Caltech Baseband Recorder (CBR)
-------------------------------


The basic idea is:

a) analog quadrature down converter
b) analog anti-aliasing low pass filters
c) custom digitizer board--Fast Flexible Digitizer (FFD)
d) EDT SBus card
e) Sun Ultra computer
f) 6 Barracuda drives
g) 2 DLT7000 drives


The FFD is controlled over an RS-232 connection from the host computer
(arm, stop, #chans, #bits, ...). And accepts an external start signal
and clock in addition to the IF signals.

The FFD has two halves with separate 16bit outputs: each half has 4 analog
channels which can be sampled at up to 50MHz with 8bits. The resulting
4x8 bits for each half are fed into separate bit packers which each output
16bits to an EDT card at the appropriate rate. For the initial Arecibo
system, we have populated half of a board and will be driving just one
EDT card.

As an example, the end-to-end system can operate as a 2poln 10MHz baseband
recording system, i.e., each of LPC-I, LCP-Q, RCP-I, RCP-Q are digitized
with 2bits at 10MHz, for a resulting 10MB/s data stream which is striped
to disk, and de-striped to tape as 1GB files so the resultant stack of
tapes can be analyzed on a system with only one tape drive.

The FFD is designed to digitize each channel at up to 50MHz, so another
mode is to do dual-polarization 50MHz baseband recording directly into the
400MBit/s Datatape recorders.

In addition, since the board has two halves, it should be possible (if
we populate the second half) to do dual band, dual polarization simultaneous
baseband recording with the use of 2 EDT cards.

Another mode is to run 10MHz dual-poln at 4bit resolution and buffer
the 20MB/s data stream on disk until it is full.

Each bit packer is a re-programmable gate array (in-situ) which currently
has been programmed to implement the following modes:
channels	nbits
-----		-----
0,1		2
2,3		2
0,1,2,3		2
0,1		4
2,3		4
0,1,2,3		4
0,1		8
2,3		8


The current FFD board has ECL drivers which we connect directly to
the RS422 receivers on the EDT SCD-20 board (nominally rated at 20MB/s but
seems to work quite well to at least 30MB/s but definitely fails before
40MB/s).


We have over designed the board to handle faster recording media as they
become cost effective.  As an example, we plan on using a fully populated
FFD outputting ~2GBit/s for a TMFE project,

The full out performance of an FFD board is 8 channels x 8bits x 50MHz coming
out as 64 ECL lines at a total rate of 3.2Gbit/sec.