SFP-5189Z specs are linked above. Is GaAs pHEMT, suppose to have 0.6dB at 0.9GHz and 18.7dB gain at same 0.9GHz.
Well just an update, 24 hours and Iāve managed to resist fiddling with things I donāt understand and concentrate more on things I do. So I thought I could quickly rig up a pole outside wedged between the house eaves and a balcony. with a PVC pipe clamped to it at a sensible offset I have a PVC pipe about 6 feet above the roof. The roof slopes so Iām a couple of feet below the ridge to the south.
Anyway I lashed up a quick ground plane. A bit different to the one wiedehopf pictured before. Iām using RG213 with 7 strands of the inner wrapped around the shield and soldered making 14 radials. The radiator I left with the dielectric in place so its a bit shorter. after trimming I applied plenty of hot glue to the top and around the radials and the gaps in the outer. A few coats of spray laquer and a check to make sure it was still resonant in the right place, s11 >25dB @ 1090MHz.
I have about 5m of RG213 down to the Airspy connected with a short 20cm length of RG178.
From an indoor message count of around 180-200 and a range of 90-100nm it immediately shot up to around 500/sec and a range around 160nm.
I have some 6 inch wide copper flashing so Iām going to try a cone next and make up something more robust. This will then go on a PVC pipe right at the apex of the roof and will be in the clear 360 degrees.
What are the lengths of the radiator and radials?
S.
I didnāt measure them after trimming. I started out with the radiator around 65mm, I knew it would need shortening from that, and the radials at 90mm.
The radials I cut to 80mm and the radiator I think ended up somewhere around 58mm.
I used a VNA to trim the antenna in almost real time and am confident in the results. If you use RG213 then these numbers should get you close enough, the antenna is quite broad band so even if you are not right on the nose it should still work pretty good.
Removing the insulator might give you a longer whip at resonance.
That in turn might be higher gain, but iām not completely familiar with the theory.
Of course you can also try building more complicated antennas with an analyzer available.
It must be an optical illusion.
They look much longer than that.
S.
Yes, I think it could be that with a higher effective velocity factor for the radiator then it might improve performance however like with noise figure itās reasonable to assume that the small difference might not be practically noticeable.
The picture was taken from the ground looking up at about 45 degrees. Antenna is about 30 feet or so. Camera is just an iPhone zoomed in. I agree the radials look long and whilst they might not all be absolutely equal to a mm they are fairly close and all around 80mm.
You could make another antenna:
Start the whip and radials at the theoretical quarter wave length.
Then try adjusting the SWR/resonance by bending the radials up/down.
I did that as a part of the process, sort of. Moving the radials up and down affects the resonant frequency minimally but plays a much bigger part in varying the resistive part of the feedpoint impedance. So to achieve a desired feedpoint impedance of 50 ohms resistive. The radiator should in any case be slightly shorter than that theoretical quarter wave due to āend-effectā.
I installed one of those LNAs and a 3.5v powered bias tee and lost pretty much all range. I played with the gain to see what I could do but nothing seem to help.
I tried t he LNA at the antenna and pi locations and it was the same result.
Iām running 3.7.1 on stretch 9.4 with a FA usb pro and FA antenna.
The blue prostick has another amplifier in it which will be completely overloaded.
You can also try a gain of 5, see if it makes a difference, but the if the pro stick internal LNA is overloaded you canāt change that with gain.
Youād have to attenuate the signal before it goes into the blue prostick+.
Could also be that your LNA is defective.
Do you have a regular rtl-sdr stick to test with by chance?
Also just try 5 V, maybe the bias-t isnāt working well.
So it could also be the bias-t that is defective, but itās unlikely.
Watching not to have too much gain is all important. The risk of overload destroying a perfectly good system is very real.
As a first step its important to realise that signals not picked up by the antenna cannot be magically produced by amplification. If they are not received by the antenna then no amount of amplification will help. Secondly if the feeder loss from the antenna to the receiver is enough to reduce the already weak signals below the background noise floor then no amount of amplification after the feeder can recover them.
There are two occasions when an amplifier might help. If the receiver noise figure is compromised such that sensitivity is not as good as it can be then a quieter amplifier can increase signals to the level the receiver can process and also an amplifier at the antenna to increase signal levels such that even with the following feeder loss they are still recoverable by the receiver. This second case also fixes the first problem as well.
Well after I fixed my little software issue the thread seems to have drifted to other things and this probably warrants a new thread but on the off chance this has been an already recurring theme I will ask here and be happy to be pointed at other resources.
On any given screen full of data such as I am now seeing with view1090-fa or the table view of Skyview I see a few RSSI values around 0 and a few as low as -20 - -25 so I suppose the range is somewhere in that order. My interest is to understand what this means in absolute terms as it seem at first glance rather arbitrary. A 20dB change represents 100 times and seems small for a receiver.
Is there perhaps a reference that explains how RSSI is being used in this system?
Thanks
Martin
RSSI readings from airspy-adsb arenāt measuring the same thing as with the rtl dongles which are dBFS. Iām not sure what the airspy values are referenced against, but in general it shows higher numbers than an rtl dongle, and signals at 0dB arenāt necessarily overloading. I am getting many signals at that level but do not get any drop outs as a result. With the rtl dongle itās very important to get the gain correct to avoid overloading, but the airspy is much more tolerant of strong signals.
Hereās my signal strength graph with the gain set to 19. I donāt seem to be getting any adverse effects from it, even though the signals appear to be squashed up against the top. If I reduce gain to get a more even spread, it reduces performance so I stopped worrying about it and left it as it is.
Itās not completely unrelated, though the calculation is different.
Most significant factor is probably that the airspy has an analog front-end thatās not as easily overloaded.
And the ADC doesnāt mind going to full scale 
This is a most crucial element in engineering 
I suppose it could stil be relative to full scale but just in what the airspy can accept which may cover a different range.
Please do explain how you produced the graph, It gives a good picture of the signal range across a good amount of time.
Martin
Excellent, many thanks 
This is the HEMT LNA that RTL-SDR uses for their one stage amplifier (I have one)
For their 2 stage filtered LNA they use the MGA-13116 (already EOL)
Uptronics is using the PSA4-5043+, about the same specs as the SPF5189Z but the IP3 is worse.
Iām planning to test with the 2 stage filtered LNA but Iām probably going to bypass the filters because I already have a cavity filter.
I tried 18 to avoid too many āstrongā signals. Then 19. Then 20. I am at 21 andā¦ even with huge amounts of āstrongā signals, the decoding is still going well. IDK.