This document describes the implementation of a higher-order IIR filter using an 8051 microcontroller. An RTOS with a FIFO structure was developed to realize the filters in a previous circuit. Software used includes Proteus, Keil uVision, and MATLAB. The circuit diagram and code with comments are provided. Observations from testing various input frequencies show decreasing output amplitude and increasing distortion as frequency increases beyond the cutoff frequency, as expected for a low-pass filter. The frequency response matches that obtained previously using MATLAB.
This document describes the implementation of a higher-order IIR filter using an 8051 microcontroller. An RTOS with a FIFO structure was developed to realize the filters in a previous circuit. Software used includes Proteus, Keil uVision, and MATLAB. The circuit diagram and code with comments are provided. Observations from testing various input frequencies show decreasing output amplitude and increasing distortion as frequency increases beyond the cutoff frequency, as expected for a low-pass filter. The frequency response matches that obtained previously using MATLAB.
This document describes the implementation of a higher-order IIR filter using an 8051 microcontroller. An RTOS with a FIFO structure was developed to realize the filters in a previous circuit. Software used includes Proteus, Keil uVision, and MATLAB. The circuit diagram and code with comments are provided. Observations from testing various input frequencies show decreasing output amplitude and increasing distortion as frequency increases beyond the cutoff frequency, as expected for a low-pass filter. The frequency response matches that obtained previously using MATLAB.
This document describes the implementation of a higher-order IIR filter using an 8051 microcontroller. An RTOS with a FIFO structure was developed to realize the filters in a previous circuit. Software used includes Proteus, Keil uVision, and MATLAB. The circuit diagram and code with comments are provided. Observations from testing various input frequencies show decreasing output amplitude and increasing distortion as frequency increases beyond the cutoff frequency, as expected for a low-pass filter. The frequency response matches that obtained previously using MATLAB.
Theory: The 8051 Microcontroller is one of the most popular and most commonly used microcontrollers in various fields like embedded systems, consumer electronics, automobiles, etc. A Microcontroller is a VLSI IC that contains a CPU (Processor) along with some other peripherals like Memory (RAM and ROM), I/O Ports, Timers/Counters, Communication Interface, ADC, etc.
OPAMP (741) stands for operational amplifier it is used as an amplifier as well as to perform operations (arithmetic operations like addition, subtraction, logarithm, exponential etc), in our project we are using it with DAC which has current as default output, OPAMP is used to convert that as voltage.
An infinite impulse response (IIR) filter is a digital filter that depends linearly on a finite number of input samples and a finite number of previous filter outputs. In other words, it combines a FIR filter with feedback from previous filter outputs. Because the filter uses previous values of the output, there is feedback of the output in the filter structure. The impulse response is “infinite” because there is feedback in the filter; if you put in an impulse (a single “1” sample followed by many “0” samples), an infinite number of non-zero values will come out (theoretically).
Circuit Description:
3 Fig 1: Circuit Diagram
The circuit is the same as the one designed in Task 1. The code has been altered to implement the Low-pass IIR filter.
The coefficients are in the range of 0-2 so they are converted to 1.7 notation of fixed precision point and the samples are already in 0.8 notation, so the output which is:
Is in the form of 1.7*0.8 = 1.15, lower 8 bits are ignored, hence we are left with output in the form of 1.7 which is converted to 0.8 form and sent to the DAC.
We are implementing a fourth order IIR filter by cascading two second order IIR filters the samples that are fed from ADC is the input to first filter, the output of first filter is input to second filter and the output of the second filter is the output of the actual filter and it is given to the DAC.
As IIR filter requires previous input and outputs and it has two cascaded filters we need more registers to store those values so two register banks are used.
Filter Parameters
4 Type : Low pass window, IIR, 4th order
Cut-off frequency = 344 Hz (lowest roll no. of group)
Sampling frequency = 5000 Hz
Code: ORG 0000H
SJMP MAIN; jump to main program
ORG 0003H
SJMP ISR; jump to interrupt service routine
MAIN:
MOV P1,#0FFH; to make port 1 as input
MOV P2,#00H; to make port 2 as output
MOV P0,#00H;
MOV IE,#81H; to enable interrupt 0
CLR PSW.3;
CLR PSW.4; to enable register bank 1
SETB P3.0; to set ale
SETB P3.1; to set sc for starting the conversion
CLR P3.0; to generate a time delay in sc pulse
CLR P3.1;
FRGRND:
SJMP FRGRND;
ISR : PUSH ACC;
PUSH PSW;
5 ; first cascaded 2nd order iir filter
MOV A,P1 ; to copy the contents after conversion
MOV R0,A; r0 stores x[n]
MOV B,#05H; an
MUL AB; an*x[n]
MOV R7,B;
MOV A,R1; r1 has x[n-1]
MOV B,#0AH; an-1
MUL AB;
MOV R6,B;
MOV A,R6;
ADD A,R7;
MOV R7,A;
MOV A,R2; r2 stores x[n-2]
MOV B,#05H; an-2
MUL AB;
MOV R6,B;
MOV A,R6;
ADD A,R7;
MOV R7,A ; an*xn+an-1*xn-1+an-2*xn-2
MOV A,R3 ; yn-1
MOV B,#0C8H; bn-1
MUL AB;
6 MOV R6,B;
MOV A,R6;
ADD A,R7;
MOV R7,A ;
MOV A,R4 ; yn-2
MOV B,#5CH; bn-2
MUL AB;
MOV R6,B;
MOV A,R7;
CLR PSW.7;
SUBB A,R6;
RLC A;
MOV R7,A ; total summation yn
MOV A,R1;
MOV R2,A;
MOV A,R0;
MOV R1,A; all inputs shifted
MOV A,R4; yn-2
SETB PSW.3; enable bank1
MOV R2,A; xn-2 of 2nd filter
CLR PSW.3; enable bank0
7 MOV A,R3;
MOV R4,A ; shifting in current filter
SETB PSW.3; enable bank1
MOV R1,A ; xn-1 of 2nd filter
CLR PSW.3; enable bank0
MOV A,R7; yn
MOV R3,A ; shifting in current filter
SETB PSW.3; enable bank1
MOV R0,A; xn-1 of 2nd filter
; first filter complete
MOV A,R0; xn
MOV B,#04H; an
MUL AB;
MOV R7,B ; xn*an
MOV A,R1; xn-1
MOV B,#08H; an-1
MUL AB;
MOV R6,B;
MOV A,R6;
ADD A,R7;
MOV R7,A; xn*an + xn-1*an-1
MOV A,R2; xn-2
MOV B,#04H; an-2
MUL AB;
8 MOV R6,B;
MOV A,R6;
ADD A,R7;
MOV R7,A ; an*xn+an-1*xn-1+an-2*xn-2
MOV A,R3 ; yn-1
MOV B,#0A7H; bn-1
MUL AB;
MOV R6,B;
MOV A,R6;
ADD A,R7;
MOV R7,A ;
MOV A,R4 ; yn-2
MOV B,#38H; bn-2
MUL AB;
MOV R6,B;
MOV A,R7;
CLR PSW.7;
SUBB A,R6
RLC A;
MOV R7,A ; total summation yn
MOV A,R1;
MOV R2,A;
MOV A,R0;
MOV R1,A; all inputs shifted
9 MOV A,R3;
MOV R4,A ;
MOV A,R7;
MOV R3,A ; output shifting in current filter
MOV A,R7;
CLR PSW.3; enable bank 0
MOV P2,R3 ; to send the contents to dac
; to start another conversion
SETB P3.0; set ale
SETB P3.1; set sc
CLR P3.0; clear ale
CLR P3.1; clr sc
POP PSW;
POP ACC;
RETI
END
Observations:
Table 1: List of output amplitude and gains for varying frequencies for the developed RTOS of Low Pass FIR filter
From the above plots, it was observed that with an increase in frequency, there was a decrease in amplitude of output waveform.
The output waveforms got progressively more and more distorted as the frequency approached half of the Nyquist frequency (2500 Hz).
Fig 6: Frequency response of the RTOS
Fig 5: Frequency response of the RTOS
12 From the semi-log plot we can observe that there is a roll-off in gain when the frequency is increased beyond cut-off frequency.
Fig 6: Frequency response of the IIR Filter obtained from the FDA Tool
13 Conclusion: The coefficients of the IIR filter were obtained from FDAtool of MATLAB. Those coefficients were converted to 1.7 form of fixed-point precision. Then using the value of samples and coefficients the output was calculated and given to DAC. For varying frequency it was observed that the amplitude of the output decreased and distortion increased with increasing input frequency beyond cut-off frequency (344 Hz). As it is a low pass filter the reduction in output amplitude is justified and the output frequency response is also similar to that of response obtained in FDAtool. From graphs we observed that as the frequency reached half of the sampling frequency (2.5KHz), the distortion increased very rapidly thus satisfying Nyquist theorem (fs >2*fm) even if fs is just greater than 2*fm there will be lot of distortion which is observed here.
Problems Faced 1. Due to the higher number of coefficients and past values of input and output that had to be stored, a single register bank was not enough. We had to use a second bank as well. This led to increased complexity of the code. 2. Distortions were higher as compared to the FIR filter, causing problems while taking readings. There were especially large distortions at f = 200 Hz. Even at the other frequencies, the distorted sinusoids were interspersed by flat lines in between.
