represented here by boxes for simplicity. The position of the switches in Fig. 4 corresponds to phase 2.

(4)

90 GY. SIMOS

t----III-I - - - - . .

IN OUT

UE

Fig. 3. Unit element (:1= 10)

3.333

Vo (the DC comDCnenl)

3~h2it ~.

out

: S'333

i ~ ~

11~3

______________

~

J I

First

V cl 10 ~delay

on ~ I stage

1 ./ U.E

out

I )

j ! i

Second in delay

stage U. E.

I '

I i 3.333

lnn

out ~

11.6s7jb ~

;.---+-I----1.{~n U. E.

I

~h~in

2.887

fC:F

out

L...--_ _ _ _ _ _ ~ U. E.

2.887

f'c-t>

I

Fig. 4. Three-point DFT processor circuit

(5)

SWITCHED-CAPACITOR DELA r CIRCUIT 91

References

1. MCCHARLES, R. H.-HODGES, D. A.: Charge circuits for analog LSI. IEEE Tr. Circuits and Systems, CAS-25, July 1978, pp. 490-497.

2. REDDY. N. S.-SWAMY, M. N. S.: Switched-capacitor realization of a discrete Fourier transformer. IEEE Tr. Circuits and Systems, CAS-30, April 1983, pp. 254-255.

3. POSPISIL, J.-DOSTAL, T.-Moos, P.: Multiple-input switched blocks and their utilization in SC network synthesis. Proc. of the Fifth Int. Symp. on Network Theory, Sarajevo, Sept. 1984.

pp. 260--265.

4. GERSHO, A.: Charge-transfer filtering. Proc. of the IEEE, 67/2, Febr. 1979, pp. 196--218.

Or. Gyula SIMON H-1521 Budapest

Hivatkozások

KAPCSOLÓDÓ DOKUMENTUMOK

The application of nullators and norators in network models is justified by the possibility they offer to reduce the analysis oflinear networks containing coupled two-poles

In the wirebound telecommunication the transmission lines have a frequency-dependent characteristic impedance and if the reflection coefficient is prescribed both for

To compute the GCD of two polynomial matrices from their generalized Sylvester matrix, a complete theory with an easy proof is offered. In light of this proof a systematic

The present paper gives a short summary of the results obtained during the last two years in the Department of Solid State Chemistry of the Research Laboratory for Inorganic

Only Wilson, Evans, Jowett and Levenson, following Capell (1768), indicate in stage direction to whom Romeo is... Williams and Gibbons remain faithful to the copy-text (Q2),

Sodium sulphate is one of the cheapest materials and by addition of other cheap materials e.g. NH 4 CI or NaCl, eutectics with different melting points can be obtained. Phase

Between the nodes and the inputs of the evevss there are veess with the transfer function Gjjl(z) where i indicates the evevs to which the output of the vees is

It is well-known that the value of r is the most dominant factor, which determines, in most cases, how many memory units are necessary for the realisation of