Virgin/powerline/WiFi/speed

kingsize

Member
My setup is a super hub 3 wired to a devolo1200+ WiFi system. I also have a sky q setup. house 2800 sq ft though rewired.

I run the 3 devolos on the 2.4 ghz band with 5ghz turned off. The router outputs both. It is I suppose a mesh system

I had VM 200mbps which usually resulted in the devolos outputting 10-30 mbps at each unit. Which was manageable I suppose, I expected some dropoff.

This is my query - and I may be totally misunderstanding everything - I upgraded my virgin and am now receiving 550 mbps but am not seeing any increase around the devolos output? Am I wrong to expect that an increase in speed coming in should be reflected , even at a reduced rate, in the output of the units connected via powerline to the router?

Or am I just missing the point entirely?

Thanks in advance to anyone who can educate me....
C1094253-CF76-4197-A733-FBD35BED4AA3.png
 

kingsize

Member
By the way I’m not getting anywhere near these speeds. This is just what the devolo app is reporting
 

ChuckMountain

Distinguished Member
Just trying to understand your setup so

You have VM SH3 router running with both 2.4GHz and 5GHz networks.

Is this a different WiFi network name (SSID) to the dLAN ones?

Then presumably you have the Study dLAN powerline point plugged into the router via an ethernet cable?

Then you have 3 dLAN 1200+ Wifi Access points in the living, kitchen and chamber rooms. Are these all on the same WiFi network name (SSID)? But these only operate on 2.4GHz? Any reason why?

Then you just have a powerline adapter 500 duo (no WiFi) in the drawing room.
 

kingsize

Member
Thanks for getting back to me - apologies for my lack of clarity.

1. You have VM SH3 router running with both 2.4GHz and 5GHz networks. - Yes, I thought I only had the 2.4GHx running, but it is in fact both.

2.
Is this a different WiFi network name (SSID) to the dLAN ones? - No, one network name for the entire house.

3.
Then presumably you have the Study dLAN powerline point plugged into the router via an ethernet cable? - Yes

4. Then you have 3 dLAN 1200+ Wifi Access points in the living, kitchen and chamber rooms. Are these all on the same WiFi network name (SSID)? But these only operate on 2.4GHz? Any reason why? - Yes, they are all on the same network name. I limited them to 2.4GHz with the hope of stopping their 5GHz signal messing with the SkyQ mesh? - which had been problematic and is generally better

5. Then you just have a powerline adapter 500 duo (no WiFi) in the drawing room. - Yes, the room gets great WiFi from the router in the study directly above so there was no need for another WiFi unit.

Thanks again for responding - I hope this makes some sense....
 

mickevh

Distinguished Member
In data networking the Link Rate (ever erroneously called "speed") of each "hop" in the infrastructure is independent of all the others. Thusly you are correct that changing the link rate of your ISP link has no effect on the link rate of any of your powerline links, ethenet links and Wi-Fi links. They are all independent of each other. Further - the link rate of each "lobe" in the ethenet is independent of all the other ethernet lobes and the link rate of any given pair of communicating Wi-Fi peers is again independent of all others (and can vary minute by minute and can be different in the "upstream" and "downstream" directions.)

When you run one of those pesky Internet "speed tests," it's results are based on the cumulative effects of all "hops" in the pathway between the source and sink devices. Essentially, it gives you an indication of what the slowest "hop" between your client and whichever SpeedTest server you test against. In olden days this used to almost always be your ISP link.

I'm fond of saying "speed test" does not actually measure the "speed" (as in Link Rate) of anything - what is does is send out a measured amount of data, time it and compute a statistical average. Like the trip computer in a car - it's not the same "thing" as the speedometer reading.

HomePlug's "cockpit" is reporting the "Link Rate" that the plugs are synching up with each other (and incidentally, not the Wi-Fi or ethernet rates.) This is not what you would expect to see if you then send data over the link. Link Rate is not a measure of "capacity" - Link Rate expresses the time it takes to transmit one bit expressed as a ratio by an arbitrarily selected time base - ie a second. So if I invented a technology called MickNet that takes 1/10th of a second to send one bit of data, that would be a 10bps "Link Rate" - it does NOT imply that I can transmit 10 bits every second. This is a much misunderstood concept, even in the IT profession. Even Wikipedia gets it wrong.

The reason I cannot send 10bps is because data networks don't transmit continuously. There are (for example) silent "gaps" between bursts of data to delimit the packets, time lost to "management" chatter that does not show up at the end stations, forward error correction, addressing information and more. The amount of the headline Link Rate "lost" to all these overheads is different for differing technologies (and varies over time.) There's a very rough rule of thumb metric called the "protocol efficiency" that compares these losses. Ethernet is very good at about 97% efficient, Wi-Fi is worse at about 55-75% (worsening as signalling conditions deteriorate) and HomePlug/Powerline is usually cited at about 45-55% (again, worsening in poor signalling conditions.)
 
Last edited:

kingsize

Member
In data networking the Link Rate (ever erroneously called "speed") of each "hop" in the infrastructure is independent of all the others. Thusly you are correct that changing the link rate of your ISP link has no effect on the link rate of any of your powerline links, ethenet links and Wi-Fi links. They are all independent of each other. Further - the link rate of each "lobe" in the ethenet is independent of all the other ethernet lobes and the link rate of any given pair of communicating Wi-Fi peers is again independent of all others (and can vary minute by minute and can be different in the "upstream" and "downstream" directions.)

When you run one of those pesky Internet "speed tests," it's results are based on the cumulative effects of all "hops" in the pathway between the source and sink devices. Essentially, it gives you an indication of what the slowest "hop" between your client and whichever SpeedTest server you test against. In olden days this used to almost always be your ISP link.

I'm fond of saying "speed test" does not actually measure the "speed" (as in Link Rate) of anything - what is does is send out a measured amount of data, time it and compute a statistical average. Like the trip computer in a car - it's not the same "thing" as the speedometer reading.

HomePlug's "cockpit" is reporting the "Link Rate" that the plugs are synching up with each other (and incidentally, not the Wi-Fi or ethernet rates.) This is not what you would expect to see if you then send data over the link. Link Rate is not a measure of "capacity" - Link Rate expresses the time it takes to transmit one bit expressed as a ratio by an arbitrarily selected time base - ie a second. So if I invented a technology called MickNet that takes 1/10th of a second to send one bit of data, that would be a 10bps "Link Rate" - it does NOT imply that I can transmit 10 bits every second. This is a much misunderstood concept, even in the IT profession. Even Wikipedia gets it wrong.

The reason I cannot send 10bps is because data networks don't transmit continuously. There are (for example) silent "gaps" between bursts of data to delimit the packets, time lost to "management" chatter that does not show up at the end stations, forward error correction, addressing information and more. The amount of the headline Link Rate "lost" to all these overheads is different for differing technologies (and varies over time.) There's a very rough rule of thumb metric called the "protocol efficiency" that compares these losses. Ethernet is very good at about 97% efficient, Wi-Fi is worse at about 55-75% (worsening as signalling conditions deteriorate) and HomePlug/Powerline is usually cited at about 45-55% (again, worsening in poor signalling conditions.)
That’s genuinely fascinating. I have been, perhaps, thinking too simplistically.

I had assumed that tripling the speed into the router would be reflected at the various plugs, and that a Speedtest was the best metric by which to measure, as it were, the endpoint.

I really appreciate the time people are taking on this
 

ChuckMountain

Distinguished Member
Thanks, @mickevh for the usual quality explanation :thumbsup:

In terms of hows this applies to you @kingsize there are a number of issues that apply to your setup and unfortunately, the news might not be what you want to hear :(

Let's take powerlines, I am not a fan of them as I am sure that @mickevh will attest to. The real-world speed of a powerline is not anything close to the "marketed speed". The biggest issue is that they quote the duplex speed, so the combined up and download speed so you can instantly half the speed. So your 1200 plugs, could only operate up to 600Mbps.

Now take the efficiency from @mickevh's post at the top end of 55% then you are looking at around 330Mbps throughput. A real-world test of them is typically much lower in reality below 200Mbps for even the top of the range ones. Again there are some wiring dependencies but even rewiring won't necessarily fundamentally impact performance that much.

This is a real-world test of your one.


Next to each on the same socket, they get a top speed of 375Mbps, so fractionally beating our earlier estimate. However, when they actually use it in an actual scenario the speed drops to 126Mbps.

I think personally powerlines advertising should be considered by the Advertising Standards Agency as if you think you are getting 1200Mbps and get nearer 120Mbps that's 10% of the advertised speed. Not sure which other adverts would get away with that, certainly, broadband can't anyway more.

Oh, and what the cockpit software suggests in terms of speed\link figures is frankly beep beep :D

So here is a problem your first bottleneck the link between the router and the rest of the WiFi network is going to be limited to no more than say 200Mbps on a good day and probably in the 100Mbps range the rest of the time.

Usually, with powerlines (like WiFi) it is only one device can talk at a time on the same frequencies. So you may find with 5 powerlines there is some contention, particularly if multiple devices are trying to all connect to the Internet etc.

On to WiFi, so you have gone with a fairly sensible approach in terms of trying to locate your WiFi access points to give you good coverage in the room.

Unfortunately, you have disabled 5GHz which is where the highest speeds will be obtained. On testing a connection I can get up to around 80Mbps on 2.4GHz depending on the conditions and I typically get up to around 600Mbps or so on 5GHz depending on the device and the bandwidth setting. By disabling it you have probably moved the bottleneck to this point and if you have multiple devices you will all be competing for the same bandwidth.

Ideally, if the dLAN is operating like a mesh network you should disable WiFi on the router so that one WiFi network controls all your network. This way clients can be encouraged to jump access points (it normally their choice), at the moment what you a bit of a hybrid system. That might work ok but it can introduce issues.

So this brings us back to the ISP, realistically unless you have a lot of devices on a good WiFi network and hardwired devices you are probably not going to make the full use of the bandwidth available and are probably overpaying. Unfortunately, VM are very good at doing marketing and encouraging people to upgrade. I think anything above 200Mbps you are into diminishing returns, very few individual services need that speed and you get 4 to 5 full UHD streams through that before you start to notice any issues. The internal infrastructure becomes a bottleneck, particularly around powerlines, which should not be used as any part of the home "backbone" in a VM or FTTP house.

Do you have any hardwired machines that would benefit?

So possible solutions depend on budget.

Personally, I would rip out the powerlines out and install hardwired cables. That probably won't go down well after just rewiring and presumably plastering etc.

Move as much as you can on to hardwired in some areas so that it is not competing for WiFi airtime.

Look at one of the WiFi mesh ones that people have used with VM. Typically they have tri-radios or hardwired base units.

Install something like UniFi and have WiFi Access points hardwired in.
 

kingsize

Member
Well, thanks for that. Very informative, if a touch disappointing.

I had been considering this conundrum from an uninformed perspective.

In light of the foregoing, I now at least partially understand the parameters within which I can manoeuvr.

my operative concern had been the devolo mesh interfering with the sky q mesh thereby disrupting the sky .

I think that both mostly work well, mostly....I’ll leave well enough alone with WiFi on 2.4 and sky on 5 - the trade off on speed, if indeed it is a trade off, is acceptable.

Thanks again for your exceptionally informed perspectives.
 

ChuckMountain

Distinguished Member
To solve the Sky Mesh issue you could try and hardwire your Sky Q and Mini boxes and switching WiFi on those boxes. That way they can only communicate by hardwired ethernet and won't interfere. That's what I do with my VM\Sky setup and it seems more reliable.
 

kingsize

Member
Thanks again.

I had the sky q hardwired (through the powerline units) with the WiFi on each box turned off so just connecting via Ethernet and it was a misery - constant disconnections and days spent trying to get it right.

so went back to sky mesh as it were and it has been much better. Not perfect but not too bad.

on a whim I enabled the 5G network on the devolos and it doesnt seem to interfere with the sky a.

having said that, the 5ghz network doesn’t seem to be any faster than the 2.4gha...

oh well, as they say, first world problems...
 

kingsize

Member
The only available relevant option is pictured below - no option to pick 80mhz on its own, and no option to do so with the powerline WiFi units...
76933842-4476-440C-96EE-640F976F53FA.jpeg
C4CE2CEE-FE3F-4C94-A2EB-3DD0EBE70E5B.jpeg
 

mickevh

Distinguished Member
The only available relevant option is pictured below - no option to pick 80mhz on its own, and no option to do so with the powerline WiFi units.

That's not unusual in cheap SOHO kit - the granularity of control is often very limited.

"80MHz" channels are not 80MHz "only" - the 40/80/160Mhz channels are creating by (kind of) "bonding" together multiple 20MHz channels. When a "thing" want to use a (say) 80MHz transmission it starts out at 20MHz and kind of has to "clear out" (or quiesce) multiple 20MHz channels before proceeding to transmit using 80Mhz.

It's not unusual to see 80MHz modes expressed as "20/40/80" or something similar in the user interface as technicaly they are all 20Mhz to start with and various bits of magic trickery are used to "step up" to 40/80/160 as required. Not least because backward compatibility needs to be maintained for older devices that don't support the newer protocols and channel widths.

Some kit doesn't offer 40/80/160MHz channels at all and it's 20MHz all the way. Cheap stuff is cheap for a reason.

80/160 are only available in the 5GHz waveband, so you'll never see them in the 2.4GHz waveband.

40Mhz is available in N in the 2.4GHz waveband, but I wouldn't use it if you have more the one hotspot. There's not enough frequency spectrum available on the 2.4Ghz waveband for more than one 40MHz channel. There's only "room" for three non-interfering 20Mhz channels or one 40Mhz plus one 20MHz. In big deployments I would only offer 20MHz channels in the 2.4GHz waveband - that effects the advertised top "speed" (basically halves it) bit it helps in reducing co-channel interference between neighbouring cells and can result in higher throughput overall as there's less "damaged"transmissions that require re-transmits, less competition for "air-time" and so one. It's difficult enough devising a channel plan with "only" three channels available - it's practically impossible with one two.
 

kingsize

Member
It's all so bewildering.

I suppose it is better to appreciate what I have, rather than what I don't have.

Wifi coverage is relatively good at avg speeds of 20mbps around the house, which covers what I need to do - PLEX etc

My query had been prompted by an uneducated view that faster in = faster out.

Now I at least appreciate it's not that simple.....
 

The latest video from AVForums

Podcast: Panasonic JZ2000 Final Thoughts - TV Calibration: Should you? And More...
Subscribe to our YouTube channel

Full fat HDMI teeshirts

Support AVForums with Patreon

Top Bottom