signal problems

[dkii]

I can't for the life of me figure out whats wrong. I have a 12dbi omni antenna on a pole 90' in the air. Connecting it is 50' of lmr-400, then to a TT 1/4 stub lightning arrestor, then to a lucent card in my star-os box. I go a little over 1/2 mile(1.03km) away, drive to the 5th floor of a parking garage, have perfect LOS and then some to my antenna. I hook my andrews 15.5dbi patch to my pigtail plugged into my lucent card in my laptop, and point it in the right direction. I can only get a marginal link(signal ~78dbm, noise ~90dbm) and the pings are dropping about 15% of the packets. Any clue as to what is wrong? I have replaced the cable, antenna, pigtails, and lightning arrestor on the omni side. I have not checked the andrews antenna to make sure it is working properly, but other than that I've run out of ideas.

[bobbyc]

Yeah I would check the antennas and polarizations first, then the connectors on the lmr-400 and the pigtails. If the tiny center pin on the pigtails isn't seated properly in the tiny connector on the pcmcia card, it will give you problems. Also when you put the connectors on your lmr400, did you crimp or solder the centerpin onto the conductor properly?
Early on we had lots of pigtail problems when we were getting them from certain vendors. Now we get all our pigtails from just one vendor who makes quality pigtails IMO.
Bob C

[dkii]

I know the omni is good. The cable was bought pre-assembled before I came to this company, and noone knows where it came from, but I have tried 2 different cables. The pigtails came from hyperlink, they appear to be good quality, they have strain reliefs and look to be assembled well. I will check to make sure that centerpin is aligning correctly. Where do you buy your pigtails from?

[dkii]

I give, I've checked and replaced everything and I still can't get squat for a signal. Maybe aliens have a rf dampening field around that location? Here is what I have checked and replaced on both ends of the link

Lucent card
Pigtail
Lightning arrestor
cable
antenna

I noticed at that particular location, there were some big powerlines between me and the other antenna, i thought those might be affecting the signal, so I moved to a different location, and still no luck. This is in a decent sized city (150,000 pop) so there is pretty much a solid cover of buildings on the ground, but most are not taller than 2-3 stories. Could they be somehow reflecting and cancelling the signal? Even in radio mobile with the urban factor on, it still says my recieved signal should be well over what is necessary. When I did the link test, the orinoco client manager reported around -78db for the recieved signal, around -90 for the noise, and around 13 for the s/n ration. The test partner(omni) showed a recieved signal around -80, noise around -88 and the snr was bouncing between 6 and 11. Ping tests dropped about 20% of the packets. If I switch the polarization on my end, it loses the signal all together. In this area I figure there are at least 500 802.11b access points(based on my netstumbling) and we have 5 breezecom 2.4ghz fhss access points active within a 5 mile radius(which work great btw) but only 3mbps bandwidth and incredibly expensive. Maybe these are interfering with my signal, but shouldn't it show up as noise if they were? I dunno I'm stumped. Any other ideas?

[bobbyc]

I guess the only other idea is that the majority of the signal and beamwidth of the omni is shooting over the parking garage. A 12dB omni would have a pretty narrow beamwidth.
Bob C

[lonnie]

This is the result of a bad cable, lightning arrestor or antenna. Power lines won't do that to you. Your noise readings are very good so we can rule out a WiLan or Tsunami close by.

Try a link from the ground with a patch antenna at the AP. This gives a duplicate try with mostly new hardware. Pigtails can be bad so easily. We make our own to ensure quality control. They guys doing them in bulk are only interested in cranking them out. When we bought them from the company we used to have a close association with, we had nothing but trouble - some of them literally fell apart. Now we have a reliable connection and we save money.

[kishvet]

Along these same lines - 4 mile link with YDI 2.4/5.8 amp/convertor. Ping times range from 1.97ms to 74ms. HPol Teletronics dish antennas. We don't seem to be losing packets, just extreme variability in ping times. I have other similar links with <5-6 ms pings and similar signal/noise ratios. This link runs about -87 dB noise and -57dB signal. The tower climber place the ends on the LMR600 and is used to the hardline stuff. Maybe it was because he has place tons of ends over the years, but it seemed like he was a little sloppy. Thanks for your opinion.

Grant

[bairdc]

I guess the only other idea is that the majority of the signal and beamwidth of the omni is shooting over the parking garage. A 12dB omni would have a pretty narrow beamwidth.
Bob C

I'd be willing to bet that this is the problem. After dealing with high-powered omnis at two POPs, I've become convinced that they are nothing short of evil. We just finished swapping out a 12 dB omni for a 120* 12dB sector on one POP the other day. The difference has been amazing. We had customers who, on the omni, were getting about -84 signal, and -91 noise. They're now seeing -71 signal and -91 noise. These customers were about 1.5 miles from the POP.

It all comes down to that "pancaked" radiation pattern that high-powered omnis have. If the customer is not in just the right place, most of the signal will go right over their head. Or in some cases, if you have customers on hills at some distance from the omni, the signal will hit the ground before it gets to them.

Don't get me wrong, I think that high-powered omnis do have their place in certain applications. However, I'd bet that in the majority of cases where someone thinks they have the right application for a high-powered omni, they'd be better served with a sector. I know I fall into this category. At the two locations where I have omnis, I thought I was doing the right thing. Turns out I was wrong on both counts. I've learned my lesson. I'll never use another high-powered omni.

Craig

[lonnie]

For Grant's question - you have very good signal and noise. In your case the only thing that will cause the variable ping times is a busy "something" in your connection chain.

Any router being maxed out will be slow to respond to a ping. Are you bandwidth controlling "ALL" users. Even a few users with no limits can give you periodic maxed out use. Just checking their email will saturate your link, but only for a small period of time.

You have to make sure that no single user can ever consume everything you have because when they do you see the situation of nothing left for everyone else.

[kishvet]

Any router being maxed out will be slow to respond to a ping. Are you bandwidth controlling "ALL" users. Even a few users with no limits can give you periodic maxed out use. Just checking their email will saturate your link, but only for a small period of time.


Lonnie,

There are no clients on either of these 2 boxes. These are P4 1.6 with 256M memory. We do not have anyone using these boxes yet as we are still testing and making sure they work ok. There is a 5.8G link from Star2 to Star3 that works fine with similar snr, distance, and lack of dropped packets. The ping times on the Star2 to Star3 5.8G link are very uniform. The link from Star3 to Star4 is the one we are having problems with. Connected into Star4 with a xcable and a laptop gets me very close to the T1 speeds we are feeding into Star1. What should I look at next? Replace the Motherboard, memory, 2.4 to 5.8 amp/converter? I need some advice. Thanks.

Grant

[dkii]

I think the problem was a number of different ones adding up. #1: Omni antenna. I replaced the 15.5db omni with a til-tek 90 degree 12 db sector and the snr almost doubled. It still wasn't near the signal I expected, but better than before and would sustain a 65k ping with ~700ms times. I also could get a weak signal with no antenna at all on the laptop after I did that. #2: Possibly bad cable. The person that assembled the cable thought you could crimp the center conductor because it was a 'crimp' connector dohhhh. But even after replacing that, snr didn't really change much. The radio was indoors on the 6th floor of a building, with a window right close to it, I had someone hold the sector antenna directly connected to the box(straight from pigtail) and point it at me out the window. That gave about a 3db rise in snr, still not what I expected though for a 1/2 mile link, but good enough for a stable link. I did a little math a while ago, for a 15db omni(7 degree beam width) if I remember correctly, would take about 1.2 miles to reach the ground level. This omni was at about 85 feet, and I was at around 45 feet, so I think I would have been in its beam. The noise on my end shot to around -91 when we connected the sector also. Now i'm back in fl and have all of these wonderful trees to deal with.. fun fun fun.