Community discussions

MikroTik App
 
taduikis
Member
Member
Topic Author
Posts: 436
Joined: Sat Jul 07, 2007 12:09 pm

Cable lenght on 5GHz links

Sun Feb 10, 2008 12:32 am

I would like to ask how badly does cable lenght affect link quality on 5GHz wireless networks. Because in my current setup I have three 23db panel antennas connected with about 5-6 meters of S400 cable (both ends are N Male).. The wifi nic's are all R52H. Two of them are working quite ok, but third one needs to point at something more distant, and signal level is very low.. I'm now currently running a x86 based PC full of R52H's, that explains 6 meters of cable to the antennas. Sadly i haven't got a possibility to check the performance with shorter cable (without purchasing an integrated RB solution), so I decided to check if someone here can share his experience.. I'm thinking of a routerboard with 3-4 minipci slots, which I would place near the panels if it has some noticeable improvement over my current setup.. But let's make that another question if it's necessary..
 
MyThoughts
Member Candidate
Member Candidate
Posts: 218
Joined: Sat Sep 17, 2005 9:07 pm

Re: Cable lenght on 5GHz links

Sun Feb 10, 2008 3:12 am

I currently use ~6m of LMR 400 to feed a couple of my backhaul 5 GHz units with no issues.
A well made cable of that length should have a max loss of ~3.5 dB including connector losses.

Cheers
 
RK
Long time Member
Long time Member
Posts: 565
Joined: Tue Nov 21, 2006 11:22 am
Location: Winnipeg, Canada and Central America

Re: Cable lenght on 5GHz links

Sun Feb 10, 2008 7:01 am

You shouldn't put too many radios next to each other.

Did you try swapping the one cable? Try it with another radio?
 
taduikis
Member
Member
Topic Author
Posts: 436
Joined: Sat Jul 07, 2007 12:09 pm

Re: Cable lenght on 5GHz links

Sun Feb 10, 2008 10:23 am

Well, that's a good idea, i'll give a try with another radio..
The note about putting radios close to each other concerned me a little bit. I am aware that signal generated by other radios is interpreted as noise, but i'm not sure what solution this might have.
And to stay on the main topic, is there a difference between cable length on 2.4GHz and 5GHz networks? I've heard somewhere that 5GHz is more sensitive to that..
 
DesertTek
newbie
Posts: 43
Joined: Sun Feb 10, 2008 9:10 am

Re: Cable lenght on 5GHz links

Sun Feb 10, 2008 11:55 am

Did you mean to say S400 cable? As in firewire cable? Maybe I am unfamiliar with S400, but we use LMR400 cable at up to 70ft runs on our towers to 5gHz antennas with no problems. However, we do use premium Times Microwave cable, so loss is minimal. Regarding your question abour 2.4 to 5Ghz cable, we us LMR400 for both with no noticable differences in signal loss.

I have noticed some vendors have advertised cable perfect for one frequency or another, but I always that that was a marketing ploy. Can anyone shed more light on the facts?

BTW, how are you attaching so many Nmale connectors to the PC? Did you modify the case?
 
RK
Long time Member
Long time Member
Posts: 565
Joined: Tue Nov 21, 2006 11:22 am
Location: Winnipeg, Canada and Central America

Re: Cable lenght on 5GHz links

Sun Feb 10, 2008 1:17 pm

is there a difference between cable length on 2.4GHz and 5GHz networks? I've heard somewhere that 5GHz is more sensitive to that..
Yes, higher frequency signals lose power more rapidly in both cables and air. But, at this length, the difference between 2.4 and 5 is minimal.
 
User avatar
tgrand
Long time Member
Long time Member
Posts: 667
Joined: Mon Aug 21, 2006 2:57 am
Location: Winnipeg, Manitoba, Canada

Re: Cable lenght on 5GHz links

Sun Feb 10, 2008 2:32 pm

If I am not mistaken S400 has a solid shield, while LMR400 has a braided shield.
Very easy to kink S type cables.
Used by cable companys
 
taduikis
Member
Member
Topic Author
Posts: 436
Joined: Sat Jul 07, 2007 12:09 pm

Re: Cable lenght on 5GHz links

Sun Feb 10, 2008 2:49 pm

The cable is marked as "S400 Low Loss Coaxial Cable" or something like that. It has no noticeable differences compared to the LMR-400, the diameter is the same, the inner conductor is cooper, the shielding is similar.. I believe it could be an identical substitude for LMR400 cable, only made by different manufacturer..

I'm running 5 wifi NIC's, three on 5ghz and two on 2ghz. 4 x R52H and 1 x Compex WLM54AG23 (which I couldn't mention with a good word).. The motherboard is actually placed in large metal waterproof box, with special mountings for everything (MB, PSU..), so N connectors are attached to it..

I now think that i'll replace all the N connectors with much more attention to quality, because some of them doesn't look so well.. So another question rises here, what N ends are the best, crimped, soldered or clamped.. :?
 
MikeMc
just joined
Posts: 4
Joined: Wed Feb 06, 2008 7:44 am

Re: Cable lenght on 5GHz links

Mon Feb 11, 2008 6:46 am

S400 is the same as LMR400. It uses the same connectors and such.

Loss in the coax is

2.4 Ghz - 22.5db/100m
5.8 Ghz - 35.5db/100m

as you can see there is a 50% increase in loss between 2.4 and 5.8 Ghz. You also need to remember that this loss is bi-directional so you have the same loss on the transmit and receive sides. And if you are using the same coax on both ends now the loss is 100%.

For example using the numbers above, the difference over the total link (don't forget the 5.8 Ghz has a LOT more path loss than 2.4Ghz) Site A transmit loss is 13 DB, Site B receive loss is 13 db so the total increase in loss over the link is 26 db. That is the equivlent of doubling the length of the 2.4 Ghz run and then some.

That's why on sites that have the dish up high, the unit has an up/down converter on the dish and the frequency going up the tower to the dish is a lot lower frequency (70 Mhz in our case). This reduces the loss at 5.8 Ghz to 1-2 db max.

If I was running for any distance at 5.8 Ghz I would consider upgrading to LMR900 at the AP site.
 
JR
Member Candidate
Member Candidate
Posts: 120
Joined: Tue Nov 08, 2005 3:27 am

Re: Cable lenght on 5GHz links

Fri Feb 15, 2008 4:16 am

re: how badly does cable lenght affect link quality on 5GHz wireless networks?
is there a difference between cable length on 2.4GHz and 5GHz networks?

I suggest you take some salt with some of these relpies.
Here's a good link:
http://timesmicrowave.com/content/pdf/broadband.pdf

JR.
 
taduikis
Member
Member
Topic Author
Posts: 436
Joined: Sat Jul 07, 2007 12:09 pm

Re: Cable lenght on 5GHz links

Sat Jul 10, 2010 1:22 am

I now think that I'll replace all the N connectors with much more attention to quality, because some of them doesn't look so well.. So another question rises here, what N ends are the best, crimped, soldered or clamped.. :?
After quite some time I have to return to this question, so without dedicating a separate thread I'd like to ask again: what are the differences in quality between soldered or crimped N type connector tips. Crimped are obviously a lot easier to put on, but does that affect signal greatly? I'm asking because ever since I've soldered these tips with quite bad fortune and transition to crimping would save me lots of pain and concerns.. When a planned change of 10 cables is ahead, I don't really look forward to soldering 20 more connectors :) If I hear positive replies here, I'm going to buy connector pin crimping tool asap :)
 
changeip
Forum Guru
Forum Guru
Posts: 3830
Joined: Fri May 28, 2004 5:22 pm

Re: Cable lenght on 5GHz links

Sat Jul 10, 2010 7:58 am

ask any car battery cable guy, he will tell you crimp is much better connection than solder. you get better conductivity with crimp when using braided cables I believe. Solder also is more brittle, it can break with bending whereas a crimp is somewhat flexible. vibration, etc is better with crimp.
 
taduikis
Member
Member
Topic Author
Posts: 436
Joined: Sat Jul 07, 2007 12:09 pm

Re: Cable lenght on 5GHz links

Sat Jul 10, 2010 11:24 am

Well, battery connectors are in completely different league. Here we're talking about high frequency which is more sensitive to everything. And I disagree that crimping is better on conductivity than soldering when speaking of ordinary current transfer, not mentioning that solder conserves joint point and and thus protects from oxidation. Crimping is easier than soldering, and is convenient where soldering is impossible or very difficult to achieve (for example high current battery cabling). Flexibility, resistance to vibration, etc I can give you. Too bad it's of no importance for N type connectors, since they are enclosed and no bending occurs for solder joint whatsoever. And you've made good point and attracted my attention mentioning braided cable. I've got one with solid core, so I don't think crimping is possible on those. Soldering is difficult mainly due to that inner insulation which is in the way and unfortunately melts and "de-centers" the copper core. Other than that it would be perfectly good joint. And I use solder with high % of silver.
 
n21roadie
Forum Guru
Forum Guru
Posts: 1949
Joined: Fri Aug 07, 2009 10:36 pm
Location: Limerick,Ireland

Re: Cable lenght on 5GHz links

Sat Jul 10, 2010 8:11 pm

Loss in the coax is
2.4 Ghz - 22.5db/100m
5.8 Ghz - 35.5db/100m
A well made cable of that length should have a max loss of ~3.5 dB including connector losses.
Cable loss is measured onto a dummy load on a test bench, and connectors have insertion loss quoted,
We have to remind ourselves that a radio card + cable + antenna form a RF circuit, which can only achieve maximum effective radiated power (ERP) when all of the components in that circuit have matching impedance at the selected frequency. This could for example be providing a excellent tuned circuit at 5550, resulting in maximum ERP but at 5180 or 5800 have a much reduced ERP, replace any of the components with a different brand could and most likely move the frequency where max ERP is?

As there is many variables involved, I take a simple approach to this “if it’s not part of the circuit then it cannot effect it” let me explain this logic, if you have a radio card, pigtail, link lead , antenna that’s four items in the chain but remove the link lead, you now have three to worry about, I have found the Integrated Antenna’s (to house the board + radio card + pigtail direct to antenna ) and now the grids with integrated box at rear of the antenna, the best solution but maybe not perfect, this approach solves two problems,
(1) reduced power loss to antenna,
(2) less components exposed to the elements,

By this approach I have reduced one of my links from -61 to -50, before I started my calculation from tech info was for an improvement of 1.5db with the removal of 1m lmr400 link leads at both ends of the link,
It could be said maybe I had a bad batch of 1m LMR400 leads; I could have taken them back to the test bench and tested them on to a dummy load but either way I am happy running without these leads.

I would welcome a discussion on SWR at 2.4 and 5.8, how to measure and reduce, link lead coiled in a service loop at antenna end – another thread perhaps

N21roadie
 
taduikis
Member
Member
Topic Author
Posts: 436
Joined: Sat Jul 07, 2007 12:09 pm

Re: Cable lenght on 5GHz links

Sat Jul 10, 2010 9:23 pm

Well sorry, but you told nothing new here. We use integrated antennas absolutely everywhere except for AP. Can't say it's impossible, but hooking up an equipment container for each sector antenna on AP tower ain't very practical nor economic, is it? I'm not even talking about omni's or difficulty of maintenance.

Today I found some useful tips in Google about crimped or soldered connectors. Keywords are 'crimped vs soldered N connectors'. Imagine that..
 
ste
Forum Guru
Forum Guru
Posts: 1924
Joined: Sun Feb 13, 2005 11:21 pm

Re: Cable lenght on 5GHz links

Sat Jul 10, 2010 9:34 pm

Well sorry, but you told nothing new here. We use integrated antennas absolutely everywhere except for AP. Can't say it's impossible, but hooking up an equipment container for each sector antenna on AP tower ain't very practical nor economic, is it? I'm not even talking about omni's or difficulty of maintenance.

Today I found some useful tips in Google about crimped or soldered connectors. Keywords are 'crimped vs soldered N connectors'. Imagine that..
We use Sektors with compartment to avoid cable. Operating 11n you get
A Lot of cables.
 
n21roadie
Forum Guru
Forum Guru
Posts: 1949
Joined: Fri Aug 07, 2009 10:36 pm
Location: Limerick,Ireland

Re: Cable lenght on 5GHz links

Sat Jul 10, 2010 11:42 pm

....I'm thinking of a routerboard with 3-4 minipci slots, which I would place near the panels ....
Well if are going to use the "equipment container" and having several radio cards in that box, just imagine how the various RF TX signals will mix with each other and most importantly how they will effect the incoming RX signals from your client stations?, In other threads posters have advised against multiple radio cards onto just one board,
We use integrated antennas absolutely everywhere except for AP. Can't say it's impossible, but hooking up an equipment container for each sector antenna on AP tower ain't very practical nor economic, is it? I'm not even talking about omni's or difficulty of maintenance.
Don't know where the equipment container comes into the setup when using Mikrotik gear after all most is very compact, and as STE replied =
use Sektors with compartment to avoid cable
but hooking up an equipment container for each sector antenna on AP tower ain't very practical nor economic
I would contest it is cheaper and very practical


And yes OMNI's can be integrated and perhaps the easiest to do,A little DIY would work OK

Difficulty of maintence,
There should be no maintence on a mast if the equipment has been properly installed and commissioned, Just regular visual checks that all is OK on site, unless you are referring to repair/adding or removing equipment off the mast, and with integrated panels it's much easier to work around cat5 than heavy duty RF cable such as LMR400 cable on the mast.
The integrated panels can be mounted on "stand-off" supports out from the mast and thus making maintence easier?
 
adrianatkins
Long time Member
Long time Member
Posts: 556
Joined: Wed Sep 05, 2007 10:34 am
Location: Spain
Contact:

Re: Cable lenght on 5GHz links

Sun Jul 11, 2010 12:57 am

For what it's worth .....

As a general rule i never have an RF cable more than 1m in length, and always LMR400.
I tend to just buy them pre-made these days, but did solder them myself previously.

Of the 70 odd APs out there, none are PC-based, they are all a RouterBoard in a box, with short cables to the antennae.

It's cheaper (and easier) to run Cat5 than RF cable, but i try to keep the Cat5 as short as possible as well.

Having the antannae 'too' close to each other has never caused me much of an issue.

I dunno what 'too' close really is, because it depends on frequency, antenna type, cable leakage etc, and the fact that the 3 stacked radios are about 3mm from each other in the miniPCI slots, and they all run from the same power rail.

So far this kind of setup works for me pretty well.
 
taduikis
Member
Member
Topic Author
Posts: 436
Joined: Sat Jul 07, 2007 12:09 pm

Re: Cable lenght on 5GHz links

Sun Jul 11, 2010 11:43 am

Using sectors with integrated containers are out of the question for me, because I simply have no means of placing them on my mast. Now when you mention, I can agree that few RB411s cost the same or even less than a RB800.

Since now, I've been using RB600+RB604 totaling 8 miniPCI slots with 6 of them used. Previously I used 20V 3.25A laptop power supply and now I feed it from industrial 27V 6.5A PSU. Power is connected directly, without any sort of PoE. Such setup has been running for more than a year now, and I can't say it had any problems of self-interference or issues with power. I do believe that using adequate spaces between channels helps to avoid such problems. They wouldn't make RB's with miniPCIs spaced that near if there were serious problems, don't you think?

So while this setup proved effective and reliable, I'm willing to stay with it in the near future. Currently project is on-going to get these antennas a little bit higher on new aluminium mast, so that includes making a bunch of new, little longer cables. That's the reason I'm investigating connector issues. And while the cable I have is with solid center conductor, I will have to solder dies to it :(
 
n21roadie
Forum Guru
Forum Guru
Posts: 1949
Joined: Fri Aug 07, 2009 10:36 pm
Location: Limerick,Ireland

Re: Cable lenght on 5GHz links

Sun Jul 11, 2010 2:48 pm

Using sectors with integrated containers are out of the question for me, because I simply have no means of placing them on my mast
Do I gather than that there is no space on the mast remaining, what type of mast are you using can you attach a picture of your current layout from which suggestions could be forwarded, until then we are only guessing what to recommend if there is space restrictions,

I do believe that using adequate spaces between channels helps to avoid such problems. They wouldn't make RB's with miniPCIs spaced that near if there were serious problems, don't you think?
No so as it’s totally dependent on usage of the radio cards used for instance following that rational the very popular Ubnt XR5 radio cards 3 should fit into a RB433AH but because of the size of the heatsinks, slots 1 and 3 can only be used ?
 
ste
Forum Guru
Forum Guru
Posts: 1924
Joined: Sun Feb 13, 2005 11:21 pm

Re: Cable lenght on 5GHz links

Sun Jul 11, 2010 3:27 pm

They wouldn't make RB's with miniPCIs spaced that near if there were serious problems, don't you think?
(
You are kidding?
 
0ldman
Forum Guru
Forum Guru
Posts: 1465
Joined: Thu Jul 27, 2006 5:01 am

Re: Cable lenght on 5GHz links

Sun Jul 11, 2010 7:04 pm

That's why on sites that have the dish up high, the unit has an up/down converter on the dish and the frequency going up the tower to the dish is a lot lower frequency (70 Mhz in our case). This reduces the loss at 5.8 Ghz to 1-2 db max.
I've googled and come up with nothing.

Where do you get your up/down converters?
 
adrianatkins
Long time Member
Long time Member
Posts: 556
Joined: Wed Sep 05, 2007 10:34 am
Location: Spain
Contact:

Re: Cable lenght on 5GHz links

Thu Jul 15, 2010 1:03 am

They wouldn't make RB's with miniPCIs spaced that near if there were serious problems, don't you think?
Price.

Fortunately it seems to work OK, which is all you can expect for the money.

If you want Radio Purity, i think you have to go into Space, or inside an Iron Asteroid.

In either case the price is astronomical.
 
adrianatkins
Long time Member
Long time Member
Posts: 556
Joined: Wed Sep 05, 2007 10:34 am
Location: Spain
Contact:

Re: Cable lenght on 5GHz links

Thu Jul 15, 2010 1:07 am

They wouldn't make RB's with miniPCIs spaced that near if there were serious problems, don't you think?
The CIA are just looking after your National Interests.
The NSA are there to Protect you.
McDonalds is a Restaurant.
Banks are as safe as Houses.
 
mars
Member Candidate
Member Candidate
Posts: 228
Joined: Sun Mar 14, 2010 1:43 pm

Re: Cable lenght on 5GHz links

Sat Jul 24, 2010 9:06 pm

hi
where can i have a look on the internet at the up/down converter
thanks


S400 is the same as LMR400. It uses the same connectors and such.

Loss in the coax is

2.4 Ghz - 22.5db/100m
5.8 Ghz - 35.5db/100m

as you can see there is a 50% increase in loss between 2.4 and 5.8 Ghz. You also need to remember that this loss is bi-directional so you have the same loss on the transmit and receive sides. And if you are using the same coax on both ends now the loss is 100%.

For example using the numbers above, the difference over the total link (don't forget the 5.8 Ghz has a LOT more path loss than 2.4Ghz) Site A transmit loss is 13 DB, Site B receive loss is 13 db so the total increase in loss over the link is 26 db. That is the equivlent of doubling the length of the 2.4 Ghz run and then some.

That's why on sites that have the dish up high, the unit has an up/down converter on the dish and the frequency going up the tower to the dish is a lot lower frequency (70 Mhz in our case). This reduces the loss at 5.8 Ghz to 1-2 db max.

If I was running for any distance at 5.8 Ghz I would consider upgrading to LMR900 at the AP site.

Who is online

Users browsing this forum: Nullcaller and 31 guests