01-16-2017 9:44 AM
Hi, I have had some very slow DSL speeds since installation, but much worse now than initial installation. I pay for the 40 Mbps service, but speedtest sites show me at about 15 Mbps. My upload seems fine, typically at around 6-8 Mbps which is good and what I pay for.
I am a relatively internet/computer savvy user. I know the connection is not being used when I run the speedtests. I also get the same speeds when connected to the router via ethernet. The routergateway page does show the full sync that I'm paying for at around 40Mbps.
The initial installation the engineer struggled to get 40Mbps. It kept going to 32-34 Mbps. He changed the router installed three times and the end I just gave up and accepted it as he had been there 5 hours. However recently it has dropped to worse and worse speeds.
I called an engineer out recently and the engineer claimed the slow speeds were caused by the powerline extender I have attached to the router. This was connecting in bridge mode to another router in the back bedroom to extend my network. That router is not acting as a DHCP server, just a wireless briedge, it creates a new network in that space with a different SSID. However even unconnecting this didn't cause the speed to go up when the engineer was there. Unfortunately I got into an argument with the engineer that the powerline was not the cause and I ended up getting charged $100 for the engineer visit without getting the problem fixed.
I would like to understand the best way forward to identify what the issue is. When I call AT&T their frontline techs don't seem very clued up, I seem to know more than them about networking. Likewise with the engineers they send out. I am very cautious of calling an engineer after getting charged $100 last time around.
Can someone give me step by step advice on any self-diagnostic actions I should follow and then the best way to contact AT&T to get a resolution?
01-16-2017 2:23 PM
May I recommend that you disable IPv6 on your AT&T provided router/gateway as having that enabled can impact speed.
What I would recommend is to minimize your home network and see if the symptom persists. A good example is what you are experiencing, apparent slow connection. Prior to engaging AT&T use an Ethernet cable and plug into a port on the back of the AT&T provided router/gateway and to the device. Then run the AT&T speed test. Should the speed test provide speeds for which you are paying then walk the device further into your home network to determine what is impacting speed.
Don't expect AT&T to fix your home network issues. However, if the speed is not what you are expecting when you plug into the AT&T provided router/gateway, then AT&T should be able to resolve.
- edited 01-16-2017 6:46 PM
As I said I am a savvy user. This is connected via ethernet. It is not a home network issue.
I have seen other users posting from UV Realtime. Please find attached two screeshots. I see UV seems to be telling me I should be using the 32Mbps profile. I am paying for 40Mbps.
But I would rather have 32Mbps if that will give me a stable connection.
I am also seeing occasional high FEC rates downstream. 1,507.3.
I disabled IPv6. I can't detect any difference.
- edited 01-16-2017 6:51 PM
UVRealtime was never tested by its author on a 5268ac and doesn't know about profiles faster than 32 Mbps, so I'd take its recommendation of 32 Mbps with a grain of salt. A max rate of 86Mbps is fine for a profile rate of 55 Mbps and a HSI tier rate of 45 Mbps. That noise ratio is adequate. I might confirm your FEC/CRC rates by going to the gateway directly and make sure that they are as reported (about 0 per time unit used by UVrealtime).
I agree with the others that I would conduct a test with only one PC connected via Ethernet to the Gateway, with no other connections (Wi-Fi, HPNA, Ethernet, Powerline adapters, etc.) and see what happens. That way we'll all be confident that it is indeed the connection and not anything in your home (other than maybe the PC doing the testing ).
Will UVrealtime give you a Bitloading graph on a 5268ac? If so, post it. There is one potential issue that doesn't always show up except in bitloading graphs.
01-16-2017 7:04 PM
Hi, I can't get a bitloading graph. I tried as you said and unplugged all the other ethernet ports. I turned off the wifi interface on the router. I got 50 Mbps speeds straight off.
I enabled the wifi interface on the router. Still plugged in via Ethernet and with Wifi turned off on my laptop I got 45 Mbps.
I unplugged the ethernet cable. I turned on Wifi on my laptop. I sat two feet from the router and got 22 Mbps. How can the wifi performance be that poor just two feet away?
01-16-2017 8:06 PM
Were you connecting via 802.11g, 802.11n, or 802.11ac?
2.4 GHz or 5.0 GHz?
What does Windows report the connection rate as (Task Manager or Network Properties)?
01-18-2017 7:08 PM
The 20Mbps speed (I get 10Mbps 15 foot away with a clear line of sight to the router), 20 Mbps right next to it is 802.11n/b/g 2.4 GHz (the default setting).
I have a 5GHz network with a separate SSID. 802.11n/ac. When I connect to that network I get 45Mbps. However I don't get such a good signal from all parts of the house.
01-18-2017 7:26 PM
Okay, that makes sense. 802.11g on 2.4 MHz is practically limited (not theoretically, that number is higher, but in actual real-world use) to about 22 Mbps. 802.11n on 5.0 MHz and/or 802.11ac on 5.0 MHz can use more channels and can get much better thoroughput. But yes, 5.0 GHz has more limited range because the higher frequencies don't carry or penetrate as well.