hi tom,
the fastest we can sample now is 6 Gsps, by interleaving
a pair of national ADC083000 3 Gsps 8 bit ADC's plugged
into a roach or mini-roach board.
agilent has offered us their 20Gsps ADC's, and we have begun
very preliminary design, working with xilinx research
and dave hawkins at caltech.
two adc's can be interleaved to get 40 Gsps,
but we aren't planning to do this.
we have also been talking to hittite, maxtex,
fujitsu, adsantec, national, that already have/will have
ADC's that range to 80 Gsps.
but this fast ADC project is still a ways off
and won't be ready this year (depends on NSF funding).
so if you need something in december, the fastest we'll have
will still be the 6 Gsps ADC.
best wishes,
dan.
Tom Kuiper wrote:
Tomorrow some of us are meeting with Sandy Weinreb and Steve Smith to
talk about a wideband, channelized down-converter. Ultimately we want
to be able to down-convert and digitize 14 - 26Ghz simultaneously.
Sandy's GAVRT design was based on the 2 Gsamp/s CASPER iADCs. Using
that approach, with two polarizations, we would need twelve
down-converter channels (both I and Q sampling). It looks like we could
safely assume that 3 Gsamp/s ADCs can be used with the ROACH boards,
which would reduce the number of channels to eight. Chuck Naudet tells
me that the next step in the VLBI world will go from the current 1 GHz
bandwidth (at one or two bits/sample) to 4 GHz. We need more dynamic
range but even 4 Gsamp/s would be attractive, reducing the number of
downconverter channels to six. So what would you do if you expected to
start building the downconverters in April? in September? in December?
In advance, many thanks
Tom
