Install "standards"

[valenti]

What are people using for minimum signal levels for client installs (w/WAR1)?

I was on a new roof this evening. The 15dB Roo was getting about 92, the 19dB Roo about 85. That's not good enough, right? I have a 24dBi PacWireless grid that I meant to use for a backhaul, but now I'm thinking it could be a customer antenna.

I tried a Trango 900, that worked well but I would prefer to use the less expensive 2.4GHz WAR1.

Link details: it is 1.13 miles to my 140' grain leg. A 15dB omni on a WAR2 there. I don't see anything but trees from the house roof (we should be fully leafed out now) The CM9 in the WAR2 is set for default power, so maybe I can pick up 2 dBm there. Hopefully will switch to sectors and maybe 2x this fall.

Oh, I ordered a pair of SR9 cards today. If those work out I will condense all the Trango down to one POP.

[wow, I just googled where this guy retired from - the NSA facility in Yakima - he had a picture of their 30' dish in his office ]

[tog]

That's not good enough, you're right. Also keep in mind when it rains and the trees get wet, your signal will drop to nothing. I do not install people on 2.4GHz whose signals are worse than -79 or so.

2.4GHz through 1 mile worth of trees when one side is an omni is probably not going to be reliable. I think you might have to go with 900MHz or get more height to get over more of the trees.

Or point a high-gain directional antenna at the AP side straight at this person.

[valenti]

Thanks, tog.

Good point on the wet leaves. I'll work on getting it down into the 70s or else install the Trango 900 for now.

I suspect the signal is running over the top of trees for the first 1/2 mile. It is just around his house and on his neighbors property. And there are some open farm fields on the path.

[soulmata]

On most client installs, we aim for anything below -80 for residential and -75 for business. It's very rare we'll let an install slide that has higher than -80, though that said there are a few in which there is no other choice. I.E. they have clear LoS, a 24 gain grid and an SR2, but there is simply too much distance to get a better signal than that.

That said, we've never used cheap CPEs.

Even on our 900MHz installs, I set -83 as our minimum for received. Due to the nature of 900, we're a lot more forgiving on the return signal, but we always will bench the connection first before letting it go. If we can't achieve <1% packet loss under a 128kbps flood, we'll usually call it a no deal, excluding certain circumstances.

[tkerns]

Signal strength is not the only thing to determine if an install is good. Look at the Quality indicator on the WAR client, should be in the 90's with occassional dip being OK. I also run ping floods and look for packet loss. Before I set the CBQ on the radio I run a speed check from Speakeasy to see what kind of BW I can pull and push from this location.

That being said, an install earlier this week had clear line of sight, sig strength in the low 70's but signal quality was anywhere from 100 to *, large packet loss and barely pulling 128 kbps from Speakeasy. After trying horiz and vertical polarity (I have 2 AP's to this location), new pigtail and radio, still could not improve the performance. I pulled out my Spectrum analyzer to take a look.... found a freq hopping 2.4 phone with about the strongest signal I have seen. Un plugged the phone and the client ran great. Showed the customer the screen with and with out the 2.4 phone. He left the phone un plugged and said he was heading to Costco to get a 5 gig cordless.

[valenti]

Soulmata, what do you mean by cheap CPEs? Not WAR1 & Rootenna, right?

I do have about five people hooked up with Ruckus Metroflex single zone (~$100), those seem to work to about 1/2 a mile with LOS. I like not having to mess around with exterior mounting, running wires, etc. No real problems with them after a few months of service. But I am thinking that a WAR1 at those spots might be nice, able to use cloaking, plus I could re-configure them remotely.

Do you do the ping floods from a laptop? I haven't seen a spot to do that directly on the WAR.
---------------------------------
tkerns, thanks!
Yes, I'm watching the Q% too. Usually it is 100, almost always 95+. One thing I've noticed is that it might stick at 100% until you send quite a few packets thru, then maybe gives a lower true value.

Oh, don't forget to tell people about the DECT phones. Get them totally out of *our* spectrum. :-)

[bobbyc]

tkerns,
Was that 2.4GHz cordless phone in use, or just sitting idle waiting to ring and be used?
Bob C

[tkerns]

It was a Panasonic base set with the handset in it. Not sure if they had another remote, but this was with the phone idle. It did get worse when we called and picked up the handset. We have a lot of clients with 2.4 phones, but have never experienced this kind of interference. Normally we have seen a wireless router be affected by the 2.4 phones, but extremely rare to see our CPE affected, especially since it was more than 50 ft away and up on the roof.

[soulmata]

Soulmata, what do you mean by cheap CPEs? Not WAR1 & Rootenna, right?

I do have about five people hooked up with Ruckus Metroflex single zone (~$100), those seem to work to about 1/2 a mile with LOS. I like not having to mess around with exterior mounting, running wires, etc. No real problems with them after a few months of service. But I am thinking that a WAR1 at those spots might be nice, able to use cloaking, plus I could re-configure them remotely.

Do you do the ping floods from a laptop? I haven't seen a spot to do that directly on the WAR.
---------------------------------
tkerns, thanks!
Yes, I'm watching the Q% too. Usually it is 100, almost always 95+. One thing I've noticed is that it might stick at 100% until you send quite a few packets thru, then maybe gives a lower true value.

Oh, don't forget to tell people about the DECT phones. Get them totally out of *our* spectrum. :-)


By "not cheap CPEs", i mean about 90% of our installs are done with 24dBi gain grids in a metal outdoor enclosure, with either a Soekris (old installs) board, WRAP (last years installs) board or WAR board (this years installs). All the wireless radios for 2.4GHz used are either the CM9, the SR2 or the WLM54G. There are a few installs with lower gain 15dBi grids or 15dBi Echo Backfires, though not many. For 900Mhz, we use only the SR9 and choose from 9 gain Yagi, a 13 gain Yagi or a 15 gain grid depending on the install. 100% of our CPEs are running StarOS as well, and all of our distros are running StarOS.

We've to date never used a rootenna for an install.


As for the floods, yes I do it from a laptop behind the client radio. I flood our datacenter with different streams based on the clients committed rate. Typically I try to flood it with 80% of what we sell the line as. If we can do that with <1% packet loss I call it good.

[go.fast]

We have a lot of clients with 2.4 phones, but have never experienced this kind of interference.

I've seen phones a time or two in the past, but like ou sa Tim, it's rare.

But look out for the 2.4 cams.

I had a customer get bit in the butt last week with cams. He had 3 or 4 sets of wieless cameras around his property. I had a helluva time getting from my laptop in his driveway to his home wireless just ustairs, never mind his 2 gig wartenna on the roof. Packet loss bonanza.

He didn't believe us at first, but show and tell works. We had him turn off his cam systems till he went smoth and explained how 2.4 gig works and channel spaces. My resolve for him is to turn him into a 5 gig feed and let him screw up his own internal 2.4

Some of the camera systems I've seen are using 4 channel spaces and not 3 like wifi. A couple cams can whipe out half the spectrum.

[soulmata]

A few months ago, we lost a customer because a local ad company put a crapton of 5GHz wireless cameras up on some billboards near the client. No amount of frequency juggling was ever able to get them back up. A shame that was.

[knolan]

All of the comments above refer to 2.4Ghz, what are the SNR levels people are happy with on 5Ghz.


Usually if the quality is up around 100% and the the signal is -83 or better we are happy to deploy.

Regards,
Keith

[therealboss]

On 5.8 I try to keep it below -81 anything above that I find can give problems from time to time. My connection was -83 and it worked most of the time, but when it was a problem that was when I needed it most. I have changed the antennena at my house and now its between -79 and -80 and never drops out. On the rest of the network everything in 5.8 is better than -78 most -65 / -70. At the end of the day, if there is a lot of 5.8 in your area I would say geep them below -80, if there is little to none in the area I guess -82 / -84 will work.

[valenti]

By "not cheap CPEs", i mean about 90% of our installs are done with 24dBi gain grids in a metal outdoor enclosure, with either a Soekris (old installs) board, WRAP (last years installs) board or WAR board (this years installs). All the wireless radios for 2.4GHz used are either the CM9, the SR2 or the WLM54G. There are a few installs with lower gain 15dBi grids or 15dBi Echo Backfires, though not many. For 900Mhz, we use only the SR9 and choose from 9 gain Yagi, a 13 gain Yagi or a 15 gain grid depending on the install. 100% of our CPEs are running StarOS as well, and all of our distros are running StarOS.

We've to date never used a rootenna for an install.

Mind if I ask what brand enclosure and grids you are using?

I just received a few PacWireless die cast aluminum boxes, they seem pretty good to me. And do you use anything between the 15dBi Backfire and the 24dBi grid?

[soulmata]

We use the pacwireless NEMA6 die-cast aluminum enclosures with the hinged door addon. I like them a lot. Not too expensive and very thick, very strong, very durable and perfect for anything WAR.

For client installs, no, the 24dBi gain grid is pretty much it, minus a few scattered 15dBi installs.

This excludes MicroPOPs, however. That's a whole different ballgame.

[kbldawg]

Out of curiosity, what do you have against Rootennas?

Most of our installs are 15dbi Rootennas, if that won't work, we break out the grid.

We've had a few issues with Rootennas, but mostly due to installation error.

[soulmata]

The majority of our customers are far away from our towers. I would say 60% of our customers are more than 10 miles from the tower, and to date I have not had any luck with any rootennas. Sure, we can get a connection - but with nowhere near the quality I want.

I am not opposed to rootennas, I just have not yet had any success with them. That said, we have a single rootenna install on the schedule for a customer that is about 7 miles away from the tower. We'll see how that does.

[tog]

Even with the larger 19dBi rootennas?

[kbldawg]

ah...ok, that makes sense.

[soulmata]

Even with the larger 19dBi rootennas?


Haven't tried them. Even then, I think i would still prefer a grid - more gain, more interference protection

However, I am definitely looking at using them for our MicroPOPs, which we have a few of in the works. They would have helped a lot at one of our first micropops, but it was before my time and customers out there ended up with backfires or omnis.

The biggest issue to us is cost - and, considerig how many 24 gain grids we already have, we'll use up all those first before we look at revamping.

I'll also add in that the majority of our customers are also in rural areas - farmland, mountains, etc. The less chance we have of ever needing to make an onsite visit, the better. A good chunk of our customers are 20 to 25 miles from the tower, and might be 50 miles or more from our office.

Our longest shot customer right now is 28 miles - using a WRAP with an SR2, 24 gain grid, pointing at a mountaintop with a CM9 in a WAR4.

Works great.

[tog]

That's crazy, I'd be using huge grids, too.

All my stuff is 5 miles or less, usually a couple at most. Most everybody cares about how it looks so huge grids are out :)

In your situation that makes a lot of sense.