Would like some feedback/suggestions on some problems I'm having creating a point to point link.

EQUIPMENT: Two RB433AH Routerboards w/ 3.27 firmware & R5H wireless cards.
PATH PROFILE: 42km (26mi). Clear Fresnel and -100 or better noise floor
ANTENNAS & CABLES: 10ft Andrew Parabolic (V pol) waveguide dishes w/ WR137 (5.9-7.0GHz) waveguide
CONFIGURATION: AP Bridge (unit #1) & Station (unit #2).
OTHER INFO: Radios share the waveguide with 6.9GHz microwave feed (single duplex). There are no bandpass filters installed. This same link w/ TrangoLink-45 radios @ 5.8GHz produces astable 28mbps link with no packet loss & 2-3ms round trip ping times.

I have been trying for a week to get this link stable with RB433AHs and R5Hs radios. I do not know alot about Mikrotiks, but I am not completely clueless either. I can get the radios to see each other at default power with a fluctuating -45 to -60 signal strength. The CCQ is the problem. The links barely stay stable and often completely die only to reconnect later. Rx CCQ (on the station) is often 90%+ but Tx is very low (20-40%). I have tried lower power, raising power, Nstreme, CSMA off, increasing time-out intervals, using only 6mbps rates, changing frequency, etc. Nothing helps. Often the radios will simply not authenticate and/or associate with each other. Other times they fail with the message "disconnected, extensive data loss"

I would be blaming my waveguide setup but the Trangos perform beautifully at 5.8GHz on this same exact link. The RouterOS reports fluctuating signal strength on one half of the link which could be slightly misaligned dishes due to the original purpose of the link only being one-way traffic. But again, the Trangos had no problem.

One issue in the back of my mind is my original test of the R5H units on my bench. I directly connected the radios to each other (with a LMR400 cable) at default power which (after reading posts on burnout issues of the R5H) makes me worry they are damaged. I just received two new R52N radios today and plan on testing them later this week. Should I be considering XR5 radios instead? I originally thought the R5H was the best for long-distance performance but maybe I was mistaken?

I can provide screenshots of my configuration or link status, etc. I am pulling my hair out trying to get this link stable. Any help would be greatly appreciated.

I attached a screenshot of my status window. The TX side is going bananas... between -70 and -35 and constantly fluctuating. The RX side stays consistent between -47 and -52.

CCQ is always low. Once, when throttling all the way down to 6mbps-only with Nstreme and max power I got 98%/60% CCQ. I cannot reproduce this though.

I cannot get the radios to see each other on 5GHZ-5MHZ or 5GHZ settings. Only 5GHZ-Turbo & 5GHZ-10MHZ will they see each other.

Freqency scan on the AP side has basically -109 or better noise readings. Frequency scan on the Station side has -96 to -99 readings. The AP side is in the middle of a city (130ft high) and the Station side is in an extremely rural area (400ft high) with basically no one living around it and no local WISPs in the area. I found it odd that I got better noise readings in the city than in the country.

Why would you directly connect two radios together with a piece of coax? I suspect their output amps are blown but it is possible they survived if they have some kind of protection built in for that.

The R5H have been great performers in their short history. I am not as fond of the R52H cards that you have coming.

Assuming the R5H cards are not damaged....

you might be seeing interference from the 6.9GHz radios sharing the waveguide. The R5H probably has a much wider comb filter than the Trango 5.8GHz radios. It could be hearing noise from the waveguide that was filtered out when you were using the Trango.

I'll be interested to hear your results with the R52H cards.

Tom

Why would you directly connect two radios together with a piece of coax? I suspect their output amps are blown but it is possible they survived if they have some kind of protection built in for that.
Tom

Because I wanted to test them so I would know how to create the link prior to going out to the remote locations so as to save time. In retrospect, it was not a smart move, I realize this. But take comfort in the fact I bought everything from Roc-Noc so it can only mean more business for you as I continue to tear up my gear.

The second radios are R52Ns, not R52Hs. Will be testing those Thursday. I will put them directly in the radios without testing so as to rule out bad radios.

Can your turn off the 6.9GHz system while you do your testing?
It's a good way to isolate the problem.

There is a high chance that you killed your radios. You should NEVER connect two transmitters to each other.
Try again with another pair of R5H or XR5.

As for your radios not seeing each other problem
Be sure you are using channel 5180, 5200, 5220 .... 5745, 5765, 5785 ...
Search the forum for more details.

Write both those Hs off. Never perform a back to back test without attenuation. 60db on a 13dbm card generally produces -45 receive, if you use an XR, it will start at +25, so you MUST use additional attenuation. Most manufacturers will list the maximum receive level before receiver damage, though some don't...... The rule of thumb is never expose your receiver to more than -30.

I have seen better performance out of SR5s than XR5s, and the SRs actually output higher power if you have to run at 24-36mbps, only when you hit 48-54 do the XRs take over. Receive performance of both has been identical from the links I've worked on.

Pop a spectrum analyzer on your port and see what is there, have a look at the broad spectrum using a 3mhz div then look at a 100khz div, is there anything over your stated noise floor? Are you sure the 7ghz system is transmitting data or is it idle?

I popped your link parameters in and there is no reason to use high power radios for this shot. A CM9/R52 will work fine, and should produce -51 even with the additional loss from the mismatched 137.

Matt

Hey Matt,

I brought a picture of the general configuration of our sites. Each site has two microwave radios operating somewhere between 6.5-6.9GHz depending on the license for each link. One receives the signal from the previous site and then the other transmits the signal to the next site. It is one way only and there is no digital component to these radios. They are 100% analog in the signal they are transmitting. If your are curious, it's purpose is to relay video/audio. I am trying to underbuild seven sites that are identical except for the first and last sites which only have a transmitter and receiver, respectively.



I have made note of the circulator I am connecting to. You will also notice the 6GHz equipment have bandpass filters on them so I suspect they are basically immune to anything I am doing.

I certainly have access to spectrum analyzers but my engineers aren't very savvy so I will probably need to tell them what the point is and what they are looking for. Is the purpose just to verify the noise readings in the 5.0-6.0GHz range that I already got from the Trango and Mikrotiks? I tested every site with both radios and found no noise anywhere at any site.

I ordered one R5H (wishful thinking) and two XR5s. Didn't see about the SR5 but they seem to be the same price as the XR5s (if not higher.) But you think an SR5 would do better for my purposes? For reference, here are the distances for each one of the links I am trying to underbuild.

Link 1: 42km
Link 2: 40km
Link 3: 38km
Link 4: 39km
Link 5: 43km
Link 6: 60km


I was planning to do a pure R5H arrangement with two radios and one RB433AH at each site. Do have a different recommendation?

My original budget was roughly $600 per site (7 sites) but with bandpasses and possibly going XR5/SR5 $800-$900 is becoming more realistic.

Can your turn off the 6.9GHz system while you do your testing?
It's a good way to isolate the problem.

I considered that, but I don't want to disrupt the link unless it is my last troubleshooting option. I'm going to try different radios and a bandpass first. Thank you for your suggestion.

Circulator isolation I guess -20 to -30dBc, 7G TX ~ 27dBm (or more) => -3dBm of 7GHz at 5.8 radio input ???

Quick test: as you have lots of 5.8 signal anyway, run both at max nominal power but insert some fixed sma/N attenuator between 5.8 radios and N to WG adaptor. May be 30 or more dB. Just to make 5.8 RSL around -70dBm. At this point, 7G out of band signal will not be -3dBm at 5.8 front-end but -33 ... -40dBm (within 5.8 HW limits I guess).

If this test improve CCQ, spend $$$ on 5.8 BPFs/LPFs or a 7GHz slug trap

Regards;

[edited] 5.8 are TDM radios ... 30dB att each side = 60dB total ... well, run some calculations to have at least 15dB S/N @ 6Mbps just for testing. The idea is reduce the 7G signal at the 5.8 front end. Or insert 35dB N/SMA attenuator one side at a time. If the RX CCQ improves where the attenuator is installed you have the answer.

Ok, just got back from testing. Here is what I have discovered.

I think at least one of my original R5H radios were bad. When I replaced the R5H in my master unit with a new R5H the issues with the fluctuating signal strengths disappeared. I also added a 5.8GHz bandpass filter to each side at the same time so I'm not sure what action helped. I just wanted to get the link working not figure out which variable was responsible. My CCQ improved to 70% and was fairly constant.

On the slave side, I replaced the other R5H with a new XR5. It didn't help the CCQ. So I started playing with settings and found that if I lock the data rate to a single value the CCQ goes to 100%. I settled on 36mbps which gave me a solid one way UDP rate of 46mbps. Not bad.

I ran into problems with my networking configuration where I can't ping the other sides. I had it working and then totally lost it towards the end of my setup. I'm sure it's not the wireless link. But I'm going to post in a different forum on this issue.

I will do more tweaking and testing with the bandpass and such later on and report back. But at least I know it is possible to get a good working link on my waveguide system alongside our 6ghz equipment. Now to test the other six links...

As the gentleman from Brazil noted, the radios we are using here, are in no way impervious to out of band noise. There is still an AC component that is present to the receiver that must be dealt with. Choosing a data rate in lieu of letting the radios negotiate - and having better success, illustrates a potential effect of other frequencies on the wire.

The 'readings' you get from internal processing are relative. Until you look with an analyzer, and look at a subset of the channel you wish to use, you won't know. I have dealt with a narrowband 75hz wide (yes, 75 hertz), signal that trashed an 802.11 radio using a 20mhz channel. That spike was enough to introduce so many hardware retransmits as to make that channel useless. That is why you need to look at problematic channels at higher resolution. If it's 'flat', relative to the noise floor, then you should be good. If there is more than a 5db (or higher) signal in one of those 100khz segments, then you may have an issue.

Here are some path calcs. I used 30db of gain at each side, which is less than your dishes have, but this is for illustration only. I used 13dbm radio output and 5800mhz. A CM9 outputs 13dbm at a 54mbps air rate.

Link 1: 42km = 26.25mi, 140.25db fspl, -67db
Link 2: 40km = 25mi, 139.82 fspl, -67db
Link 3: 38km = 23.75mi, 139.38 fspl, -67db
Link 4: 39km = -67db
Link 5: 43km = -67db
Link 6: 60km = 37.5mi, 143.34 fspl, -70.34db

Since you indicate that 36mbps is something you seem amicable to... a CM9 outputs 17dbm at 36, thus giving you another 4db. An SR5 at 36mbps yields 24dbm, and an XR5 26dbm. Falling back to 24mbps gives both cards an additional 2db of output.

If a bpf or lpf eliminates the 7ghz issue, then another consideration is that an SR5 at 54mbps has a receiver threshold of -74, whereas at 36mbps it increases to -83.

I can only give my opinion regarding the R5H - and that is that I can not recommend them per my personal experience.

XR cards tend to not work well with non-XR cards....

Matt

Are you sure my dish is only adding about 30db? Specs I had from Andrew show it is a 41db dish in the 5.9-7.0 range. Would there be that big of drop off down to where my Mikrotiks are living in the 5.7-5.8GHz range?

I'm actually getting -53 through -57 on my link @ 36mbps w/ full power. So your free space loss calculations are sort of close.

I am going to go scrap my original R5H plans in favor of a pure XR5 arrangement. I had high hopes for the R5H and maybe they would work fine if I applied myself to it. But I just want to reduce my exposure to possible problems so I'm going to pay a premium for the XR5s. I have no real reason to believe R5H wouldn't work, but I just want to play it safer.

I only moderately understand the point you are trying to make on the spectrum analyzer. As I understand you, the purpose of it is to get a more precise and scientific look at what noise/frequencies are making it in to my radio? But I'm confused as to what benefit this will give me now that I have bandpass filters on both sides of my link. Even if noise is making it through and disrupting the full potential of my link, what could I possibly do about it once I know it is there? I have already tried changing channels and the same result occurs on both 40MHz blocks. I don't mean to sound critical of this idea, I have the tools to do this. I just don't know how I would act on any information the spectrum analysis would give me.

I am testing my second link with the Mikrotiks today. Personnel issues may prevent me from getting to my third link which produced terrible results (6mbps.) with the Trango and no bandpass. I will update the thread when I have more tests computer.

Thanks! Please keep the advice coming as it is all very helpful.

Tested Link #2 and Link #3 yesterday

Both ran solid 100% CCQ @54mbps.
Tested with TCP & 128bit AES WPA2 encryption enabled

LINK #2 (40km): AtoZ=46mbps / ZtoA=46mbps / Bidirectional=60mbps total / power levels between -49 & -53
LINK #3 (38km): AtoZ=46mbps / ZtoA=46mbps / Bidirectional=60mbps total / power levels between -55 & -58

Link 3 was the one that had given me only 6-10mbps on the Trango with a stable connection only achieved @ 6mbps. But adding the mikrotiks w/ bandpass seems to have resolved that issue. Both links produced basically the same results with slightly different power which leads me to believe that this is the max performance I'll see from the units in my situation.

It also makes me less happy about the 1st link only being stable @36mbps when it basically has the same specs as these other two.

It turns out our hugely expensive spectrum analyzer we recently got only goes up to 3GHz so we will need to contract out for that part.

I mentioned I was using less gain than what your dishes had....

Having an indepth look at your spectrum would definitively indicated if you had an issue in one of your 40mhz blocks.

Want to update this. It is funny how much 7 months of tesing and experience with underbuilding mikrotiks changes your perspective.

First, as others mentioned, I needed to filter the very powerful licensed 6GHz signals away from my 5GHz radios. So I bought a bunch of l-com bandpass filters and the problem was solved. I still keep the data rates locked to a single speed. All my links are essentially 100% CCQ. I also have a 97km NLOS link that is pulling down a solid 24mbps link at 90%+ CCQ with power fluctuating between -60 and -80 throughout the day.

It also turned out that I blew up the R5H radios by impulsively testing them back to back at full power. I'm sure subjecting them to high power 6GHz radios didn't help the situation. I have since opted for a pure XR5 environment with the previously mentioned bandpass filters.

The 42km link I originally spoke of is running solid at 48mbps and -53db.

I am probably 3 months away from putting this entire relay network into production so I will use this thread to update with pictures, details, and final performance benchmarks. (Not alot of waveguide underbuilding discussion here so hopefully this can become a resource to others.) It should end up really nice. 7ms pings end to end as of this writing. I'm sure there will be more problems before it's all over but so far so good.

Thank you all for the help.

I am now using two RB411AH w/ matching SR71-15 on this link. Link is not perfect and signal strength bounces around. Not sure if it's an issue with the cards or what. But if I cap each side at 30M in bandwidth test then it stays solid. If I go higher the cards start out good and then bomb down to a bad CCQ and lower rates.

Almost forgot, passing 30M/30M I get 3ms to 8ms ping times consistently. No traffic I get 1 to 2ms consistently. Very impressive, Mtik!

Out of curiosity, did you try the sr71 link with vpls in lieu of wds? You should see a nice increase in throughput. WDS in 4.x is problematic and n cards typically see better results when not using wds.

what do you mean not using WDS.
I have a link with R52n with EoIP tunnel. If i use WDS is it going to drop some throughput?

I am now using two RB411AH w/ matching SR71-15 on this link. Link is not perfect and signal strength bounces around. Not sure if it's an issue with the cards or what. But if I cap each side at 30M in bandwidth test then it stays solid. If I go higher the cards start out good and then bomb down to a bad CCQ and lower rates.

Almost forgot, passing 30M/30M I get 3ms to 8ms ping times consistently. No traffic I get 1 to 2ms consistently. Very impressive, Mtik!

Hi, this is a capture of my configuration, is a 44km PTP with rb433 r52hn antenna Kozumi (28db V / H).

disabled csma
protocol= nv2
tdma period size = 10 (is 10 because 2 was unstable and 7 smaller bandwidth)
Extension Channel = HT BELOW CONTROL
distance = 44
adpative noise inmunity = ap&client I can not get a link like yours!
sorry for my English

We are also trying to build a link with RB433AH's / SR71-15's and getting very lopsided results, both on the tower and on the bench. Transfer rates 70M x 8M. colebert, could you share your (sanitized) config?

Thanks