ISSN 1063-7397, Russian Microelectronics, 2017, Vol. 46, No. 3, pp. 171–179. © Pleiades Publishing, Ltd., 2017.
Original Russian Text © V.A. Marfin, P.V. Nekrasov, O.A. Kalashnikov, A.Yu. Nikiforov, 2017, published in Mikroelektronika, 2017, Vol. 46, No. 3, pp. 187–196.
Functional Testing of Digital Signal Processors
in Radiation Experiments
V. A. Marf in
*, P. V. Nekrasov
**, O. A. Kalashnikov
***, and A. Yu. Nikiforov
National Research Nuclear University MEPhI, Moscow, 115409 Russia
Specialized Electronic Systems (SPELS), Moscow, 115409 Russia
Received October 12, 2016
Abstract⎯This paper analyzes the specifics of digital signal processors’ (DSPs) radiation-induced failures.
The general methodology, as well as the hardware and software, for the functional testing (FT) of DSPs in
radiation experiments is developed and implemented. The experimental results are presented to demonstrate
the effectiveness of the solutions proposed.
Digital signal processors (DSPs) are widely used in
modern equipment for information processing and
communication, as well as in control systems for filter-
ing, spectral and time-frequency analysis, convolu-
tion, etc. [1, 2]. This equipment and these systems
often operate under conditions of ionizing radiation
(IR), being employed in space hardware, power com-
plexes, and nuclear physics experiments. Particular
DSPs for such equipment can be adequately selected
based only on their radiation hardness, which assumes
certain radiation tests and experiments .
During the radiation tests of the electronic compo-
nents (including DSPs), a performance check is car-
ried out to verify the compliance of all the parameters
of the test object with the requirements specified. In
addition to checking the electrical parameters (cur-
rents and voltages), functional testing (FT) has to be
carried out to diagnose various operating modes of an
object (microchip). Checking the electrical parame-
ters (parametric check) of the microchips of different
classes is generally carried out by universal standard-
ized procedures, which reduces the functional speci-
ficity of each microchip class to the requirements for
the measuring devices (current and voltage measure-
ment ranges and measurement accuracy).
However, the FT of the microchips has to employ
completely different algorithms (and even different
principles) to take into account the functional speci-
ficity of a particular test object [4–6]. For example,
the FT of memory chips is generally carried out using
algorithmic functional tests (with various sequences of
read and write operations on memory arrays); the FT
of logic gates is based on iterating through all possible
logical states and checking the truth table; and the FT
of the analog-to-digital converters (ADCs) involves
precise measurements of the conversion response with
the subsequent evaluation of the accuracy parameters
[7, 8]. In terms of FT, microprocessors are the most
complex test objects with heterogeneous functional
blocks (arithmetic modules, memory units, registers,
clock units, control units, interfaces, etc.) and tens or
hundreds (or even more) of functional modes. The FT
of microprocessors requires specialized test programs
that enable the complete checking of all microproces-
sor blocks and modes.
Radiation tests impose the following specific
requirements on FT methods .
(1) The FT equipment must be compatible with the
radiation test equipment: accelerators, isotope
sources, laser and X-ray sources, etc. First of all, this
implies remote FT with the corresponding constraints
on frequency, output capability, and noise. In addi-
tion, it is important to ensure the protection of both
staff and equipment from radiation.
(2) The FT cycles must run as quickly as possible
while guaranteeing a reliable and statistically signifi-
cant result under real IR conditions. As a rule, the run
time of an FT cycle should not exceed several minutes
(sometimes, several seconds).
(3) It is required to enable automatic and (possibly)
remote control of the FT process, processing the
results, and storing large data arrays. An object is gen-