Jump to content
Sign in to follow this  
Master_Scythe

Smallest, most direction antenna

Recommended Posts

Dammit.  Wish @Phoenix135 was here.  

Antenna design and testing is this dudes job. 

However, Yagi antenna is probably the most directional antenna I can think of.  

If you have enough elements in front of the reflector you can increase the gain whilst tuning it down to be very directional.

 

AD

Share this post


Link to post
Share on other sites

Wayyyyyyy too big.

I'd love to use a higher frequency, but the availability and legality of these little transmitter\recievers is hard.

Share this post


Link to post
Share on other sites
5 hours ago, Master_Scythe said:

Wayyyyyyy too big.

I'd love to use a higher frequency, but the availability and legality of these little transmitter\recievers is hard.

If 35x20cm is too large, you're going to have a hard time with DF.

A mangentic loop antemnna may get the size a little smaller, but best DFing is with the null, which is supposedly counter intuitive for beginners.

Without knowing the usecase, I can't say of there are other technologies applicable to your task, that may not even need to use DFing technology.

Share this post


Link to post
Share on other sites

ok stadl tell me everything wrong with this harebrained idea :P

3 monopole antennas at the vertices of an equilateral triangle with a height at least, say, 1/8 λ

a bearing is arrived at when two antennas show the same positive (near 90°) phase delay

Share this post


Link to post
Share on other sites

Bearing accuracy will be dependent on a lot of things. The 2 most common ways of configuring/implementing that sort of approach are: 

1. Switched antennas and a single receiver - look at 'Doppler RDF kits' for details -  I've DFed with the equipment - and got solid bearings  to around 10 or 20 deg with 4 antennas on a larger scale (approx λ apart) - But never looked much into the maths/implementation details to understand how it works.

2. Multiple receivers, with time of arrival calculations / beamforming. Antennas need to be rigid (wobble relative to each other = errors), phase matched/calibration on all cabling and synchronised receivers required. My dated and edge experience* was with acquiring the signal with ADCs and processing in the digital domain, but I believe there are also options for analogue mixing of the signals to get a blip.

* worked as a software dev  on a few Radar and DF systems, in the late 90s - not in signal processing subsystems, but needed to understand the basics of the physics, system design and software concepts. I'm not qualified to answer the details, but have a good generalist/systems engineering view of these sorts of systems to be able to identify the challenges based on past work with a bunch of people far smarter than me.

Share this post


Link to post
Share on other sites

woah, hadnt even thought about 1.   i assume that is like the reverse of a VOR beacon for aircraft navigation — which has a ring of transmitters that take turns, effectively applying FM to a carrier.  makes me wonder about how fast the switching would need to be and the analogue side of that.

as for the DSP of 2, i guess i am supposing that correlation would provide ample latitude for noisy/weak and time-smeared signals (mechanical or digital judder) and some of the timing error being reigned through deduction based on triangulation, but then again, they are probably too close together and i am imagining the receivers magically sitting on the same 2D plane as the transmitter. 

Share this post


Link to post
Share on other sites
12 hours ago, @~thehung said:

woah, hadnt even thought about 1.   i assume that is like the reverse of a VOR beacon for aircraft navigation — which has a ring of transmitters that take turns, effectively applying FM to a carrier.  makes me wonder about how fast the switching would need to be and the analogue side of that.

as for the DSP of 2, i guess i am supposing that correlation would provide ample latitude for noisy/weak and time-smeared signals (mechanical or digital judder) and some of the timing error being reigned through deduction based on triangulation, but then again, they are probably too close together and i am imagining the receivers magically sitting on the same 2D plane as the transmitter. 

ToA systems are non trivial. The correlation algorithm is a brute force of Multiply-Accumulates that's pretty easy to implement. The issue is what are you performing the analysis on? Raw RF or downconverterd/IF. There are SDRs around these days that will output the I&Q samples of the IF, so that is best. 

Sample time and rate are a balancing act - bigger sample set (increased by sampling faster or taking more samples - longer capture) will give more data for correlation to reinforce. But you ideally want the Rx and TX  stationary during the measurement, so short capture time is desirable. A preference to higher sample rates  if you can get there. The best outcome is from adding channels - but every channel is an antenna, phase calibrated feeder, high quality phase aligned receiver an ADC channel, more data bandwidth, and more processing - so more $s and more size/weight & power - TANSTAAFL 🙂

Triangulation is not going to happen in the DF receiver. That is a higher level tracking function performed by the person or software that extracts multiple Lines of Bearing from different locations. The underlying DF technology is independent of the higher level tracking systems. I've worked on systems (as a hobbyist) combining doppler receivers and small beam antennas being swept. Include your GPS position on a map, and you have a simple but effective mobile vehicle mounted DF system.

The 2D plane part is not really an issue. Yes, your Rx Antennas need to be in the same plane - but easily achievable. The Tx antenna, assuming a single antenna, won't make much difference, other than 3dB loss if it's alternate polarisation (horizontal vs vertical). It won't mess with the phase detection/ToA calcs that much. 

Share this post


Link to post
Share on other sites

i didnt mean triangulation quite so literally.  i meant, for any configuration of at least 3 antennas with a triangular relationship, you could think of the total +/- timing error* as positional error in terms of something like a radius around the idealised positions of each antenna.   the actual measured 'triangle' represented by the phase data has to conform to points within those three circles.  the maximum error between each pair of measurements is always constrained by the third, if they are to collectively be consistent with a known signal coming from a singular physical source.  so, there could *potentially* be a probabalistic 'line of best fit triangle' drawn to apply fuzzy weighted corrections to the measurements.  <-- already stupidly complicated in 2D.  probably impossible in 3D.

*lets just blithely assume that random mechanical error is known to make up the highest proportion.

Share this post


Link to post
Share on other sites
Posted (edited)

Just consider the back of an envelope calculation for what the ToA difference is for a pair of antennas 30cm apart: 1 nanosecond - and that is for a signal in line with the antennas - so you are measuring angle by comparing differences between 0 and 1ns - needing you need to measure many times faster than 1ns - or downconvert in a phase matched way.

Possible, just not simple or cheap 😞

Edited by stadl

Share this post


Link to post
Share on other sites

ha yeah  TANSTAAFL

now i am curious about what you would see if you just had 2 antennas 1/2λ apart, and then you summed them and eyeballed a spectogram.  complex data and noise, but maybe you could identify a signature visual pattern — which is what our brains are good at — in the form of comb filtering either side of the bearing that you could keep sweeping past and hone in on.

 

Share this post


Link to post
Share on other sites

Really not sure what that would do - That is a question for people with more brains than me.

Share this post


Link to post
Share on other sites
On 10/2/2018 at 5:54 AM, @~thehung said:

ha yeah  TANSTAAFL

now i am curious about what you would see if you just had 2 antennas 1/2λ apart, and then you summed them and eyeballed a spectogram.  complex data and noise, but maybe you could identify a signature visual pattern — which is what our brains are good at — in the form of comb filtering either side of the bearing that you could keep sweeping past and hone in on.

 

Physics is a bitch:

For unambiguous results, the antennas should be spaced half a wavelength apart, or less. However, this can result in significant mutual coupling between elements, which means that each antenna's phase measurement will be corrupted by the other's.

Share this post


Link to post
Share on other sites
On 10/12/2018 at 2:58 PM, ResidentNeville said:

yes it seems like that could be somewhere between significant and deal breaker

as far as i can tell 'mutual coupling' is expressed as an impedance but its unclear to me how this effects magnitude/phase and hence adds ambiguity

however - what i proposed above *might* diminish its relevance.

suppose you can to rotate the orientation of the two antennas while mapping the spectra of their summed signal for all compass directions.  even with a fairly FUBAR signal you should be able to identify some kind of pattern of symmetry either side of the points where total cancellation would (ideally) be expected.  and drastically more precisely than you could vs merely watching a meter. 

Edited by @~thehung

Share this post


Link to post
Share on other sites
On 10/13/2018 at 5:38 AM, @~thehung said:

as far as i can tell 'mutual coupling' is expressed as an impedance but its unclear to me how this effects magnitude/phase and hence adds ambiguity 

Mutual coupling means that the signal received in one antenna will be transmitted to the other nearby antenna through electric and magnetic fields. Basically, two antennas in close proximity act a bit like one antenna.

 

On 10/13/2018 at 5:38 AM, @~thehung said:

suppose you can to rotate the orientation of the two antennas while mapping the spectra of their summed signal for all compass directions.  even with a fairly FUBAR signal you should be able to identify some kind of pattern of symmetry either side of the points where total cancellation would (ideally) be expected.  and drastically more precisely than you could vs merely watching a meter.  

This exacerbates the problem. There is still a lot of crosstalk between the two antennas, except now the crosstalk is a few orders of magnitude more complex.

 

On 10/13/2018 at 8:40 PM, Dasa said:

How would a cantena go with FM? works well for WiFi 

Unlikely to be useful here. In general, in WiFi antenna is going to need to be 6 times bigger to work for FM. The wavelength of WiFi is 12cm. The wavelength of this FM is 70cm.

  • Like 1

Share this post


Link to post
Share on other sites
5 hours ago, ResidentNeville said:

Mutual coupling means that the signal received in one antenna will be transmitted to the other nearby antenna through electric and magnetic fields. Basically, two antennas in close proximity act a bit like one antenna.

This exacerbates the problem. There is still a lot of crosstalk between the two antennas, except now the crosstalk is a few orders of magnitude more complex.

maybe you missed my point.  i understand well enough what it is in general terms.  and after some reading i am under the impression that the mutual coupling effect on two receiving monopoles is independent of transmitter direction.

it will surely play havoc on SNR regardless of what exactly it does to phase/magnitude, so there is that — but my point was there will still likely be symmetry either side of the bearing to the transmitter (pointing towards or away).  if you expect otherwise for some specific reason, then do elaborate. 

i can see that accounting for the errors added to any two omnis rooted to the ground would result in calculating a useless range of maybes, but propose that when data from all orientations is mapped, a true bearing could be inferred.

Share this post


Link to post
Share on other sites
On 10/15/2018 at 7:17 AM, @~thehung said:

and after some reading i am under the impression that the mutual coupling effect on two receiving monopoles is independent of transmitter direction.

I do not know if that's true or not, but I know its irrelevant.

The mutual coupling between two receivers is heavily dependent on receiver direction. Spinning the antennas will fuck your shit up. It doesn't matter if the transmitter is fixed or not.

Share this post


Link to post
Share on other sites

perhaps i should have said "...incidence of incoming signal direction of arrival" instead.  i wasnt talking about the orientation or type of transmitter.

"Note that in general, the definition of the receiving mutual impedance implies that the receiving mutual impedance is dependent on the direction of the plane wave, which is used as the excitation source.  But for omni-directional antennas such as dipole and monopole  antennas, the receiving mutual impedance is independent of the azimuth angle of the excitation plane wave." https://studylib.net/doc/18215614/mutual-coupling-in-antenna-arrays

still not convinced you grasp what i am proposing about the symmetry of the spectral information.  do you have anything more elucidating to add than 'it will fuck your shit up'?

 

Share this post


Link to post
Share on other sites

Still not convinced that you grasp what mutual coupling is. Still not convinced that you know about about RF.

You just repeated the exact same nonsense. "Incidence of incoming signal direction of arrival" uses more impressive words but it says the same thing, and reiterates a flawed understanding of RF.

Your idea requires the two receiving antennas being totally independent, but they will not be, because they will be coupled. Please don't make me say it a fifth time.

The receiving antennas mutual coupling might very well be independent of the azimuth angle of the excitation plane wave, BUT THE ANTENNAS MUTUAL COUPLING WILL NOT BE INDEPENDENT OF THE RELATIVE POSITIONS TO EACHOTHER.

Please don't make me say it a sixth time.

Share this post


Link to post
Share on other sites

LOL nice one.  so maybe you know little beyond generalities then?  because youve had ample opportunity to add something pertinent.

if you check the relevant section of the technical paper i quoted from there is actually a graph showing the amount of mutual impedance due to coupling as a function of antenna separation in fractions of a wavelength.  yeah, somehow that hadnt escaped my attention, and has never been in dispute. just so we are clear, i was supposing that the distance between the rotated antennas would remain constant.

IF it is your contention that no useful phase information (ie. ZERO) could possibly survive the presence of ANY amount of coupling (Z > 0), then why not educate me on how and why this is the case?  shouldnt be too hard if you actually know what you are talking about and are actually interested in a civil discussion. 

 

Share this post


Link to post
Share on other sites
5 hours ago, @~thehung said:

i was supposing that the distance between the rotated antennas would remain constant.

This is a deeply flawed supposition. Physics is a bitch. RF physics is particularly nasty.

To show how deeply flawed this is, start by imagining that we have two cars spaced 10m apart, and we start rotating the cars. You can't just keep the cars the same distance apart when you rotate them. That phrase has no specific meaning. It could mean:

1) The distance between the two nearest points of the cars remain the same

2) The distance between the center of mass of the cars remains the same

3) The distance between the center of area stays the same

4) And so on and so on

So what do you mean when you say that "the distance between the rotates antennas would remain constant"? What kind of keeping the distance constant would prevent you from adding a mindblowing layer of complexity to the mutual coupling? Not only do you have to choose between the examples I gave for cars, but in the RF world there's a magnitude more complexity and possibilities. Did I mention that RF is particularly nasty?

5 hours ago, @~thehung said:

IF it is your contention that no useful phase information (ie. ZERO) could possibly survive the presence of ANY amount of coupling (Z > 0)

That is not my contention. My two contentions so far are:

1) Putting two antennas 0.5lambda apart is not likely to be even remotely straight forward due to mutual coupling (not to mention a whole host of other real-world complexities). You agreed with this contention and said "yes it seems like that could be somewhere between significant and deal breaker."

2) Your proposed solution to the coupling issues, that is rotating the antennas, does not solve the issues. It makes it worse. Much worse. You'll start to see why this is the case when you consider how its impossible to even "keep the antennas at a constant distance apart".

^^ Someone with more braincells than me would've used the word "centroid" instead of "center of area".

Edited by ResidentNeville

Share this post


Link to post
Share on other sites

ohhh righhhhht ... i think i see the cause of misunderstanding now.  that car analogy OMG WTF! :P

well i cant blame you since i never stated it all explicitly in one chunk. 

- the antennas are monopole omnis. 

- when i originally said "rotate the orientation of the two antennas while mapping the spectra of their summed signal for all compass directions" i was talking about rotating the array — ie. two antennas both affixed to a single rigid structure at around 1/2λ apart — so that a measurement is made at 'every' bearing represented by a straight line between antenna A and antenna B.  

so for example, ignoring coupling, if the transmitter was due north, you would expect near cancellation when the array is pointed at either 0° or 180°.  yes?  

 

 

Share this post


Link to post
Share on other sites

Haha. It took us way too long to figure out we were talking about different things.

On 10/18/2018 at 6:42 AM, @~thehung said:

so for example, ignoring coupling, if the transmitter was due north, you would expect near cancellation when the array is pointed at either 0° or 180°.  yes?   

I'm not 100% sure of that, but I can't right now think of any reason that wouldn't be true in an ideal world on a 2D plane. I'm happy to continue on the assumption its true...

 

Edited by ResidentNeville

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×