IEEE Solid-States Circuits Magazine - Winter 2022 - 96

scheme, compared to the traditional
FeFET scheme. Lastly, Prof. Li gave an
outlook of FeFET-based computing-inmemory
(CIM) solutions after presenting
a few FeFET-based CIM designs.
The second technical talk, " Key
Technologies of Heterogeneous Computing
Chips, " was held on 18 October
2021, by Prof. Zhiyi Yu, from the
School of Microelectronics
Science
and Technology, Sun Yat-Sen University,
China. Prof. Yu introduced the
advantages of heterogeneous computing
systems by analyzing the
challenges and requirements faced
by traditional processor architecture.
Prof. Yu explained that the basic form
of the next-generation heterogeneous
computing system is a large-scale
heterogeneous multichip system enabled
by an advanced interconnection
network. Therefore, finding new
interconnects, clocks, and devices is
the key to the development of nextgeneration
heterogeneous chip processors.
Prof. Yu also described how
to realize a new interconnection mode
through an efficient on-chip network
with an interchip interconnection. He
presented his proposed design methodology
of data-driven asynchronous
circuits with innovative design processes
based on existing electronic
design automation tools. Prof. Yu discussed
how to use nonvolatile registers
based on hybrid magnetic tunnel
junction-CMOS technology to realize
fine-grained power management and
achieve low power consumption. During
the question and answer session,
the audiences had a vibrant discussion
with Prof. Yu. The lecture was
concluded with a certificate award as
a recognition token to Prof. Yu.
The technical talk series this October
was very successful. Every lecture
had a full house of attendants from
HUST and neighboring universities in
Wuhan. More than 200 people, including
faculty members, researchers, and
students, attended the talk series.
-Fei Liu
(graduate Student Member)
-Chao Wang
(Chapter chair, IEEE Senior Member)
SHOP TALK: WHAT YOU DIDN'T LEARN IN SCHOOL (continued from p. 20)
in the all-channel FFT is the average
sine wave.
Using the linear regression slope
from Figure 5, it's an easy mathematical
step (even for me) to calculate
the time skew of this particular
channel versus the average. Using MS
Excel and just 512 points, one might
expect only a rough approximation
of the timing error. Surprisingly, this
method duplicates the results of a
more sophisticated calibration algorithm
to the femtosecond. Impressive,
no?
10
15
20
-20
-15
-10
-5
5
-4E+13
-2E+13
There is another way to do the skew
measurement that you might like better,
but it doesn't reveal anything about
jitter, and it doesn't result in any cool
graphics. Take the FFT of the entire
data record (i.e., all interleaved channels).
Now edit or filter the same data
record, leaving only the data from just
one channel and take the FFT again. Finally,
compare the phase of the fundamental
in both FFTs. The time skew of
the single channel is
skew = D ^h phase in radians /
(2r # input frequency).
For my data, the linear regression
method gives a skew of 46.23 fs. This
phase method gives 46.25 fs. Damn,
I'm good!
Linear Regression:
y = -4.62E-14 x -1.51E-01
R2 = 5.76E-2
2E+13
Derivative of Fundamental (LSBs/s)
FIGURE 5: Error values versus input signal derivative for one channel of an interleaved
ADC. The input signal is estimated by the FFT fundamental using all ADC channels. Errors
are calculated by a filtered IFFT where the dc, the fundamental, and the first four harmonics
have all been removed. Linear regression of the result indicates this channel is 46 fs early
compared to the average of the other channels. LSB: least significant bit.
96 WINTER 2022
IEEE SOLID-STATE CIRCUITS MAGAZINE
4E+13
Time's Up
There are lots of other fun, nonspectral
things to do with ADCs, if you have
the ...um... time. But I need to share
this magazine with the other authors.
If you'd like to see more time-domain
ADC tricks, send your obsequious entreaties
to shifobrains@ieee.org. Be sure
to include flattering comments on this
article and others in this series.
Error (LSBs)

IEEE Solid-States Circuits Magazine - Winter 2022

Table of Contents for the Digital Edition of IEEE Solid-States Circuits Magazine - Winter 2022