I have heard reports the some Mikrotik devices only work at 10 and 20Mhz bandwidth (not 5Mhz)? Is this true for all Mikrotik devices that support the AREDN firmware? I have a local ham trying to connect to the mesh using the 27dB gain Mikrotik dish and it's not working.
thank you Bill NG1P
thank you Bill NG1P
https://www.arednmesh.org/content/hap-ac-lite-hardware-quirks
https://github.com/aredn/aredn_ar71xx/issues/301
just trying to understand if this is across multiple Mikrotik devices. It's good to know yours are working.
Jim W8ERW
I tested early versions of the AREDN software for a 2 GHz MikroTik "BaseBox" and it produced a really weak and funny-looking output (as seen on the spectrum analyzer) when operated on 5 MHz. As I recall, it barely linked with another BaseBox and would not talk to Ubiquiti gear. With 10 and 20 MHz BW it worked fine. I do not know if the AREDN developer team ever followed up on that.
Ken
I just returned from a short "drive test" session with my portable Mikrotik SXTsq Lite2, which is the same Mikrotik board shared by the SXT, LHG, and LDF devices. I understand that this was not a controlled bench test, but just to add some anecdotal data to the mix, it does not seem like a generalization can be made that "Mikrotik radios cannot communicate with Ubiquiti devices using 5 MHz channel width." In about an hour of driving around my area I was able to make direct RF contact with almost a dozen Ubiquiti Bullet and AirGrid devices. There was not a single Ubiquiti node within RF range that I could not make RF contact with, so I saw no failures using 5 MHz channel width.
Thanks Bill NG1P
Generally 10Mhz is going to better perform at most distances over 5Mhz channel widths. By "perform" better, this means higher data throughput measured by iperf. Is 5Mhz in use due to limited RF space? If not, have you tested and compared with a 10Mhz channel width to determine which yields better performance? How significant is it to invest in working around the symptoms?
Joe AE6XE
thank you Bill NG1P
When we do cut the channel in half, the same 64 carrier waves and xmit power are squeezed into half this bandwidth, and then the symbol length is doubled. The difference is:
1) half the bandwidth can only do maximum half the bit rate, so max MCS15 rate drops from 130Mbps @ 20Mhz channel to 65Mbps @ 10Mhz channel.
2) the symbol is transmitted twice as long. This longer symbol time and built in guard gap between transmitted symbols gives more time for fading or reflections that could trash the successive symbol for larger bandwidths and shorter guard gap. If this guard gap is too short for how long it takes a signal reflection to be received, data throughput is trashed -- inter-symbol interference is occurring.
Unbiquiti gives many channel width options in AirOS. If we did the math, we could determine the optimal channel width, or minimum amount of time in the
symbal gap interval necessary for a given distance for how long reflections would take to be received, to determine optimal channel width or symbol gap timing. This is a big reason to set channel width, based on optimal timing for a given distance of the link.
Generally, the lower SNR for the larger bandwith can still achieve higher thoughput than half the bandwidth and higher SNR that cuts the max rate in half. If you can already achieve 32.5Mbps link rate or greater in 10Mhz channel, it's unlikely to improve performance n 5Mhz channel, with max 32.5Mbps protocol rate. But check iperf thoughput as best measure, not raw link rate.
Joe AE6XE
I switched over to 10mhz this morning. Will run some tests. I did install iperf but noticed on the node up on the 250 tower it ran out of space when I tried to load Iperfspeed but still loaded. I then removed SNMPD to get some more space then did a reboot but it never came back. Will need to make a trip out and see if a power cycle brings it back. (Hopefully I don't need to climb the tower to hit the rest button)
Denis
thank you Bill NG1P
I appreciate the clarification and additional information.
Thanks,
Jim W8ERW