Diversity Operations

User avatar
EA1DDO
Posts: 5
Joined: Sun Apr 09, 2017 4:21 pm
Contact:

Re: Diversity Operations

Postby EA1DDO » Fri Jan 26, 2018 5:54 pm

SDA3.png
SDA3.png (23.56 KiB) Viewed 6260 times
User avatar
w-u-2-o
Posts: 1739
Joined: Fri Mar 10, 2017 1:47 pm

Re: Diversity Operations

Postby w-u-2-o » Fri Jan 26, 2018 9:20 pm

Maximo,

To be coherent, or phase and sample synchronized, you have to be much, much better than that.

Let's take a 20M frequency for example, say 14.2MHz. At that frequency, a 10 degree phase shift is equivalent to approx. 2nS.

That is why both ADCs need to be clocked from the same sample clock, and clocked at the same time, i.e. there should be no differential delay between samples.

Now our radios do allow for an external frequency reference, so you could feed each one from the same reference using a set of matched length cables. This would at least get everything on the same timebase. BUT...the radios have no mechanism for synchronizing their sample times.

That said, if you had two radios, both locked the same 10MHz reference, and two data streams, theoretically they would be phase locked at some random phase. You could then simply guess at the phase difference you needed to steer a beam by trail and error until you got what you wanted. With only two radios and two antennas this would not be too difficult. Indeed, that's what we currently do in our current dual ADC radios. Even though they are phase and sample locked, we just guess. However in your picture you have three antennas. That would be darn near impossible to guess at.

73,

Scott
User avatar
EA1DDO
Posts: 5
Joined: Sun Apr 09, 2017 4:21 pm
Contact:

Re: Diversity Operations

Postby EA1DDO » Fri Jan 26, 2018 10:15 pm

That is why both ADCs need to be clocked from the same sample clock, and clocked at the same time, i.e. there should be no differential delay between samples.


So, there isn´t any way to clock two separated SDR ?

Perhaps.... a NCO ?

https://en.wikipedia.org/wiki/Numerically_controlled_oscillator
User avatar
w-u-2-o
Posts: 1739
Joined: Fri Mar 10, 2017 1:47 pm

Re: Diversity Operations

Postby w-u-2-o » Sat Jan 27, 2018 1:49 am

Certainly there are ways to do what you want, just not with the hardware and firmware that exists in any openHPSDR architecture radio, including Apache Labs radios.

There would need to be a sync signal in addition to the 10MHz, and it would need to be accurate to around +/-1nS, and/or there would need to be a way to calibrate the various receivers using a calibration signal to measure what the differentials are.

73,

Scott
User avatar
EA1DDO
Posts: 5
Joined: Sun Apr 09, 2017 4:21 pm
Contact:

Re: Diversity Operations

Postby EA1DDO » Sun Jan 28, 2018 3:36 pm

Then, an easier way; 2x SDR boards (Hermes, Orion, etc.), sharing clock, in same enclosure, let´s say back-to-back.
This way you get two coherent boards, with two coherent receivers and two coherent transmitters, starting point for beamforming.

GPS is not mandatory.

SDA4.png
SDA4.png (17.21 KiB) Viewed 6223 times
User avatar
w-u-2-o
Posts: 1739
Joined: Fri Mar 10, 2017 1:47 pm

Re: Diversity Operations

Postby w-u-2-o » Sun Jan 28, 2018 5:08 pm

Yes, you'd have to modify the boards to allow a single board to supply the same ADC clock to all ADCs. I have not studied the problem. Surely not a difficult technical challenge, however you may have to devise a buffer/driver circuit to ensure signal integrity.

Then there is the problem of writing firmware and software to exploit this hardware. There would need to be sufficient time tags embedded in the IF data streams to be able to match up the streams on a sample basis coming from multiple radios. Again, not a difficult technical challenge, there are thousands of DF systems around the world that do exactly that .
K9RX
Posts: 275
Joined: Fri Apr 14, 2017 3:47 pm

Re: Diversity Operations

Postby K9RX » Sat Apr 13, 2019 1:02 pm

It's been awhile since I last visited this particular subforum ... I have 2 antennas on HF - for example on 15 I have 6/6 long boom monobanders and also a 4 element custom SteppIR.... so I once again revisited diversity. I have these nasty signals on 15 towards the west making difficult copy at times on SP/Asia. I'd LOVE to find out what the cause of this particular one is - its shaped like an obtuse triangle and the peak is S7 - S8 with the base being broad - maybe 5 - 8 Khz.

So I tried diversity again and WOW. FINALLY I have a way to get rid of these things .. AND reduce the baseline as well. I made a video - I'll need to figure out some way to post it to Youtube. There's no sound (I don't port sound to the PC) ... but it's pretty obvious what is going on. I found a signal that was off the center point of noise (the noise repeats as well - all up and down this and other bands) close enough such that when properly set up/nulled using diversity you can see a) that the signal does not change in peak amplitude and b) the noise floor goes WAY down - nearly 25db down! From -90 dbm to -114 (2.6Khz bw)!

I have automated all my antennas and had planned on carrying this through to allow full diversity when desired ... I hope to finish it up in the next week or two pending work load. Just need to carry this ability through to 160 and 80.

Gary, K9RX
K9RX
Posts: 275
Joined: Fri Apr 14, 2017 3:47 pm

Re: Diversity Operations

Postby K9RX » Sat Apr 13, 2019 1:15 pm

sorry - misspoke there - the noise floor does not go down by 24db ... that nasty noise signal PEAK goes from -90 down to the (new) noise floor at 0-114db. Since these interference signals repeat and are broad the overall noise floor does go down a decent amount ... looking at the video it changes about 5 db.

Gary
User avatar
w-u-2-o
Posts: 1739
Joined: Fri Mar 10, 2017 1:47 pm

Re: Diversity Operations

Postby w-u-2-o » Sun Apr 14, 2019 3:29 pm

Gary,

It sure is nice to have an antenna farm <Jealous>!

Did you get the video on Youtube yet?

73,

Scott
K9RX
Posts: 275
Joined: Fri Apr 14, 2017 3:47 pm

Re: Diversity Operations

Postby K9RX » Mon Apr 15, 2019 12:21 pm

I've not had a chance to look in to how to do it ... besides the fact that there is no audio... so I'll work on getting something better.

One reason why I did this is because the only video (on UT) that shows coherent beamforming doesn't show it WITH a signal! So I for one was left asking : but how does it work with a signal there? Thus my desire to confirm it not only eliminates the noise but it doesn't impact the signal.

here is a Dropbox link for anyone wanting to see it now... note a couple of things - although there is a signal there (for the first 2/3 of the video) you can clearly see if it happened to be where the peak of my noise is it would have been completely unreadable, not even know someone was there... 2) even though with the noise there as is - i.e. not on the noise peak but offset on other 'crap' ... although you can see peaks I can say that that signal was still difficult to copy. 3) You can clearly see the signal is completely out of the noise when "ENABLE" is hit - easy copy!

data: when I centered on the noise peak (not quite there in the video) the signal went from a -91dbm to -115dbm for a 24 db drop... AND the noise floor dropped by 5 db. One has to keep in mind - this is MY particular ugly scenario ... the improvement isn't that it made the floor any better , rather it is that the noise was very broad and clearly raised the noise floor everywhere...

https://www.dropbox.com/s/mohqju0an7fmv ... .webm?dl=0

Gary , K9RX
K9RX
Posts: 275
Joined: Fri Apr 14, 2017 3:47 pm

Re: Diversity Operations

Postby K9RX » Mon Apr 15, 2019 12:25 pm

One other comment: just to be clear - this feature is only "local" in nature - in 2 ways: 1) it acts on local noise, not sky wave and 2) frequency-wise it is somewhat narrow in response so one has to optimize at a particular frequency (the one of interest)... I don't know that it is overly broad once tuned at a particular point.

Gary
User avatar
w-u-2-o
Posts: 1739
Joined: Fri Mar 10, 2017 1:47 pm

Re: Diversity Operations

Postby w-u-2-o » Mon Apr 15, 2019 2:44 pm

The dropbox link worked perfectly, I was able to play the video directly from dropbox. And it was very clear what was going on even if I couldn't see you manipulating the diversity controls. Next time it would be handy to see them. You can also change your display settings to move the panadapter amplitude scale to the center if you want to make more precise measurements.
K9RX wrote:One other comment: just to be clear - this feature is only "local" in nature - in 2 ways: 1) it acts on local noise, not sky wave and 2) frequency-wise it is somewhat narrow in response so one has to optimize at a particular frequency (the one of interest)... I don't know that it is overly broad once tuned at a particular point.
And these are all necessary conditions, and they mirror my first post in this topic back on page 1.

People have difficulty understanding exactly what all this is and what it is capable of. First, there is a deficiency in our amateur radio lexicon. Most people will call the signal you wanted to get rid of "noise". But noise is random. Atmospheric noise coming from antenna 1 is not the same as atmospheric noise coming from antenna 2. The signal you dealt with is not noise. In the parlance of professional RF engineers, it is an "interfering signal", or simply an "interferer". It comes from a point source (or a relatively small distributed source like a section of power line). That signal will be exactly the same at two receive antennas in the same proximity, simply offset by some amount of time defined by the angle of arrival relative to the arrangement of the antennas. This set of properties is also known as "coherence". And since "diversity" is really coherent beamforming as implemented in PowerSDR and Thetis, what you wind up doing is steering an antenna pattern null in the direction of the interferer while still having enough of a lobe in the direction of the signal you want to receive. And since you've got a massive antenna farm there <jealous> you are likely to be able to find a combination of antennas that have the correct geometry to get this done.

Cool stuff, isn't it?

73!

Scott
K9RX
Posts: 275
Joined: Fri Apr 14, 2017 3:47 pm

Re: Diversity Operations

Postby K9RX » Tue Apr 16, 2019 12:53 pm

Yes - very cool. I played with it on FT8 ... that 'interferer' that i have I think you'd agree is a nasty one. I'd love to find out what is causing it - Plasma TV, grow lights ... haven't a clue ... It is worst on 15 and in that direction covers the whole band with the repetitive nature seen there (the next "peak" is just off the right at about 84Khz separation from the one seen - both up and down).

anyway on FT8 it would increase some signals (S/N as reported) and strangely decrease some. Not sure what is happening there. The antennas are at different heights with the 4 element SteppIR currently at 37' and the stacks for that band at 45/90. I probably should have used the bottom antenna alone with the SteppIR ... might have been more informative... the plan is (and has been), if I can get it done the SteppIR will be at 86' and I'll add 7/7 on ring rotors at 40/75' for 10M and an 8 element LFA 6 meter in the doing as well. Just need to come up with a reasonable 'reference' antenna for the WARC bands as trying to use one of the other yagis just doesn't work. The requirement that BOTH antennas see the offender well is paramount. I have parts for a 2 element SteppIR that I might put together in to a 2 element yagi and mount it on the rotating tower for this purpose.

Gary
K9RX
User avatar
w-u-2-o
Posts: 1739
Joined: Fri Mar 10, 2017 1:47 pm

Re: Diversity Operations

Postby w-u-2-o » Tue Apr 16, 2019 3:13 pm

anyway on FT8 it would increase some signals (S/N as reported) and strangely decrease some.
No, this is not strange at all. You really have no idea what the composite antenna pattern is due to the beamforming. The two-element antenna array has not been engineered. You, me, and most people are just rolling the dice and hoping we get lobes and nulls where we want them. There is generally only one, sometimes two, phase and amplitude values that cause a deep null to form in the direction of an interferer. If the resulting lobes are not commensurate with the angle of arrival of the desired signal it's not going to work very well.

The easiest array to model azimuthally is a pair of verticals (dipoles or monopoles). If you are using a set of stacked Yagis that point in the same direction your are moving things in elevation. And so on. Simple two-element array stuff.

73!

Scott
Joe-W4WT
Posts: 83
Joined: Sun Apr 09, 2017 5:27 pm
Location: Cumming, GA

Re: Diversity Operations

Postby Joe-W4WT » Tue Apr 16, 2019 5:56 pm

I tried out the diversity option on Thetis 2.6.4 last night on 160m and it seemed to me to be easier to find the null than Protocol 1 versions. I did notice an anomaly though. Several times, when I had nulled out interference, I found if I pressed the Disable button to disable diversity then pressed enable to turn it back on I found the null point had moved. I had to re-null the interference and it usually meant moving the phase control about 45 degrees or so; usually clockwise. Once I did this, toggling enable/disable didn't cause a problem again until I had moved to a different part of the band and nulled out another problem, toggled enable/disable, and once again saw that the null was lost and had to be found again.

The nulls produced seemed to be very good, deep, and eliminated the interference completely! I was able to null out everything I tried.

Joe W4WT
User avatar
WB4AIO Kevin
Posts: 13
Joined: Sun Apr 09, 2017 3:28 pm
Location: Western Pennsylvania
Contact:

Re: Diversity Operations

Postby WB4AIO Kevin » Sun Jun 16, 2019 9:53 pm

w-u-2-o wrote:Good idea, Bill! I will make this thread a "sticky". Here's what I've posted in the past, perhaps it will be of help...

Diversity can only be achieved using radios that have two separate ADCs. Currently this would be the 100D, 200D and 8000.[...]



Can coherent beam-forming diversity be done with my Anan-10 and OpenHPSDR mRX 3.4.9 ? I believe my Anan-10 has 2 ADCs, or am I wrong? But it does not have an "RX 2" port -- only a TX port and an RX port.


73,

Kevin.
Better fidelity means better communications.
User avatar
w-u-2-o
Posts: 1739
Joined: Fri Mar 10, 2017 1:47 pm

Re: Diversity Operations

Postby w-u-2-o » Sun Jun 16, 2019 11:02 pm

As the post you quoted already points out, the answer is no. Your radio only has a single ADC.

Return to “Diversity Operations”