The VNA speaks and the results surprise
I've been using the 5B4AGN filters for years with excellent results. So when a friend of mine was talking about setting up So2r, I offered to check out his filters so he could have some measured data to plug into his attenuation calculations. When I stacked the 4 filters on the bench up, the results were surprising. The Array Solutions really looked good.
A BPF needs to do two things. First and most importantly, protect the 2nd radio from being damaged by RF out of the first. Here the adjacent band attenuation is the key metric. For this one, "enough to avoid damage" is just as good in the practical sense as an infinite attenuation. And secondly, it needs to attenuate the rig's second harmonic so you can for example work 40m and 80m without having the harmonic content of the 80m transmission wipe out the noise floor on 40m. While the attenuation needed varies with the specific setup, more may not be needed but it definitely does not hurt.
Filter loss is another factor, but it's more of a meter mover on the transmit side than big factor on the receiving end. Less loss is better of course, but if you generally run an amp, then a few watts lost to heat in the filter won't matter. I say on the transmit side because the difference between 90w and 100w will not be discernible on the receiving end. Of course some of the biggest reflector battles have been over that statement.
Compared in the graphs below are the Array Filtermax, the 5B4AGN and the Dunestar 600 6-band auto-switching filter sets, as well as some 5B4AGN single-band filters I had on the bench at the time from Field Day.
The Filtermax performance owes two debts to it's nice plots - first to the design topology, an elliptical (Cauer) type which takes advantage of putting the filter's notches right on the adjacent bands to gain a very nice attenuation boost where it's appreciated most - and secondly they are well trimmed as is reflected in the low loss measured. Having built quite a number of the 5B4AGN filters in the past, I thought perhaps the first unit measured may have not been representative of the normal filter. A second Filtermax purchased a month later had numbers very similar to the first so at least based on those two units, the performance seems to be consistent.
On to the data. For each plot, click the graph to view a high resolution view.
5-BAND COMPOSITE PLOTS
Array Solutions Filtermax
SINGLE BAND COMPARISONS