The problem of further reduction of dc power dissipation in RSFQ devices was studied theoretically, on a simple model of the power supply line, and experimentally, on the main building blocks of the autocorrelator. The study gave the following new results:
Two different designs of an RSFQ autocorrelator for radio astronomy applications were proposed and studied experimentally with the following main results:
In the nearest future we plan to focus on demonstrating a completely
operational, fully integrated prototype autocorrelator with 16
channels and beyond, complete with the double oversampling quantizer,
delay line, and prescalers. Projected parameters of this
autocorrelator include 4 GHz input signal bandwidth, 16 GHz clock
frequency and 16 Mbps-per-channel output rate. It will be integrated
with a specialized room-temperature interface, allowing real-time data
acquisition. Currently, physical dimensions of the existing layout
permit to place up to 64 channels on a standard
chip. Further increase of layout density (for the HYPRES' standard
-
-
Nb-trilayer process) is
possible and should allow up to 100 channels on a
chip, or up to a 400 channels on a single
chip. Straightforward pickup of signals from each channel in this case
might cause serious problems with pin count and heat load from the
leads. However, multiplexing of the outputs can be readily
introduced, slashing the number of pins by a factor of 16 or more.
Going beyond
channels to achieve even higher frequency
resolution would require development of special multi-chip modules
[54]. Further improvement in performance of auto- and cross-
correlators for radio astronomy can be achieved by changing the bit
representation of the sampled data from 2-level to 3- or 4-level.
With minor modifications, ultra-wide-band correlators can also be used
in spread spectrum communications and Doppler radar systems.
Measurement of error rates in low-power gates carried out at speed further builds our confidence that RSFQ technology offers unparalleled performance, particularly in unique applications such as space-borne missions or Petaflops-scale computing (see, e.g, [39]) where low power dissipation (combined with high clock speed) could be a crucial advantage.