improved analog circuits using … the toolbox are ready, ... our hspice coding script, we set out...

International Journal of Engineering Technology and Scientific Innovation

ISSN: 2456-1851

Volume:03, Issue:01 "January-February 2018"

www.ijetsi.org Copyright © IJETSI 2018, All right reserved 9

IMPROVED ANALOG CIRCUITS USING OPTIMIZATION

ALGORITHMS AND HSPICE

Reyhaneh Pedram, Mohammadreza Amini

Department of Electrical Engineering, College of Engineering,

Borujerd Branch, Islamic Azad University, Borujerd, Iran

ABSTRACT

In this essay, we first review the definition of a rational framework for optimization and its

methods. The expression of the sample problem, the PSO algorithm and its application n

electronics is one of the issues discussed here. Therefore, in this essay we will see the

improvement of the output parameters of an electronic circuit by applying the proposed method.

Evolution is an appropriate optimization process. Scientists use PSO mechanisms to create

powerful and effective optimization tools for PSO Algorithm. In this essay, after selecting the

topology of the circuit by the designer, the values of circuit parameters including the transistor

size (length and width) of the bias capacitance and resistive circuit resistance of the circuit are

proposed by the optimization algorithm and the requirements of the problem, including the gain

values, the circuit power lost circuit, are optimized. The use of PSO and Beehive ABC

algorithms is a new method for designing these circuits. This simulation is done using MATLAB

and HSPICE software at the same time. And these two algorithms are compared with other

methods performed on a circuit and finally, using evolutionary algorithms, the values are

calculated for the parameters of a circuit to reach important metrics such as gain, bandwidth,

power consumption, etc.

Keywords: improved analog circuits, optimization algorithms, HSPICE.

1. INTRODUCTION

Electronics integrated circuits have

introduced analogue and digital System

together with the introduction of centralized

systems on a chip. Although the analogue

part contains a small percentage of the entire

circuit, it is considered to be a vital part of

the integrated circuits. The design of

analogue part’s circuits is more complex

than digital parts, and in addition to high

cost and time-consuming design, it requires

a high level of design expertise. Hence,

optimal design of analogue integrated

circuits is called the bottleneck of

electronics circuit design [1-3].


ISSN: 2456-1851


www.ijetsi.org Copyright © IJETSI 2018, All right reserved 0

The design of analogue circuits does not just

include the design of the original design

topology, but also the size and setting of the

elements. Therefore, it is a complex and

challenging process[4, 5]. The method of

sizing elements is often a slow, frustrating,

and repetitive process that depends on the

designer’s knowledge, freshness and

experience. For this reason, different

methods have been proposed for the

automatic design of analogue integrated

circuits by researchers. The automatic

design methods of the analogue integrated

circuits have two important aspects: from

the perspective of the first aspect, using

these methods is no longer required in the

manual, slow, and boring way of the past,

and can easily be used in the design of large

scale analogue circuits. Also, high accuracy,

ease of use, genericity flexibility and design

time are among the other benefits. But in

terms of the second aspect, the advancement

in the use of automatic design methods of

analogue circuits and the CAD tools in this

field is far behind digital circuits, and very

few are commercialized. Therefore, the

development of new CAD tools to increase

design efficiency, reduce the time to market

and control the cost of designing integrated

circuits analogue and mixed signal is

essential [6-9].

New tools for automatic design of integrated

circuits analogue are comprised of two basic

components: the synthesis and optimization

section. In the synthesis section, response or

responses are introduced as output, but

usually the answer provided is not the best

answer possible, with some changes there

may be better results. Hence, in the

optimization section, the responses obtained

from the previous section to each other will

approach the best possible response[10-12].

Today, researchers use different

optimization methods in the tools of

automatic design of integrated circuits

analogue. The new approach of researchers

in recent years is using of innovative

algorithms, especially collective intelligence

algorithms. The most prominent feature of

the collective intelligence algorithms is their

social behaviors. These algorithms consist of

members that display simple behaviors, but

the collective effect of these behaviors is

such that they result in collective

intelligence. Collective intelligence

algorithms team members collaborate to

achieve the ultimate goal. Of course, this

method is more effective than when

members act individually. Recently,

collective intelligence optimization

techniques have been increasingly

welcomed due to their high performance

functionalities and numerous pristine fields

in circuit design applications. One of the

newest collective intelligence algorithms is

the GSA Gravitation Search Algorithm[13,

14].

This algorithm is devoted to the concepts of

mass and gravity and the simulation of the

relevant rules. But since the design of the

analogue is a multi-objective problem, and

the various indexes of the circuitry , such as

gain, bandwidth, phase margin, and power


ISSN: 2456-1851


www.ijetsi.org Copyright © IJETSI 2018, All right reserved Page 21

consumption, cannot easily be answered, the

best answer[15-17].

Hence, there must be compromises between

the existing solutions that this reconciliation

is possible with the help of several goal of

making the optimization algorithm. In this

essay, a new optimization tool is proposed

for the automatic design of integrated

circuits analogue based on the new

algorithm PSO and ABC. This tool has

extensive search capabilities within

predetermined interval for each design

criterion. The mentioned tool simulates the

circuit in the beginning by using the

HSPICE simulator software. Then the

simulation results are optimized by

optimization algorithms PSO and ABC, and

the optimal results of this process with

continue to be obtained.

Optimization-based methods for

synthesizing automatic design use an

optimization engine instead of a design

topology. The optimization process is a

repeatable process that updates every design

variable.

This process continues until they reach a

point of equilibrium. In terms of

performance evaluation, the evaluation

engine is usually implemented using

equations based optimization techniques,

simulation-based optimization or behavior-

based optimization.

Equation-based methods use analytic design

equations to evaluate circuit performance.

These equations are created manually or

automatically by symbolic analysis tools.

Then the problem is formulated as an

optimization problem and is usually solved

using the numerical algorithm method[18-

21].

In general, these categories of methods in

determining the optimal performance of the

circuit, are used as an analytical tool in the

internal loop of the optimization process.

2. RESEARCH METHODOLOGY

We analyze two amplifiers with two beehive

and particle mass algorithms using software

MATLAB and HISPICE and compare the

results. The problem of optimizing the

differential amplifier circuit is solved by

considering single-objective and multi-

objective methods. The solution method in

both cases is the definitive method

Geometric Programming. The characteristics

that are considered as target functions in this

optimization are power, UGBW, DC Gain.

The unit gain bandwidth function (UGBW)

and (DC Gain) dc are the optimal

performance of the circuit for the high value

of these functions. Therefore, minimizing

target power functions, as well as

maximizing target bandwidth functions and

gaining profit, are the main objective of

optimization.

In the meantime, the problem has also been

solved in the multi-purpose situation. In this

sense, instead of considering only one

objective function, weighted composition of

each of the four defined objective functions


ISSN: 2456-1851



is considered at the main evaluator function.

The weight composition considered in this

essay is a function of the summed functions

and also the product of functions.

Obviously, the goal is to minimize the main

evaluator’s function. It is necessary to

reverse the function of the bandwidth of the

unit and dc gain, which is intended to

maximize them. Therefore, the main

objective function in the total weighting is

according to formula (1). Normally, values

must be set to normal values. The main

objective function is defined multiplicatively

be four objective functions defined

according to formula (2).

In the partied, due to the fact that the tools

of the toolbox are ready, there is no

possibility to change them in general and

that they are not applicable to all circuits,

the method is difficult and the speed of

response to some is very slow. First, using

our HSPICE coding script, we set out list to

simulate our circuit and get the amount of

open circuit gain we get. For this purpose,

for each node in our net list file, we select a

number that uses these numbers (common

nodes have the same number because they

all have a given voltage and current) to

determine the voltage and current of each

cut.

For example, for MOSFETS, with these

values we will be able to calculate the

MOSFET work area and, consequently, the

amount of gain for each MOSFET. We need

to combine all of the MOSFETS in the

saturation bias region here, and because we

want to use our circuit as the analogue

circuit.

The L and W values are in the gain and all

other characteristics of the transistor. For

example, they are involved in noise or the

amount of slew rate or velocity, but

according to our application, we use this

channel length to provide some conditions,

and we must know that we are always able

to meet multiple demands and can not afford

all the benefits or properties to hare together.

For example, in a circuit we want to be

faster and less noise at the same time.

Whatever our demands are, the ration of our

productivity will be reduced to a relatively

small proportion, that is, when we do some

optimization, we are surely at the cost of

losing some of our other demands.

In fact, in the social and natural

environment, people and natural creatures

do not enjoy all the benefits at the same

time, and they will always lose cost for an

advantage. It may be optimized in the

direction of gain in the circuit, but at the

same time there is an opposite course of

motion with the slope of the output. But we

must always make sure that the value of the

demands that matter most to use is first and

foremost to handle the process of forming

the circuit for our important purposes. We

must say that in this essay, the values of L,

W and the input current of the source and

reference voltage, we will be able to easily

and quickly achieve our coding, using the

two important applications of MATLAB and


ISSN: 2456-1851



HSPICE, and we will be able to when the

device itself is produced, it’s easy for the

user to decide on the length and width of the

channel, and even use the power to draw the

layout for the design of the integrated

circuits. Also, we can even use this

technique for digital circuits, and the speed

with which digital circuits can be connected,

which will result in the speed of our output

segment.

3. RESULTS AND DISCUSSIONS

The findings have all come to mean small

dimensions of the elements, less and

acceptable time to carry out the analysis

process and, of course, lower production

costs. As far as ABC is concerned, in the

result of the AD8531 sound amplifier to

Gain we get about twice as much in the PSO

method. Or in the results obtained from the

two-stage diffential amplifier circuit, the

power consumption in the PSO method is

very valuable and abnormally high with the

ABC method. Also, by comparing the

optimization results in this circuits, the past

analyzes of GA and DE (Barros et al, 2010).

The superiority of the PSO method in all

target functions is observed. It is noteworthy

that the convergence graph (the process of

PSO algorithm in 80 cycles and the ABC

algorithm in 20 cycles) represents all of the

above items.

Table 1: The results of the ABC algorithm and the PSO algorithm

in the amplifier circuit AD8531

Variable PSO ABC

W1=W2 (um) 1.5137 24.9

W3=W4(um) 2.2706 33

W5(um) 2.6490 33

W6(um) 3.0274 16.6

W7=W10 (um) 1.1353 11.6

W8(um) 1.8921 16

W9(um) 2.6490 26

W11(um) 3.7843 16

W12(um) 3.0274 17

W13(um) 4.1627 24.3


ISSN: 2456-1851



W14(um) 4.5411 13.66

W15(um) 5.2980 21.32

W16(um) 0.3784 20.23

L1=L2 (um) 2.2706 6.4

L3=L4(um) 3.4059 8.5

L5(um) 3.0274 8.5

L6(um) 5.6764 6.7

L7=L10(um) 2.2706 17.9

L8(um) 2.6490 20.1

L9(um) 3.0274 22.11

L11(um) 1.8921 18.65

L12(um) 1.5137 30.3

L13(um) 5.6764 15.3

L14(um) 4.9196 12.56

L15(um) 6.0549 20.12

L16(um) 0.7569 17.54

Iref1 4.5411 uA 43 uA

DC Gain 25.54 db 14.14 K

UGBW 694.6 MHz 612.55 MHz

Power(mW) 20.433 34.35


ISSN: 2456-1851



Table 2: The results of Differential amplifier two classes circuit

by the PSO and ABC algorithms

Variable PSO ABC

W1=W2 (um) 16.5000 16.7000

W3=W4(um) 22.9000 26.1000

W5(um) 40.5 76.6197

W6(um) 36.9 208861

W7=W8 (um) 96.3 45.0244

L1=L2 (um) 4.5 9.3000

L3=L4(um) 57.7 72.1000

L5(um) 16.4 82.9000

L6(um) 60.3 15.50000

L7=L8(um) 99.2 49.1070

Iref1 372.2 uA 43 uA

Cc(pF) 17 26.3076

DC Gain 81.756 54.54

UGBW 772.33 653.54

Power(mW) 0.6 1.534


ISSN: 2456-1851



Table 3: Comparing results with different methods

DE GA ABC PSO Constraints Specifications

85.92 78.436 54.54 81.756 >=8.5 DC gain (dB)

2.7684 6.6431 6.535 7.7233 >=2.5 GBW (MHz)

2.1298 2.4344 1.534 6/0 minimized Power(mW)

11037 10533 9876 4534 - Total run time (s)

Figure 1: The ABC algorithm runs in 20 cycles


ISSN: 2456-1851



Figure 2: Convergence of the PSO algorithm in 80 repetitions

Figure 3: Sound amplifier circuit AD8531


ISSN: 2456-1851



Figure 4: Differential amplifier circuit of two floors

Table 4: Specifications of the amplifier design

4. FORMULAS AND MATHEMATICAL

RELATIONS

The objective function is defined in the

following.

1. 𝑀𝑖𝑛𝑖𝑚𝑖𝑧𝑒 𝑤1 . (1

𝐵𝑊𝑛𝑜𝑟𝑚) + 𝑤2. (

1

𝐺𝑎𝑖𝑛𝑛𝑜𝑟𝑚) +

𝑤3 . 𝑁𝑜𝑖𝑠𝑒𝑛𝑜𝑟𝑚 + 𝑤4. 𝑝𝑜𝑤𝑒𝑟𝑛𝑜𝑟𝑚

2. 𝑀𝑖𝑛𝑖𝑚𝑖𝑧𝑒{(1

𝐵𝑊) . (

1

𝐺𝑎𝑖𝑛) . 𝑁𝑜𝑖𝑠𝑒. 𝑃𝑜𝑤𝑒𝑟}

5. DISCUSSION AND CONCLUSION

This essay presents a framework for

optimizing circuits. By using evolutionary

algorithms, the values for the parameters of

a circuit were calculated to reach important

metrics such as gain, bandwidth, power

consumption, and soon. The ABC and PSO

algorithms were programmed into

MATLAB software and connected to

HSPICE software. By extracting the netlists

from the software, in the MATLAB setting,

the above algorithms were executed and the

Required Specification

90> Gain(dB)

>700khz UGBW(MHz)

<50nv Noise

1 Vdd(v)


ISSN: 2456-1851



appropriate parameters were extracted to

achieve the desired target function. Due to

the fact that several criteria (gain and

bandwidth and power consumption) were

considered in optimization, the objective

function was written as a combination of

metrics expressed as math. As a result, a

multi-objective problem became a single-

objective problem. The results show that the

PSO algorithm works much better than the

ABC algorithm in terms of speed and

efficiency (reaching the desired goal). At the

end, the results of the other methods in the

comparison of GA and DE studies and the

better performance of PSO were confirmed.

In order to improve the research, the

following suggestions are presented blow:

1. Finding a method for MATLAB

software link with other software for

analyzing and optimizing IC design can be a

new way to optimize circuits.

2. The use of algorithms and newer and

faster methods for faster analysis, because

for circuits with more parameters, such as 30

transistors. It takes about 4 to 5 hours for

this. For example, a new method has been

proposed by an Iranian researcher called the

stock market algorithm, which is suggested.

3. One of the challenges of hardware

programming languages is the optimization

of digital circuits designed by programmers

in these languages. Linking these software

with MATLAB programs to improve these

circuits is one of the most valuable work that

is suggested as a continuation of the work.

REFERENCES

[1] B. A. A. Antao and A. J. Brodersen,

"A framework for synthesis and verification

of analog systems," Analog Integrated

Circuits and Signal Processing, vol. 8, pp.

183-199, 1995/09/01 1995.

[2] G. Oltean, S. Hintea, and E. Sipos,

"A Genetic Algorithm-Based Multiobjective

Optimization for Analog Circuit Design," in

Knowledge-Based and Intelligent

Information and Engineering Systems: 13th

International Conference, KES 2009,

Santiago, Chile, September 28-30, 2009,

Proceedings, Part II, J. D. Velásquez, S. A.

Ríos, R. J. Howlett, and L. C. Jain, Eds., ed

Berlin, Heidelberg: Springer Berlin

Heidelberg, 2009, pp. 506-514.

[3] M. R. Amini, M. Mahdavi, and M. J.

Omidi, "Analysis of a multi-user cognitive

radio network considering primary users

return," Computers & Electrical

Engineering, vol. 53, pp. 73-88, 7// 2016.

[4] R. A. Thakker, M. S. Baghini, and

M. B. Patil, "Low-Power Low-Voltage

Analog Circuit Design Using Hierarchical

Particle Swarm Optimization," in 2009 22nd

International Conference on VLSI Design,

2009, pp. 427-432.

[5] H. Xu and Y. Ding, "Optimizing

Method for Analog Circuit Design Using

Adaptive Immune Genetic Algorithm," in

2009 Fourth International Conference on

Frontier of Computer Science and

Technology, 2009, pp. 359-363.


ISSN: 2456-1851



[6] J. Yanfeng, J. Jiaxin, Z. Xiaobo, and

Y. Bing, "Automated analog circuit design

using Genetic Algorithms," in 2009 3rd

International Conference on Anti-

counterfeiting, Security, and Identification

in Communication, 2009, pp. 223-228.

[7] M. Barros, J. Guilherme, and N.

Horta, "Analog circuits optimization based

on evolutionary computation techniques,"

Integration, the VLSI Journal, vol. 43, pp.

136-155, 2010/01/01/ 2010.

[8] M. Fakhfakh, Y. Cooren, A. Sallem,

M. Loulou, and P. Siarry, "Analog circuit

design optimization through the particle

swarm optimization technique," Analog

Integrated Circuits and Signal Processing,

vol. 63, pp. 71-82, 2010/04/01 2010.

[9] Mohammad Reza Amini, Einollah

Balarastaghi, "Universal neural network

demodulator for software defined radio,"

International Journal of Engineering and

Technology, vol. 3, 2011.

[10] A. Jafari, M. Zekri, S. Sadri, and A.

Mallahzade, "Design of Analog Integrated

Circuits by Using Genetic Algorithm," in

2010 Second International Conference on

Computer Engineering and Applications,

2010, pp. 578-581.

[11] J. G. Manuel Barros, Nuno Horta,

Analog Circuits and Systems Optimization

based on Evolutionary Computation

Techniques: Springer-Verlag Berlin

Heidelberg, 2010.

[12] Mohammad Reza Amini, Einollah

Balarastaghi,"Improving ann bfsk

demodulator performance with training data

sequence sent by transmitter," presented at

the Machine Learning and Computing

(ICMLC), 2010 Second International

Conference on, 2010.

[13] C. Ngai, "Optimization in switched

circuits design using constraint

programming," in 2012 International

Conference on System Science and

Engineering (ICSSE), 2012, pp. 143-146.

[14] MR. Amini and A. Mirzavandi,

"Phase-Type Model Spectrum Sensing for

Cognitive Radios," IETE Journal of

Research, vol. 61, pp. 1-7, 2015 2015.

[15] M. R. Amini, F. Hemati, and A.

Mirzavandi, "Trilateral Tradeoff of Sensing,

Transmission, and Contention Times in a

Multiuser Split-Phase CR Networks,"

Sensors Journal, IEEE, vol. 15, pp. 6044-

6055, 2015.

[16] M. W. Cohen, M. Aga, and T.

Weinberg, "Genetic Algorithm Software

System for Analog Circuit Design,"

Procedia CIRP, vol. 36, pp. 17-22,

2015/01/01/ 2015.

[17] D. Joshi, S. Dash, U. Agarwal, R.

Bhattacharjee, and G. Trivedi, "Analog

Circuit Optimization Based on Hybrid

Particle Swarm Optimization," in 2015

International Conference on Computational

Science and Computational Intelligence

(CSCI), 2015, pp. 164-169.


ISSN: 2456-1851



[18] G. Berkol, E. Afacan, G. Dündar,

and E. V. Fernandez, "A hierarchical design

automation concept for analog circuits," in

2016 IEEE International Conference on

Electronics, Circuits and Systems (ICECS),

2016, pp. 133-136.

[19] S. Mallick, R. Kar, S. P. Ghoshal,

and D. Mandal, "Optimal sizing and design

of CMOS analogue amplifier circuits using

craziness-based particle swarm

optimization," International Journal of

Numerical Modelling: Electronic Networks,

Devices and Fields, vol. 29, pp. 943-966,

2016.

[20] E. Afacan and G. Dündar, "Inversion

coefficient optimization assisted analog

circuit sizing tool," in 2017 14th

International Conference on Synthesis,

Modeling, Analysis and Simulation Methods

and Applications to Circuit Design

(SMACD), 2017, pp. 1-4.

[21] M. Amini, M. Mahdavi, and M. J.

Omidi, "Energy Efficiency Optimization of

Secondary Network Considering Primary

User Return with Alternating-Phase-Type

Traffic," IEEE Transactions on

Communications, vol. PP, pp. 1-1, 2017.

improved analog circuits using … the toolbox are ready, ... our hspice coding script, we set out...

Documents