Don’t forget to install graphs1090 as well if you haven’t already done so, it’s by far the best tool for monitoring your feeder.
2 Likes
Looks OK. You could probably turn the gain down a notch or two without losing any range.
Mine is set to 14 and looks like this:
2 Likes
I still wonder why most of us use a two stage 38dB LNA and then turn down Airspy’s build-in LNA about 19dB meaning that a one stage 19dB LNA with the Airspy at LNA level 21 would do the same job with less noise or am I missing somewhat?
You might want to read up on the Friis Noise Equation which is a method of describing the contribution of each stage of a receiver chain to the overall noise figure. The short version is that the first couple of stages in the chain have a much larger impact on the overall noise figure than later ones, so using a good LNA next to the antenna gives the best performance.
If we take a very simplified model, with some arbitrary figures:
Assume we have an antenna feeding a length of coax that has 3db loss and a noise figure of 3dB, feeding a filter with 2.5dB loss (like the FA one, with 2.5dB NF) the receiver which has its own internal amp set to 40.5dB gain and let’s guess a noise figure of 1dB. The overall gain is 35dB with an overall noise factor of 6.5dB.
If we now put an LNA such as the rtl-sdr one in the chain, which has a gain of 26dB and noise figure of ~1dB, before the coax, then reduce the gain on the in built amplifier to 14.5dB so that the overall gain is the same at 35dB. The noise figure has dropped to only 1.03dB.
There is a calculator here which you can put your own figures in to see how they affect things. In reality it’s more complicated since amplifiers and filters are not perfect, receivers usually have a few more stages than just an amplifier, and an amp like the rtl-sdr one has several stages of filtering and amplification. It should serve to illustrate the point though.
There’s also this article which talks about positioning of amplifiers and the effect on performance.
10 Likes
That’s why I said that the rule of thumb, in analog signal path, is “amplify before attenuation”.
I keep the Airspy at 21. There is no need to try to keep the samples at -3dB because that “measurement” is not correct for Airspy.
1 Like
That isn’t a reason to keep the gain crazy high.
You can get overload with the airspy as well even though it’s rare.
It might be worth experimenting. I got a measurable increase in performance by reducing the gain on mine a few months ago. I did have it at 17 and reduced it to 14. It wasn’t immediately obvious from the graphs however, but it was clear when comparing relative performance to nearby receivers.
1 Like
Define “crazy high”.
I have a very long coax cable feeding the Airspy, so that might make a difference too.
I just wanted to say that the graph1090 doesn’t seem to be useful. I did vary the gain, reset the airpspy service, and watched in real time the message numbers for change.
I’d say if you don’t see a decrease it doesn’t hurt to reduce the gain 
With airspy i typically just tweak so the weakest signal in the graphs is at -30 dB.
Well actually i even go a bit lower than that for me personally but that’s what i like to recommend.
Setting the weakest signal anywhere from -20 dB to -30 dB is what i’d call reasonable.
If the average weakest signal is above -20 dB on the airspy … i’d call that crazy high gain 
Just don’t think there is any advantage of having a signal stronger than that in regards to the decoding.
2 Likes
Well that’s not crazy high gain at all, rather near the bottom end of typical gain settings i’d say.
Gotta get more amplification up your tree
I remember you having much higher signal levels.
2 Likes
I was using a cheap eBay LNA, with two filters, and the first stage replaced by me with an SPF5189Z.
Then I replaced it with RTL-SDR Blog LNA and indeed the signal got lower.
I might revert to my former LNA, just need some will to get inside that outdoor box. Plus… it doesn’t look pretty.
Since you reminded me of that, I’ll replace it 
Yeah you’re loading down the output signal with that modification.
The piece of wire going to the coil surely matters for 1090 Mhz.
Or maybe i’m misjudging it.
Maybe just add one of the PCB biastees instead of making your own bias-t on that board, should improve signal integrity maybe? (to exfiltrate the DC you’re stepping down)
I was more joking, the signal is probably still perfectly sufficient.
Didn’t you get an improvement switching to the rtl-sdr LNA?
Sadly the currently available rtl-sdr LNAs seem to have wider SAW filters and depending on the interference you face aren’t as good anymore as the ones produced half a year ago and earlier.
1 Like
I un-coiled that bit in purpose, to create an air inductance for the 1GHz blocking. Kept it away from other components to minimize capacitance coupling.
The ferrite core might not be useful, but it’s there…
I need that 5V stabilizer because I push 7.5V on my 45m long coax to compensate for any voltage drop.
Bought mine on Oct. 26, 2019. It has SF9175 filters, alleged passband 8 MHz.
I ordered the same board for my experiments last week!
I suppose you use a low drop regulator instead of a plain 7805?
The filtered RTL-SDR LNA is using a two stage MGA13116 with an amplification of 38dB minus 2x 5dB insertion loss for the saw filters equals about 27dB remaining amplification.
I have a bias-T and a variable supply, so I can feed as much as I need to get that 5V…
Used a LDO inside the RTL-SDR one, because of space.
The generic board SAW filters are TA0970A, same grade IMO with the others.
The SPF5189Z has a gain of 17.9dB at 1GHz.
My RG6 cable attenuation is -5.9db / 100 foot, for 150 ft is approx 9dB. Losses in change-over connectors and gas protector probably add another 1-3 dB.
9dB loss is huge, at least you have the LNA in front and that’s something I can never do with 105dB@960MHz blasting from a nearby cellular phone tower!
Not really. Probably I don’t have any strong interferences close-by to make a difference.
That’s the difference between the two LNA:
