Maximum likelihood estimation of ADC parameters

László Balogh, I. Kollár, Attila Sárhegyi

Research output: Chapter in Book/Report/Conference proceedingConference contribution

19 Citations (Scopus)

Abstract

Dynamic testing of analog-digital converters (ADC) is a complex task. A possible approach is using a sine wave because it can be generated with high precision. However, in the sine wave fitting method for the test of ADC's, all the available information is extracted from the measured data. Therefore, the estimated ADC parameters (ENOB, linearity errors) are not always accurate enough, and not detailed information is gained about the nonlinearity of the ADC. Generally, maximum likelihood (ML) estimation is a powerful method for the estimation of unknown parameters. However, currently it is not used for the processing of such data, because of the difficulties of formulating it, furthermore because of the numerically demanding task of the minimization of the ML cost function [9]. We have succeeded in formulating the maximum likelihood function for a sine wave excitation, and in minimizing it. The number of parameters is frightening (all comparison levels of the ADC plus parameters of the sine wave plus variance of an additive input noise), but proper handling allows to determine the best values based on the data. The proper definition of the ML function and formulation of the numerical method are presented, with results using simulation and measurement data. To our knowledge, this is the first case to solve the full maximum likelihood problem.

Original languageEnglish
Title of host publication2010 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2010 - Proceedings
Pages24-29
Number of pages6
DOIs
Publication statusPublished - 2010
Event2010 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2010 - Austin, TX, United States
Duration: May 3 2010May 6 2010

Other

Other2010 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2010
CountryUnited States
CityAustin, TX
Period5/3/105/6/10

Fingerprint

Maximum likelihood estimation
Maximum likelihood
converters
sine waves
analogs
Cost functions
wave excitation
Numerical methods
linearity
nonlinearity
Testing
costs
formulations
Processing
optimization
simulation

ASJC Scopus subject areas

  • Instrumentation

Cite this

Balogh, L., Kollár, I., & Sárhegyi, A. (2010). Maximum likelihood estimation of ADC parameters. In 2010 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2010 - Proceedings (pp. 24-29). [5488286] https://doi.org/10.1109/IMTC.2010.5488286

Maximum likelihood estimation of ADC parameters. / Balogh, László; Kollár, I.; Sárhegyi, Attila.

2010 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2010 - Proceedings. 2010. p. 24-29 5488286.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Balogh, L, Kollár, I & Sárhegyi, A 2010, Maximum likelihood estimation of ADC parameters. in 2010 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2010 - Proceedings., 5488286, pp. 24-29, 2010 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2010, Austin, TX, United States, 5/3/10. https://doi.org/10.1109/IMTC.2010.5488286
Balogh L, Kollár I, Sárhegyi A. Maximum likelihood estimation of ADC parameters. In 2010 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2010 - Proceedings. 2010. p. 24-29. 5488286 https://doi.org/10.1109/IMTC.2010.5488286
Balogh, László ; Kollár, I. ; Sárhegyi, Attila. / Maximum likelihood estimation of ADC parameters. 2010 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2010 - Proceedings. 2010. pp. 24-29
@inproceedings{aff6af173fe549eea894c99990d432fd,
title = "Maximum likelihood estimation of ADC parameters",
abstract = "Dynamic testing of analog-digital converters (ADC) is a complex task. A possible approach is using a sine wave because it can be generated with high precision. However, in the sine wave fitting method for the test of ADC's, all the available information is extracted from the measured data. Therefore, the estimated ADC parameters (ENOB, linearity errors) are not always accurate enough, and not detailed information is gained about the nonlinearity of the ADC. Generally, maximum likelihood (ML) estimation is a powerful method for the estimation of unknown parameters. However, currently it is not used for the processing of such data, because of the difficulties of formulating it, furthermore because of the numerically demanding task of the minimization of the ML cost function [9]. We have succeeded in formulating the maximum likelihood function for a sine wave excitation, and in minimizing it. The number of parameters is frightening (all comparison levels of the ADC plus parameters of the sine wave plus variance of an additive input noise), but proper handling allows to determine the best values based on the data. The proper definition of the ML function and formulation of the numerical method are presented, with results using simulation and measurement data. To our knowledge, this is the first case to solve the full maximum likelihood problem.",
author = "L{\'a}szl{\'o} Balogh and I. Koll{\'a}r and Attila S{\'a}rhegyi",
year = "2010",
doi = "10.1109/IMTC.2010.5488286",
language = "English",
isbn = "9781424428335",
pages = "24--29",
booktitle = "2010 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2010 - Proceedings",

}

TY - GEN

T1 - Maximum likelihood estimation of ADC parameters

AU - Balogh, László

AU - Kollár, I.

AU - Sárhegyi, Attila

PY - 2010

Y1 - 2010

N2 - Dynamic testing of analog-digital converters (ADC) is a complex task. A possible approach is using a sine wave because it can be generated with high precision. However, in the sine wave fitting method for the test of ADC's, all the available information is extracted from the measured data. Therefore, the estimated ADC parameters (ENOB, linearity errors) are not always accurate enough, and not detailed information is gained about the nonlinearity of the ADC. Generally, maximum likelihood (ML) estimation is a powerful method for the estimation of unknown parameters. However, currently it is not used for the processing of such data, because of the difficulties of formulating it, furthermore because of the numerically demanding task of the minimization of the ML cost function [9]. We have succeeded in formulating the maximum likelihood function for a sine wave excitation, and in minimizing it. The number of parameters is frightening (all comparison levels of the ADC plus parameters of the sine wave plus variance of an additive input noise), but proper handling allows to determine the best values based on the data. The proper definition of the ML function and formulation of the numerical method are presented, with results using simulation and measurement data. To our knowledge, this is the first case to solve the full maximum likelihood problem.

AB - Dynamic testing of analog-digital converters (ADC) is a complex task. A possible approach is using a sine wave because it can be generated with high precision. However, in the sine wave fitting method for the test of ADC's, all the available information is extracted from the measured data. Therefore, the estimated ADC parameters (ENOB, linearity errors) are not always accurate enough, and not detailed information is gained about the nonlinearity of the ADC. Generally, maximum likelihood (ML) estimation is a powerful method for the estimation of unknown parameters. However, currently it is not used for the processing of such data, because of the difficulties of formulating it, furthermore because of the numerically demanding task of the minimization of the ML cost function [9]. We have succeeded in formulating the maximum likelihood function for a sine wave excitation, and in minimizing it. The number of parameters is frightening (all comparison levels of the ADC plus parameters of the sine wave plus variance of an additive input noise), but proper handling allows to determine the best values based on the data. The proper definition of the ML function and formulation of the numerical method are presented, with results using simulation and measurement data. To our knowledge, this is the first case to solve the full maximum likelihood problem.

UR - http://www.scopus.com/inward/record.url?scp=77957857037&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77957857037&partnerID=8YFLogxK

U2 - 10.1109/IMTC.2010.5488286

DO - 10.1109/IMTC.2010.5488286

M3 - Conference contribution

AN - SCOPUS:77957857037

SN - 9781424428335

SP - 24

EP - 29

BT - 2010 IEEE International Instrumentation and Measurement Technology Conference, I2MTC 2010 - Proceedings

ER -