1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Antenna dilemma

Discussion in 'Tomato Firmware' started by mvsgeek, Jan 15, 2012.

  1. mvsgeek

    mvsgeek Networkin' Nut Member

    I finally took the plunge and replaced my "main" WRT54GL with an Asus RT-N16 running Toastman 7493. In all the excitement, I forgot to change the TX and RX antenna settings from "auto" to "A". The router has a single Hyperlink mast-mounted 120-degree 17dBi sectorial antenna. In testing, inSSIDer shows that the "A" antenna is the one closest to the power jack. (It's "B" on the WRT54GL). Despite my oversight, everything seems to be working OK, apart from occasional random connection drops of some of the WDS slaves. I'm hesistant to make any changes to the main router, because every time I do, some of the slaves don't reconnect, and I get mildly irate phone calls from frustrated users whose slaves stubbornly refuse to reconnect even after a power cycle.

    Can somebody please enlighten me as to how the firmware handles antenna assignment? I removed the 2 extraneous rubber duckies from the Asus, because the nearest AP is 0.3 miles away, so everything is being handled by the Hyperlink on the "A" connection, with router set to "auto". Could my oversight be the cause of the random disconnects?

    It's a G-only network, 2.4ghz band only. TX power is set to 60, per Toastman recommendation.

    Thanks as always to Toastman et al for their outstanding work.
     
  2. backwoodsman

    backwoodsman Networkin' Nut Member

    With MIMO, the antennas aren't simply selected based on strongest signal like with the diversity used by 2-antenna non-MIMO routers. MIMO does much more sophisticated signal processing using multiple paths and interference patterns, and can be a huge advantage, particularly if the signal has to go through much of anything. By using only one antenna, you're losing that advantage. Put the 2 omnis back on and see if they help. You might be surprised at what they'll do, even at 0.3 mile. Currently I have one user connected at 1/3 mile using a 5dbi omni, going through 3 walls, getting a decent connection.
     
  3. mvsgeek

    mvsgeek Networkin' Nut Member

    I can give it a shot, but the router is in a metal-walled windowless building. We even have to step outside to use cell phones. The Hyperlink antenna is mast-mounted above the roof line with an extension cable to the router.
     
  4. backwoodsman

    backwoodsman Networkin' Nut Member

    Oh.. well that's a problem. In that case, if you're getting poor connections, I'd suggest one of: a second outdoor directional antenna, or a single directional antenna with a narrower beam width, or better antenna(s) on the other end(s).

    17dbi gives a narrow enough vertical beam width that you really have to pay attention that your targets are inside it.

    What type of coax are you using, and how long is it? At 2.4ghz it makes a big difference.
     
  5. mvsgeek

    mvsgeek Networkin' Nut Member

    I've considered adding a second outdoor antenna, but they're not cheap, and it will be tough to install on the existing mast. On the main router's device list page, signal strength of the WDS slaves ranges from -51 to -68 dbm, so I don't think that's my problem. Can't remember the exact specs of the coax, but it's pretty heavy duty, length at least 10'. It was the shortest I could get that would fit the environment.

    I have to reboot the main router this evening, so I'm going to change "auto" to "A" and see if that helps the problem of the failed slave connections.

    Thanks for the feedback, I wasn't aware of the vertical beam issue. So much to learn about this stuff, and it's not easy for an old mainframe dinosaur:confused:
     
  6. ntest7

    ntest7 Network Guru Member

    You're probably losing a huge amount of signal over the 10' cable run. You can google the RG-whatever cable type printed on the outside jacket and find it's dB loss per foot at 2.4ghz to calculate what you're losing.

    Consider mounting the whole router in a weatherproof box on the mast -- even using the standard omnis. Use a power-over-ethernet adapter so you can cycle power without climbing. Or just mount it on the side of the building.
     
  7. Toastman

    Toastman Super Moderator Staff Member Member

    I was about to suggest that also. It's a complete waste of time to use antenna connected by a cable run of normal crap coaxial cable at these frequencies. Cable loss is just horrific - try Andrews Heliax :)

    (As an aside, I have always goggled at the addon antennas that you see with around 6 feet of what basically amounts to the cheapest and lossiest coax available with a plug on the end for the router!)

    I'd also think it might be better, if long range is the aim, to use an AP dedicated to that antenna, and force it to use "B" speeds. "B" can reach pretty impressive distances and it will use twice the power. My own choice would be to place the WRT54 in a box as ntest7 suggests, right next to the antenna, set to "B" only, wire by CAT5 to your RT-N16.

    -68dbm is just a tad low, this might bring it up by several dB.
     
  8. backwoodsman

    backwoodsman Networkin' Nut Member

    You're right, at those signal strengths, antennas aren't your problem. It's gotta be something else.

    For a 10+ ft. cable run, you definitely have to know what type of cable it is. It's printed on the cable every 2-3 ft. LMR-400 will have a loss of 0.6 dbi for a 10ft run, which is fine. LMR-240 would be OK in a pinch at about twice that attenuation. Anything else, don't bother, you lose too much of your signal. I use LMR-195 or LMR-240 up to 6ft and really try to avoid going farther than that, but if I have to, it's LMR-400.
     
  9. Toastman

    Toastman Super Moderator Staff Member Member

    Drop of WDS has always been a fairly common occurrence. jsmiddleton4 may be able to offer some advice on that. He's been pretty successful with WDS. But at -68dbm it wouldn't take much interference to knock the connection down. Is it the weakest one that loses connection, is there a pattern?
     
  10. mvsgeek

    mvsgeek Networkin' Nut Member

    There doesn't seem to be any pattern to the drops. Sometimes it's the slave with the strongest signal (~50), other times it's the weaker ones. The one with consistently the weakest signal (~68) almost never drops!

    Since last night's reboot, with the antenna changed from "auto" to "A", I haven't seen any drops. Most of the slaves failed to connect initially, but I run a little script on each of them every 10 minutes to reboot if the main doesn't respond to a ping.

    Long range isn't as important as speed and reliability. Over the last couple of months, traffic has increased significantly due to proliferation of iPads, kindles, smartphones, online gaming, video streaming etc. Hence the deployment of the RT-N16 with Toastman and QoS, which has certainly improved the overall performance at the expense of reliability.

    One of the slaves is indeed in a weatherproof box next to its antenna, connected via cat5 to another WRT54GL AP inside the house.

    Main antenna cable is 10' LMR-400.
     
  11. Toastman

    Toastman Super Moderator Staff Member Member

    All seems OK to me. Drops could be due to many things, and so hard to find out what. Suggestion - try changing country.... (in the wireless driver setup, that is :D ) That has worked for a few people. Then I expect you've already been following the Wireless Driver issues thread :)
     
  12. tvlz

    tvlz Serious Server Member

    I have noticed that the slave router loses the ability to connect to DHCP on the main router if there's a dropout(or reboot) and only if there are no devices connected to it.
    Have you tried "Use user-entered gateway if WAN is disabled" on the slave routers to see if that fixes the problem?

    If it works maybe Toastman can change the name to "Use Gateway IP for DHCP if WAN is disabled" & add "Required for WDS Slaves" to notes.
    change status-overview to display the DHCP IP instead of disabled when checked.
     
  13. mvsgeek

    mvsgeek Networkin' Nut Member

    Hmmm...interesting observation, thanks for that. The dropped slaves may indeed have been dormant at the time of the dropout/reboot. I've checked the option you suggested on a couple of the slaves, and rebooted one of them successfully with no ill-effects. I'll be monitoring closely to see what happens (as will my users:confused:).

    I notice that Vanilla Tomato 1.28 (which I was running on the old main router) doesn't have that option checked, but who knows what's happening behind the GUI...

    Maybe this thread should be renamed "WDS dropouts and how to live with them;)"
     
  14. crashnburn

    crashnburn Networkin' Nut Member

    Toastman. I believe you wrote the article on TomatoUSB about increase Output Power etc and that you are an RF engineer by education/ work background.

    I've had a few questions around Antenna's and Radiation Patterns.

    I've studied Antenna design back in engg college at Uni. Whole bunch of TV, Video, & Satellite stuff. Different Antenna Designs and Signal Radiation Patterns & Lobes. Omni v/s Directional Antennas. Its been ages since I looked into that stuff.
    Recently, I re-study online on google: the Radiation Patterns for single Diploes and the differences in Directional v/s Omni.

    #) - I am wondering whats the cumulative Radiation Patterns of the 3 x 2 dBi antennas on the ASUS RT-N16 together in a line (An Antenna Array?).
    Or Any typical 3 x Omni-directional Antennas that are together? Are there any Free Apps S/W to model this easily?

    #) - And, what would be the effect, if one of those Antennas was extended by a Low Loss Cable and taken away by couple of meters?

    MOTIVATION & LOGISTICAL ISSUE:
    I live in a 2 Floor Condo/Flat in a Re-inforced Steel/Iron Concrete building. The 2 floors are separated by such Concrete floor/ceiling.
    I cant move the router beyond a point, but the good part is its centrally located on the lower level near the Concrete Stairway, such that it would only take a Cable of 1.5/2.5 meter to extend one (or more if you recommend so), out of the 3 Antennas to a suitable point on the upper level (near Stairs).

    #) CABLE SPECIFICS ??
    I read the points about getting the right kind of Coaxial Cable.
    Can I get one of these 1.5-2.5 m for plugging behind Asus RT-N16 or do I have to make one myself using the Cable & Adapters? Any Specifics that you'd recommend.
    I searched Andrews Helix on E-bay. They are damn expensive!
    LMR-400 Seem more affordable. How much of a DBi drop difference would they be?
    My local cable TV guy provides us with STB with Digital Cable 200+ Channels - Would that kind of cable qualify -What cables are those?
    PS: God, it might be cheaper to get a TP-Link TL WR-710N (Airport Extreme CLONE) for the other Condo Level - http://www.ebay.com/itm/330671751693?ssPageName=STRK:MESINDXX:IT&_trksid=p3984.m1436.l2649

    #) Maybe once extended, I can also add a Higher dBi Antenna? (I've already put Tomato Power to 100+)

    Reference:

    http://www.data-alliance.net/Page.bok?template=cables-connectors - LMR200 cable: 0.54dBi of loss/meter (Is this Accurate? Good enough? )
    www.data-alliance.net/Page.bok?template=antennas%20directional
    www.data-alliance.net/Page.bok?template=antennas-omnidirectional-dipole - This also shows why some people might experience WORSE signal / coverage after going for a HIGHER dB Antenna. The Antennas Vertical radiation decreases.

     
  15. Toastman

    Toastman Super Moderator Staff Member Member

    LMR-400 isn't too bad if you keep the length short. Heliax was more of a humorous comment than a serious suggestion. But - it's great cable!

    Ref. the other points. Really, trawling through the source code for hints on interfacing to these drivers, there is not enough information to be able to comment. Firstly, the "beam forming" stuff - I would doubt very much if this is much more than a marketing gimmick. There's nothing I've seen to suggest that it really exists in consumer routers except for some occasional marketing hype. Anyway, you can't model a system that you have no information on. And a decent modelling package is extremely expensive.

    Most people are making brief tests and then posting in forums without really understanding that moving the router (and hence the antennas) by several millimeters can make great changes in signal strength. Anything in the room that changes or moves can do the same. A story here. A long time ago, I remember a student working on a design on the bench who had played around for hours trying to find out why his newly built prototype didn't actually receive any signal from a local TV repeater. I moved his test jig a few inches on his bench and voila - there it was at great strength. He had quite coincidentally parked his jig in the exact spot where it didn't receive anything at all. The same thing happens to routers, and that is why they are vastly better with two antennas in a diversity setup.

    Re the general operation of MIMO, it relies on two (or more) different signals being received at the other end, that can be separated (mathematically) from one another due to differences between them caused by changes in signal path. As the two transmitting antennas are very close together in a typical router, only separated by around 6" on 2.4GHz - this difference usually isn't very great. Quite possibly moving one of the antennas away a little may improve things. This has been pointed out by many engineers in technical discussion forums regarding antennas on consumer routers. But nobody would buy a router a meter long, and people don't like plugging in external antennas. If you do try it, expect to find a slight improvement, but you probably won't see anything startling.

    The situation is much better on 5GHz because of the decrease in wavelength, making the small separation of 6" much more effective on that band. This also explains why a USB stick can't really implement MIMO effectively as the two antennas are VERY close together, and the designers or users can change only the orientation of the antennas.

    But let's be honest - is MIMO itself really doing very much for most of us? I think not. Firstly, the throughput increase on a single channel is rather higher for "N" modulation systems than for "G" - that is where most of the performance increase really comes from. Then you double the channel width - and you have pretty much the throughput that most people actually achieve in practice. (At the expense of stomping on other band users of course, but that's politics eh? )

    In tests I've done here, under typical use in a typical building with typical users, and typical interference from the 30+ other wifi routers (and all of their "invisible" users too) on 2.4GHz, I don't see anything that would actually convince me that MIMO implementation is doing anything particularly useful. However, I'm a cynic :D But reading through the various discussion on the forums and looking at people's real-life experiences, anybody can see that "G" can often achieve better speeds more reliably, at least on 2.4GHz, than "N". Personally, I am not a fan of MIMO as used in wifi systems in its present form.

    Now 5GHz is a rather different matter. The almost total lack of interference due to more channels, few users, and decreased range, gives it a great advantage. But - it is still not very common to see the claimed speeds achieved in practice.

    Now, funnily enough, a 2 meter length of CAT5 plugged into the router achieves the best speeds, costs a few cents, is secure, cannot be interfered with, and has no dropouts. Unless you absolutely HAVE to use wireless, for a phone or something, you'll always get better results from simple technology like a bit of wire. :p If our situation were that we had invented WiFi first, and then someone came along with a cable and LAN socket, everyone would be raving about the vast improvement in throughput and reliability, no?

    The world can be a funny place, at times :D
     
  16. crashnburn

    crashnburn Networkin' Nut Member

    I have CAT6 Cabling all through the Condo. Something that was planned and done before it was painted etc.

    But, I'd like to have an "optimal" Wireless environment for the Phones and the times I lounge around using laptop & tablet.

    Well, in terms of Wire/Wireless, they both use the same fundamental precept - Medium Access Control - MAC. In WiFi, the MEDIUM is Air, and give its unbounded, open nature leads to MORE interference from Surroundings & Other Communications. Like a crowded room where everyone's screaming. (CSMA/CA)
    And, Ethernet, the CABLE is MEDIUM, which is bounded & closed where its interference characteristics are relatively better with lesser cross talk (CSMA/CD).

    Well the Heliax is $$ expensive. If that was a joke :) then which would you recommend? LMR-400 or something better along the same lines?

    I've had experiences like yours (the story where you moved the students experiment) and its part of the experiences when Wireless transmissions take on a life of their own. I've had the early years of "adjusting" the TV antennas (Mickey Mouse, 2 Extensibles and Yagi o n roof), where picture & sound would be clear if I am standing next to the TV, and crap out when I moved back to my couch. Lol!

    I believe you're talking about Phase Shift / Phase Delay/ Differences? Its been a long time since I read those books.
    I'm glad its an improvement and not worse scenario if the antennas are spaced out more.
    Do point out "which forums & threads" I could find this discussion on Consumer Router Antennas.

    So, given that the RT-N16 has 3 Antennas, do you think its better to:
    - just to extend out the "one" antenna I really need by 2/2.5 m to where I need, or
    - even better to just separate "all three" antenna (including other two) by certain amount? -How much distance would be optimal / would you suggest? Or just enough around the Wooden Cabinet that houses my Router & Network Equipment?

    Interesting stuff. I will read this in more detail and revert back.
     
  17. Toastman

    Toastman Super Moderator Staff Member Member

    To be honest, I wouldn't bother. In my opinion, it won't make any worthwhile difference. But if you want to see if spacing improves MIMO it would make sense to leave one on the router and move the other two out. I'd move only one antenna and set the router to "G" personally. No, that's not true, I'd probably forget about it and use another AP - depends what you're trying to do. :)

    A dB of loss isn't much. For a few meters LMR-400 is fine.
     
    crashnburn likes this.
  18. crashnburn

    crashnburn Networkin' Nut Member

    I'm glad its an improvement and not worse scenario if the antennas are spaced out more.
    Do point out "which forums & threads" I could find this discussion on Consumer Router Antennas. I'd just like to have a reference to it for academic reading. I am curious.

    lol! You're confusing me with all 3 scenarios.. Lol ! Nice one buddy :p I guess I got to find some of that Cable to test out these scenarios.. crap! No easy way out.. LOL.

    PS: Why set to G?
     
  19. crashnburn

    crashnburn Networkin' Nut Member

    Toastman?
     
  20. Oltion Nikolla

    Oltion Nikolla Network Newbie Member

    Hello! I am new here. I have used hyperlink tech antennas and other different antennas for a few years. Had a site with 3 ->> 17 dBi 120° (HG2417P-120) hyperlink antennas with 360° of coverage. I had so much trouble with them because they have small Vertical Beam Width of 6.5° so i decided to remove them some days ago. On an other site have 2->> 14 dBi 90° (HG2414SP-090) hyperlink total 180° of coverage. I am very much satisfied with this site and never had problems. they have 15° Vertical Beam Width. hope this helps
     

Share This Page