4 ms·

I have tried MOCA, but it runs at higher frequencies than cheap European coaxial was designed for (lower frequencies for terrestrial television) and had stabili

I have tried MOCA, but it runs at higher frequencies than cheap European coaxial was designed for (lower frequencies for terrestrial television) and had stability/performance issues due to this. There's an alternative called G.hn over Coaxial, that runs at 100-200Mhz which seems to fair much better on crappy coaxial, at the cost of max bandwidth (500Mbps-1000Mbps) and harder to get hardware. The manufactures of G.hn2 Coaxial adapters seem to be stingy about selling them to individuals.

sidewndr46 3y ago
Given that WiFi is meant to work over the air, you could probably get a set of access points, RP-SMA to type-F adapters, and a set of attenuators and connect them back to back over basically any random coax. The loss is higher in the coax than in the air but there should be a noise floor consistent with transistor noise at room temperature. You'd want to start with as many pads as possible, so something like >80 dB of pad. Remove them one on by one until the reported signal level is in the middle of the acceptable range. From the device perspective it won't really know the difference. You might need to disable antenna diversity as otherwise the APs might attempt to use one of the antenna ports with nothing connected to it.
- Dylan16807 3y ago
  > The loss is higher in the coax than in the air Oh, I did the math on this before. So you technically don't have much loss from the air itself, but you lose 6 decibels every time you double the distance from the antenna. So at the distances people normally use wifi, the loss from distance alone is higher for air than for coax. And that's not counting the 40+ decibels of loss you start with for adjacent devices with antennas.
  - sidewndr46 3y ago
    it really depends on the coax involved. The old stuff put in homes is abysmally bad for cable TV at 700 or so MHz. So at triple the frequency it is really bad. The APs are nominal 50 ohms, and the coax is 75 ohms. So that alone is a huge loss that is present twice in the system. Also sometimes the coax takes insane routes because the installer couldn't be bothered to run it the shortest distance. It isn't hard to find a scenario where two APs with dipoles technically have less loss than a run of coax. The advantage the coax has is noise floor, which is what actually determines the bandwidth of a communications channel.
    - Dylan16807 3y ago
      > The APs are nominal 50 ohms, and the coax is 75 ohms. So that alone is a huge loss that is present twice in the system. Is the impedance mismatch worse than the 40 decibels of loss you get just by having antennas? Shouldn't it be pretty small? And you can correct for it, can't you? > The old stuff put in homes is abysmally bad for cable TV at 700 or so MHz. So at triple the frequency it is really bad. What kind of cabling should I assume? RG-59 loses about half a decibel per meter at 2.4GHz. So if someone is 30 meters away, but it's 60 meters via coax, both methods should give them 30 decibels more loss than at 1 meter away. The total loss is still over twice as bad for the antenna connection. And that's with no walls. What does a situation look like where the antenna technically has less loss?
- ssl-3 3y ago
  Wifi radios are generally designed for 50 Ohm impedances. But household TV coax in the States (RG-59, RG-6, RG-11, and friends) is 75 Ohm. And impedance matters. How much it matters varies by application, but it can be very important. There's more going on in bit of coax than a shielded conductor[1]. (Relatedly, twisted pair cabling like Cat5e has more going on than a few pairs of wires that happen to be twisted.) But the concept of having a pair of radios connected together with coax does, of course, work fine as long as the I's are crossed and the T's are dotted. I've set up PtP Wifi-ish wireless links on the bench with attenuators and coax jumpers instead of antennas, but I kept the characteristic impedance consistent throughout. (Many-to-many networks are also possible, with similar caveats.) [1]: Similarities of Wave Behavior, as presented by Dr. J.N. Shive, 1959. https://www.youtube.com/watch?v=DovunOxlY1k https://www.youtube.com/watch?v=DovunOxlY1k
shellfishgene 3y ago
> The manufactures of G.hn2 Coaxial adapters seem to be stingy about selling them to individuals. I can find several at Amazon (Germany). For example https://www.amazon.de/GIGA-Copper-Ethernet-Netzwerk-Koaxialkabel/dp/B08YXZ1W69/ https://www.amazon.de/GIGA-Copper-Ethernet-Netzwerk-Koaxialk...
- PlutoIsAPlanet 3y ago
  I wanted to grab a pair of Nexuslink/Comtrend adapters.
mrkstu 3y ago
I have ancient coax in the US on my house built in the 70s, leading to syncing at lower bandwidth often on my moca, sometimes as low as 100Mb. Luckily it runs outside the house, so at some point I’ll replace it with newer cable and actually get the theoretical 2.5 Gb.
- Symbiote 3y ago
  If you're replacing the cable, you may as well install an ethernet cable.
  - hughesjj 3y ago
    Or fiber if you don't care for poe
- PlutoIsAPlanet 3y ago
  I would suggest trying G.hn over MOCA on old coax networks. When I tried MOCA, disabling the upper frequencies on the devices would lead to a much stable performance, at the cost of speed. MOCA seems designed with cable networks in mind. G.hn had no issues, and the latency is much lower than Moca (1-2ms vs 6ms).
jauntywundrkind 3y ago
G.hn also is available as a power line technology too. I'd used HomePlug powerline for years in various houses/apartments. Some links would be 150Mb/s, others struggled to keep 20Mb/s, depending. I had high hopes G.hn2 was going to be a lot better, but both Nexuslink & Comtrend 2Gb/s failed to make a significant difference for me versus the HomePlug, still usually under 40Mb/s. That house had particularly bad wiring it seems. Dedicated coax could be way better. Interesting to hear a moca failed story! I didn't consider that as a possibility.
- PlutoIsAPlanet 3y ago
  I'm in the UK where powerline doesn't fair well agaisnt the UKs ring circuit style electrical networks.

sidewndr46 3y ago

Given that WiFi is meant to work over the air, you could probably get a set of access points, RP-SMA to type-F adapters, and a set of attenuators and connect them back to back over basically any random coax. The loss is higher in the coax than in the air but there should be a noise floor consistent with transistor noise at room temperature. You'd want to start with as many pads as possible, so something like >80 dB of pad. Remove them one on by one until the reported signal level is in the middle of the acceptable range. From the device perspective it won't really know the difference. You might need to disable antenna diversity as otherwise the APs might attempt to use one of the antenna ports with nothing connected to it.

Dylan16807 3y ago
> The loss is higher in the coax than in the air Oh, I did the math on this before. So you technically don't have much loss from the air itself, but you lose 6 decibels every time you double the distance from the antenna. So at the distances people normally use wifi, the loss from distance alone is higher for air than for coax. And that's not counting the 40+ decibels of loss you start with for adjacent devices with antennas.
- sidewndr46 3y ago
  it really depends on the coax involved. The old stuff put in homes is abysmally bad for cable TV at 700 or so MHz. So at triple the frequency it is really bad. The APs are nominal 50 ohms, and the coax is 75 ohms. So that alone is a huge loss that is present twice in the system. Also sometimes the coax takes insane routes because the installer couldn't be bothered to run it the shortest distance. It isn't hard to find a scenario where two APs with dipoles technically have less loss than a run of coax. The advantage the coax has is noise floor, which is what actually determines the bandwidth of a communications channel.
  - Dylan16807 3y ago
    > The APs are nominal 50 ohms, and the coax is 75 ohms. So that alone is a huge loss that is present twice in the system. Is the impedance mismatch worse than the 40 decibels of loss you get just by having antennas? Shouldn't it be pretty small? And you can correct for it, can't you? > The old stuff put in homes is abysmally bad for cable TV at 700 or so MHz. So at triple the frequency it is really bad. What kind of cabling should I assume? RG-59 loses about half a decibel per meter at 2.4GHz. So if someone is 30 meters away, but it's 60 meters via coax, both methods should give them 30 decibels more loss than at 1 meter away. The total loss is still over twice as bad for the antenna connection. And that's with no walls. What does a situation look like where the antenna technically has less loss?
ssl-3 3y ago
Wifi radios are generally designed for 50 Ohm impedances. But household TV coax in the States (RG-59, RG-6, RG-11, and friends) is 75 Ohm. And impedance matters. How much it matters varies by application, but it can be very important. There's more going on in bit of coax than a shielded conductor[1]. (Relatedly, twisted pair cabling like Cat5e has more going on than a few pairs of wires that happen to be twisted.) But the concept of having a pair of radios connected together with coax does, of course, work fine as long as the I's are crossed and the T's are dotted. I've set up PtP Wifi-ish wireless links on the bench with attenuators and coax jumpers instead of antennas, but I kept the characteristic impedance consistent throughout. (Many-to-many networks are also possible, with similar caveats.) [1]: Similarities of Wave Behavior, as presented by Dr. J.N. Shive, 1959. https://www.youtube.com/watch?v=DovunOxlY1k https://www.youtube.com/watch?v=DovunOxlY1k

shellfishgene 3y ago

> The manufactures of G.hn2 Coaxial adapters seem to be stingy about selling them to individuals. I can find several at Amazon (Germany). For example https://www.amazon.de/GIGA-Copper-Ethernet-Netzwerk-Koaxialkabel/dp/B08YXZ1W69/ https://www.amazon.de/GIGA-Copper-Ethernet-Netzwerk-Koaxialk...

PlutoIsAPlanet 3y ago
I wanted to grab a pair of Nexuslink/Comtrend adapters.

mrkstu 3y ago

I have ancient coax in the US on my house built in the 70s, leading to syncing at lower bandwidth often on my moca, sometimes as low as 100Mb. Luckily it runs outside the house, so at some point I’ll replace it with newer cable and actually get the theoretical 2.5 Gb.

Symbiote 3y ago
If you're replacing the cable, you may as well install an ethernet cable.
- hughesjj 3y ago
  Or fiber if you don't care for poe
PlutoIsAPlanet 3y ago
I would suggest trying G.hn over MOCA on old coax networks. When I tried MOCA, disabling the upper frequencies on the devices would lead to a much stable performance, at the cost of speed. MOCA seems designed with cable networks in mind. G.hn had no issues, and the latency is much lower than Moca (1-2ms vs 6ms).

jauntywundrkind 3y ago

G.hn also is available as a power line technology too. I'd used HomePlug powerline for years in various houses/apartments. Some links would be 150Mb/s, others struggled to keep 20Mb/s, depending. I had high hopes G.hn2 was going to be a lot better, but both Nexuslink & Comtrend 2Gb/s failed to make a significant difference for me versus the HomePlug, still usually under 40Mb/s. That house had particularly bad wiring it seems. Dedicated coax could be way better. Interesting to hear a moca failed story! I didn't consider that as a possibility.

PlutoIsAPlanet 3y ago
I'm in the UK where powerline doesn't fair well agaisnt the UKs ring circuit style electrical networks.