Finite impulse response (FIR) filters are characterized by the fact that they use only delayed versions of the input signal to filter the input to the output. For a causal discrete-time FIR filter of order N, each value of the output sequence is a weighted sum of the most recent input values:

Where:

x[n] is the input signal,

y[n] is the output signal,

N is the filter order

bi is the value of the impulse response, also a coefficient of the filter

Filter design

FIR filter designing is finding the coefficients and filter order that meet certain specifications. When a particular frequency response is desired, several different design methods are common:

Window design method

Frequency sampling method

Least MSE (mean square error) method

Parks-McClellan method

DFT algorithms

There many software such as MATLAB, GNU Octave, Scilab and SciPy (Python). In this topic, I only share how to implement FIR filter on FPGA. So the detail of filter designing, finding coefficients will shared in the later topics.

Implementing a FIR filter on FPGA for slow signal

FIR filter for slow signal will designed by using special way to save the number of multipliers.

Finding coefficients using window method

Using Python

from future import print_function from future import division import numpy as np

fS = 48000 # Sampling rate. fL = 22000 # Cutoff frequency. N = 13 # Filter length, must be odd.

The R-2R resistor ladder network directly converts a parallel digital symbol/word into an analog voltage

(24 bit R-2R resistor ladder)

The R–2R
network causes these digital bits to be weighted in their contribution to the
output voltage V_{out}. Depending on which bits are set to 1 and which to 0, the
output voltage (V_{out}) will have a corresponding stepped value between
0 and V_{ref} minus the value of the minimal step, corresponding to
bit 0. The actual value of V_{ref} (and the voltage of logic 0) will depend on the type of
technology used to generate the digital signals.

For a
digital value VAL, of a R–2R DAC with N bits and 0 V/V_{ref} logic levels, the output voltage V_{out} is:

Vout = Vref * VAL / 2^N

3. Digital Design (Verilog) FPGA-based R-2R resistor ladder DAC

(R-2R
DAC block diagram)

(R-2R DAC RTL
diagram – Quartus Software)

I2S Receiver module:

The I2S
protocol is a common standard used to send audio data. It is a serial protocol
very similar to SPI, but it is a streaming protocol. That means it is always
transmitting data. I2S sends a stream of stereo audio data. For each audio
sample there is a left channel and a right channel value. The values can be any
number of bits, although 16, 20, 24, and 32 bit values are the most common

(I2S
Protocol timing diagram)

Clock Control Module:

This module includes sampling
rate detect and PLL module.

Sampling rate detect operates
base on the period of Word Clock of I2S (WS) to determines the sampling rate of
I2S. It will be 44.1khz, 88.2khz, 176.4khz, 352.8khz… or 48khz, 96khz, 192khz,
384khz, 768khz. Then this module will choose suitable input OSC (49.152Mhz or
45.1584Mhz) to feed to PLL module.

PLL module uses external clock
that is choose to generate few internal clocks. These clocks are used for
oversampling clocks, filter clocks, R-2R transmitter clocks

(PLL module RTL diagram – Quartus Software)

Interpolation filter module:

Digital data transmitter:

Digital data transmitter module
receives data from Interpolation filter module, converts digital data from
single-end to balanced data and transmits R2R driver module (Altera Max V CPLD)

Data transmit protocol is my
protocol, it is not standard protocol. It is a bit stream protocol and Left /
Right are transmitted separately.

(Digital data transmitter RTL diagram – Quartus Software)

4. PCB Design FPGA-based R-2R resistor ladder DAC

PCB includes 2 parts : FPGA using Altera Cyclone IV and CPLD using Altera MAX V

FPGA Part

FPGA Part uses Altera Cyclone IV EP4CE6 –
144TQFP package

Oscillators use CRYSTEK CCHD-957. Crystek’s
Model CCHD-957 HCMOS CLOCK oscillator family has been designed specifically for
High Definition Audio (HD Audio). It features a typical low close-in phase
noise of -100 dBc/Hz @ 10 Hz offset, and a noise floor of -169 dBc/Hz. With
this extreme low phase noise performance, you will “Hear the Difference”.