Co-phasing LPDA questions

[benbenrf]

zguy1234

When you say custom made - what exactly do you mean?

Other than of course the frequency/band-width (225 - 400), did you give the "maker" a Tau or Sigma figure to work with, or a 3dB beam-width you wished to have(?), were these exsisiting antennas you have had modified in some way ........ just what was the objective in having custom work undertaken?

Going to the effort you have, and I am speaking for myself now (i.e. personal opinion - nothing more) I would have thought "in for a penny, in for a pound" - why not have your antenna potential max'd out?

225 - 400Meg's is bang on Fleetsat comm's (as I am sure you know), and while there is nowhere near as much analogue Tx to listen to today as there was in the 80's-90's, there is a ton of digital stuff - much of it in clear (i.e. not encyphered), and relatively easy to de'mod (demodulate). But a lot of it is sent out in circular polarisation - and receiving Circular polarised Tx with a conventional LPDA can be a pig at the best of times (especially when it comes to Sat Tx).

At min you will need to mount your 2 LPDA's with motors to re-orientate the element planes from vertical to horizontal. but better still would be to add another set of elements - as in the attached photo - with switchable connections, which will give you the potential to receive horizontal/vertical, as well as both left & right twist signals - to exploit left/right twist transmissions you will need the additional coax connection as in the 2nd photo (... and of course you will need to be able to switch them). You'll now have the flexibility in one antenna [type] that would otherwise require at least 2 antenna's to achieve - an advantage over even a pair of custom made axial mode helical antennas!

..... just my ideas - you've gone to this effort of getting bits made up specificialy, you've got receivers well capable of exploiting such an antenna arrangment, and the extra cost wouldn't be much [more].

Max out the oppurtunity ......(?)

 

[prcguy]

Yes, advertised two way splitter loss is in addition to the 3dB theoretical, but combining two in phase signals with a two way splitter backwards (combiner) will give you 3dB more signal at the common port minus the slight insertion loss of the combiner. That's usually less than .5dB for a Wilkinson type splitter. TV splitters are another story.

You can even split a signal then recombine with another backwards and only incur minimal loss instead of 3dB loss + another 3dB loss for a total of over 6dB loss as some would think.

I have a bunch of high power filters that consist of a two way splitter feeding two identical filters then recombined again with a backwards splitter. The total loss for the entire assembly is about .5dB and the reason they were split is because a single filter will not handle the required power level.

So, if zguy can space two log periodic antennas properly he will get close to 3dB more gain and SNR plus a narrower pattern on the antennas by combining.
prcguy

What splitter/combiner has a loss of only 1.68 db? Mini-Circuits sometimes specifies loss as "loss above theoretical", so a 1.68 db spec really means 1.68 db ABOVE 3 db.

Example: http://www.minicircuits.com/pdfs/ZAPDJ-2.pdf

Your splitter likely has a loss of 4.68 db total.

The loss of a practical, well-made splitter/combiner will be in excess of 3 db, therefor the theoretical gain of your stacked antennas will be less than with a single antenna.

If you were to phase the antennas with phased cables, no combiner, you could see close to the theoretical 3 db maximum gain (minus cable losses), but only on one specific narrow range of frequencies.

Bottom line: Your idea simply will not work without violating the laws of physics.

Edit: Many of us have seen phased LPDA's, particularly at cable tv head end sites. The caveat there is, they're generally being used on a much narrower range of frequencies, and are 'cross-eyed' in the manner that prcguy stated above. They are also NOT using a splitter/combiner

 

[zz0468]

Yes, advertised two way splitter loss is in addition to the 3dB theoretical, but combining two in phase signals with a two way splitter backwards (combiner) will give you 3dB more signal at the common port minus the slight insertion loss of the combiner. That's usually less than .5dB for a Wilkinson type splitter. TV splitters are another story.

You can even split a signal then recombine with another backwards and only incur minimal loss instead of 3dB loss + another 3dB loss for a total of over 6dB loss as some would think.

I have a bunch of high power filters that consist of a two way splitter feeding two identical filters then recombined again with a backwards splitter. The total loss for the entire assembly is about .5dB and the reason they were split is because a single filter will not handle the required power level.

So, if zguy can space two log periodic antennas properly he will get close to 3dB more gain and SNR plus a narrower pattern on the antennas by combining.
prcguy

Thanks for refreshing my memory...

I did some experiments, results posted in another thread here, maybe a year ago. If I recall correctly, with two combiner inputs, fed perfectly in phase, overall loss was the published insertion loss above 3 db. So, in other words, a combiner rated at 1 db above 3 db loss, with two 0 dbm signals fed in phase, combined to an output of 0 dbm minus the insertion loss, or -1 db, for no net gain. Drop one input, and the loss goes from 1 db, to 4 db.

So, I disagree about the power gain, but do agree about the pattern... that is, if he can get the spacing right.

 

[prcguy]

This morning I grabbed two random 50ohm 1-2GHz splitters, connected them back to back and the total insertion loss measures .5dB across most of the band on my network analyzer.

That was .5dB loss for splitting the signal, the loss of two 4" long .141 flex cables between the splitters, the second splitter as a combiner and an additional 6" long cable.

If you consider only half that mess would be used for combining only, the net gain would be around 2.75dB. Lower frequencies would have slightly less loss.
prcguy

Thanks for refreshing my memory...

I did some experiments, results posted in another thread here, maybe a year ago. If I recall correctly, with two combiner inputs, fed perfectly in phase, overall loss was the published insertion loss above 3 db. So, in other words, a combiner rated at 1 db above 3 db loss, with two 0 dbm signals fed in phase, combined to an output of 0 dbm minus the insertion loss, or -1 db, for no net gain. Drop one input, and the loss goes from 1 db, to 4 db.

So, I disagree about the power gain, but do agree about the pattern... that is, if he can get the spacing right.

 

[zguy1243]

Great Information. I have two of the logs on the tripod now and I have rigged a T support with extra room on them to change the spacing of the antennas. It works. I have spent about 2 hours with it last night. I gotta run to work now. I will post some more info on it tonight.

This morning I grabbed two random 50ohm 1-2GHz splitters, connected them back to back and the total insertion loss measures .5dB across most of the band on my network analyzer.

That was .5dB loss for splitting the signal, the loss of two 4" long .141 flex cables between the splitters, the second splitter as a combiner and an additional 6" long cable.

If you consider only half that mess would be used for combining only, the net gain would be around 2.75dB. Lower frequencies would have slightly less loss.
prcguy

 

[zz0468]

This morning I grabbed two random 50ohm 1-2GHz splitters, connected them back to back and the total insertion loss measures .5dB across most of the band on my network analyzer.

That was .5dB loss for splitting the signal, the loss of two 4" long .141 flex cables between the splitters, the second splitter as a combiner and an additional 6" long cable.

If you consider only half that mess would be used for combining only, the net gain would be around 2.75dB. Lower frequencies would have slightly less loss.
prcguy

Sounds like high quality transformer type splitters. There are some out there with loss less than .25 db above 3 db splitting loss. But the OP is stating 1.68 db.

I wish I could find the previous thread from a year ago, I'll just have to duplicate the tests again. One thing that's unresolved is whether the OP is using a resistive splitter, or a transformer type.

Another issue that will come to play in a practical application of splitters is that they only meet published specs when properly terminated. Neither the stacked LPDA's, nor the receiver itself, is going to be 50 ohms across the band of interest, so the performance of the splitter will only get worse.

 

[zguy1243]

Sounds like high quality transformer type splitters. There are some out there with loss less than .25 db above 3 db splitting loss.

Was Mike including the assumed 3 db splitting loss in his results? His numbers looked like 1/2 db loss in total.

Last night using what gear I have talked about with the antennas 25 inches apart, equal length jumpers to the MC combiner and into my R8500 I tested on 236.7 Mhz, 307.0Mhz, 363.1Mhz with weak S-1 or less ARTCC ground controller signals. I found better signal quality with the 2 antennas on all freqs with gain at 236.7 being the best. Again this is all by ear and the Smeter on the 8500. The Smeter was about 1/2 S unit to 3/4 S unit higher on all signals. The antennas spacing is really sensitive. The antennas beam width is much smaller now too I can tell. I had the antennas about 4 feet off the ground, I need to get them in the air to really test. Will update soon with some photos.

I

 

[zz0468]

Was Mike including the assumed 3 db splitting loss in his results? His numbers looked like 1/2 db loss in total.

He was describing actual measurements, so that would take into account the 3 db. So... here's the story:

Use one input and and terminate the other. Loss WILL be 3 db plus resistive losses. Now, feed both with identical in-phase signal. Take combined output, and split again for 3 db loss. He's seeing .5 db down end-to-end, so he's seeing signals summed at the common port, but he didn't state what he's actually measuring at that point. I did a similar test once, for a similar thread, and I seem to recall seeing essentially 3 db loss with one port terminated, and 0 db loss (neglecting the resistive losses) with both inputs fed in phase. But then, splitting the combined output again like prcguy did, there should be an additional 3 db loss, not .5 db total.

So, there is a conflict in what he's reporting, and what I recall from my test. What bothers me about his test is, he has signal passing through TWO passive devices with no gain, and he's reporting only .5 db down for the whole thing.

Last night using what gear I have talked about with the antennas 25 inches apart, equal length jumpers to the MC combiner and into my R8500 I tested on 236.7 Mhz, 307.0Mhz, 363.1Mhz with weak S-1 or less ARTCC ground controller signals. I found better signal quality with the 2 antennas on all freqs with gain at 236.7 being the best. Again this is all by ear and the Smeter on the 8500. The Smeter was about 1/2 S unit to 3/4 S unit higher on all signals. The antennas spacing is really sensitive. The antennas beam width is much smaller now too I can tell. I had the antennas about 4 feet off the ground, I need to get them in the air to really test. Will update soon with some photos.

Got a 3 db attenuator? Make some measurements... two antennas with the pad, and one antenna without. The measurement should be about the same if the antennas are phased correctly.

I'll try to test a couple of splitters tonight to confirm whether or not prcguy is right, and I've lost a few brain cells along the way. =)

 

[prcguy]

I measured .5dB total loss for a signal passing through one splitter then both outputs run into both inputs of another splitter and then out the common port.

These are Wilkinson type dividers which are commonly sold by MiniCircuits and other microwave/RF vendors. These have two 1/4 wavelength matching sections of about 75ohms each and usually a 100ohm resistor across the two ends of the matching sections. The resistor helps with isolation by applying some signal from one output port directly to the other which will be 180deg out of phase from a signal that passes from an output port through one of the internal 1/4 wave matching sections, then the other to arrive at the opposite output port 1/2 wavelength down the path. 1/2 wavelength = 180deg phase shift. Hope that all made sense.

When combining two in phase signals the resistor is canceled out and all you have is two 1/4 wave matching sections connecting to a single point with very little loss. Splitting a signal equally will leave you with no more than 1/2 power at either port because you can bend the laws of physics. But there are no lossy paths inside the splitter and when using it as a combiner, its more of an impedance matching device to connect two 50ohm lines to a common 50ohm point.

There are other types of splitters like TV types which are more of a transformer with two secondaries. These should act very similar to a Wilkinson when combining if the phasing is connected correct inside.

Then there are broad band resistive dividers which have about 6dB loss through all ports and would not be suitable for lossless combining.
prcguy

He was describing actual measurements, so that would take into account the 3 db. So... here's the story:

Use one input and and terminate the other. Loss WILL be 3 db plus resistive losses. Now, feed both with identical in-phase signal. Take combined output, and split again for 3 db loss. He's seeing .5 db down end-to-end, so he's seeing signals summed at the common port, but he didn't state what he's actually measuring at that point. I did a similar test once, for a similar thread, and I seem to recall seeing essentially 3 db loss with one port terminated, and 0 db loss (neglecting the resistive losses) with both inputs fed in phase. But then, splitting the combined output again like prcguy did, there should be an additional 3 db loss, not .5 db total.

So, there is a conflict in what he's reporting, and what I recall from my test. What bothers me about his test is, he has signal passing through TWO passive devices with no gain, and he's reporting only .5 db down for the whole thing.



Got a 3 db attenuator? Make some measurements... two antennas with the pad, and one antenna without. The measurement should be about the same if the antennas are phased correctly.

I'll try to test a couple of splitters tonight to confirm whether or not prcguy is right, and I've lost a few brain cells along the way. =)

 

[zz0468]

Ok... you're right!

I just did the identical test. In my case, I ended up with .6 db. So much for those brain cells.

I did a couple of other things to refresh my memory... Remove one side of those connected splitter/combiners and loss goes to 6 db. I also introduced some delay on one side. It doesn't take much mismatch between line lengths before the loss goes all over the place.

 

[Token]

zguy, a question. Have you considered spacing the stack so that the distance varies? ie, at the front of the antenna (shorter, higher freq, elements) the antenna booms are closer together than at the back? Forming a partial V when viewed from the ends of the elements?

A picture of an HF application of the general concept here:
http://www.pust-norden.de/oloog/lp1_gr.jpg

And a pdf of an R&S here:
http://www2.rohde-schwarz.com/file_12438/HL046E_brief_e.pdf

In this manor you can optimze the spacing for the frequency range of each element of the LPA. And both of those imaged examples have much greater bandwidth than you are looking for, so the build would be less radical. They cover an octive or more of bandwidth, but you are looking for less than 50%.

T!
Mohave Desert, California, USA

 

[benbenrf]

Token - the configuration you are describing adds little to nothing in terms of bandwidth. The objective is to shape radiation pattern/angle.

As far as gain is concerned - yes, as a dual polarised feed element for a parabolic dish, some gain benefits can be realised relative to the overall space/volume the feed takes up, but as a stand alone antenna element, nope.

... and it's worth noting that SNR is generally improved as well.

 

[Token]

Token - the configuration you are describing adds little to nothing in terms of bandwidth. The objective is to shape radiation pattern/angle.

As far as gain is concerned - yes, as a dual polarised feed element for a parabolic dish, some gain benefits can be realised relative to the overall space/volume the feed takes up, but as a stand alone antenna element, nope.

... and it's worth noting that SNR is generally improved as well.

I am familiar with this application, having used it (and measured it on a range) several times, and did not mean to imply it would improve bandwidth. My comment about bandwidth was only to point out that the antennas I was showing images of work across a wider bandwidth (as a percentage of freq) than the relatively narrow goal that zguy is after so the physical angles would not be a pronounced. Gain flatness and consistent radiation pattern across the operating range of the antenna was what I was after when suggesting it.

Keeping two log periodics at a constant physical distance apart (vs electrical distance based on lambda) results in optimal gain at one frequency with a gain fall off as you depart from that freq. Gain is, naturally, directly related to beamwidth, anytime the beamwidth increases gain must decrease. Using the approach I outlined can flatten the gain response by keeping the beamwidth more constant, it will not make the peak gain any more, but it can (when done correctly) cause the gain of the combined pair to be more even across the bandwidth of the antenna. When done correctly it can make very near the optimized gain across the usable bandwidth of the antenna, instead of at one point in the bandwidth.

T!
Mohave Desert, California, USA

 

[zguy1243]

I am familiar with this application, having used it (and measured it on a range) several times, and did not mean to imply it would improve bandwidth. My comment about bandwidth was only to point out that the antennas I was showing images of work across a wider bandwidth (as a percentage of freq) than the relatively narrow goal that zguy is after so the physical angles would not be a pronounced. Gain flatness and consistent radiation pattern across the operating range of the antenna was what I was after when suggesting it.



Keeping two log periodics at a constant physical distance apart (vs electrical distance based on lambda) results in optimal gain at one frequency with a gain fall off as you depart from that freq. Gain is, naturally, directly related to beamwidth, anytime the beamwidth increases gain must decrease. Using the approach I outlined can flatten the gain response by keeping the beamwidth more constant, it will not make the peak gain any more, but it can (when done correctly) cause the gain of the combined pair to be more even across the bandwidth of the antenna. When done correctly it can make very near the optimized gain across the usable bandwidth of the antenna, instead of at one point in the bandwidth.

T!
Mohave Desert, California, USA

I have been doing some hands on testing here with some different configurations. I got my support structure up this weekend that will house my rotor so I can make some test videos showing the antennas in a co phased setup vs. standalone and a few different configurations of each at select frequencies across the 225-400Mhz band. I should have some of this theory put in into practice on video by the end of the week. All of the great information from all the guys out there has been very informative.

 

[zguy1243]

Keeping two log periodics at a constant physical distance apart (vs electrical distance based on lambda) results in optimal gain at one frequency with a gain fall off as you depart from that freq.
T!
Mohave Desert, California, USA

I have noticed this and it changes based on antenna spacing but is not very pronounced over the 175 bandwidth. There is a sweet spot where I can keep all my test freqs up in signal strength across the band within reason. I am going to fab some spacers that will allow me to angle the antennas at each other and be adjustable and see how it affects my test freqs across the band. Once I prove the best setup in regard to spacing I will need to go further with proper test equipment to fine tune these things and put some paper with my results.

 

[zguy1243]

Some exciting test results

Spent most ofthe day today getting the antennas mounted on the tower. Right now the antennas are at 43 feet above the ground. I have mast mounted the 400Mhz low pass filter and Angle Linear Preamplifier near the top of the tower in a weather proof enclosure. The results of the entire project are better than I had expected. I am doing all of the video editing now that show the setup and more comparisons. Here are some highlights of signals I am detecting with consistency from Gordon county Georgia.

285.5 S-1 Memphis center at Joelton controller ground signal audible at 190 miles
348.625 S-1 or less Indy center controller ground signal audible at better than 200 miles. Indy center airspace is 200 miles from here and I am unsure of what sector this is in Indy airspace so I know its better than 200 miles.
363.0 S-1 Memphis center controller at Nashville. 190 miles

All of these signals are very weak, at some points the S meter does not even move but I can make out what is being said. Of course these signals are undetectable on any of my other antennas. The radio used was a Icom R8500. The main feedline is LMR600. Video to come soon showing comparisons and the reception of these signals above. And I will be showing the reception of the above signal after 24hours to show this was not a atmospheric enhancement situation.

Jody

 

[zz0468]

Excellent! Glad to see it worked out for you.