CHAPTER 5

Chopping: a technique for noise and offset reduction

5.4. Noise modulation

From the previous paragraph we have seen the modulation effects of chopping on a input process x(t). It is important to notice the difference between a narrow-band process and a broadband process after chopper modulation. The input stationary random process x(t) can be white noise or 1/f noise applied to the amplifier A(f) which has a band limiting effect on the noise.

5.4.1. White noise modulation

For the beginning, assume that x(t) is a broadband random process white noise like with a power spectral density Swhite. In order to simplify the analysis, the gain of the amplifier is taken to be 0dB and the frequency transfer has a first order behavior:

(5.12)

The power spectral density of the noise at the output can be found from (5.11) with assumption that Sxx(f)=Swhite:

(5.13)

The series from (5.13) can be computed by using Poisson summation rule [9]. For large values of fLPT (fLPT>>1) we get:

(5.14)

In conclusion, the chopper modulator has a small influence on the white noise when the bandwidth of the amplifier is larger than the chopping frequency. This is not the case for sampling where undersampling phenomena actually increases the noise in the baseband [1]. The power spectral density of the white noise will be unchanged as long as the bandwidth of the amplifier is larger than the chopping frequency. Chopping at frequencies higher than fLP will reduce the power spectral density of the white noise as explained in reference [4]. Although white noise is a fundamental limitation it can be reduced by chopping. Oversampling in D/A and A/D converters has about the same effect on the baseband white noise.

5.4.2. 1/f noise modulation

When narrow band random processes are applied at the input of the chopper modulator the situation will change. The input power spectral density of 1/f noise is:

(5.15)

In the constant c we have included the 1/f noise constant k1/f process dependent and the geometry factor, dependent on the dimensions of the transistors. After chopping the PSD of the modulated 1/f noise becomes:

(5.16)

Fig.5.5 shows the normalized PSD of the flicker noise after chopping when fLPT>>1. The effect of chopper modulation on the flicker noise will be the reduction of the PSD of the output noise at low frequencies. At odd multiples of the chopper frequency the

 Fig.5.5: PSD of 1/f noise after chopper modulation

PSD of the flicker noise increases. Thus, under the same condition fLPT>>1 the PSD of the white noise after chopping remains the same and the PSD of the flicker noise at low frequencies will be reduced.

Featured Video
Editorial
 What Would Joe Do?by Peggy AycinenaComputer History Museum: the Future of War is Here
 IP Showcaseby Peggy AycinenaIP-SoC 2017 – Headliners include de Geus, Saxby
More Editorial
Jobs
ASIC Design Engineer for Infinera Corp at Sunnyvale, CA
Principal PIC Hardware Controls Engineer for Infinera Corp at Sunnyvale, CA
REVISED***Director Product Line RF/IC for EDA Careers at San Jose, CA
Senior PIC Test Development Engineer for Infinera Corp at Sunnyvale, CA
Senior SW Developer for EDA Careers at San Jose, CA
Upcoming Events
“Empowering Leadership with WIT and WISDOM” at SEMI 673 South Milpitas Blvd. Milpitas CA - Nov 28, 2017
Artificial Intelligence and Convolution Neural Networks Discussion at San Jose State University Student Union Theater San Jose CA - Dec 4, 2017
Silicon Valley's Only Comprehensive Embedded Systems Conference at San Jose Convention Center 150 W. San Carlos St. San Jose CA - Dec 5 - 7, 2017
Oski Technology’s Decoding Formal Club Meeting at The Conference Center San Jose CA - Dec 7, 2017