Comparators are a critical component in sigma-delta A/D converters. We have
studied experimentally the sensitivity of a quantum flux parametron (QFP)
comparator operated over a range of sampling frequencies F-S from 40 Hz to
40 MHz, In one experimental method, following Ko and Lee, we measure the fi
ring probability of the QFP as a function of applied nux The sensitivity ca
n be derived from the slope of this curve. In a second method, we measure t
he sensitivity directly by observing the spectrum of the QFP output while t
he amplitude of a small sinusoidal applied flux is adjusted to exceed the n
oise floor by 3 dB, The two methods were found to be in good, but not perfe
ct agreement. The sensitivity was measured as we varied both FS and the clo
ck rise time. The spectrum of the quantization noise exhibits, as expected,
a flat floor whose level is inversely proportional to Fs. The best energy
sensitivity that we observed was for a clock frequency of similar to 20 MHz
with a 10 ns rise time. The measured sensitivity was about 1500 h (Planck'
s constant). The readout circuit prevented us from clocking the comparator
into the GHz range for even greater sensitivity. We also believe that the c
omparator could be optimized to improve sensitivity further. The noise floo
r was low enough that we could observe excess low-frequency noise below 5 H
z. We have not yet determined whether it is intrinsic to the comparator or
originates from our test electronics. We hypothesize that the noise floor w
ill continue to fall as Fs increases until we reach the speed limit of the
comparator, at which point successive output samples will no longer be unco
rrelated, or until we reach the uncertainty-principle Limit (h) in the 100s
of GHz range.