Lou, Many high-speed data converters actually interleave samples from multiple lower speed ADCs. They delay clock, rather than the signal, which is much easier. Even then, unless the delay is perfect, you get really bad spurs. Mismatches in delay on the order of 100 fs can be a big problem.
So what you want to try should be possible, but the results aren't going to be stellar. The ADCs are set up for simultaneous sampling at 200 MHz or 5ns. You would need to split the signal and then delay one leg by 2.5ns. No hilbert transforms or phasing networks are required. You *might* also be able to get the clock generator chip to generate an inverted clock to the 2nd ADC chip, in which case you wouldn't need the delay. In both cases, you're going to need to do some FPGA work to get the samples lined up in the right order. Matt On Wed, Jul 20, 2016 at 12:32 PM, madengr <rfeng...@me.com> wrote: > Posting this to GR (as opposed to USRP) since it seems more theory. > > Say I have an X300 and want to get DC - 200 MHz BW from a single daughter > card by combining two real sampled streams. The X300 ADCs (or within the > dual channel ADCs themselves) clocks can't be interleaved. > > An argument is that one of the two RF input channels can simply be delayed > 90 degrees with a length of coax. This gives 90 degree shift at 200 MHz, > 45 > at 100 MHz, etc, with the argument being the lower frequencies become more > over-sampled, so no information is lost. Seems too simple, but I can't > mathematically show it won't work, since the oversampling makes some sense. > > My argument is that this won't work, and it would require a nasty SSB > phasing network on one of two input channels (analog equivalent of Hilbert > transform) to provide a constant 90 degree shift over the entire band > (since > sample clocks can't be quadrature), with equivalent delay on the other. > Then a Hilbert transform at baseband in GR before they are summed into a > 200 > MHz complex baseband. > > In either case the anti-aliasing filters would need cutoff increased. > > Lou > > > > > > > -- > View this message in context: > http://gnuradio.4.n7.nabble.com/Interleaved-sampling-without-interleaved-clocks-tp60986.html > Sent from the GnuRadio mailing list archive at Nabble.com. > > _______________________________________________ > Discuss-gnuradio mailing list > Discuss-gnuradio@gnu.org > https://lists.gnu.org/mailman/listinfo/discuss-gnuradio >
_______________________________________________ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
