Members of the HomeGrid Forum (HGF) and its president Donna Yasay have directly challenged the conclusions we came to after we finished the tests of powerline adapters that have HomePlug AV2 chips and G.hn chips.
The dispute is with the very first tests we conducted where we found that in the master bedroom the speeds in Mbps were:
|Comtrend’s G.hn||D-Link’s AV2|
|Test 1 Test 2||Test 1 Test 2|
|15.3 15.0||No connection|
|7.4 7.4||No connection|
|Total 22.7 22.4||No connection|
We reported that one network technology showed what we considered “poor” results and that the other did not connect at all. We did not identify which was which.
We did not want to embarrass either technology because, in fact, we subsequently found there were two major sources of interference: A broken ground fault interrupt outlet in a bathroom that we immediately replaced and an ancient Radio Shack powerline-to-phoneline adapter that we had used over a decade ago to connect a DirecTV STB to DirecTV for the purpose of downloading updated TV guides.
We immediately unplugged the RadioShack adapter and retested. The results were the ones that we reported in last week’s issue.
Secondly, having dealt so much with broadband networks that have speeds that are pre-set separately for up and down, we presumed the same was true for home networking. For example, a broadband speed that is expressed as 100 Mbps down and 10 Mbps up is exactly that. It is not 110 Mbps. When faster uploads speeds are needed, such as for uploading videos or medical images, it is still only 10 Mbps up.
You can’t “borrow” some of the download speeds.
However, HGF said, “Not so fast.”
Without being told, HGF was sure that it was the G.hn adapters that made the connection, albeit slow, in the master bedroom and not AV2 because that had been its experience in hundreds of similar tests.
HGF said that G.hn has access to the total available bandwidth for up- or down-loading. With G.hn, the 22 Mbps total it can access is more than sufficient for streaming one, two or possibly three HD videos from Netflix and other OTT services.
The end result is that a customer who had purchased the adapters would have been satisfied with and kept the G.hn adapters but, having seen no connection being made with the HomePlug AV2 adapters, would have returned them to the retailer.
Few consumers would have done what we did — hire a technician to come troubleshoot and find and fix the problems.
On an even larger scale, a pay TV service, instead of shipping customer-installable adapters, would have been forced to send a technician into the home to troubleshoot and complete the installation. That could end up creating ill-will and costing it billions.
The G.hn adapters worked where AV2 adapters did not — and at that location they provided speeds that would have enabled 1080p HD videos to be streamed flicker-free.
Yasay said, “G.hn produced 20 Mbps speed on the ‘connected’ line that was not conditioned. Most HD video is at 6-8 Mbps, which incidentally is what telcos’ use in their IPTV pay TV technology that is based on [Ericsson’s] Mediaroom. Netflix HD is at 5 Mbps, while the recommended 4K HD is at 25 Mbps. So the 20 Mbps line that connected, assuming it was G.hn, would have been able to receive three HD streams from AT&T U-verse for example.”
Netflix’s recommended speeds are at its Web site
3.0 Mbps Recommended for SD quality
5.0 Mbps Recommended for HD quality
25 Mbps Recommended for Ultra HD [4K] quality
Yasay also cited a more detailed test that Small Net Builder conducted with equipment that costs thousands of dollars.
She said the highlights of the Small Net Builder tests are:
– Even though G.hn is at 50Mhz and HPAV2 is at 86Mhz, which physics shows is at a higher throughput, we [G.hn] still beat out HPAV2 at mid to longer loop testing, 148 Mbps.
– (page 8): “the Comtrend adapter does quite well in the mid and far-range Location C test and turning in its best comparative performance in worst-case Location E with 148Mbps. This is actually the best throughput we’ve ever measured in Location E for any power line adapter tested!”
– (page 20): “G.hn looks focused mainly on service providers, much as MoCA is.”
Sigma Designs adapter used in test – front and rear
Another G.hn backer said, “You mentioned that G.hn could connect where the HPAV2 could not, but at ‘almost’ unusable speeds. I read in the article that you said ‘about 20 Mbps.’ Since any but the most demanding 4K video would be under that, I think it could be considered very useful. A standard definition OTA (over an antenna) or AT&T U-verse stream is only about 4 Mbps and an HD stream is between 6 and 12 Mbps depending on the content supplier. The most demanding HD stream is actually ATSC at between 8 and 18 Mbps, depending on the action in the video. This would leave another 2 Mbps for best effort data.”
Developers of the specifications for G.hn and its chip and equipment builders are not sitting still.
They say they are working on improving G.hn’s performance with changes that are within G.hn specifications and also changes to the circuit board on which the G.hn chip resides.
Newer firmware, it said, will be made available to increase the performance of existing G.hn gear.
Referring to the tests that Small Net Builder conducted, a G.hn backer who has been involved in many tests of powerline network technologies said:
“Here is an in-depth comparison of power line that is not skewed by a limited test plan [referring to the one The Online Reporter conducted although by and large both tests came to the same conclusions]. This review does not stop at the easy loops in easy conditions. It goes beyond and tests in the real world with real noises and setups. In the end, it shows G.hn does perform much better on difficult loops but it still doesn’t complete the picture on error rate and QoS.”
One person with absolutely no vested interests in either powerline technology said, “This sounds like differing definitions of conditions supported. In a clean environment one performs better than the other. In a ‘dirty’ environment the other performs better. So I don’t think the test or assumption was wrong, but rather focused on a specific condition. Now you are considering two conditions.”
So, our initial findings, although unpublished, were correct but our conclusions about G.hn were wrong.
Why Does It Matter?
Here are some of our assumptions:
– Consumer decides to buy a 4K-capable UHD TV because of their picture quality, price and promotional campaigns.
– Consumer finds that most 4K content is only available online.
– Customer is already having problems with Wi-Fi so decides to buy a pair of powerline adapters rather than install Ethernet cables.
– Customer installs the powerline adapters, connects the UHD set and finds either a) 4K streams are flicker free and keeps the adapters or b) they aren’t flicker free and returns the now used adapters to the store. His alternatives then are a) install MoCA’s coax adapters, which are horribly under-promoted, b) install Ethernet cables or c) try a new model of a Wi-Fi router or buy a Wi-Fi extender. In no case will he try another make of powerline adapter.
Does the world really need two incompatible powerline network technologies?
Or would that be as confusing as having two incompatible Wi-Fi technologies? Or having two incompatible cell phone networks? Or having some cars with the steering wheel on the left and some with it on the right?
The sensible thing seems to be to have only one powerline network technology — so which will it be?
HomePlug is long-established, universally available, offers more models, has a much larger installed base and the AV2 version produced the best results in our tests.
G.hn says that ultimately its technology will produce better results in more situations, particularly difficult ones for AV2. G.hn backers also point out its capability to be used over coax and telephone wires.
One telco, Korea’s KT, is using G.hn in place of G.fast to provide high-speed broadband over existing telephone wires. As reported last month, Spain’s Jazztel uses Comtrend’s G.hn Powerline Ethernet Adapter with electrical pass-through sockets to allow Jazztel’s subscribers to extend a local area network via existing power lines, eliminating the need for extra wiring between FTTH (the ONT) the home gateway.
HomePlug’s main focus has been on the retail market. G.hn seems to be focused on selling its powerline networking to service providers, especially to telcos — cablecos having largely standardized on the coax-based MoCA. In that case, G.hn’s claim to provide access to every electrical outlet provides a meaningful benefit to service providers in terms of reducing the number of service calls.
Who will win? HomePlug offers higher speeds now, although not necessarily at every outlet. G.hn promises access at every outlet and that it has higher speeds coming in future versions. HomePlug is not talking about future versions but surely HomePlug backers are not sitting still, especially the two big ones — the chipmaking giants Broadcom and Qualcomm.
The answer is: We don’t know.
What about 4K?
The fact remains that in our tests AV2 outperformed …
For the complete article and latest edition, please write email@example.com or click here to register for a four week free trial