BiggarDigger

Excellent! It will be fascinating to hear about the development of the system and your correlations against video detected meteors, particularly those over central England versus central or southern France. Have you been able to measure or calibrate the radiation pattern of the array - that will yield important additional information to feed into the cross channel correlations. It is similar, but my contention is that the head echo exhibits a slow speed Doppler shift, suggestive of a sub-orbital speed object as opposed to a meteor travelling perhaps an order of magnitude faster. Do the majority of your captures exhibit a slow head echo? In general, one needs to be a bit careful about the spectral interpretation. With not much resolution presented in the time domain, the trace looks like it has the typical spikes occurring at the phase switching points of the radar array. These spikes are I believe caused by aliasing in the FFT employed in Spectrum Lab. As such, they do not necessarily tell us much about the nature of the object causing the echo or scattering, but indicate the response of the FFT to a high amplitude impulsive change when the transmitter phased array switches. Their presence however does indicate a scattering or reflective medium at a stationary point in the sky, as one would expect from an over dense plasma field. Certainly, I agree that it is an outlier on the SNR chart that you present, and all the more fascinating to understand what it may be. Richard

That's a nice antenna array Mike and could be very useful in calculating further parameters, if you can correlate against known trajectories and signals, for example from the ISS. The orthogonal and vertical components may yield interesting results. The mast head preamps will assist with weak signal detection and overcome feeder loss too. I was perhaps not specific enough when I mentioned space debris. In this context, I meant incoming man-made orbital debris entering the upper atmosphere at sub orbital velocities of maybe 7km/s as opposed to a meteor which may have a velocity of 40-60km/s. The radar itself is presumably tracking orbital parameters of objects in low earth orbit and hence looking for direct scattering off the object itself, such as we might see on the ISS from the radar. However, your observation has a head echo suggestive of low incoming velocity followed by an ionisation trail, which would be reasonable to ascribe to an over-dense field in the ionosphere caused by the incoming object. In other-words, I suspect it was in the atmosphere at a similar altitude to a meteor and not in orbit. Due to the slow Doppler it was possibly not meteroic, but a de-orbiting piece of orbital debris. I don't have the mathematics to verify, but it seems reasonable that the SNR would be related to the intensity of the ionisation and that in turn related to the velocity and mass of the object. All other things being equal, the intensity of the SNR and slow Doppler suggest to me that it was a high mass object entering the upper atmosphere at sub orbital speeds rather than a grain of sand entering at cosmic speeds. Of course, I could be wrong and would be interested in hearing opinions. Richard

I didn't capture that one Mike, but my detector has been offline for a little while as we have been doing some work on the house. Two thoughts come to mind: the Doppler shift on the head echo is slow. Although it's difficult to estimate the absolute velocity from a single observation, in this case, you can clearly see the phase switching every two seconds in the head echo indicating its frequency is changing slowly relative to time. That is suggestive of a slow entry, meaning space debris rather than meteoric. The second thought is related to the location. You speculate it being close to your site, presumably due to the very strong signal to noise ratio. However, if near overhead, your antenna would be unlikely to see a common volume to the GRAVES radar. More likely, is that is is south of the radar and the debris has a large mass resulting in an intense ionisation. This could build upon an already elevated ionisation of the E-layer that occurs during summer months and in particular late June/early July. Sporadic -E propagation can be seeded by incoming debris or meteors and perhaps this was the case here? Richard

I'm working mobile and will be away until sometime on Saturday now, so may not be able to add much detail, but I use a slightly modified (tuned for my setup and site) version of the detection script presented by IanL here:

Yes, that looks very much like the ISS scattering off the western facing frontal lobe. If you can see a pass over central England, that will be really interesting. I made a fascinating observation this evening. I detected what I would characterise as an incoming large piece of space debris. It had the classic head echo caused by the radial velocity Doppler shift, followed by a long tail generated by ionised gas ionosphere. This initially looks like a big meteor, but had a decidedly lower Doppler shift on the head indicating much slower incoming radial velocity. Compare this trace with Stormchaser's detection of a meteor at 21:30 yesterday. The phased switching of GRAVES is visible within the head echo, but the frequency shift is only around 200Hz over about 3 seconds, very roughly the same as I see for satellites and markedly slower than for meteor strikes. Based on the intensity of both the head echo and intensity & duration of the tail it appears that it had a fair amount of mass. Spectrum lab recorded that trace at a peak of 46.7db above noise, some 25db stronger than average detections. Assuming that the ionised tail indicates an altitude in the ionosphere, it appears to have come in somewhere over Southern England, the English Channel or Northern France. It may have been a slightly scary yet very interesting sight had it been dark. Richard

Apologies, I'm slightly late to the party - I've been inactive for a while as there's very little deep sky visual observations available from my site at this time of the year and my meteor detector was offline as we performed some maintenance at the property. Anyway, those look like excellent out-of-shower detections and bode well for the Perseids in August. Given the geometries and your radio horizon, I can't see how they would be anything other forward scatter from radiation in the (expectant weak) northern facing lobes. Your new antenna seems to be performing well - I used to run an array of 4 such long boom yagis in the 1990's for EME on 144MHz. They work very well. I have tried on several occasions to detect the ISS from these lobes, but I fear the ISS elevation may be too high for me to detect scatter from them. I might have detected a single very weak signal from the north facing lobes, but it's far from convincing. All of my other detections of the ISS are when it is very low to my horizon, transiting the South of France and south of the radar. Yet given my range to GRAVES, and even more so with your range, we still have strong circumstantial evidence of north facing radiation. There is an ISS pass at around 09:45 tomorrow morning that reaches 41 degrees elevation from here and may yield a good common scatter medium from Stornoway. If you detect that transit (or any like it), the case is closed because the ISS is well north of the radar, being over central England at the time. Have a look on Heavens Above for ISS passes over central and southern England and you may just detect some. Richard

At that band, a 3.2m dish is probably the smallest practical size and even then, depending on the mount and base, you may be in danger of giving yourself additional engineering challenges that could be removed by going to stacked and bayed yagis. A 3.2m dish should yield a forward gain of around 20-22dBi at those frequencies, which is probably a couple of dB less than an array of four 15 element yagis stacked and bayed, or a multi-element phased array. Given the size of the yagi, the array is pretty small, lightweight and has very little wind loading, so a simple ground pole with az-el control should match the dish. Better still, if you can just arrange elevation control, you can run drift scans across your chosen target, so the mounting arrangement becomes considerably less complex. However, don't underestimate the stacking and baying phasing and accuracy to achieve an optimum solution and if you want to go for circular polarisation, then such a solution is not viable. If you decide to go with the dish and it's already mounted and driven, a small 2 or 3 element yagi, centred on the 432MHz amateur band may suffice, mounted on a suitable A-frame. You'll need to get the spacing correct from the front of the dish to the driven element and a worm drive may help tune that for optimum gain/bandwidth. The smoothness or finish of the dish is surprisingly tolerant to surface defects and abnormalities, especially at such low frequencies, so if you need to attach additional brackets for the mounting frame, these should make little difference to the performance. Circular polarisation should give you good coverage and some years ago there used to be amateur radio satellite antennas available in that band. I see no reason why these shouldn't still be available from various suppliers. Try to use the lowest loss co-ax feed as you can afford, even with the LNA mounted at mast head. Good luck! Richard

That's great James, really pleased for you. My heat map looks a little different, but I'm cautious: the numbers seem quite high, even for the peak of a shower. The peaks in the evenings of the 20th and 23rd may be noise radiated by next door's heating system. That said, beyond the obvious curious evening peaks in my data, there are some similarities. The "switch on" of incoming strikes on or around the 17th/18th, the background rates and the relative lack of strikes in the evenings of the first half of the month are similar in both sets of data. I'll need to look more closely at my data to see what those evening peaks represent. Richard

Yes, a warm welcome Geminids! I'd second what Ian says, a higher rate sampling is needed to accurately record the transient head echoes. That implied a shorter FFT window and more experimentation with Spectrum Lab. Unfortunately, lack of time over the last couple of months meant I didn't progress the idea. There may be issues with aliasing at high sample rates and fast transients: Could be interesting nonetheless. Here's a couple of incoming strikes with clear head echoes that recorded as I type: The problem here is noise. Taking the first return (on the left of the image), at what point does an algorithm detect the start of the head echo... 1, 2, 3 or 4? Point 4 looks like noise, but is on a possible extended line from point 3. There is "something" at points 3 and 2, but are they signal or noise? Point 1 is clearly in the head echo on this return, but the signal is now strong enough to blur into adjacent pixels (time), reducing the accuracy of any measurement. Much higher resolution is required, but, as noted, that may make the FFT sensitive to the discontinuities in the GRAVES Radar signal. I suspect a quantitative rather than qualitative approach is needed to save being bogged down in noise and spikes. Richard

Now, that is definitely a meteor! Well done!

My guess is that they are predominantly as a result of rear lobes. The front lobes illuminate the sky at an altitude of 90km well over your southern horizon. It's possible anomalous propagation could result in back scatter reaching you from the frontal lobes, but there are two generally accepted modes that would provide that path: Tropospheric ducting or Sporadic E. Neither modes are open right now between Scotland to southern France. In fact, I'm not sure tropo would even allow for such propagation because the incoming signal originates in the ionosphere, above the troposphere and may not be able to break into any super-refracting duct from above (the process that keeps the signal leaking from a duct may prevent an external signal breaking into a duct). On a positive note, I find the number of detections recorded at my site drops dramatically in the presence of disturbed weather, such as we have just now. This suggests that when weather conditions are more favourable (a calmer troposhere), your detection rate could increase. It will be very interesting to see how your detections develop as we proceed through April to the Lyrids. If you can run one of the published algorithms on Spectrum Lab on the laptop, you can build a record af background rates, which could be useful if you are able to get the Pi based detector running in the future. Richard

Certainly does look like a meteor. Well done, congratulations! It's difficult to completely interpret without time and frequency scales, but the apparent Doppler shift of the head followed by the intense body is classic meteor. That's encouraging to see. Hopefully you get a few recorded events overnight when the background rate peaks. Richard

The range to the hills from Stornoway shouldn't give rise to too many issues. The signal propagating from the reflection media will have been refracted and dispersed through the ionosphere and troposphere already. The may be some additional distortion, but the Scottish mountains are generally flat topped rounded hills (and quite wet most of the time too!). It's possible more bifurfication of the signal could occur however. The Icelandic VOR beacon is a very interesting proposition. The beacon at Akureyri is on 113.6MHz which is a nice frequency to use for meteor detection as it will scatter well. It's radiation pattern is ideal for these purposes and, additionally, it's a good range being approximately 1100km, which is a magic number in this context. It means you will be able to see a 90km altitude at Reykjavik on your horizon and a forward scatter common volume will exist all the way from your horizon to your zenith, giving you a large reflecting media to capture incoming meteors. A broad beamwidth vertically polarised antenna pointing northwest from Stornoway should have a decent chance of detecting meteoric reflections from that beacon. You will undoubtedly have interference from auroral scatter which will lead to blurring. Auroral scattering will be quite prevalent at your latitude, but auroral scattering is fairly wide wideband and quite different to the narrowband meteor scatter you are looking for. Audibly the narrowband Morse or voice signal will be blurred into a broad hissing noise. Even though your current antenna is tuned to approx 144MHz, it may be worthwhile turning the antenna to the northwest and see if you can hear that beacon via the Aurora, taking into account auroral forecasts and timing it to an increase in auroral activity. We are close to sunspot minimum, but there are still plenty of visual Aurora reported visible from Skye and other northern locations. If you can hear the beacon via Aurora, ther is a very good chance of obtaining meteor reflections off it. I would be tempted to look for it myself, but my antenna is mounted on the SouthEastern side of the house and totally blocked to the NorthWest. Richard

Apologies, I've been out of circulation for a few days. I don't think the gain of the antenna is a major part of the engineering problem when listening for GRAVES: the antenna I use is a simple 2 element phased array (an HB9CV design https://www.qsl.net/dk7zb/HB9CV/Details-HB9CV.htm) with forward a gain of 4dBd (around 7dBi). Clearly the GRAVES radar transmits are very strong signal, so a 3 element yagi will do fine if you can see a main lobe common volume via line of site. Your topographic profile shouldn't present much obstruction with a clear line of sight over water to the mainland. The arithmetic however shows the forward lobes of the radar are well over your horizon, and for me too. From other work, we have deduced there are likely to be reasonably strong rear facing lobes and sufficiently high angles for me to detect forward scatter in a northerly direction. However my site is 350km further south than yours, which significantly alters the geometry of the common volume compared to that for Stornoway. The maximum range to a 90km altitude is 1100km (line of sight) which, for you, is over Southern England and Northern France. Although there appear to be rear facing lobes, I suspect these are not strong enough at the required angles to achieve echoes that are resolvable with entry level equipment at Stornoway. Having now established your dongle is quite well tuned, you could try look for meteor scatter signals from GB3VHF. Although this is running much lower power than the expected signal from GRAVES, that could be compensated for by increased gain in the receive antenna. I wouldn't recommend stacking or baying antennas for 144MHz in the first instance: doing so poses real engineering challenges getting the phasing correct. It will also significantly reduce the 3dB beam-width and place deep notches not far from the centre line of the antenna. A small azimuth or elevation misalignment of the antenna could easily reduce the received signal dramatically. A modest single 9 element yagi (https://www.radioworld.co.uk/220309_tonna_2m_9_element_yagi_antenna_144_to_148_mhz) will give decent forward gain and broad beam width to be able to look for GB3VHF. A mast head preamp could also be beneficial as the signal may be quite weak. These should give a reasonable opportunity to receive meteoric signals from GB3VHF. The challenge will then be to adjust the detection algorithms for take account the non continuous nature of the signal broadcast from GB3VHF. To capture signals from GRAVES, I use SDR# and feed the audio into Spectrum Lab. The primary reason for this is that Spectrum Lab refuses point blank to drive my dongle. I suspect there is a non-standard build in my dongle firmware, but it's overcome by using SDR#. To connect SDR# to Spectrum Lab I use a free virtual audio cable from https://www.vb-audio.com/Cable/. This arrangement works well, even though SDR# uses more CPU than I would like. There is an advantage to this method too: you are operating a configurable receiver via SDR# and can optimise parameters such as gain, bandwidths, noise blankers and notch filters using a much simpler GUI than through the configurations in Spectrum Lab. Here's a screen grab of my SDR# configuration: I now use a tweaked version of IanL's Spectrum Lab detection algorithm which gives good results for me. Overall it works well, as can be seen with a satellite detection as I type: My honest opinion, given what we know and/or have deduced about GRAVES, is that its south facing lobes will be too far away and well over the horizon from Stonorway. Any north facing lobes will be difficult to use at such extreme range and a closer north facing beacon such as GB3VHF may yield better results even if there are still limitations. Richard

It is possible to use amateur radio beacons for meteor detection. In the past, I used to operated amateur meteor scatter and would often hear signals from such beacons. However as has been noted, amateur beacons are typically low power and illuminate smaller volumes of the sky. Reliable detection of signals from these beacons may need higher gain antennas. The non CW nature of these beacons does mean you may miss a number of incoming meteors, including the head echo of long duration events. But if your range to GRAVES is too far, they are an alternative worth trying. Richard

The detection of Morse signals on 144MHz amateur band is encouraging. You could take a look for specific beacons listed here: https://www.microwavers.org/maps/2m.htm to see if you can identify the station and use that as to verify the SDR tuning accuracy. Background meteors will peak in the Lyrids shower on the 23rd April which should give a good indication of any possible common volume for GRAVES or even GB3VHF as the transmitter station. Richard

I think you're correct in your approach: setup the tuning first so you have echoes, either from ISS or meteors, then look at your detection software. All things being equal, the ISS would have been detectable from your site last night, being only 3 or 4 degrees lower than for me. Perhaps experiment with Spectrum Lab and the published scripts to correlate against your Pi software. Bear in mind, the majority of echoes I recieve out of shower peaks are very short duration (between 0.1 and 0.5 seconds) and may be almost inaudible, so you may need some form of automated logger to see them just to get the tuning right. Take heart though, even though your range to GRAVES is extreme, you may still receive signals. If there is backfire radiation illuminating a common volume north of the site providing the path for the signals that I detect, it follows there is a good chance you may have a similar path. If GRAVES turns out to be too far over the horizon for you then wxsatuser presents an alternative of using GB3VHF, which is radiating in your direction, albeit with less ERP. Richard

Did you detect the ISS pass this evening? The pass at 19:24 was detected here at the extreme range: The zero radial Doppler shift occurred at 19:24:18, which corresponds to the the ISS at the maximum elevation according to Heavens above: The footprint here is where the ISS is 10° above the horizon, which should still be above your horizon in Stornoway and hence should be detectable. I'm interested in understanding more about your meteor common volume simulations. If my arithmetic is correct, the 95km altitude range is ~1100km with the antenna pointing at the horizon. Perhaps a badly drawn sketch may help explain my arithmetic: Therefore, to receive meteoric reflections from a common volume 95km above the surface of the Earth, that common volume should be approximately 1100km distant. There will be some blurring in this as the GRAVES radar signal is not uniform and the scattering media not constant, but it will do for approximations I think. Normally, expected ranges for forward scatter is up to ~2200km where both stations are pointing at each other and the common volume bisects them. However, that is not the commonly accepted position for GRAVES which, based on articles and photographs of the site, illuminates a semi circle from West through South to East. If this commonly accepted position is correct, we are receiving backscatter echoes from south of the site. However, that would be outside the 1100km range to the common volume from my site. My contention is therefore that there must be significant rear lobe radiation at sufficiently high angles to illuminate the sky over northern France in order for me to receive meteoric reflections. It follows that for your site at Stornoway, the common volume would be over Southern England and the English Channel and definitely require significant rear lobe radiation. Have you seen any meteor echoes yet, as these would be very interesting at such range. Richard

As has been noted, you'll be too far north from GB3VHF to receive tropospheric ground waves unless there is ducting underway. You may detect meteor echoes from it however. Many years ago, I would often hear echoes from the now defunct Lerwick beacon from my (then) site in central England, without tropo ducting. Your range to GRAVES is pretty extreme. I have postulated elsewhere that there may be significant rear lobes from the radar as the geometry requires forward scatter off rear lobes from the radar to be able to detect a signal at my location near Biggar and I am some 350km south of you. Remember, your range to Dijon is just the start. You need to be able to see a common volume of the sky at about 95km altitude. For south facing lobes this would be over Iberia, southern France, Italy, the Aegean Sea, northern Balkan states etc. I suspect these areas will be over the horizon from Stornoway, so it will be very interesting to see what signals you detect off any north facing lobes, which would be over the Channel and southern England. Detection of these signals will provide further evidence of significant rear lobe radiation. I would ensure the antenna is pointing at the horizon to maximise the chance of illuminating the common volume of the sky. My antenna is a simple HB9CV 2-element phased array, with around 10m of RG58u coax feeding an RTL-SDR and achieves good results. There's not much shower activity just now, so you'll need to rely on background rates, which peak in the early hours of the morning. Try tuning to somewhere between 143.048 to 143.049MHZ on USB, depending on the accuracy of your SDR. That should give you sufficient offset to see echoes at around 1 to 2kHz. My SDR is tuned to143.048850MHz USB and yields 1.2kHz baseband echoes. Good luck, Richard

Ahh, I didn't catch that rather important observation Ian! Taking your trace on the 11th, there is a 2Khz change in frequency in 17 approximately seconds. Cropping mine to the same 2Khz vertical window (centred on the 0Hz Doppler shift point), I recorded it for 40 seconds. In the image below, I've scaled my horizontal axis to the same as your's and crudely overlaid the two images. My image is very wide at that scale, so I've aligned the two at the -1Khz point (the point at which it drops off the bottom of the waterfall plot). That point is somewhat dependent on the accuracy of the RF tuning - hence the misalignment of the GRAVES azimuth switching pulses. Also, there's some scaling errors left to right across the image, but even with those and the crude overlay, you can see a very large difference in radial velocity between your trace and mine. Perhaps with more time (and an earlier evening) it may be possible to make a better job of the alignment. Visually, I'd estimate that the ISS ran across about 90° of your sky compared to approximately 60° of mine at the same absolute velocity, meaning the magnitude of your radial velocity was much higher than mine. I think! Richard

That's an interesting one Ian. I captured the same two passes, but didn't notice the difference in Doppler until I saw your post. I will need to think about this a bit: the gradients (i.e. the rate of change in Doppler shift), appear on face value to be different to yours. However, I think this is to be expected. Since my range to the ISS was greater, yet the absolute path taken by the ISS was the same, it traversed a smaller angle of my sky than yours in the same amount of time, so the radial velocity would have been lower and the gradient less steep.....I think. Richard

Thanks Ian, Yes, now I've added a definition for the cfg.SpecFreqMax and cfg.SpecFreqMin constants in the conditional actions script, the logger isn't missing much. In fact, arguably, for my site and system, it's too sensitive. Generally speaking, I don't think I've got too much RF noise. However, in addition to events of the order of a second or more, I recorded a fair number of extremely short duration and weak events overnight (<0.1s and ~15dB SNR), My original script didn't detect many of these. Given my range to GRAVES, I have no reason to think these are not meteoric in nature, but I suspect they may be very low in mass, which almost instantaneously vaporise creating a very weak and short duration events. I suppose that for a complete picture these should be included, but it would make post analysis of the events more difficult. As a result, I've tweaked the SNR threshold and duration up to 16dB and 0.1s. I'll see what effect that has on filtering out these extremely short duration events over the next day or so. If I can find the time, I might also have a play with a second spectrum, with a narrow window, to record the head echoes and see if I can deduce information about incoming velocity. Richard

Progress! Reviewing the .USR file, I can see there are a number of configuration variables which are passed to the conditional actions script. Primary for my concern is the cfg.SpecFreqMax and cfg.SpecFreqMin. These were set outside the tuning range of my dongle - remember I'm running the dongle driven by SDR# and not Spectrum Lab. Adding those variables into the conditional actions text file plus tweaking some minor SNR and duration settings and I'm now recording meteors with this new script: I'll monitor the output for a few days and see what the script records, particularly in the small hours when background rates and duration tend to peak. Richard

Ian, A brief update on some tests I ran with the script. Since my RTL-SDR dongle appears not to like Spectrum Lab, or more specifically, I haven't been able to find a driver dll to interface the two and need to drive it from SDR#, I had to fall back to testing only the conditional actions in the text file. These imported fine, but I couldn't get the script to trigger correctly. I can see it reading the average SNR, but upon an incoming meteor, it steadfastly refuses to trigger any further. I have adjusted the SNR trigger point down to a couple of dB, at which point it should be triggers almost continuously, but still nothing. Oddly enough, when there is an incoming strike, I see the green icon on the main display briefly flash to yellow and then back to green, just as it does when executing my current script: I need to spend more time reviewing the new script to see if there's a configuration in SL that I need to adjust to get the new script to recognise the peaks in the spectrum. Given that I'm only using the conditional actions and not your full setup, is there something in the initial configuration of SL using your method that specifies the incoming frequency range over which the conditional action script should search for a peak? Anyway, I'll try find some more time over the next week to run some more tests and checks. Richard

I had a little play with this just now and encountered the usual pernickity Spectrum Lab. I've never been able to get SL to drive my RTL-SDR directly, and so it was today. I use a virtual cable (from vb-audio.com) to connect the SDR output to SL and use SDR# to drive the rf functions of the SDR. Nonetheless, I'm interested in the features in this script and tested it with a 1kHz local oscillator in SL to fine tune the conditional actions. For now, I've reverted back to my original script as I don't have enough time to tweak much further tonight. However, with a bit of luck I can find time in the week to merge the conditional actions text file into my hardware specific setup. I'll report back in a few days. Richard