SILVA, V. M. L.; SILVA, VERONICA MARIA LIMA.; SILVA, VERONICA M. L.; http://lattes.cnpq.br/8036572249593151; SILVA, Veronica Maria Lima.
Resumen:
Integrated circuits testing is an important stage in the manufacture and fundamental in
the commercialization of these components, because in this stage are evaluated which
circuits have defects and which are able to be commercialized. In mixed integrated
circuits, such as analog-digital converters (ADC) that interface analog and digital parts
of electronic systems, the test is performed by measuring parameters such as linearity,
noise, among others. For ADC testing, there is the IEEE 1241 standard that establishes
test methods and their procedures for evaluating static and dynamic parameters of ADC.
One such method is the sine wave fit method. However, there is a new type of converter,
Analog-to-Information Converter (AIC), which does not yet have a defined test pattern.
The AIC implements the Compressive Sampling Theory in practical terms and provides a
digitized and compressed output (y) corresponding directly to the analog input signal (x).
A reconstruction algorithm is needed to obtain a reconstructed signal xˆ that computes x
from y. This algorithm is generally not part of the AIC itself and may even be distant from
the AIC interconnected by a communication channel to it. In scientific papers focusing on
AIC testing, in general, the reconstructed signal xˆ is evaluated to test the AIC itself and
as a consequence, the performance of the reconstruction algorithm may affect the result of
the AIC test. The reconstruction algorithms implement the solution of an optimization
problem and are quite sensitive to the input parameters, so it is not interesting that the
test result depends on these algorithms. Few papers propose test methods that consider
only the direct output of the AIC (y). However, they take into account a single test signal
(the sine), which is a reference signal as described in the IEEE 1241 standard for ADC
testing. In this thesis, two new test methods for AIC are proposed, in which only the
AIC outputs, that is, the compressed outputs y, are taken into account. In the first one,
multi-tones (multi-sines) are applied and, in the second, an adaptation of the sine wave fit
method of four parameters, also recommended by the IEEE Standard 1241, was developed.
Experimental and simulation results were obtained for two architectures of different AICs
(Random Demodulator - RD and Random Modulation Pre-Integrator - RMPI) and the
Signal to Noise and Distortion Ratio (SINAD) of both were measured directly from the
AIC output confirming the functionality of the proposed AIC test methods.