Hey all, I’ve been trying to optimize use of 5GHz spectrum by bringing our AP’s from 20MHz channels up to 30MHz channels. In doing so I try to leave 5MHz of spacing between the designated channels. However - I’ve found that rolloff is not as sharp or sudden as I would have expected, so what looks good on paper actually results in overlap of adjacent channels. I set up a bench test to determine whether there was a problematic variable between the Radio, Antenna, or LMR cables. I couldn’t make any correlation to hardware but it did provide an interesting insight…
I powered 4 AP’s using a Packetflux Rack Injector on my bench (see photo), set in freerun with their tx power reduced and ran spectrum analysis on an SM. My findings were that a 30MHz channel looks more like a 40MHz channel before dropping off sharply. I tried 20MHz channels and found that they ate up 30MHz of spectrum. Firmware version had no effect on the analysis.
I understand that a “perfect” channel is not physically possible… but should I expect my channels to be more concise than I am finding? Does synchronization make this issue null? is the 10-15dB of loss at the edges of the channels enough to keep my colocated AP’s from hearing each other? I hope this is not the case as it would eat up a bunch more spectrum, but could it be better practice 40+ MHz spacing between 30MHz channels?
Any insights, or clarifications of my misunderstanding are much appreciated.
If your 450 radios are sync’d then you don’t need a guard band. It also looks like you’re using horn antennas… which typically have excellent F/B ratios and no side lobes… so all that is good for adjacent AP’s and back to back reuse… or even offset reuse. For even further benefits use 450i radios which have filtering in them.
Yep all my 450/450i’s are getting sync from multiple sources. I used the RF elements horns briefly but they weren’t appropriate for the application so they just happened to be on the shelf. I set them up in this lab to see if the rolloff was more concise than the sector’s, which it was not… pointing me back to the radio. 100% of my 5GHz AP’s (besides 450M’s) in the field are on the Cambium PMP 450 60˚ antennas.
I’m still inquisitive that the overlapping rolloff could be an issue… one of my towers has (7) 5GHz AP’s covering roughly a 180˚ swath. 5 out of 7 AP’s started going into a DFS hit frenzy on adjacent 30MHz channels with 5MHz buffers. I used STA 5.9 to free up some space in 5.4 and gave the AP’s a much larger buffer between channels and the issue dissipated. I’ve got things happy now, but it is just concerning to me that I might be leaving some should be usable spectrum on the table…
Here was the response I got from cambium support. I replicated this on my bench with the SM scanning @ 5MHz and did indeed see more square response which was a relief.
Cambium - “The built-in SM spectral scan is not designed to act as a precision spectrum analyzer, thus, the result will not be a perfect representation of the actual occupied channel bandwidth using this technique. However, with a correct scanning bandwidth setting, the result should be pretty reenable. Thus, to get a better representation of the occupied channel bandwidth, the SM scanning bandwidth should be set as narrowest as possible, which is 5MHz is the narrowest setting available.
Below is a sample of SM scan with scanning bandwidth set to 5MHz, the results from 5740 @20MHz bandwidth and 5770 @30MHz bandwidth show an expected squarer response of the occupied channel bandwidth.”
Correct, the 450 platform’s Spectrum Analyzer is a “Peak Power” reading, meaning if channel bandwidth is 20 MHz, it reports the peak power found anywhere in that 20 MHz. As you’ve shown here, it’s pretty misleading as if it overlaps a receive signal by 5 of the 20 MHz, it will report the power at that 5 MHz for that frequency’s power reading.
So like you found out, it is highly recommended to run 5 MHz for Spectrum Analyzer to get most precise reading. Unless you’re looking for any interference anywhere in that Channel Bandwidth’s operation.