Sporadic Meteor Scatter rate?

[james_screech]

I've just got a meteor radar scatter system setup using Graves, however I'm getting very few detections (1 or 2 a day) from sporadics. I thought I would be getting more than this an hour. What is the rate I should expect outside of showers?

Thank you
James

[BiggarDigger]

That's a difficult one to answer James.

It depends on the sensitivity of your receiving system and on the detection algorithm you are using. 

If you are receiving only 1 or 2 per day, there may be a problem with your antenna or feeder as that rate is two orders of magnitude lower than the rate I detect per day.  I often exceed 300 echoes logged per day, but by far the majority of them are very short duration, so you also need to check the algorithm and adjust any parameters that may be filtering out responses.

This is a chart of duration of returns logged so far today from my site:

What system are you using?

Richard

[james_screech]

13 hours ago, BiggarDigger said:

That's a difficult one to answer James.

It depends on the sensitivity of your receiving system and on the detection algorithm you are using. 

If you are receiving only 1 or 2 per day, there may be a problem with your antenna or feeder as that rate is two orders of magnitude lower than the rate I detect per day.  I often exceed 300 echoes logged per day, but by far the majority of them are very short duration, so you also need to check the algorithm and adjust any parameters that may be filtering out responses.

This is a chart of duration of returns logged so far today from my site:

What system are you using?

Richard

I'm using a NooElec NESDR SMArTee  RTL-SDR with a commercial 4 element 2m Yagi in my loft. I was using Ian's configuration of Spectrum Lab (https://www.blackwaterskies.co.uk/2019/01/radio-meteor-scatter-improved-spectrum-lab-conditional-actions/). Though shortly after making this post I noticed a meteor on the waterfall that was not registered. So switch back to the MetScat_starter_v1.USR configuration, slightly modified to give a 75% overlap in the waterfall and a bandwidth of 2kHz starting at 143.049 KHz. However even with this I've had no detections over night.

[wxsatuser]

Not done anything lately but yesterday afternoon I had the IC7100 on GRAVES for about 2 hours.

I could hear it on occasions direct but no meteors in two hours, normally I would get a few pings.

GRAVES is not a good direction for me as there is 400feet of chalk in the way.

[BiggarDigger]

My count stands at 115 since midnight, so a few follow up points and questions:

Double check the coaxial connection to the antenna just in case there's a fault.

The 4 element yagi will be starting to get sensitive to its environment.  Is there any obstruction or nearby large metallic objects in the loft (like a water tank).  The beamwidth will be narrower, more sensitive and more difficult to align in the loft.  Is there any chance you could mount it outside, even temporarily to see if you receive echoes?

Do you see sporadic echoes in say SDR# or HDSR (without Spectrum lab driving the dongle?).

Do you receive any amateur radio transmissions: the GB3VHF beacon should be easily detected on a ground wave at Bedford.  Use that to verify the accuracy of the dongle tuning.

Richard

[james_screech]

I'll double check the cable tonight when I get home. There are some metallic objects in the loft (boil flue, hot air heating ducting), the yagi is pointing away from the flue and is laying on fiberglass insulation on top of the ducting. Unfortunately there is no way I can mount it outside. The house is on reasonably high ground for the area and the nearest hills in the direction of GRAVES are about 8 miles away.  I did check the tuning using a FM transmitter and it looked good, less than 100 Hz off.

James

[The Admiral]

James, I've not done any meteor scatter work for the last 3 years, and even then I only monitored the showers. Initially I used a home-made 3-element Yagi in my loft, and it worked reasonably well, feeding in to a FunCube dongle. I must admit that there is not a lot of metalwork in my loft, save for the tangle of power cables, and I did mount it on a short pole to keep it off the rafters. I was getting quite a bit of interference. I later changed to an Innovantenna 3-element LFA-Q with some success, possibly higher sensitivity than the straight Yagi. My frequency offset was about 190Hz.

I've just set it running and over the last hour it's recorded 9 hits, one of which was a decent one, rather than the usual blip:

I'd also strongly recommend that you try to detect the VHF beacon GB3VHF (http://www.gb3vhf.co.uk/) on 144.43MHz. That's there on a cycle all the time and so you can confirm that things are working without waiting for a random event to occur.

Good luck

Ian

[james_screech]

I'm getting a faint signal from GB3VHF (just over 1000 Hz below), not much more than an enhancement to the background noise though. I would expect the direction to be OK as it's not far off my line to GRAVES.

 

 

 

[BiggarDigger]

That looks a little weaker than I might have expected from Bedford (I lived in the Milton Keynes and Northampton area for 15 years), but it's difficult to know for sure as each site is different and the beacon has changed a lot since I was active on 144MHz.  You could try adjusting the antenna matching or tuning, if you have any, to see if you can obtain a stronger signal.  Also try lift it away from the metallic ducting as that could easily be de-tuning it - a bit of heavy duty string dangling it from the rafters will suffice for testing.

Assuming however that is the beacon, and you should be able to tell by listening to the audio via Spectrum Lab, then is shows the tuning of the SDR is ok, which is good.

 

Richard

[The Admiral]

1 hour ago, james_screech said:

I'm getting a faint signal from GB3VHF (just over 1000 Hz below), not much more than an enhancement to the background noise though. I would expect the direction to be OK as it's not far off my line to GRAVES.

 

Hmm, I would have hoped for a better signal than that. You're on the centre-line of the North-pointing antenna of GB3VHF, and I'm on the CL of the North-Westerly one, and my signal is somewhat better. Of course, I'm not sure what effect the terrain between you and the Tx is. Here's my trace, for what it's worth:

 

Ah, Richard's pipped me to the post!

Ian

[james_screech]

Thanks for the suggestion,  I've adjusted the trimmer and tied the antenna up with bits of string and the signal from the beacon shows a huge improvement. Now to try for the meteors. 

James

[BiggarDigger]

That sounds much better James.  Fingers crossed you should now be seeing some background rate meteoric signals.

You may still need to tune the conditional actions script to optimise detections.

Richard

[james_screech]

Well overnight I had 9 events in the log files  all with a duration of 0 or 1 and no screen captures.

21:03:23,1,-105,-117.1,2868.6,1
22:10:39,2,-104.5,-117.7,2853.6,1
22:18:32,3,-103.5,-117.9,2861.5,1
22:22:40,4,-107.2,-118.1,2866.5,1
22:45:55,5,-106.4,-118.1,2852.5,1
00:06:43,6,-104.6,-117.9,2889.3,1
02:29:16,1,-96.6,-117.8,2025.3,0
04:26:11,1,-96.7,-117.2,2002.5,0
05:19:19,1,-98.1,-118.1,2000.9,0

The last three might have been meteors going by the frequency. This is getting as frustrating as cloud when you've bought a new telescope.

James

[BiggarDigger]

I assume the log reads as follows:

Time, event count, peak signal, noise floor, peak signal frequency, duration.

If so, the logged event count looks odd for the last three.  If the assumption is correct, event count should increment once per echo logged, as per the first six events, then resetting to zero on midnight (which it appears not to do - the 6th event is after midnight).

What frequency are you tuned to on the VFO? - that will tell you if the 2kHz signal is likely to be GRAVES.

Also, what units is the duration measured in and can it be made more granular?  Depending on the continuous actions script, you may have 9 events: 6 between 1 and 2 seconds duration and 3 events between 0 and 1 seconds, or something similar.

However, it could well be that the script is not functioning as it should - no screen captures and odd looking event counts.

I'm not sure if your script takes into account that Spectrum lab requires double backslashes \\  to represent a single backslash in the path variables, i.e.  C:\Users\Richard\Desktop is represented as C:\\Users\\Richard\\Desktop.  A typo there could easily break the ability of the script to write screen capture to the disk.

And yes, I completely sympathise, sometimes it can be very frustrating getting everything setup just right with no reference to work against: but it does looks promising, and perhaps with a few tweaks and a bit of debugging you may have a functional system.

Richard

 

[james_screech]

The event count may be odd as I stopped and re-started several times and deleted some spurious events that occurred when re-starting.

I'm tuned to 143,048,000 Hz

The configuration is based on MetScat_starter_v1.USR that I obtained from the BAA radio section website (I think). I changed the audio settings as my SDR provides samples at 2,400,000 sample per second. I've set the FFT input size to 524288 and the Spectrum waterfall scroll interval to automatic with 75% overlap.

The conditional actions are as follows:

if( initialising ) then A=0:Sig=0:C=0:E=0:t2=0:Z=0:low=1000:high=3000:MName="Meteor":Mtime=0:MSig=0:MFreq=0:Freq=0:MNoise=0
if( always ) then A=noise(low,high):Sig=peak_a(low,high):Freq=peak_f(low,high)
if( always ) then q2=str("YYYYMMDD",now):h1=str("mmss",now):h2=str("hh:mm:ss",now)
if( Sig>(A+17) ) then C=C+1:D=D+1:timer0.restart(2)
if( C=1 ) then Mtime=h2:MName="Meteor"+h1:MSig=Sig:MFreq=Freq:MNoise=A
if( C=3 ) then timer1.start(30)
if( C>Z ) then Z=C
if( timer0.expired(1) ) then E=E+1:sp.print("                 Meteor ",E," Dur",C):fopen2("c:\\Spectrum\\"+"event_log"+str("YYYYMMDD",now)+".txt",a):fp2(Mtime+","+str(E)+","+str(MSig)+","+str(MNoise)+","+str(MFreq)+","+str(C)):fclose2:C=0
if( timer1.expired(1) ) then capture("c:\\Spectrum\\screenshots\\"+"event"+str("YYMMDDhhmmss",now)+".jpg",75)
if( val(h1,"####")=5959 ) then t2=t2+1
if( t2=1 ) then C=0:E=0:Z=0
if( val(h1,"####")=0000 ) then t2=0
 

The odd thing is I did initially get a few correct meteor detections with screen captures, however I then tried someone else configuration (IanL on this forum)  and now nothing, despite a complete re-install of Spectrum Lab.

 

James

[BiggarDigger]

OK, that makes much more sense James: I assumed you wouldn't be up overnight starting and stopping the script.

So, anyway, it looks like the trigger for the event is set at 17dB above noise, that may be quite high, especially out of a shower.  Try changing the line

if( Sig>(A+17) ) then C=C+1:D=D+1:timer0.restart(2)

to

if( Sig>(A+12) ) then C=C+1:D=D+1:timer0.restart(2)

That's 12 dB above noise.  Experimentation is important here, the lower the value, the more false triggers you may experience.

You can, if you want, also tweak the "low" and "high" values to be tighter to your expected 2kHz nominal value, again reducing the likelihood of spurious triggers.  At present it's quite wide looking over 1 to 3Khz for anything above 17dB SNR.

The other thing to check is does the c:\spectrum\screenshots folder exist after the re-install of Spectrum Lab?  If it doesn't the screen capture will likely fail.

There's a whole discussion to be had on what is a spurious trigger.  Some people like to have close to zero background and only capture the large, loud and long echoes during a shower.  Others (like myself)  like to see the background rate too.  I have a lowish signal to noise trigger limit but a tight frequency band.  There's probably no right or wrong, just different.

Richard

 

/edit

I just looked a bit more closely at the script James.  You could also tweak the timer values if necessary.

timer0 appears to be used to prevent multiple triggers on events that follow each other less than 2 seconds apart.  Probably OK but you can experiment with the value.

timer1 appears to be a delay of 30 seconds after the trigger before a screen shot is taken - depending on your waterfall plots, that may never capture a signal.  Set the delay to be suitable for your waterfall scroll rate.

Hope that helps.

 

[james_screech]

Thanks for the suggestions, I found a more fundamental issue with the FFT setup and when the conditional actions run and I'm now detecting week events on average every 12 mins. I've tweeked the trigger down a bit but going too low gave me too many false triggers. I've also changed the screen capture trigger so that it always captures when C=1. 

I'll probably tighten up the low/high values as you suggest.

Thanks for the help

James