dslreports logo
 
    All Forums Hot Topics Gallery
spc
Search similar:


uniqs
1432
wirelessdog
join:2008-07-15
Queen Anne, MD

wirelessdog

Member

Nanobeam Packet Loss Poor Connection Low TX Rate

Just posting this to save others headache.

There are issues with the Nanobeams. Auto-ack does not seem to work properly. Also having issues with port negotiation. Having to hard code ACK and port speed for them to work properly. Hopefully this will save others some headache.
LLigetfa
join:2006-05-15
Fort Frances, ON

LLigetfa

Member

Which model of 'beam? Some of the early releases supposedly shipped with gig capable ports but then the PoE injector was not gig and that broke the speed negotiation on some switches.
wirelessdog
join:2008-07-15
Queen Anne, MD

wirelessdog

Member

M5-400. POE's are labeled gig. Speed/duplex mismatch errors and dropped packets to both Mikrotik and Cisco devices.
plugnothing
join:2014-09-03
US

plugnothing to wirelessdog

Member

to wirelessdog
This is addressed on the UBNT forums. It seems, indeed, to have something to do with ACK; auto does not find the distance properly. There are beta firmware versions especially for non beam-to-bridge links.

The athstats shows up to 50% or more aggregated long retry percent (or more) for my beam-beam PTP link, which is paired NBE M5 400s over three mile CLOS span, running at a consistent -63 dBm and 99% quality. I set the transmission speed at MCS 12 because it was jumping around a lot, and that seemed to help.

I really wonder if it is not still a power supply problem. Inconsistent power, poor grounding for the mains, etc gives me all kinds of headaches out here in the countryside; but I think it is cheap POE units. Anyway when the throughput really builds up, as during a speed stresstest, that's when the quality numbers go down.

Could be self-interference, too, according to some threads.

All that being said, I have actually gotten consistent 85 Mbps throughput (measured on speedtest) through this link, with a jitter of 1ms and total latency of 19 ms through two routers, two radios, and three consumer-grade switches (I know, I know, don't start on me; this is just how I have to do it in this case). It has not burped even once in the two months since it was fired up. It is resilient--it operates unshielded, and is less than two feet from a shielded nanobridge ptp link in the same UNII band. Overall, I could not be happier with it, as it provides internet nearly comparable to what I used to get from FIOS.

I wonder what the link would be capable of if the "aggregated retry percent" were much lower. But then, I don't know what athstats really truly means. "Go read the code" (the usual google reply to this query) does not help me much, poor ignoramus that I am. Anyone who can link to a discussion of athstats' practical interpretation would have my sincerest thanks.

I know wirelessdog has mentioned the Cambium line favorably in a few recent threads, so he seems to have good knowledge of those, and way way more experience than I do in this area. Could he give me some idea of a comparable-speed PTP link from Cambium? What would they cost compared to the UBNT Beams, and what's the latency (very important to me)?

Inssomniak
The Glitch
Premium Member
join:2005-04-06
Cayuga, ON

Inssomniak to wirelessdog

Premium Member

to wirelessdog
Ive had the ACK problem, the beta firmware fixes it.
Also seen link issues with a nano beam to a mikrotik RB600.
Everywhere else I have them in 100 meg ports or wired for 100 meg only on Gig ports.
wirelessdog
join:2008-07-15
Queen Anne, MD

wirelessdog to plugnothing

Member

to plugnothing
said by plugnothing:

what's the latency (very important to me)?

Add 20-30ms. I don't use them to backhaul with for this reason.
plugnothing
join:2014-09-03
US

plugnothing

Member

Thanks wirelessdog. Important, because latency is so rarely mentioned--one can really only get this kind of info from someone who has actually spent the money and implemented the radios.