I use the 2.4GHz or 5GHz frequency on my "repeater" (a TP-LINK Archer C7), connected to my main router (another TP-LINK Archer C7), using the other radio for WiFi. The Internet just feels slow! The latency seems poor and often the bandwidth is extremely variable.
And yet when I connect via ethernet to the repeater, the Internet feels fast and zippy and no bandwidth issues. What is going on here?
There's limited bandwidth on a wireless channel compared to a wired connection. Further, that bandwidth is "time-sliced" between basic operation (such as beacons and announcements) and sending data. A single radio is either transmitting or receiving, not both at the same time.
Past that, if you're using the same radio for "repeating" that you are for "connection" then you immediately suffer at least a 2x reduction as every packet you send to your repeater, it then needs to send from that repeater to your "main" router. If you use two different radios then there will be more latency but not as much bandwidth reduction.
Thanks for the response.
So is there anything I can do or am I going to have to accept a poor experience using the other radio on a dual-band repeater?
Run wires. Repeaters are inherently lower bandwidth and higher latency. There might be things you could do to configure things better (such as choose a less congested channel for example) but it only goes so far.
You can mitigate it somewhat by using 5 GHz for your backhaul and one of the higher-bandwidth (80-MHz) channels, if your equipment and the RF environment permit it. DFS, which is not "optional", often limits the use of 160-MHz channels in the US.
Powerline modems (Homeplug) work well in most cases if you can't run a cable easily.
I use a powerline device for my media PC, and an access point, it does work well, has some latency issues so I wouldn't use for gaming but better than wifi.
How bad is really the latency? Have you measured it, or it's just the feeling?
While wifi repeaters do add some latency, it should be unnoticeable for normal browsing; that is, unless there is something wrong happening in your network.
These are known to (potentially) interfere with VDSL-modems, so in case you use a VDSL link make sure to separate the powerline devices from the dsl modem....
as @jeff said; use the 5ghz radio for connecting the the two archers and the 2,4ghz for your access-wifi. i run this setup with the same devices, works well.
regarding powerline, as @dlakelan said; it usually is between "real" cables and wifi from a qos point of view. just avoid drawing power on the same fuse (as the powerline) which does not go throu the powerline adapter.
I did measure at one point, I think it had something like 40ms delay, compared to less than 1 over cat5e but the delay was very consistent.
As for drawing power off the circuit, I suspect this doesn't matter much with modern home plug AV stuff but haven't tested it. I suspect noisy switching power supplies would be the biggest issue. Plugging your wall warts into good quality surge protectors would likely help keep noise off the powerline.
I just did a ping test from the router to device attached to poweline modem and got < 7ms
I'm using a homeplug AV2 device
Yes, I should have said it was latency under load. Without load it is about 3 to 5ms but if I do an iperf test to my machine and run the pings it goes up to around 40ms and stays there. There doesn't seem to be any QoS possibility inside the device, and it doesn't seem to help to tag the ping packets with DSCP values, it just builds up a small queue and seems to have FIFO type behavior, but at least limited to around 40ms
One thing to check with power-line adapters is if you are on the same "phase" or not, at least in the US. At least when I looked at them, they relied on "stray" coupling between the phases. You may have better connectivity when the outlets are on the same phase.
There isn't a simple way that I know of to determine if two outlets in your house are on the same phase or not. At least in the US, 120-V power is supplied as two "hot" lines, 180° out of phase from each other, typically colored red and black at the panel, and a common neutral, typically colored white. By the time you get to the 120-V outlets, the distinction between the "red" phase and the "black" phase typically has been lost in terms of color-coding of the conductors.
My understanding is that modern powerline adapters do works better on the same phase, but really not too bad on cross-phase, earlier technology worked really poorly on cross phase. Unless you need to squeeze maximum bandwidth out of it, I'd say just plug em and go and then test to see if they're performing well enough. They are surprisingly good. The difficulty of getting onto the same phase is now you have to run some Cat5e/Cat6 over to a different outlet, and you have to figure out which outlets are on the same phase etc... I think for typical consumers this is not easy.
People in other countries with different wiring standards may be better off.
Powerline unfortunately destroys my VDSL connection and makes it very unstable!
Is your vdsl modem on a good surge protector or UPS? Just wondering.
Also potentially try adding a clip on ferrite choke to the power supply lines: on amazon there are a number of sets of ferrite chokes you can find by searching for "clip on ferrite" It won't help if the issue is the powerline noise bleeding over onto the copper phone lines directly, but if the issue is powerline noise coming into the modem through the power supply it will help a bunch.
@Gigabit A surge protector that advertises RFI filtering, often available for musical instruments and recording gear, could be useful eg. https://www.amazon.com/Furman-SS6B-Plug-Surge-Protector/dp/B0002D017M
Both phases should be fused individually, so just remove the fuse for one of them - the outlets that are still powered are on the same phase.
No, in the US the main breaker to your panel must be a dual pole breaker, it cuts power to both phases simultaneously. (anything else would be quite dangerous in case someone thought they'd cut power but it was only half true...)
Once it hits the panel, each circuit is separately breakered unless they're for a 220 V line or a multiwire branch (two 110 lines on a single 3 conductor line).
The main way to figure out if you have two circuits on the same phase is to figure out which breakers they're connected to, and then measure the AC voltage across the two breakers. If it's 220 they're on opposite phases, and if it's 0 they're on the same phase. Requires an AC multi-meter and knowledge of electronics and power safety though.
THIS IS DANGEROUS and inaccurate, at least homes in the US wired to the National Electrical Code!!!
Correct!!!