Page 5 of 5 FirstFirst ... 345
Results 41 to 50 of 50

Thread: NooElec with small antenna ---> FA Pro Stick + FA 1090 Antenna = worse results?

  1. #41
    Flight attendant Strix technica's Avatar
    Join Date
    Apr 2017
    Posts
    94
    Quote Originally Posted by RockRidge View Post
    I’m using the Pro Stick with FA filter and antenna and think that the next step is to to optimize my hardware and experiment a little.
    In case you've missed it elsewhere, I've observed before that the FA filter's bandwidth is too wide to exclude some of the major sources of noise, notably GSM. If it's a problem in Europe (GSM900), it'll be worse in North America (GSM950) because the spectrum is shifted 50 MHz closer to our frequency of interest.

    You might be better off to replace the filter with a decent cavity or SAW filter (more on why later). I haven't yet had the opportunity to experiment, so I can't make any specific recommendations, only point out that the issue exists.

    Quote Originally Posted by RockRidge View Post
    Being in the hills, line of site may be a little challenging and distance may not be the best measurement yet.
    So this complicates things a bit. At 100 nm out, a/c at cruise altitude make about 3° with the horizon which means most a/c will be on your horizon, just where the hills are. Distance is definitely not your best metric.

    You might consider looking at www.heywhatsthat.com to see what your terrain limits actually are. If you are running mutabilities's dump1090 and have its web interface configured, follow abcd567's instructions from step 8 to superimpose those limits on dump1090's output. If not, that same link explains how to install dump1090-mutability.

    Quote Originally Posted by RockRidge View Post
    I changed the gain setting from “max” to 30 at 0600 this morning and found that my max range stayed about the same but Tracks Seen dropped and my Single Level changed.

    Can you folks help me interpret the data and tell me how you assess your setting such as gain? Is the noise change a good or bad thing?
    The noise floor doesn't tell you a whole lot. It'll mostly be proportional to gain. Signal to noise ratio might tell you something of interest, but really only if you can control for distance for a given a/c. Track count is rather less important than the ratio of total tracks to single-point tracks or the absolute number of multi-point tracks because single-point tracks indicate that signal is too weak or the plane too far away to get more than a single position report from that a/c.

    You can either optimise for a/c count (or maybe message count) or multi-point tracks. What is optimal for one may or may not be what is optimal for the other, but multi-point tracks is probably what you should care about from the point of view of FR24 statistics.

    A couple of relevant points about radio theory that might help you understand what's going on:

    Superheterodyne radio receivers work by mixing a local oscillator (how you tune the radio), applying a filter around the IF (intermediate frequency) and then amplifying the result (your gain control). There are a bunch of reasons why it's done this way which all boil down to simplier, easier and therefore cheaper and better quality design. Nearly every radio receiver of nearly every type and application works this way.

    A radio receiver has a minimum sensitivity below which SNR goes to zero and a maximum rx power before the RF amp starts to become saturated. Note that the RF amp can be driven into saturation by any frequency within the bandwidth of the IF stage filter, not just the frequency of interest, and note that the IF filter bandwidth only has to be narrow enough to exclude the unwanted LO mixer products and is almost certainly much wider than the bandwidth of the RTL digitiser so you can't assume that the waterfall seen in SDR# tells you much about whether nearby frequencies could be driving the RF amp into saturation. The effect of RF amp saturation depends on modulation of the most powerful signal, but it's never good and you'll lose all of the weaker signals of interest. That's one of the two main reasons for using an external bandpass filter — to exclude hot signals in the vicinity of the signal of interest. These are fixed characteristics of the receiver; the only controls you have are LO frequency and gain.

    Given that there is some filtering in the tuner, hot frequencies far away from 1090 MHz oughtn't be a problem, but the FA Prostick has an additional wideband, low noise [RF] amplifier (LNA) between the antenna input and the tuner which introduces a new problem: intermodulation distortion. This is too complex to explain succinctly, but suffice to say that unfiltered, wideband RF amps can mix together frequencies, however distantly separated from 1090 MHz, producing new frequencies which can end up smack in the middle of the quiet band you're interested in. The hotter the other frequencies, the more noise introduced into your quiet band. That's the other reason why external bandpass filters are used before any active amplifier (and yet sadly, the Pro+'s BP filter is installed between the tuner and LNA ).

    Since the frequency is fixed, you're down from two controls to just one, gain. As you know, increasing gain can bring distant signals above the minimum sensitivity enough for it to be decodable, but it also increases noise. All things being equal, the SNR of signals that remain within the dynamic range of the receiver won't change so, for them, gain doesn't matter. Aside from intermod noise, the problem with choosing the right gain is balancing the quiet signals from afar with the loud signals from nearer transmitters. It helps to remember that a/c density increases with distance, which means that even with gain is set high enough to drive the receiver into saturation for a/c within 10-25 nm, there are many more a/c >100 nm than in the 10-25 nm range.

    ADS-B is not designed to even attempt to cope with simultaneous transmissions (collisions). Those messages simply get corrupted, so it doesn't matter if a nearby transmitter drowns out a distant transmitter because the distant message will be garbled anyway, even if it's not lost to saturation. Same thing applies for two distant colliding messages, and the probability of message collisions increases with a/c density and therefore with distance.

    If gain is too high, you'll pick up occasional messages from a/c too distant to receive reliably. They'll show up as single-point tracks, but they're pretty much useless for tracking so FR24 excludes them from your statistics.

    Putting it all together: when gain was higher, your track count was inflated by useless single-point tracks. The resolution of the plot is too poor to estimate the proportion tracks that were single-point, but it seems a fair bet that when you reduced gain, you picked up fewer useless messages but also reduced the number of good messages that were lost to saturation and/or intermod noise, hence why the message and a/c count went up even as track count when down.

    The number of messages > -3 dBFS together with your peak power plot are clues that you were losing a lot of messages to saturation. Remember that dump1090 only reports valid messages so, where gain is too high, messages > -3 dBFS actually represent weaker messages because the strongest were lost and weren't reported. More significantly, the plot of peak signal power didn't change appreciably when you reduced gain (though average power reduced, which is to be expected). This implies that the true peak was (and sometimes still is) somewhere up in saturation-land. There will always be some messages lost to saturation. The trick is to find the balance between losing messages to saturation and losing messages to insufficient gain.

    Automated scripts like the one abcd567 provided are a good starting point for figuring out gain, but there is no substitute for understanding what's going on and making a judgement call based on all available data and your specific circumstances.

    If you're running dump1090-mutability, are up to installing and configuring Munin, and want more and better data (Munin lets you zoom in on graph data and might keep better historical data), have a look at my Munin plugins. There is some experimental stuff in there for determining signal quality. Some of it still needs work, but I haven't the time to take the next step in some of that work just now.
    Last edited by Strix technica; 2017-07-28 at 16:24. Reason: correction of technical detail (saturation)

  2. #42
    Passenger RockRidge's Avatar
    Join Date
    Jul 2017
    Location
    Cloverdale CA USA
    Posts
    21
    Thanks for the very helpful advice and the great information.

    There are no nearby cell sites but I have a low powered network extender at my house. So far, I cannot see any difference having it on or off. I will read up on filters.

    I did use www.heywhatsthat.com to give me a good base for line of site. But, my property is so steep and there are so many hills nearby, I am going to need to move my receiver around to really know the best locations.
    Any threads or examples on remote placement of the raspberry would be welcome. I have a high spot 450 feet away that would be perfect.

    I will reread your post and study up on the great info.

    Thanks!
    Last edited by RockRidge; 2017-07-29 at 13:56.

  3. #43
    Flight attendant Strix technica's Avatar
    Join Date
    Apr 2017
    Posts
    94
    Quote Originally Posted by RockRidge View Post
    There are no nearby cell sites but I have a low powered network extender at my house. So far, I cannot see any difference having it on or off. I will read up on filters.
    If there's nothing of any strength near in the 900-1200 MHz range, your existing filter is probably good enough. Can't hurt to try if you want to eek out the last bit of performance, though, and can spare a bit of cash. Even some of the el-cheapo SAW filters from China on Ebay are probably better than the FA discrete BP filter.

    That said, intuiting about RF power is tricky. Don't forget that power falls off with the square of distance, so even though a cell tower may be in the 10s or maybe 100s of watts (in aggregate), your 50mW transmitter is much, much closer. A difference of, say, 10 metres to 1 km is a factor of 10^2 which, when squared, makes a difference of 40 dB! (It begins to get a bit complicated when you consider how that power is distributed across space, but I've probably made my point.)

    Some TV and FM radio stations can be surprisingly powerful. They may not be nearby, but they can be 10s-100 kW. Our nearest DVB-T transmitter is 180kW per MUX (and last I knew there were about 8 or 9 of them), though it's in the middle of nowhere. Droitwitch (still!) transmits 198 kHz at 500 kW. (Fascinating history behind that involving the Cold War, but that's another story.)

    Quote Originally Posted by RockRidge View Post
    my property is so steep and there are so many hills nearby, I am going to need to move my receiver around to really know the best locations.
    No question, but hills are a bugger. Experimentation is about all you can do, maybe with a gel-cell and a DC-DC converter in the first instance. this'd be ideal.

    I guess I got lucky to end up in a place near the top of a hill otherwise surround by flat land for a considerable distance. Still, my stats aren't great because I haven't got my antenna on the roof yet.

    Quote Originally Posted by RockRidge View Post
    remote placement of the raspberry would be welcome. I have a high spot 450 feet away that would be perfect.
    Running mains is probably not wise. Power over Ethernet (extra hardware required) might do, but be aware that not all PoE solutions are created equal and if you're finding your pi to be unreliable, you might have to augment filtering with extra inductance and capacitance. Look up, hah, construction of a pi filter (no connection with the Raspberry Pi).

    The alternative is to run some low-ish voltage (no more than 50V, and not with any significant current capability) cabling out to your pi and then downconvert to 5V DC), which is a massive PITA. Don't forget a minimum of IP67-rated waterproofing for all parts.
    Last edited by Strix technica; 2017-07-29 at 14:59.

  4. #44
    Captain abcd567's Avatar
    Join Date
    Sep 2013
    Location
    Toronto CYYZ
    Posts
    2,862
    @Strix technica:
    A hobbyist "keithma" has tested Flightaware filter using his VNA, and today has posted results here:
    http://discussions.flightaware.com/p...4.html#p209274

  5. #45
    Purser
    Join Date
    Feb 2012
    Location
    ESGG1
    Posts
    190
    Quote Originally Posted by Strix technica View Post
    In case you've missed it elsewhere, I've observed before that the FA filter's bandwidth is too wide to exclude some of the major sources of noise, notably GSM. If it's a problem in Europe (GSM900), it'll be worse in North America (GSM950) because the spectrum is shifted 50 MHz closer to our frequency of interest.
    No, it's the other way around. GSM850 in North America is further away, and also reversed duplex. GSM900 has BTSes up to 960 MHz, where the FA filter is wide open.


    Also summarized here

    http://discussions.flightaware.com/p...1.html#p203201

    /M
    Last edited by MrMac; 2017-07-29 at 19:52.
    F-ESDF1, F-ESGG1, F-ESGP1, F-ESNK1, F-ESNV2, F-ESNV3 F-ESSL4, F-LFMN3
    P-ESGR, P-ESIA, P-ESIB, P-ESGF

  6. #46
    Flight attendant Strix technica's Avatar
    Join Date
    Apr 2017
    Posts
    94
    Quote Originally Posted by abcd567 View Post
    @Strix technica:
    A hobbyist "keithma" has tested Flightaware filter using his VNA, and today has posted results here:
    http://discussions.flightaware.com/p...4.html#p209274
    I've seen plots like this before, and it kinda proves my point: the lower knee is at around 925 MHz and considering GSM900 downlink reaches up to 960 MHz, the FA BP filter is not sufficient to suppress GSM900 signals.

    It is doubtless even more of a problem with North American GSM950, so I don't know why he says it has good attenuation on cell frequencies. It's not as if 2G or 3G have gone away, even if most are using 4G these days.

  7. #47
    Flight attendant Strix technica's Avatar
    Join Date
    Apr 2017
    Posts
    94
    Quote Originally Posted by MrMac View Post
    No, it's the other way around. GSM850 in North America is further away, and also reversed duplex. GSM900 has BTSes up to 960 MHz, where the FA filter is wide open.
    I could have sworn that I'd seen references to GSM950, including in the US, but a new search gets nada so either I remember wrong or what I read was a typo.

    In any event, the FA BP is no good for Europe, Australasia or anywhere that uses GSM900.

  8. #48
    Purser
    Join Date
    Feb 2012
    Location
    ESGG1
    Posts
    190
    Quote Originally Posted by abcd567 View Post
    @Strix technica:
    A hobbyist "keithma" has tested Flightaware filter using his VNA, and today has posted results here:
    I summarized the FA filter problems back in the beginning of 2016, and posted my VNA measurements:

    https://forum.flightradar24.com/thre...ll=1#post79610

    /M
    F-ESDF1, F-ESGG1, F-ESGP1, F-ESNK1, F-ESNV2, F-ESNV3 F-ESSL4, F-LFMN3
    P-ESGR, P-ESIA, P-ESIB, P-ESGF

  9. #49
    Purser
    Join Date
    Feb 2012
    Location
    ESGG1
    Posts
    190
    Quote Originally Posted by Strix technica View Post
    In any event, the FA BP is no good for Europe, Australasia or anywhere that uses GSM900.
    It works reasonably well if the interference is from FM or DVBT transmitters, but not against GSM900 downlink. So it varies a lot with location.

    /M
    F-ESDF1, F-ESGG1, F-ESGP1, F-ESNK1, F-ESNV2, F-ESNV3 F-ESSL4, F-LFMN3
    P-ESGR, P-ESIA, P-ESIB, P-ESGF

  10. #50
    Passenger
    Join Date
    Feb 2015
    Posts
    28

    opinion

    Hi everyone,

    I'm using this USB dongle for FR24 feed: https://postimg.org/image/rbo4jtsrh/
    Also I'm using the FA antenna 3m high above ground. I got around 200nm range with this setup.
    I wonder if it will be better (to increase the range) to purchase a FA Pro Stick plus and replace my USB SDR dongle or buy any LNA+bandpass attached to my actual dongle?

    Any comment on this please?

    Thank you and best regards!

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •