Backscatter Radar

[VK2ZRH]

Hi Andrew,

The 21 Jan 06UTC chart expanded to 30,000 km is curious, indeed.

The return at around 13,000 km is most likely supermode TEP, where the reflections/refractions via the two peaks of the equatorial ionospheric anomaly occur fairly high in these structures out above the mid-Pacific, and/or the two peaks were spaced comparatively far away from the geomagnetic equator (possible, given the geomagnetic disturbance of 20 Jan UTC). VK4WTN hearing propagation from the USA at around the same time would corroborate that.

The return peaking at about 27,000 km is propbably NOT conventional long path (as encountered on 20m, for example), but rather a non-great circle TEP path via the equatorial anomaly, perhaps in the easterly direction or the westerly direction via the EIA above the Indian ocean (from the sides of the rear lobes of each of your's and David's antennas). Then again, I can't rule out some sort of long path propagation supported by polar electron precipitation extending northward across the auroral zone . . . given the geomagnetic conditions of 20 Jan UTC.

To reach the southern peak of the EIA, clearly the outward and return signal elevation would be very low, possibly under one degree, depending on how far south of the geomagnetic equator the southern peak of the EIA was positioned.

What is needed now is a way to add azimuth readout . . . followed by elevation

The IF of the receiver will add a fixed range error to the results, arising from group delay and directly related to the IF bandwidth.

At the time I was involved in the commissioning of the 4A~4B ionosonde when I worked at IPS (mid-late 1970s), Leo Macnamara explored sources of error in the machine and its implementation. The biggest source was the IF delay, which I think was something like either 1.5 km or 15 km . . . but I can't, for the life of me, remember which. If recall serves me correctly, efforts were made to measure this characteristic of the 'sondes, calibrated against a delay line of known length (a drum of coax ).

73, Roger Harrison VK2ZRH

[VK3OE]

Hello Roger,

Thanks again, I am finding the results astounding, I certainly did not expect to get such large signals back. However, it is now obvious that this would happen as the area from which the signals reflect can be very large thus providing a large amount of reflector gain, which is what makes it work so well.

As to the delay, a few 10's of ms will not make much difference in this system as it can easily be compensated for by ensuring that the direct signal is a 0 km. The bigger problem is the group delay of the transmitter and receiver filters which smears out the received signal and reduces the system gain. I have a scheme where the system can learn what the group delay is (it turns up as a error if the sent chirp is compared with the direct signal) and can use this to compensate the system. Luckily the system range gates are about 12.5 km each at a 12000 sampling rate so that errors of 100km will not make much difference. Similarily for frequency errors, they can be easily compensated for up to about 50Hz, the key issue is sampling stability and frequency stability during the measurement.

I am investigating a means by which the receiver can be publically available within a reasonable time frame so that anyone can use it, chirps everywhere, DC to daylight!!!

73
Andrew VK3OE.

[VK4WM]

G'day Andrew & David

Have you considered chirping at the moon when a good moonrise/time is suitable, a good way to prove calibration and quash doubters maybe?

Excellent work, looking forward to see what comes out of this, very exciting.

73 Wade VK4WM

[VK3OE]

Hello All,

Firstly, thanks to those who have helped me get this far, particularly David VK3AUU for being "on the receiving end" for the initial testing. Frank VK3OP has also assisted with a couple of receiving efforts. Thanks also to to all those who have offered positive comments.

In the latest round of testing I have attempted to find out where the limits of the system are, if there are any spurious effects, and if a group delay equaliser is required.

To the group delay, the off air test signals with no propagation present are so close to those signals simulated without the radios in place that it is evident that the radios have no effect except to shift the audio to RF and back. Further, the direct signal is the limiting factor so lowering this signal by having stations further away for receiving would be most advantageous in increasing the system sensitivity.

I have included a picture (second) comparing the no propagation off air signal with one of the previous pictures for the multihop Es. The backscatter is very evident from 900km on. Up to the 900km point there is no return signal. After 900km the backscatter signal (red) is well above the system limit.

At a distance of 7400km there is a response about which I am skeptical. Some testing with Frank VK3OP and David VK3AUU is shown in the first picture. The spurious response is not evident in Franks' received signal so it is almost certainly an artifact in Davids receiver. Further, Franks received signal is above Davids' (I have normalised the direct signal) indicating the better noise performance of Davids receiving system.

There is some progress in working out how to get the software out in a useable package, I will update this page as there is progress.

73
Andrew VK3OE

[VK6KDX]

Hi
Please excuse my ignorance as it has probably covered before
But what is the "source" of the backscatter radar ie are you generating it or
is a commercial signal?
In "Amateur Radio Astronomy" by ZS5JF he shows how to build your own meteor scatter
radar.
Jack

[VK3OE]

Hello Jack,

The source is Spectrum Lab signal generator via a sound card into a 756PIII at 100 watts. The receiver is VK3AUU 60 km away using a IC706MkIIG to sound card and Spectrum Lab for recording.

The resulting .WAV file is emailed to me and I process it in a complex receiver in Matlab... = = bistatic backscatter radar !!!

Andrew
VK3OE

[VK3OE]

Hello,

Just some more checking and verification. This time I am comparing the basic system noise when no signal is present, a signal when no propagation is present and the backscatter signals received at 0610UTC on 21 Jan 2010.

The system noise in black sits at about 0dB. The chirp signal with no propagation present is shown in blue. The direct signal sits at about 65dB above the noise. The two spurious responses at 7500km and 36000km are evident. The red signal is the backscatter signal which was obtained from a direction of 60 degrees, and is normalised to the non propagation signal for comparison.
There is no backscatter up to about 1000km where signals start to return. There is then continuous backscatter out to 30000km. The large peak at 13500km corresponds to the West coast of the USAand sits at about 60dB above the system noise. The next peak is between 25000km and 30000km and is still at least 30dB above the system noise. There is some conjecture about this but given a claim by VK4WTN that he heard a Texan station at the time calling towards South Africa, it may have indeed been open to South Africa from the USA and the return is from South Africa on the long path at 60 degrees heading. Thus the peak at between 25000km and 30000km probably corresponds the the location of South Africa on the long path and is backscatter from the South African land mass. Past 30000km there are probably no further returns and the signals there probably correspond to the lift in noise from the propagation to about 20dB above the system noise.

So, apart from the spurious signals, I can not find any reason or effect that could call into doubt the results shown. Indeed, there is substantial corrorboration of the results with stations heard and conatcts being made at the times of the various observations that serve to verify all the results shown previously.

Andrew Martin
VK3OE

[VK2ZRH]

Curiouser and curiouser.

Do the "spurious" peaks at ~3600 and ~7400 km appear when 3AUU has his antenna set in a different direction, or if he uses another antenna (eg. a dipole) ?

73, Roger Harrison VK2ZRH

[VK3AUU]

I think I have resolved the problem of the spurious peaks or at least the 7500 Km one anyway. I use two computers and the output of the secondary one feeds audio into the primary one. The input to the primary from the secondary is via the microphone input which was turned down but not muted and there was just enough getting through to cause a false peak which the 50 millisecond processing time of the secondary sound card introduced.

That is the theory anyway. We have to run a test to prove it. There are always traps for young players and tricks for old dogs to learn.

David
VK3AUU

[VK7JG]

Can you teach an OLD dog new tricks ? :wink

[VK4XA]

Can you teach an OLD dog new tricks ? :wink

Whilst I dont really qualify as 'old' (I'm only 51) I am certainly learning new tricks......this is bloody interesting!

[VK3OP]

David,

It's great that you have isolated the source of the spurious, and that it is not the 706. Hopefully we can arrange another test
when you Andrew and I are all available for more comparison. Andrew let me know when you wish to run another test.

Cheers

Frank VK3OP

[VK3OE]

Hello Frank,

The 7500km is a spurious, but it seems to be something to do with the hardware/software. I have run a local test using the 756 and 7000 and it is evident that although at a much lower levels than David sees. Could be in the Spectrum Lab or soundcard or transceivers, more testing is required. It is internal to the system.

The great thing is that it is the only spurious we see, otherwise ????? So return signals look to be te real thing. Always a problem with a new system such as this, what is real and what is spurious.

Thanks again for your help.

Andrew
VK3OE.

[VK3OE]

Hello all,
This is version 4 of this post, I have corrected some first errors, included the antenna patterns, included a better estimate of the path loss VK3AUU to VK3OE on 50MHz, added the radar equation and lastly proposed some propagation effects.

Some further checking is in order, particularly for the very high amplitude return signal shown in the second picture of Jan 26, 7:43 pm which shows the probable return signal from the West Coast of the USA at 13500km being equal to the direct signal. This still seems surprising to me so a few calculations are in order. The direct signal between David VK3AUU and me with antennas at 60 degrees is about S5 which is about -105dBm. For the return signal to be about the same level as the direct, the total path loss to the reflector must also be the same as the path loss VK3OE to VK3AUU. The transmit level is +50dBm so the total path loss is 155dB. The antenna patterns result in a loss of 26dB on the direct path VK3AUU to VK3OE ( from RMD), the antenna gains combined are 22dB so that the difference is an increase in gain between the direct and reflected signals of 48dB. This reduces the required return path loss in total to 155-48 = 107dB. The one way distance to VK3AUU is 60km and the ONE WAY distance to the West coast is 13500km. The ratio of the distances is 225 times further which is roughly 2^8 times so that the free space path loss is 8*6 = 48dB more than the loss between VK3AUU and VK3OE. Assuming the worst case requirement for reflector gain at the distant end, assumes free space loss so that the reflector gain at the other end is then required to be 107-(2*48)=11dB. (The 2 * 48dB come about because of the radar equation).

I used the calculation used for large passive reflectors at microwave frequencies to calculate the area for a passive reflector required to give 20dB gain at 6m using gain= 20*log10(4*pi*area/(wavelength^2)). It turns out that a passive reflector at 6m for 20dB gain (being very conservative) is roughly 30 square m or 0.00003square km. Assuming that each point of the reflection in the returned signal trace is about 10km deep, and if we assume a width of 10km is involved in the reflection then an area of 100 square km is illuminated for each point in the trace and just 0.0000003% of the illuminated area is required to reflect to get the measured signal to reflect. Given the nature of the mountains of the US West coast, then this result seems entirely within the realms of of a probable explanation for the result. A higher path loss requires even less reflection gain to get the levels received.

If the signal travels via the f layer it will be constrained in the vertical direction so it may be better to assume 4dB for every double of the distance so that the additional one way path loss (above the loss VK3AUU to VK3OE) to the US West Coast is 32dB. This then gives a reflector gain requirement of 107-(2*32)=43dB for the reflection gain and not 11dB if 6dB for every double distance as previously used. This requires a reflector of 400 square meters, still just 0.000004% of the estimated reflection area of 100square km.

The signal at between 25000km and 30000km is twice as far and is 30dB lower in level than the signal at 13500km. Given that the free space path loss increases by 6dB for a doubling of the distance and that a similar area could be illuminated in South Africa (long path) then it is probable that a similar explanation to that for the 13500km return above would explain the result at 27000km, although with more path and/or reflection loss.

The checking continues and it seems to be reducing any doubt about the performance of this backscatter radar system. Checking for moon echo signals is a further possibility but I am not sure that the timing and frequency stability we currently have is sufficient to achieve a good result as averaging is required over several minutes.

Maybe I have it correct this time, third attempt and put in the radar effect as well, fourth attempt!!!

Seems too good to be true!!!

Andrew
VK3OE.

[VK3OE]

Good News on progress for an application.....

Wolf, DL4YHF of Spectrum Lab fame is incorporating the chirp technology into Spectrum Lab, some of the modules are already there in the black-box section. Chirps can be generated, received and displayed in the time domain scope. There are several extra bits still required to make it completely functional such as range instead of time for the x axis, timing inputs from the computer clock and converting the data to dB to get sufficient dynamic range. Early testing looks very good as once set up it is easy to operate. Just watch the scope for radar returns.

An updated version of Spectrum Lab should be available soon (weeks) so keep an eye on the developments.

A full description of the chirp backscatter system together with initial results will be published in DUBUS 2/2010.

73,
Andrew
VK3OE

[VK5PJ]

Andrew,
would it be possible to make this function from a single site, given that you do not want to Tx and Rx at the same time?

(I had deleted my previous post as had reconsidered what it contained)

Regards,

[VK4WDM]

Ok, let me get this straight without the technobabble. You fire a signal NE across the Pacific. It hits the coast of California and is reflected back to another receiver thus showing that the path is open. The time taken for the return gives you the distance and that confirms it is California. Has this old guy got it right? If so WOW!

Some questions:

1. How narrow is the beam? Can you tell if it is the coast of Mexico, California or Alaska?
2. How big does the target have to be? What about Hawaii of some of the smaller Pacific Islands?
3. I presume that the signal gets up so high that intervening land masses don't get in the way so you could, say, check a path from VK4 to ZS?
4.Will it work on 2m? I have often thought that VK to KH6 on 2m is possible. This might tell us when it might be possible. Also other paths such as: VK6 to Reunion or even Africa, or VK7 to Vk0.

This might seem nerdy but an explanation for the technology-challenged would be useful.

73

Wayne VK4WDM

[VK3OE]

Hello Wayne and Peter,

I will post a detailed reply on the weekend.

Thanks for your questions.

73
Andrew
VK3OE

[VK4EKA]

I am only watching the developments, and not participating, so don't take my comments as gospel.
It is not necessarily land the signal reflects off.
On 1280MHz the radar shows backscatter from rain.
On Six metres I am sure you have worked backscatter from Sporadic E patches.
From here the classic is Brisbane to Sydney by beaming at a Sporadic E patch out near Norfolk Island.
No land reflection needed.
Similar happens with the Aurora on 6 and 2 metres.
I look forward to the explanation on the weekend.

[VK4TU]

I would say that the reflections are off everything. By knowing the expected distances of the hops, if it's mid-point then it's off the ionosphere, if it's full-distance it's from whatever's at the end. The important point isn't knowing "hey there's propagation to XYZ and it's X Km wide" but rather we have 2 hops open towards the North East, let's start calling. It should also let us know the geometry of the path so we can tell whether they are Es or F2 hops. At least that's the way I understand it. It's certainly more informative than waiting for beacons or activity from a part of the world and very exciting work in its possibilities