 Mike2009
join:2009-01-13
Ottawa, ON
kudos:3
 ·TekSavvy DSL
1 edit | [DSL] 19 meg Profile with TP-Link 8840T Is there anyone on a 19 meg profile with one of the TP-Link Trendchip modems? Let me know what stats you have and how stable it's been. My downstream stats suck compared to two Broadcom modems that I've tested. On the 14 meg profile my stats were good with the TP-Link. I'll test it out and see how stable it is with these downstream stats.
Another modem I tested was the 2wire 2700 HG-B which also had better stats.
DSL Down Up
Current Rate: 19452 kbs 798 kbs
Max Rate: 22613 kbs 1268 kbs
Current Connection:
Current Noise Margin: 10.9 dB 20.0 dB
Current Attenuation: 8.6 dB 5.0 dB
Current Output Power: 11.0 dBm 11.9 dBm |
|
|
|
Mike2009
join:2009-01-13
Ottawa, ON
kudos:3 Reviews:
·TekSavvy DSL
1 edit | For anyone interested this is what my stats looked like on the 14 meg profile with the TP-Link. Lots of room. |
|
| reply to Mike2009
For people who doubted me when I said that some modems/tools reported higher theoretical speeds than the technical maximum for the spec, here is an example of this happening on ADSL2+... max theoretical speed in excess of 26Mbps on a standard that is technically limited to 24Mbps simply because the spec says so. |
|
Mike2009
join:2009-01-13
Ottawa, ON
kudos:3 | Very true. |
|
kovy
join:2009-03-26
kudos:8
1 edit | reply to InvalidError
said by InvalidError:For people who doubted me when I said that some modems/tools reported higher theoretical speeds than the technical maximum for the spec, here is an example of this happening on ADSL2+... max theoretical speed in excess of 26Mbps on a standard that is technically limited to 24Mbps simply because the spec says so.
It also depends on the SLAM. Old DSLAM, I have never seen a sync attainable rate higher then 8128... those slam only have ADSL... it doesn't matter what modem is in use.
In this case it does look like a bug reading from the tool or modem. Heck, I'd like to see what happens with DMT. |
|
| said by kovy:In this case it does look like a bug reading from the tool or modem. Heck, I'd like to see what happens with DMT. As I said in the original thread where I mentioned this, this is not a bug, it is a straight computation of maximum theoretical bit loading based on measured SNRM on a per-tone basis regardless of higher-level spec limits on the standard... the wire-level part of the spec could theoretically reach that speed if the MAC-layer specs allowed it.
This is a bit like if someone reported line speed based on the Shannon-Hartley theorem instead of any particular wire-level signaling protocol. |
|
kovy
join:2009-03-26
kudos:8 | said by InvalidError:said by kovy:In this case it does look like a bug reading from the tool or modem. Heck, I'd like to see what happens with DMT. As I said in the original thread where I mentioned this, this is not a bug, it is a straight computation of maximum theoretical bit loading based on measured SNRM on a per-tone basis regardless of higher-level spec limits on the standard... the wire-level part of the spec could theoretically reach that speed if the MAC-layer specs allowed it. This is a bit like if someone reported line speed based on the Shannon-Hartley theorem instead of any particular wire-level signaling protocol. This would require more testing to prove your argument.
Im willing to bet that on the old ADSL1 dslam, there's no higher rate then 8128... if you find one, please let me know. |
|
| said by kovy:Im willing to bet that on the old ADSL1 dslam, there's no higher rate then 8128... if you find one, please let me know. There has been people posting OrbMT screen shots on ADSL1 with theoretical rates over 12Mbps.
The DSL spectrum is 1.5MHz wide, the first ~400kHz are voice+upstream which leaves ~1MHz for downstream. This gets split into ~250 tones. Each tone operates at 4ksym/s and up to 14bits per symbol, which yields an absolute maximum theoretical wire rate of 14Mbps.
The 8Mbps "limit" on ADSL1 is not a wire-level signaling limit even within the ADSL1 PHY spec, it is an arbitrary limit at higher levels of the spec. |
|
kovy
join:2009-03-26
kudos:8 | said by InvalidError:said by kovy:Im willing to bet that on the old ADSL1 dslam, there's no higher rate then 8128... if you find one, please let me know. There has been people posting OrbMT screen shots on ADSL1 with theoretical rates over 12Mbps. The DSL spectrum is 1.5MHz wide, the first ~400kHz are voice+upstream which leaves ~1MHz for downstream. This gets split into ~250 tones. Each tone operates at 4ksym/s and up to 14bits per symbol, which yields an absolute maximum theoretical wire rate of 14Mbps. The 8Mbps "limit" on ADSL1 is not a wire-level signaling limit even within the ADSL1 PHY spec, it is an arbitrary limit at higher levels of the spec. I'm not talking about ADSL1 speed or profile, I'm talking about adsl1 SLAM only. They don't have higher rate then 8128... Like I said, prove it to me, and I'll be willing to bet that the OrbMT shoot were ADSL1/2/2+ SLAM on adsl1/2 profiles. If you find any... i'll upgrade your dsl profile if you have dsl. lol
For what its worth, the guy posted shoots with 2 different modem, yet both show similar yet very different stats... what are those reasons? |
|
| said by kovy:I'm not talking about ADSL1 speed or profile, I'm talking about adsl1 SLAM only. They don't have higher rate then 8128. So what? The point is that the ADSL1 PHY spec can technically go well beyond 8Mbps while the higher level parts of the spec do not allow it. DSLAMs and modems are besides the point.
Maximum theoretical != maximum attainable != maximum actually deliverable and stable. |
|
kovy
join:2009-03-26
kudos:8
1 edit | said by InvalidError:said by kovy:I'm not talking about ADSL1 speed or profile, I'm talking about adsl1 SLAM only. They don't have higher rate then 8128. So what? The point is that the ADSL1 PHY spec can technically go well beyond 8Mbps while the higher level parts of the spec do not allow it. DSLAMs and modems are besides the point. SLAM and modems are not beside the point... it's very much depend on how the attainable rate will go.
Look, it's fine and dandy to talk about the theoretical side... it's the concrete practical that counts here.
I guess we're not talking about the same thing... I taught we were talking about the attainable rate here... guess not. |
|
| said by kovy:I guess we're not talking about the same thing... I taught we were talking about the attainable rate here... guess not. Go re-read my first post in this thread... I clearly said I was writing about theoretical speed and have consistently stuck to the theoretical side, not attainable. Speeds beyond the standard's arbitrary limits are obviously not attainable, which is why I am explaining OrbMT's impossible "attainable" rate as a theoretical one.
Under real-world conditions, even "max attainable" rates when calculated correctly taking all standard limitations into account are rarely actually achievable... which makes them quite theoretical as well anyhow.
As for why two similar modems would report max theoretical rates ~10% apart, can you imagine how tight the phase and amplitude noise tolerances are at 14bits/symbol? I doubt a $25 modem is built to particularly tight tolerances so process variations between component and PCB batches could quite easily explain that. |
|
kovy
join:2009-03-26
kudos:8
1 edit | said by InvalidError:said by kovy:I guess we're not talking about the same thing... I taught we were talking about the attainable rate here... guess not. Go re-read my first post in this thread... I clearly said I was writing about theoretical speed and have consistently stuck to the theoretical side, not attainable. Speeds beyond the standard's arbitrary limits are obviously not attainable, which is why I am explaining OrbMT's impossible "attainable" rate as a theoretical one. Under real-world conditions, even "max attainable" rates when calculated correctly taking all standard limitations into account are rarely actually achievable... which makes them quite theoretical as well anyhow. As for why two similar modems would report max theoretical rates ~10% apart, can you imagine how tight the phase and amplitude noise tolerances are at 14bits/symbol? I doubt a $25 modem is built to particularly tight tolerances so process variations between component and PCB batches could quite easily explain that. I don't know if my eyes are playing tricks... but the first graphics looks like there's more bit loading/tones... yet has less attainable rate... it also seems to go as high as 15bit...
Speed going over beyond the arbitrary limit may be explained by other things... Like I said, why on a certain DSLAM it never surpasses 8128kbps? |
|
| reply to Mike2009
Did Bell not like your 96% upstream RCO and clamp you down to 797 when you upgraded? Or are you now on a remote?
I'm currently 797k too (66% RCO w/ broadcom) straight from CO. Annoying.
From what I read the new tplink (-T modems) trend chips don't boast much strength when SNR ratios nearing the default rock bottom of 6. I'd also like to see some screenshot comparisons. |
|
Mike2009
join:2009-01-13
Ottawa, ON
kudos:3 Reviews:
·TekSavvy DSL
1 edit | said by gslice:Did Bell not like your 96% upstream RCO and clamp you down to 797 when you upgraded? Or are you now on a remote?
I'm currently 797k too (66% RCO w/ broadcom) straight from CO. Annoying.
From what I read the new tplink (-T modems) trend chips don't boast much strength when SNR ratios nearing the default rock bottom of 6. I'd also like to see some screenshot comparisons.
I was on 14271/1085 and then upgraded to the the 16 meg plan and got 19455/797. I was told I risked having Bell decrease the download speed if I requested the full upload. I decided to leave it as is. On the 14 meg profile I was at 82% RCO on the upstream with any of my Broadcom modems. The 96% was with the TP_link which was stable nevertheless. |
|
| reply to kovy
said by kovy:Like I said, why on a certain DSLAM it never surpasses 8128kbps? To achieve the 14bit/symbol level, you need SNR in excess of 80dB... the noise from the DSLAM's DACs alone can be enough to make it impossible if the DSLAM's design was never intended to go that far. Since higher-level limitations make the higher speeds that would require 14bits/symbol impossible, there is no reason for ADSL1 DSLAMs to be designed for it, which leaves plenty of room for manufacturers to cut corners. |
|
 dillyhammerSTART me upPremium,MVM
join:2010-01-09
Scarborough, ON
kudos:9
·Start Communicat..
 ·Cogeco Cable
·TekSavvy DSL
·Caneris
| reply to Mike2009
said by Mike2009:I was on 14271/1085 and then upgraded to the the 16 meg plan and got 19455/797.
Damn.
I was on 14272/797, requested full upload and got 14272/1085. Now I've requested a bump to the 16meg service. I wasn't warned I could have my upload bumped back to 797. That'd be a fscker. The 797 is pathetically low. Hell it was low in 2002. In 2012 it's just bloody embarassing.
Mike
--
Cogeco - The New UBB Devil -»[Burloak] Usage Based Billing Nightmare
Make The Switch - »openmedia.ca/switch |
|
Mike2009
join:2009-01-13
Ottawa, ON
kudos:3 | They'll likely give you the full upload but it's hit and miss if they give you the 16 meg or 19 meg sync. I can live with the lower upload. |
|
| reply to dillyhammer
said by dillyhammer:The 797 is pathetically low. Hell it was low in 2002. In 2012 it's just bloody embarassing. 1Mbps does not really enable you to do anything you couldn't already do at 800kbps so I would not call it that much of an improvement... but that is as fast as standard ADSL1/2+ goes, can't do much about it. |
|
| said by InvalidError:said by dillyhammer:The 797 is pathetically low. Hell it was low in 2002. In 2012 it's just bloody embarassing. 1Mbps does not really enable you to do anything you couldn't already do at 800kbps so I would not call it that much of an improvement... but that is as fast as standard ADSL1/2+ goes, can't do much about it. It helps when local-hosting a minecraft server. That'd be one extra friend that can connect without causing lag for everyone else.
I'm on 640kbit up - I'd be happy with 800kbit or 1.1mbit. Percentage wise that's a huge increase! |
|
