yes - it is just a prove of his curve - but this is every test where a porsche, ferrari etc. is tested. they prove whether the manufacturers data is true and what results these specs mean in reality while driving.
p.s. but i’m totally openminded to better and more comprehensive tests that will be posted here
Just to be pedantic (or to help anyone wanting to build one), the Jetvision filter is not a cavity filter, but rather it’s a Three Pole Interdigital Filter. Internally it’ll look something like this.
yep - it’s a three-pole-filter and that is how we call it over here in germany. but in english internet they mostly sub-sum this in the cavity filter class - that’s why i used this wording which is common here in the forum too 
p.s. here is another link that maybe is helpful for those who try to tinker a homebrew solution: A homebrew 1090Mhz ADS-B interdigital bandpassfilter | Projects by Keptenkurk
Thanks for this interesting report. It does highlight an issue with the strong signals around 950MHz that even the ceramic filters struggle to deal with.We are constantly trying to improve our filters and we do have a new design which has more gain (~18dB) has less insertion loss and more out of band attenuation.
We have prototypes which look amazing on paper and on the spectrum analyser but don’t perform as well as the existing products with the ADS-B signals. We are working to ascertain why!
Red = existing SAW HAB Amp, Blue - new design with SAW.
great thanx for the information! but i in no way would say that the saw and ceramic filter struggle - they have to fight there. at least at my very country-side ads-b site the 960mhz gsm is the biggest challenge for an mode-s receiver - and both of your filter/lna combos improve the ads-b reception to nearly the same results (aircrafts seen and message-rate) the cavity does - even if the scan of cavity filter looks cleaner at 960gsm …
fwiw - i’ve added a scan without filters or amplifier with gain levels 50, 40, 30, 20: The Final Filter Shootout. but to be honest i do not really see/understand what the advantage for the test i did here is …
edit: the proportion between noise and 1090mhz remains the same
Ah yes, but the internet is a zoo full of lies and “alternative facts”!
Built like the Jestvision filter, interdidital closely resemble cavity filters, but may also be printed on circuit board - which cavities can’t (at least not until the freq gets high enough to bury it in a multi layer board).
Once you’re using PCB, the interdigital filter easily becomes a stripline or hairpin. The insertion loss starts to rise, so best used with an amp.
RF (particularly as the freq. gets higher) closely resembles voodoo.
Perhaps you need a fresh chicken!
Theory and Practice.
In theory, theory and practice are the same.
In practice, they aren’t
A 1/4w stub. Just to show how easy it is.
I used RG59 so I could solder the copper at the short.
Edit: Used a camera this time
hahaha - while your electronic skills look decent - your photographic skills may have room for little improvement
i’d love to see your filter graphs taken with your homebrew analyzer - and moreover it were super-interesting to see your frequency scans with rtl_power with and without this filter!
The script you’re using to plot the spectrums scales the intensity range to the strongest signal, so in theory changing the gain should not really change the way the spectrum looks. The raw signal levels should drop evenly, then the script scales them back up again when plotting the output. In practice, since you’re seeing the FM and 950MHz bands stay about the same but the other signals fade as you drop the gain, that probably means that (as expected) the higher gain values are causing those strong signals to be over-amplified so you’re not measuring their strength correctly.
ah - ok - understand! to be honest i never looked into the script that plots the data but in the end the result remains - the three other filters clearly outperform the fa-filter at 960mhz AND the jetvision filter has the by far best overall performance within all tested filters AND much less passband attenuation than fa-filter …
tl;dr the filters are operating as designed?
and the fa-filter design does not fit to a large number of sites - especially outside usa don’t you agree it would be better if fa would communicate this behavior more clear to customers outside us?
I don’t know that teaching customers about how RF works is really in FA’s remit. The filter is sold at cost in the hope that it will be useful; if you know it’s not going to be useful in your particular noise environment after looking at the spectrum, don’t buy it! But it is hopefully cheap enough that “try it and see if it helps” is a good option for those who don’t know (and don’t have the technical ability to measure) what their noise environment looks like.
As a reference, perhaps half of our flightfeeder installs don’t need a filter, and the other half are OK with the FA external filter. I don’t think we’ve ever needed a tighter filter to get acceptable performance on a FF. Note that we don’t put many FFs in western Europe, which is where most of the 950MHz problems are.
i do understand your point - and flightawares goal is not to create some sites with best possible results but a huge number of sites with decent results at low costs and easy to setup. my tread is more from a tinkerers perspective - but i tried to be fair to each menufacturer. in fact i like all three companies - flightaware, jetvision and uputronics
anyways i hope that threads like this improve this forum and help those who love to tinker and reach out for best possible results at their sites. this was at least my goal …
btw did you see this video i posted in another thread - i hope this will be also possible with the lime mini?! LimeSDR all in one lab from Crowd Supply on Vimeo
perfect! muuuch brighter and sharper looks nearly 100% like my site looks besides i do not have this fancy filter. did you really get any advantage from using the fans and all the cabling ferrite. i gave things like that a try two years ago but never was able to find real improvement in any way?
Thx for you effort Tom. I have looked at filter response using a Airspy mini with the Spectrum Spy program with good results. Here are 2 screenshots of examples where I was looking at LTE filters to help my OTA TV reception out in the boonies. Looking forward to your response. While not “LAB” quality it gave me the information I was looking for.
The Pi’s live in the roof space where it’s currently 34C and 80% humidity (the sun set four hours ago), so I think the fans are a good idea.
I bought 20 clip-on ferrites for $2, so I’ve been putting them on everything in sight.
Have I noticed a difference? - No. But I figure even if they don’t help, they won’t make anything worse either.
One change I’m going to make is to put the shortest USB cable I can get between each Pi and its RTL.
One of my RTL’s failed today and it turned out to be a fractured solder joint on the USB connector (the strain relief pins hadn’t been soldered).
Edit:
I just found a datasheet for the ‘white’ clip-on ferrites I have
They ought to help with switch mode PSU noise and anything else under 1MHz, but at higher frequencies they have no appreciable effect at all (>100MHz).
It’s hardly surprising no one is raving about their benefits.
(the black ferrites have no part number, but are probably made from a similar mix).
hello dunnsville - thanx for reading
this is really cool how you and some others here in forum are already using sdr devices as analyzers! up to now the only tool i have used was the cheap nooelec dongle in combination with rtl_power. but some days ago i ordered a lime mini sdr and as soon as it arrives (end 2017) i was wanting to try out these things too.since your filter starts at 700mhz with blocking it looks that it really does a good job because the strong lte signal is completely wiped out in the second chart. the question is if always lab quality is needed. when you read anthonys post above (he’s the guy behind uputronics) they develop a new product and found out that there is significant difference between their lab measurements and real tests. and obviously your tests above also led to a good working solution!
my pis also live in the unused attic - only difference mine lay on an alu-box and yours on a cardboard-box in summer i have often about 35-40 degree celsius - but over the last 2 years never had a problem. not a sinle sd-card, dongle or rpi2 died. my dongles stick directly into the pi but when i tested the zero w i used a 4" usb cable and saw no difference at all …
