[TowerTalk] Actual LP Performance vs Tribanders

[Jim Lux]

At 02:08 AM 6/28/2004 -0400, Larry Phipps wrote:

>Jim, I'm not that familiar with the beacons, but since your post I did a 
>little research. There are a couple of major problems.
>
>First, the transmissions are very short... there wouldn't be time for more 
>than one sample per beam heading.. and it would take almost 2 hours just 
>to gather the samples for one rotation (36 samples). The signals are going 
>to be all over the place during that time frame... and that doesn't take 
>interference into account. Timing would also be critical... your computer 
>clock would have to be dead nuts on. There's also really no accurate way 
>to correlate the signal strength to anything else minute-by-minute, so the 
>levels would be more or less meaningless.

Yes, I agree that you'd have a problem doing the measurement in one pass 
through.  You'd have to do a statistical approach over many days and hours.


Timing you can get from the signals themselves, and the PC clock is good 
enough to sequence it. As long as you're within a few seconds of the "real" 
time, it's fairly straightforward to track (programs like BeaconSee do it, 
for instance).

You could correlate the signal strength on the antenna under test to a very 
short monopole antenna  or small non-resonant loop (which will be fairly 
consistent in azimuth and elevation response).




>Even with a 20 minute continuous carrier at 100W, I doubt the received 
>strength of the beacons would be enough to be useful for plotting the 
>pattern of a beam with 30dB F/B ratio.  You would need a stable signal 
>about 50dB above the noise floor... probably something around S9... and 
>you'd have to listen to make sure there is no interference while the 
>samples are being taken.

Kind of depends on the accuracy you require.  If you want tenth dB accuracy 
on something that is 30 dB down, it would be a challenge (I doubt there are 
many antenna ranges or network analyzers that can do that well).  If you 
are willing to tolerate 1dB errors (out of 30), and, say, 0.5 dB out of 
10dB, I think you could do it with, maybe, 10-15 dB SNR. Depending on the 
integration time.  If the beacon is sufficiently narrow band (which the 
NCDXF beacons are, during the test tones), your measurement bandwidth 
(which sets the SNR) can be quite narrow (probably limited by ionospheric 
doppler spread). Maybe 5-10 Hz?

Using  the noise level in VOACAP for residential areas (-145 dBW/Hz), 
you're only going to be seeing something like -105 dBm noise 
floors.  Running a quick VOACAP from a monopole in Tangier to a swwhip Los 
Angeles, (june SSN=100) it looks like you'll have SNR (in 1 Hz) of above 15 
dB for about 4 hours.  That's going to be a bit marginal.

However, as you point out, big broadcast stations would be a better bet.




>My tests used commercial stations with BIG signals that transmit 
>continuously, like WWV. I was testing a very broadband beam, so actual 
>frequencies weren't that important. Testing a LPDA or SteppIR should also 
>work OK with my software. You could probably make it work with a ham with 
>a strong signal though, who would be willing to make several  5 minute 
>continuous transmissions while the data was gathered.
>
>If you happened to have a neighbor within a few blocks, you could get 
>meaningful results, especially if your antennas are high. You could use 
>the TRX-Meter utility that comes with TRX-Manager since only one pass, and 
>one sample per 10 degree heading would be needed. I got the idea for my 
>program from trying TRX-Meter, but added the ability to integrate multiple 
>samples over time to smooth out the effects of fading on skywave signals. 
>Further away than about 10 wavelengths would give you a nice pattern, but 
>I doubt that it would correlate into similar skywave performance.
>
>Larry N8LP