How many HomePlug units do I need?

[fizikz]

The configuration I have in mind is as follows:

Modem ---(powerline)---> Router ---(powerline)---> PC

In particular, how many units are needed at the router? I would hope I can get away with just one there, but the modem needs to plug into the router's WAN port, while the PC needs to plug into the LAN ports.

The purpose of this setup is to put the router in a central location since it is also a wireless AP.

 

[relax]

You need 4 units:
One at the modem
One at the wan port of the router
One at the LAN port of the router
One at the PC

The first two need to be setup so that they create a first network
The second two need to be setup so that they create a second network different from the first one.
Remember that the electric wiring will be shared between the two networks, so performance will suffer from the coexistence of the two networks on a same physical line.

 

[fizikz]

Four units... I was afraid of that. What are some ways of making better use of resources? What about:

Modem ===(ethernet)===> Router ===(ethernet)===> Wireless AP ---(powerline)---> various PCs

Any other better setups?


EDIT: Actually, this is more what I had in mind:

Modem ===(ethernet)===> Router ===(ethernet)===> Wireless AP

and in parallel from the same router:

Router---(powerline)---> PC

OR

Modem =====(long ethernet)=====> Wireless router ---(powerline)---> PC

 

[thiggins]

You are going to have the easiest time by keeping your powerline connections on the same subnet, i.e. on the LAN side of the router. Plug one powerline adapter into a router LAN port, then use another powerline adapter to connect each device that you can't / don't want to connect via Ethernet or wireless.

 

[fizikz]

Thanks, I'll choose a configuration that keeps the powerline units on the LAN side of the router.

 

[relax]

Just for the sake of completeness, there are available powerline units which are also wifi access point. That might solve your issues.

 

[fizikz]

That's a good point. If I were making the network configuration from scratch, I think that would make it simplest. Right now, I'm going to try to make the best configuration I can while minimizing new purchases.

 

[fizikz]

Another idea: what about using an extension cord for easier location selection for the router and its LAN-side Homeplug adapter? Am I correct in thinking that a regular extension cord should not have any significant impact on the HomePlug performance?

 

[thiggins]

Another idea: what about using an extension cord for easier location selection for the router and its LAN-side Homeplug adapter? Am I correct in thinking that a regular extension cord should not have any significant impact on the HomePlug performance?

Homeplug adapters must be plugged directly into wall outlets for maximum performance. Even plain extension cords can attenuate the high frequencies that HomePlug uses.

 

[fizikz]

Even higher gauge extension cords? I guess it's back to planning configurations for me.

 

[fizikz]

I got a starter kit (TP-Link TL-PA511KIT) plugged it in (one in the basement, the other on the main floor), and had my expectations fall rather flat. On one hand it does what it needs to do, which is share the internet connection (5Mbps, so shouldn't be that challenging). The powerline status indicator light is red (green > 80Mbps, amber 40-80Mbps, red < 40Mbps) and the link rate shown in the management utility is about 30Mbps. How they get this number, I do not know. Transfering a large test file got me about 14Mbps. At least if the connection stays stable, then it might still be better than a wifi adapter.

Just to test things in closer to ideal settings, I also tried plugging them both in the same room, and saw the management utility report around 350Mbps. I also updated the firmware, but I didn't notice any appreciable change.

I don't really enjoy seeing the red status light and matching performance when I opted for the "500Mbps" gigabit model, but at least it solves the home network wiring problem. All my pre-purchase concerns regarding gigabit vs 10/100 LAN ports seem utterly amusing now.

 

[thiggins]

Check the tips here
http://www.smallnetbuilder.com/lanw...ways-to-boost-powerline-network-speed?start=1

How fast was your file transfer when the adapters were in the same room?

If you want to experiment, I would try a pair of 200Mbps adapters from NETGEAR. 200 Mbps technology is more mature.

 

[fizikz]

I finally got around to testing more throroughly, and it seems overall, I'm getting about 5-10Mpbs with the HomePlug kit in most places in the house, some a bit more, some a bit less.

Testing was done using one laptop (Asus G1) plugged into the router (Engenius ESR9850). The HomePlug adapter was plugged into the router (LAN side). I then took the other HomePlug adapter and tested with my netbook (Acer Aspire One). Testing was done with iperf, with the -d flag for simultaneous bi-directional testing. Without simultaneous testing (-r flag) the speeds were a bit higher.

In the same room, I saw about 90Mbps, with a max close to 95Mbps. Anytime the path involved going through different breakers, the speeds dropped a good deal. From the router's location to a power outlet near the electrical box, I saw around 15Mbps. To the basement location where it is meant to stay, I get 5-10Mbps. I noticed that the results are not static and change a bit on every test, sometimes by as much as 1-5Mbps.

I also tested from the same iperf server plugged into the router to various wireless G clients and was surprised to see how low the results were, about 5-10Mbps on the main floor pretty close to the router, and around 3Mbps in the basement. What confuses me is that when doing internet speedtests, none of the computers have difficulty getting 5.X Mbps download results. I wonder how that is. Perhaps because of the fluctuations in bandwidth and signal strength, whether with wireless or HomePlug. HomePlug is at least more stable and still faster compared to wireless in most locations except those closest to the router which would be on a different breaker.

Checking for noisy devices plugged into outlets would be quite a daunting task, and I doubt my family would take kindly to me unplugging everything and testing. My conclusion is that with my current network setup, there is little point in upgrading the 6Mbps DSL internet connection as the network does not support it for most of the devices. The next upgrade might have to be a strong wireless AP. In the meantime, the HomePlug does the job and is better than wireless in the basement location.

I was considering another HomePlug adapter for the home media area which is at the other end of the room where the router is located. Interestingly, since the outlet is on a different breaker, the HomePlug adapter actually performs worse there (~3Mbps) than in the basement, and clearly worse than wireless for that location. However, on an outlet on the opposite wall it would perform much better at 80-90Mbps.

EDIT: I checked in the electrical box, but there was no indication of AFCI breakers. I would test with other adapters if I had access to them. Could a 200Mbps adapter actually outperform a 500Mbps one?

EDIT2: One important point is that the bandwidth reported by iperf may be different from the actual throughput. The file transfers I had done before were completed more quickly than the iperf results would indicate.

 

[thiggins]

I recommend using either drag-and-drop file transfers or LAN Speed Test or NetStress. Easier to use and more reflective of actual performance.

Yes, powerline does take some work sometimes. Do you have any 220V appliances in your home (water heater, electric range)? These appliances help increase phase-to-phase coupling which can help improve throughput.

Yes, a 200 Mbps adapter can outperform a 500 Mbps one. 500 Mbps adapters use an expanded frequency band (see http://www.smallnetbuilder.com/lanwan/lanwan-reviews/31624-homeplug-av-500-adapter-roundup). Higher frequencies don't travel as far as lower ones, so the higher speeds that 500 Mbps adapters produce are mostly obtained over short distances (in-room).

Before you give up, I'd try a pair of 200 Mbps adapters. Get them from Amazon and you have 30 days to try and return. I'd also look for small "wall-wart" style adapters plugged in anywhere in the same room. Cellphone / tablet chargers are notorious noise producers and will knock down powerline throughput.

 

[fizikz]

I decided to rearrange the network configuration, and moving one of the HomePlug adapters close to the electrical box improved throughput dramatically throughout the house. Previously, each adapter was quite far from the electrical box. A few quick iperf readings showed 40-60Mbps compared to the previous 5-10Mbps.

For the actual manual file transfer test, I got about 30Mbps, and the status light was green instead of amber or red. The link rate was reported as 150Mbps by the utility software.

As it turns out with this new configuration, the DSL modem (Thomson ST-516) needs to be plugged into the same outlet. This dropped performance significantly. Iperf (30Mbps); manual file transfer (23Mbps); utility software (50-75Mbps). The status light turned red with the modem plugged in. I will need to either find a filter, or plug it in elsewhere, or use an extension cord so it does not disturb the HomePlug. However, at least now the performance is much improved and much better than wireless.

For comparison, I also tested the wireless (wireless netbook -> wireless router -> wired network computer) using the manual file transfer and the best result I got was about 15Mbps when right next to the wireless router.

I wish I had read the article you linked to before my purchase. It could have been cheaper and perhaps more performing as well.

 

[thiggins]

Thanks for reporting back. See how a little work yields big benefits with powerline? And your throughput is going to be much more reliable than with wireless.

Try plugging the modem into a 6 ft or so simple extension cord. Or a cheap surge-protected outlet strip. That should provide enough filtering for the noise generated by its power supply.

Enjoy your new network.

 

[fizikz]

Thanks, a simple old extension cord did the trick. With the modem plugged into it, the file transfer rate is indistinguishable from the rate when the modem is not plugged in at all. The status light is green, but sometimes changes to amber and back. With the modem plugged into a surge protector, the rates are the same as with the extension cord, but the status light does not change. Either way the rates are the best I've managed so far.

One thing to note about the transfer rates I mentioned is that they were measured using an old netbook, which may have been a bottleneck. A newer netbook I tested had faster rates at least for the wireless.

Moving forward, I may be able to improve things even more by filtering out some wall-wart adapters in the room with the other HomePlug adapter.