Re: WiFi out to 800 feet

Those are fair assumptions. However, I've been surprised a few times. For example, I couldn't figure out why I was getting miserable 2.4GHz performance in an isolated farm house that was 2 miles from the nearest neighbor or potential source of RF interference. I finally got around to doing a site survey with a spectrum analyzer and directional dish antenna. I wound that there was a point to point

2.4GHz wireless link between an office building about 5 miles away, and an isolated pump house about 3 miles away. The farm house was directly in the line of sight. At first, I simply changed channels (1, 6, or 11), but the pump house link changed channel every time the link faded or was obstructed. So much for adaptive channel selection. So, I switched to 5GHz, and avoided the problem. Yes, interference can be a problem in the middle of nowhere.

Correct. However, 5MHz is not the occupied bandwidth of the signal. It varies by modulation mode and type. For example, conventional

2.4GHz 802.11b/g is typically about 22MHz wide and occupies four 5MHz channels. The 2.4GHz band is 83.5MHz wide. Therefore, if it is only possible to fit 3 non-overlapping 22MHz wide signals in the band before running out of bandwidth. This is where the recommended CH1, 6, and 11 comes from. Incidentally, picking a channel that lands in between CH1, 6, or 11 will end up overlapping the two adjacent channels and interfere with both.

On 5 GHz, it's the same story. You divide the available bandwidth by the occupied bandwidth of the signal to get the number of available non-overlapping channels. Diagrams such as these show how it works:

You can use 20, 40, 80, or 160 on *PARTS* of the 5GHz band. 10 MHz is available but I don't know any situation where it might be useful. The bandwidth situation is a mess on 5GHz. I don't have the time to explain where all the various protocols, power levels, bandwidth restrictions, and standards, DFS radar protection, etc, fit together. Also, things get really strange with 802.11ax. See Fig 9:

In what country? See: Go to the column marked United States. Notice that 20 MHz is the minimum allocated occupied bandwidth. 10 MHz is on the chart, but it look like no country is using it. 5, 8, 30, and 50 MHz are not on the chart.

Yep. However, if a wide bandwidth is such a great solution, why doesn't everyone just setup their routers to use as much occupied bandwidth as possible, or perhaps just use the entire band? Sure, there are benefits, but compromises must be made to use a larger part of the band? Hint: Think about how long a radio needs to be transmitting in order to deliver (for example) 1 MByte of payload data. If it can deliver the data twice as fast and therefore uses half the air time, that's that much more air time for other users of the bandwidth used.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

The question was for an 800ft link.

800ft / 5280ft/mile = 0.152 miles Please adjust your computation accordingly.

That depends on whether the intruding material is absorptive or reflective. You can get a way with much less clearance if the signal is absorbed. Yes, the signal level goes down, but it also stays down and does not vary. However, if it's reflective, then it will refract (bend) part of the signal, creating the opportunity for fades, nulls, cancellation, etc. It can also create reinforcement and stronger signal levels, but those tend to change radically if anything moves.

40% intrusion is a usable number for real links, but only works if you have a sufficiently large fade margin, also known as SOM (system operating margin). 20 dB would be a good minimum. I carry 20dB and 30dB attenuators in my toolbox. If the system still works reasonably well with 20dB loss inserted at one antenna, it will probably be reliable. If it dies completely, you need a bigger antenna or more transmit power.

While I'm ranting on the topic, fade margin (or SOM) is related to reliability (or downtime):

SOM dB Reliability % Downtime per year 8 90 876 hrs

18 99 88 hrs 28 99.9 8.8 hrs 38 99.99 53 minutes 48 99.999 5.3 minutes 58 99.9999 32 seconds 99% reliability might sound great, but that means your link will be useless for 1% of the year, or 3.6 days per year. Don't go below 20 dB fade margin, which is 70 hours of downtime per year.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

Yep. And the calculator I used could handle tenths.

That's why I calculated for .2 miles. I couldn't do 0.152 miles.

nels.

The United States. I'm using official firmware on a US radio. I had heard that newer radios were limited to 10MHz as the smallest slice, but older gear is grandfathered in.

As for the usefulness of 10Mhz, well... Seriously?

I can think of all sorts of things.. Namely anything where you need

70mbps or less.

Or, in a really crowded area, you might be able to find 10Mhz of clean spectrum..

My own link to my WISP is 10Mhz (I have my own dedicated AP). Delivers me everything I need without having another 10Mhz just polluting the area= =2E

I could probably get away with 5Mhz, but I've upgraded to the AC line and that is no longer an option. It is, however, still an option in the M series.

I'll assume a Ubiquiti M5 radio. I have some really old M5-Bullet radios, with firmware that can't be upgraded to the latest greatest. However, my house is a mess resulting from my office move, and I'm not inclined to dig one out and check what it can do. I did some Googling and found that the Rocket-M5 does support 5 and 10 MHz channel bandwidth, so I'll assume that your unspecified M5 version also does the same.

The article conveniently explains part of the logic behind using wider channels and mostly answers my question from my previous rant, which you deleted and/or ignored. Basically, the approximate math is simple. If your WISP configures their access point for a 40 MHz bandwidth channel and the ISP has 10 full time connected users, the system can deliver no more than 4 Mbits/sec to each user. If the WISP reduces the occupied bandwidth to 5 MHz, and still has 10 full time users, each one will only get 0.5 MBits/sec, which is inadequate. If your WISP doesn't have much of a user load, or doesn't overload the channel with too many wireless users, 5 MHz occupied bandwidth will work just fine. Note that this simplistic channel loading estimate ignores various factors that will either increase or decrease the channel loading. For example, I'm assuming that the channel usage is sustained at the maximum available rate, which is sometimes a bad assumption. This becomes really messy if the streaming media provider adjusts their deliver rate based upon error rate levels returns from the viewers computer or media player.

Also, there is a problem. This assumes that the WISP has exclusive use of the channel and that there are no other users on the same channel. Any co-channel users will appear as interference causing the WISP access point to lower the data rate to a level where the BER (bit error rate) is high enough to produce usable throughput. In many cases, this throughput reduction can be drastic, but for this discussion, I'll assume it reduces throughput to half. That means that delivering a given amount of data will double the air time (how long the transmitter occupies the channel) and delivery will therefore take twice as long. Actually, it's longer because the packet size is also reduced, but to keep things simple, I'll ignore that. The result of slowing down due to interference is that every users connection slows down, and data takes twice as long to deliver. Instead of ten happy Netflix viewers, the ISP support phone will have 10 irate customers complaining of buffering.

So, what can an WISP do? Well, it could not load the channel to the maximum capacity for a given occupied bandwidth. It could add another radio on a different channel and move some of the customers there. Or, it could just size the occupied bandwidth setting to match the actual channel loading with some overhead left for interference and high usage peaks.

So, why did your WISP use 5 MHz. None of the advanced 5GHz mode beyond 802.11a are going to work well crammed into a 5 MHz occupied bandwidth channel. I'm not sure if 802.11a will work in a 5 MHz channel. 802.11ac requires an 80 MHz channel. It would be interesting to sniff the traffic between your Ubiquiti M5-something radio and the WISP access point with a Wi-Fi Analyzer (Android) or something similar. My guess(tm) is you're running 802.11a.

So, what kind of performance can one expect in a 5 MHz wide channel compared to a 20 MHz channel? That would 1/4th the speed *OR* double the range due to increase in power density (dBm/Hz). That's why it was attractive to your WISP. Cut the data rate in half yields a range increase of sqrt(2) or 1.414.

That's also why the FCC and other regulators seem to have purged 5 and

10 MHz occupied bandwidth from the rules-n-regs. It's much too close to narrow band modulation and carries some of the detrimental effects of narrow band modulation. It was fine when the typical 5 Ghz signal used 20 MHz modulation. However, with 40, 80, and 160 MHz now available, the narrower occupied bandwidths had to go.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

Yeah, that does. But I was referring to the glassware and the HCl.

-sw

That's an input error of (0.2 - 0.152) / 0.2 = 24% Perhaps using the online calculator which I provided might have been a better idea? Or maybe a different Fresnel Zone calculator?

This is interesting and might explain a few things: Since the F2 zone is detrimental to receive signal level, antenna heights are often selected so that F1 is an unobstructed path and F2 is obstructed by a hill or the earth bulge along the path. In other words, the area around the F1 line is where you get your usable signal, while the area around the F2 line is where you get your problems. The reason you can get away with 40% incursion into the F1 zone is that reflecting objects on or near the F1 line will add, not cancel. I guess it really should be something like: 0.0 to 0.6 F1 = OK. Direct path. 0.6 F1 to 1.4 F1 = problems due to destructive cancellation. 1.4 F1 to 0.6 F2 = OK 1.4 F2 to 0.6 F3 = problems due to destructive cancellation. I'm not too sure the exact coefficients are correct. I'll check (later). In other words, there is a "band" straddling the various odd numbered Fresnel Zone lines which define areas that should not contain reflective objects.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

To the best of my knowledge, access to chemistry glassware is not restricted in California. In the United States, some regions have stringent regulations concerning the ownership of chemicals and equipment. For example, Texas once required the registration of even the most basic laboratory glassware.[20] However, this requirement was repealed on June 6, 2019.[21]

Just buy what you need online: Or, you can make your own: I have some glass tubing which I use to make the small stuff. However, I haven't had much luck with glass blowing.

For HCL, if you can't get the real stuff, buy some muriatic acid and distill it: I've never had to do this but it doesn't seem difficult. (famous last assumption).

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

Pennsylvania is a tax-friendly state for retirees. The second link is for a $200,000 house in our general neighborhood (note the high taxes and that it is a twin, however). b

Few wildfires. Quite diverse in the best sense of that word. And a 2,936 mile move would definitely scrape off the barnacles. If you prefer more country-style living, but with college-town amenities nearby, there is central PA (near our summer house).

Bloomsburg is a college town.

Then, of course, there is always Ranger, TX.

Peter Wieck Melrose Park, PA

Which is where I ended up. Real estate is cheap. Albeit most of it is going to need work to some extent.

Added bonus: You get to make my life miserable.

--
"I am a river to my people." 
Jeff-1.0 
WA6FWi 
http:foxsmercantile.com

Interesting video, thanks, I live near downtown SC I wonder the same.

Yeah, but the whole point was that the OP didn't have to guess nor did I need to educate him/her on the formulas for calculating the Fresnel. There are online calculators for it. Hell, I have an app for it on my phone.

This was what I was replying to, in the first place:

be.

r.)

io.

I replied with:

the

t.

The first bit of text I gave was the information for him to do it himself. First and upfront.

I also ran a quick calc with the values I could. Clearly not the 800 ft required (I made this blatantly obvious), but close enough (on the safe side) that if the OP didn't want to bother looking up the precise amount, my calculation was on the "safe side" and would work just fine.

I wasn't trying to expend the calories to give a precise answer, otherwise I'd have gotten up, walked across the room, and used my app.

I also figured that if the OP wanted the precise calculation, that I had not provide, he/she could burn the calories to type "fresnel zone calculator" in google and have at it.

I wasn't deliberately ignoring anything. I was just picking/choosing what to reply to. (Limited time and all that jazz)

I disagree with the 4mb/s for each user though. Clients with less than ideal signals should be put into low priority on the AirMAX scheduling priority. This prevents them from hogging up transmission time. (for M radios - AC radios apparently are able to handle that with whatever programming logic UBNT has come up with as you no longer have to specify priorities)

There area also various modulating schemes to help with bottlenecking (TDMA, CDMA, and various new ones I'm sure).

My own tests flat out contradict that 4mb/s bs.. I can deliver a lot more to customers than that.. And yes, these are netflixers so they're using bandwidth constantly. Maybe back in the day this was true, but that post you referenced is 5 years old and it is no longer the case.

Just out of curiosity, where are y'all getting this idea that a 40MHz signal can only deliver 40mbps? The 40MHz signals I use will deliver at

300MB/s (roughly)