I have been forced to remove AREDN nodes from 2 sites for interference issues with commercial equipment. Both sites were using Rocket M5 with shielded Cat5 cable. Ferrites and RF armor cases on the rockets. at both locations we tried changing frequencies and bandwidth on the Rockets with not fix to the repeaters until the power was cur to the Rockets.
The effects were seen as low as 403 govt repeater as well as up into the GMRS bands. The Rockets were replaced with the same effect.
Mike
You are here
Ubiquiti interference with UHF repeaters
Mon, 02/19/2018 - 16:05
#1
AE4ML
Ubiquiti interference with UHF repeaters
Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer
So your M5 Rockets at two different sites were each interfering with 460 and 403 MHz repeaters with which they were collected? How far from these systems were the Rockets? Are you thinking IMD or harmonics? Were there any changes noted and measurements taken before and after?
I would love to discuss this with you on the phone if possible.
73,
Gordon Beattie, W2TTT
201.314.6964
w2ttt@arrl.net
I noted in another thread recently to watch out about ferrites, the network cable is the method the devices have to dump to ground, putting a ferrite on it could choke noise into the device making it radiate.
However there was zero change with adding the Armor I would lean more to it coming off the Cat 5 cable vs the device.
Was this single or double shielded network cable?
I have heard that Ubiquiti has supplied a firmware tweak (for the OEM firmware) to change some clock frequency inside the radio, in order to move radiated birdies one way, or the other. I wonder if the developers are familiar with this approach, or could research it? If it's just a single channel that is being interfered with, this might alleviate the problem.
If it is Ethernet related, I imagine that using a 100 Mbps Ethernet switch might be preferable to using a gigabit switch in most instances ...
Seem I had forgotten the processing frequency is 400MHz on many of the supported Ubiquiti devices, that might make sense for the 403Mhz repeater, Perhaps some rolloff or some spurs from the internal oscillator?
Did a quick pull of the FCC filing, nothing obvious sticks out to me in the graphs shown but I only did a really quick look over.
Does have me wanting to power up a node next to my spectrum analyzer and see what it shows.
As to if we can do anything about it inside AREDN. Well, maybe, It probably needs a feature enhancement ticket in bloodhound to move it forward if it is thought that this would solve it. I'm not sure how accessible the setting for CPU speed is on these devices inside the kernel used as they are quite different than a desktop. I would worry if the oscillator noise is all the way from at least 400Mhz to 470 how much we could really move it, but that is where I would need to see the spectrum to know for sure.
That said you might be on to something Ken.
I did a broad band spectrum and posted it here quite some time ago (2015):
https://www.aredn.org/content/rocket-shields#comments
Today I had a closer look.
The M5 XM chassis has a nice oscillator at 390 MHz with prominent harmonics on 780 and 1170 MHz - these are "dirty" with significant sideband noise.You can see the 390 and 1170 quite clearly in my original posting.
The M5 XW, on the other hand, has a somewhat cleaner oscillator at 400 MHz. Harmonics are there, but less prominent. What the XW oscillator has, however, are sidebands at +/- 5, 10, 15 MHz. These appear when the CPU is running its signal processing routine (goes away during a reboot). So it puts out signals on 390, 395, 400, 405, 410 MHz. These sidebands fade down into the general noise level as you move away from 400 MHz. I did not see discrete carriers in the 462/467 MHz frequency range though ... a faraday cage would be required to get good readings for this type of thing at my QTH - there is a lot of RF around that is not coming from the M5 (also visible in my original posting, although I have found a way to reduce it somewhat).
Well at least we know where it is coming from (at least for the low end at 400Mhz say +/-20)
And that is right CPU speed got bumped 10MHz between XM and XW that explains the differences on carrier location.
Seems by the width its not going to be easy to clear out its going to need it seems at least 20Mhz if not 50Mhz of offset, something I'm not sure how possible it is to do and what impacts it would have (eg did Ubiquiti have to adjust timings or anything else where) The core kernel should handle it fine, it would be any low level chip parts that might pull off that oscillator that would be a concern.
I see shield kits were already tried, the only thought I can have is I wonder once we get Titanium rockets in main line if they will be any better (more metal) or if spray painting the plastic case with something like MG Chemicals SuperShield would help. Ultimately I'm personably not holding my breath on either, since were talking an oscillator near a receive antenna which is designed to receive very low power signals I think ultimately it probably involves the CPU changing its core frequency.
I'll leave it to the community to open an enhancement ticket to look into this if it is something they want pursued deeper.
I have to say that I do not know if the change in CPU clock frequency is due to the XW hardware design, or firmware. But it did fix a problem with a cell phone provider (2nd harmonic). In any case, the area just above 400 MHz (400-415) shows some pretty wide band noise content that is unlikely to be fixed by clock shifting. Above 425 MHz, the level is below my local noise floor.
There are actually 2 changes, a firmware change for XM devices, I don’t know if that was a move up or down but I suspect it was from I could find,390MHz to 400MHz. The second change is new XW devices all the datasheets now read 400 MHz so it looks like they upped it as part of the refresh.
The CPU as I understand it is speced 300-400MHz controlled by an internal PLL.
Down clocking would slow down a node performance but maybe would move that noise out of the 400 band (depends how wide it is, obviously at least 20Mhz by your tests and we should assume wider probably at least 50MHZ for safety)
Flip side: this also opens a question on 6th harmonic, it looks like your main spike is 401 which may make 6th harmonic at 2406 but if it’s actually 400 on some other device that could put it at 2400 flat (hello problem with negative channels) with the older 390
clocking it was 2340 way clear of us. I wonder if this is why some have reported issues with 2.4GHz sites being deaf.
NOTE: Ubiquiti forums deserve credit for pointing out 6th Harmonic, I’m just bringing up the issue that impacts us that normally UBNT users will not hit.
I would think if this gets moved forward it probably needs to drop to 350MHz or maybe even 325. To avoid interfering with cell sites and to avoid interference to -1 and -2.
Don't pay much attention to the 401 on the graph - it's not calibrated. The marker generator on the SA shows that the oscillator is actually very very close to 400.
Tonight I had a look at 2400 MHz radiation and it is very low - below my noise level and many dB below a typical signal. I don't think the rise time of that logic family supports generation of any significant harmonics that high in frequency.