Antenna placement matters here too. Make sure you don't have two antennas that are mounted so close together that their signals are bleeding into each other. If you have signal strengths that are insanely high (stronger than -40), we have found it can frequently be attributed to a combination of radios mounted close together in a board along with antennas that are mounted too close together. In these situations just moving the antennas farther apart from each other can be enough to improve the quality of a link to a point where it will function at the levels you need.

what about a dual polarity antenna that you have no way to separate? With -20s since they are on the same antenna. is it just not feasable to use a HDDA-5W dish and expect them to not interefere with each other?

I have a spectrum analyzer. I will be hooking it up to the antenna port on a stacked card in a RB433 to see what is visible between the two cards. I will upload pictures. I tested this once before and I can confirm there was significant bleed over.

check the following images - test done with one card transmitting into a dummy load and spectrum analyzer hooked up to the antenna port on the stacked card that is not transmitting.

Tests below are all 20mhz channels with different power settings. at full power 25db you could only run two cards if they were on on channels 5745 and 5825 and you woud "get by." Even then they are still seeing each other at -93 on the edges.

Now this is 10mhz channels. Now we are able to put two cards on top of each other with full power and not bleed over too much. You could run 3 non overlapping channels at full power 25db or use all 5 if you are running a lower power setting such as 15db.

Side by side comparision. I didn't have a RB600 laying around at the moment so i just put two RB433/333 beside each other and did the test again. Bleed over was almost completely gone. Very minimal bleed over at 25db power. If i dropped back to 20db power it was gone compleatly.

my problem is that even if i use 5200 and 5805 (600+ mhz separation) I still get half speed. I dont think its intereference really, i think its a software problem possibly. or a power problem on the minipci riser ... who knows.

ps - thank you for doing some testing and posting pics. thats how we will finally solve this.

Now try with High power cards, such as XR2s And as well R52N.. I think the results would be interesting to see.

We always sell 1 radio card in one board on the same freq, we typically don't have much issue with putting a 5 gig and 2.4 gig card in the same box though.

wirelesswaves

I too am a ham and have run this RB433AH/2-XR5 config. Without spectrum analyzers, just intuition--and just plain old RF experience DXing and running VHF/UHF and above contest stations that have come in #1 or close to it (AA9D) I have come to some conclusions. I worried about the use of two high power, especially XR5's, cards in the same box before I even powered it up. It just felt wrong--and certainly doesnt work.

Some in this thread seem to want to squeeze more blood out of this rock than could possibly be available. I think many are trying to solve the wrong problem, my opinion only.

The first hint at what I am taking an educated guess on is IF connectivity. I had two XR3 cards connect on a 5GHz frequency, i.e. one not convertable to the offset 3.65GHz frequency. Seems like simple proximity allowed connectivity between the two cards. Who knows at what frequency? Dont even want to know. What would you possibly expect with two or more stacked cards? I too was contemplating putting in a 'ham engineered' shield of aluminum foil inside of the static bags that cards or boards come in. Seemed a little too rigged up before I proceeded. Of course that begs the question, why are there three mini-pci slots. Another topic all together.

However......I already had two XR5's in the top and bottom slot to separate them by a slot (wow). Trying to get something out of a config that was mounted in a place not readily accessible, AND, coincidentally I had the cards connected to a dual-pol Gabriel 2ft 'drum' dish I surmised that a workable config could be had. The cards have pretty good receive characteristics, and ample power for sure, and I if I operated with only one of them transmitting, maybe I had a chance. Two cards transmitting produced the expected results--de-reg, de-reg. Of course you know where I am going. Dual Nstreme with 220MHz separation actually worked--to a fashion. A short 7 mile link, more than 700ft off the ground at each end with nothing in between was able to connect.

Where is the catch? Band was pretty crowded, so results must be interpreted with a grain or more of salt. Attempting to config the correct polarity/frequency/direction combo is very critical in the ability to obtain reasonable bitrates and stability of the link of course. I was able to operate turbo in one direction (downstream the desirable direction if you can make it work), and 20MHz channel on the other. The isolated feedhorn collared dish provided excellent near field ingress from many other co-located APs. However, putting out copius power into the cross-pol feedhorn has the expected effect. Lowering power (1dBm increments) gave me the best possible compromise.......

Now, for the real world performance of this highly utilized backhaul. The more traffic either direction sustained, had a direct relation to the CCQ of the 'other' direction. CCQ of course is estimated by performance and jitter only, since it doesnt seem to be a metric available in dual-nstreme. This was expected, and could even make this config unusable, but I believe it provides some empirical evidence that multiple cards can be used in thee 433 class environment, with the caveat that you might not get maximum bitrate. However again[!], I was able to run the turbo channel at 54Mbps@2, with the cost of lowered stable bandwith in the other direction. With careful tuning of power and micro adjustment of channel (5MHz increments), a full-duplex, albeit asymmetric backhaul can be operated. Many times the asymmetry matches the real world bandwidth requirement.

The moral of this verbose ham-engineered story is that I got what I expected out of something that was deployed with a useless second card. As with DXing and contesting, it all seemed to be in the antenna guys. Put up 4-18el beams on 144MHz, and you know its going to burn through a lot more distance than the same EIRP obtained using a single 18el beam with 6dB more output power. Im sure that could be disputed. Thats why we used 4 different yagis on six meters. Some could hear a certain distant station, while other antennas could not hear a peep out of the same station....

Bottom line is that I believe IMHO that the wrong problem was attempting to be solved. For those that just want it to work with all three slots filled, eventually the stress will beat you.....BTW, an RB600 with two cards doing the same thing doesnt have these problems...Hmmm....Is there a significant null to the side of the cards that takes you below the threshold of pain?

Thats my story, and Im stickin to it...Hope this in some way helps someone leverage what they have invested and obtains the benefit of an asymmetric--and very fast (done properly) backhaul that leverages just the benefits and good characteristics of each component of the total link........If I missed something obvious, forum, or helped in any way, please advise...

Hi Znet

To be sure it'll be us hams that cure the problem. I will read your post a few times more when I am less under the influence of Spanish liquer, but its a friday night and in this tiny village, 1 nights bar life is all I have to look forward to.



kblazk
Can you run the test again, with the 2 cards stacked on top of one another.
Can you put a Tee piece on the 50ohm port and insert 2 x 50 ohm loads to create a VSWR of 2:1.

kblack, you will need to give the card some gusto and get it to pass 20mbps to a neighbouring client unit ie bandwidth test.

Then can you somehow couple the analyser onto the pigtail of the second radio card or use an RF probe at the ground point of the antenna port.

I want to know what happens when the card is subject to VSWR. And its effects on the second card.

Are there any circulating RF currents on the outer of the sheath?

My gut feeling is that the cheap pigtails radiate when mismatched.

73's de Simon EA5/G0MGA

kblazk

One more thing, current probe and voltage probe!!!!!

kblazk

Just looked at your test again.. And not so sure I agree with your method.

I assume the card connected to the SA is disabled, but in this state what is Zsource? I do not have access to a circuit for the PA and I am not sure where the diversity switch is, are we sure that the pigtail to the SA is terminated under these conditions?

Or was the card enabled? and PA powered up?

znet

Yes, I see where you are coming from, IMHO and going slightly of thread, there seems to be a gang of folk who are hell bent on trying to get 108Mbps throughput at any costs on a backhaul yet at the expense of quality, and at higher frame retries.

Yet with narrower bandwidths and hence increased SNR, quality improves and retries are less.

Why spend 20% of my day trying to figure out why I have 20% packet retries on my backhaul, reduce the bandwidth and enjoy the 20% wasted time with a paella and sangria.


My caveat, sometimes quality beats capacity, you are 5-9+20 old man, but can you repeat your callsign again

BTW, an RB600 with two cards doing the same thing doesnt have these problems...Hmmm....Is there a significant null to the side of the cards that takes you below the threshold of pain?

I've got 600's with the same problem. The setup I have that seems to work well is an XR5 and R52 or R52H combination. R52s or R52H's in the band cause problems, -20dB signal inside the box. XR5 with either card doesn't seem to have the same issue.

As long as my signal levels are in the low 70's or better, I've had good CCQ @ 54mb, even with R52H. Mid 70's or worse, then I have problems with two radios on the same band.

Well at least we are getting somewhere. Maybe there is a problem when 2 of MT designed cards are connected on the same backplane.

Interesting that a Ubiquity mixed with an MT does not give issues. That is the first comment I have heard so far relating to mixed manufacturers on same backplane.

But I surely would like to know if the problem is related to radiated RF cos of a mismatch. At this point your feedback is great, but no doubt your antennas are perfectly matched.

WHAT HAPPENS WHEN ONE RADIO FACES A MISMATCHED LOAD.

Wish ZNET would confirm and further tests as there are very few of us with access or even own an SA at these frequencies.

Well at least we are getting somewhere. Maybe there is a problem when 2 of MT designed cards are connected on the same backplane.

Interesting that a Ubiquity mixed with an MT does not give issues. That is the first comment I have heard so far relating to mixed manufacturers on same backplane.

But I surely would like to know if the problem is related to radiated RF cos of a mismatch. At this point your feedback is great, but no doubt your antennas are perfectly matched.

WHAT HAPPENS WHEN ONE RADIO FACES A MISMATCHED LOAD.

Wish ZNET would confirm and further tests as there are very few of us with access or even own an SA at these frequencies.

Do you want me to do the same test with two Ubiquity XR5's ???

Yeah, we bought 2 of the MPCI-FLEX from http://www.adexelec.com as well, and then realized that they overcharged us when it's almost half the price here in Taiwan . I admit, it's not the easiest thing to find, probably why adexelec charges so much.

In any case, the MPCI-FLEX type extender is convenient

Can you send me 10 of this to Spain at Taiwan cost? : D!

okay, im following up to the original topic here. I previously had 2 XR5 cards in rb433ah boxes providing ptp backhaul about 22km. Anytime the second card associated I would get bad CCQ and bandwidth would drop in half or less. Originally I thought it was interference with multiple cards in the same box. Today I finally was able to test the second link with 2 additional rb411ah boxes. Completely separate routerboards enclosed in their own metal enclosures. The only thing being shared still is the HDDA-5W dual polarity dish, one radio link on vertical and the other on horizontal. End result, same result as before... if both are associated I get horrible CCQ and less than half the bandwidth. Even if the second link isn't doing anything but associated. So, my conclusion is that the HDDA-5W dish isn't usable - does this sound plausable?

For testing I had one radio at 5845 and the other at 5210. I figure thats plenty of separation. I moved the second radio from 5210 to a handful of other freqs and no noticeable changes. Does anyone have any other ideas? Even with nstreme2, one radio doing tx and the other doing rx I had the same problem.

So, my conclusion is that the HDDA-5W dish isn't usable - does this sound plausable?

Yes. There is a 32dB difference between the two polarities, but with 2 separate radio cards we have seen some interference issues between the two pols due to that not being enough separation. This isn't an issue on 802.11n 2x2 links due to the way the standard works, so you could go to a single 802.11n radio on each side using 2 antenna connections to resolve the issue. Another option is multiple single pol antennas with some vertical separation to help reduce the interference.

You may have significantly better luck if you properly line the enclosures with Microwave absorber material.

hxxp://www.cumingmw.com/Product%20Applications ... ssion.html


My Appologies:
I did not read changeip's post about the seperate enclosures.
Perhaps someone else may benefit from this.

If I perform a scan I see the other radio at -23db. One horizontal the other verticle. Is this a normal value when using a dual polarity dish?

Another option is multiple single pol antennas with some vertical separation to help reduce the interference.

Is vertical separation better than horizontal?

By vertical seperation the assumption must be made that it is on a vertical structure ie. tower.
On building tops the option may be either vertical or horizontal.

Spacial diversity is the key.

Which is better is dependant upon the environment.

Definitely, according to the results of the spectrum analyzer, two high-powered cards in a rb433ah, overlap among themselves.
Solution db down at least tolerable?
Use rb433ah in two different bands 2.4 and 5.8 GHz. to avoid overlap?

kblazk thanks for your great contribution.!