[time-nuts] filtering a 10Mhz frequency standard?
eb at telight.com
Tue Dec 13 12:07:29 EST 2011
I reported a filter design I was working on back in November, using
10 base T LAN filter modules. The first experimental air-wired unit
with two modules got to about -75 dBc at 30 MHz, rising to -65 dBc at 70 MHz.
After building the same circuit with shielded compartments, it ran
about -85 dBc flat over this range, but it should have been over 100.
The weak link turned out to be the cross-talk within the modules
themselves - there are two 17 MHz LPF sections in each, and I
cascaded them all.
Four separate modules cascaded, using only the Tx portion of each,
should reach about 120 dBc rejection, with about 6 dB insertion loss.
The original two-module circuit (or maybe even a single one) would
probably suffice for most applications.
I haven't yet impedance matched the 100 ohm differential filters to
the 50 ohm cable environment - I'm thinking it may not even be necessary.
If you have any old LAN cards, hubs, or routers around, you may
already have some nice "free" LPFs for this purpose. The box to put
them in (packaging cost) is another matter.
More information about the time-nuts