Actually i'm looking for a higer IP3 value: something above 35 or 40 dBm.
At 9MHz, 2.4KHz band, 1.5 form 6:60 factor.
Any advice?
Thanks
Mauro

Well, you can either be prepared to pay a pretty penny for a custom unit and
specify the Ip3 or go to a lower frequency. The best I've ever seen are some
3.75 MHz 10 pole filters made by McCoy. You can see their shape factor and
phase and group delay on N2PK's web site for his Vector Network Analyzer.
Those are my filters shown there. They exhibit +45 dBm Ip3 as singletons,
and drop like a rock to -130 dB and don't pop back up. I have no idea what
they were made for or what they cost but you can bet they were for a severe
environment. Second best that I'm acquainted with are some 10 pole
monolithic filters at 5.2 MHz, made for AT&T by Motorola. I couldn't measure
them but a friend in a professional lab did and a hybridized pair of them
measured +46 dBm or presumeably, +43 for each one. Again, about the only
thing I know about them is that the application was for FDM of microwave
links, where they had to pass a voice bandwidth through perhaps 8 or 10 of
them and still be intelligible. I suspect that if I had to pay for them when
they were new, I couldn't have afforded them. Gorgeous group delay,

At +35 or 45, you're getting into the realm where if you can put a signal
into them that will exercise the max Ip3, you're probably in danger of
breaking something with the power. Keep the gain down ahead of the filter.
Short of doing moonbounce where Noise Figure is everything, you can probably
get by with a much poorer NF than you think. ~20 dB NF, is -120 dBm in a
3KHz bandwidth, and on anything below 20 meters is more than enough
sensitivity if you have anything like an antenna. You put 20 dB of
attenuation in the front end filters, the mixer and anything else ahead of
the narrow filter and a spec of +25 dBm for the filter will do a
surprisingly good job in any application short of a laboratory requirement.

Regards

W4ZCB