…I want to do this now. I have so much unused coax in my house and Powerline works but not as well as an 867 Mbps 802.11ac single antenna link could work?
You might also look into MoCA adapters for converting coax to Ethernet, e.g. https://www.instructables.com/id/How-to-set-up-a-coax-MoCA-network/
I’ve done a few point-to-point MoCA using adapters from GoCoax (2.5Gbps, newer) and Actiontec (1Gbps). The GoCoax are the only Moca 2.5 adapters I’ve found for retail sale.
Wow, those aren’t cheap (the actiontec is $80/ea). How reliable are they at hitting the advertised speeds?
I looked through my notebook and found the MAC’s and other info but not my iperf speed tests. I have them deployed around the house and when I wired them up (fresh runs of RG-6 with high quality connectors) I got the expected speeds. The GoCoax units are faster and step down to talk with the Actiontec’s that I already had from a previous project.
Compared to running new cat5 wiring, it’s probably a great deal. I just did this recently, so I can’t comment on long-term reliability… also YMMV depending on the complexity and quality of your coax cabling… but for me, it’s got way better throughput and latency than WiFi. It works just like gigabit ethernet, and frees up the wireless net for mobile uses. I’m very happy with MoCA so far.
I used MoCA in a prior house (middle floor of a 3 floor condo, so no easy way to run cat5) and got the advertised speeds and it was rock solid for YEARS.
Being a radio person, when I first read this, I was honestly a little bit surprised that it worked at all.
If you know the frequencies involved (almost certainly 2.4 GHz here, since it’s a very old article), you can calculate the loss for a given type of coax. In this case, a 2.4 GHz signal in RG-6 loses 98% of its power over a 100-foot run. Most houses are smaller than that, so a more reasonable run of 25 feet turns out to be a loss of only 4 dB or 63%. Very acceptable given the rather insane sensitivity of wifi radios. Even 5.7 GHz would only present a loss of 7.4 dB over 25 feet. I’m not sure how or whether any of this work work with the fancy beam-forming and MIMO stuff that’s out now, however.
There is also the matter of a potential impedance mismatch. I don’t know for sure what impedance wifi radios operate on, but I assume it’s 50 ohms. RG-6 is 75 ohms so there will be some loss at both ends of the cable.
MIMO will fail to find multiple spatial channels, unless of course you do multiple coax runs. But if you wired one antenna port up to the coax and left antennas on the rest, it would probably manage to use both :)
Nice hack, I’d like to see someone benchmark it, and compare it with powerline approach on latency and throughput
Ethernet going full circle.
Or: Thiccnet with three Cs.
This is a great hack.
I wonder what the range is like when you do that. Seems like it could be pretty good, assuming suitable cabling?
Quick and dirty link budget:
TX power: 100mW (20 dBm)
RX sensitivity at the lowest speeds: about -90dBm
LMR-600 loss at 2.4GHz: 14.5 dB per 100m
LMR-400 loss at 2.4GHz: 22 dB per 100m
Belden 9913 loss at 2.4GHz: 25.5 dB per 100m
So depending on the cost and thickness of the cable you want to deal with, somewhere between 425 and 750 meters would be the outside limit. Subtract several tens of meters for every adapter/filter/diplexer/etc. in the path.
You can actually go a lot further than that in free space, with good enough antennas on each end and no obstacles in the way. Coax loss is a killer at microwave frequencies.