Bicephale
join:2005-09-24 | reply to dslrocker3
Re: Help/Advice with my Line Statistics please
Well, all you've got is a couple static numbers so i
guess Bell can pretend almost anything they want.
Nothing is known about noise performance, isn't it? |
|
dslrocker3
join:2002-05-26
Toronto
1 edit | reply to dslrocker3
Funny how Velcom complained to managment and an hour later, I at least got the speed back before that I had before the problems started. At 2496/800, the noise margin flutuates between 9.5 and 11.0 dB. They left the interleaving on though but that's not that big of a deal to me.
Noise margin doesn't tell you the occupancy but I would think that it can give you idea of what percentage you are at (ie. 6.0 dB pretty much being 95-100%). Between 9.5 and 11.0 dB, I really doubt that the my RCO is 90+% like Bell claimed it would be.
Now my only problem is what to do with my ST516 (sell or keep because it is valuable for troubleshooting). I know that will probably barely be able to do the 2496 and will probably be around 6 dB with the ST516. I don't want to plug the Speedtouch in yet though because Bell might see the higher RCO. |
|
Bicephale
join:2005-09-24
 ·TekSavvy Solutions..
| Hi,
I don't concur about the value of SNR Margins but i do
share your concern about which unit should be on-line
while Bell looks around. Perhaps DeadPool can provide
some advice here - i've hoped that he might, actually.
 |
|
dslrocker3
join:2002-05-26
Toronto
1 edit | said by Bicephale  :Hi, I don't concur about the value of SNR Margins but i do share your concern about which unit should be on-line while Bell looks around. Perhaps DeadPool can provide some advice here - i've hoped that he might, actually. I'm not saying that you can always go by the noise margin but in my case, I know that the Speedstream has consistently been able to connect at a higher download speed than the Speedtouch (when I was having problems Speedtouch would connect at 1856 Kbps but the Speedstream would connect at 1984 Kbps both reporting noise margins of about 6 to 6.5).
When I was at 1728 and after the connection had improved again, the noise margin was often at 14 for the Speedtouch with a RCO of 62% but the SNR went as high as 18dB in a few instances with the Speedsream. Since the the Speedstream has been consistently been able to connect at a higher attainable rate when I couldn't achieve the full profile speed and always reported higher SNR values than the Speedtouch, I am speculating that the RCO on the Speedstream was lower than the 62% reported by the ST516 at 1728Kbps. |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| I've taken a quick look at the RCO records i published and my ST516
won the contest against my ST546 which won against my SS4200...
On the other hand, i know from memory that the SS4200 still worked
with somewhat lower SNR Margins which made any of those STs fail.
But... The SS couldn't be tweaked while the STs could be forced to
stay connected long after that SS got overwhelmed by strong noise.
...
 |
|
dslrocker3
join:2002-05-26
Toronto
| reply to dslrocker3
It's a good thing that I had my Speedstream 5200 hooked up and not the Speedtouch 516 when Bell did the test to change the profile. Velcom tells me that I am at 87% occupancy with the Speedstream but when I hooked up the Speedtouch modem, it was only able to connect untweaked at 2432 Kbps. I have also noticed that the Speedtouch 516 does something odd when showing RCO on an interleaved line. It will never connect using more than 90-92% of the downstream part of the line. This restriction doesn't seem to be there with a fast path lilne. I beleive that ST516 miscalculates the RCO on an interleaved line by not factoring in the extra overhead for error correction.
By the way, isn't it possible that turning interleaved mode on actually drives up the occupancy of the line up?
Based on the way the old ST Home calculates occupancy, it was actually calculated based on the amount of line rate used up, rather than the ATM rate. Since error correction uses some of the line, more of the line would have to be used to connect at any given speed. I'm just wondering out of curiosity because it seems like to me fast path seems to generate a lot less errors than the number of corrected errors in interleaved mode on the same line. I'm just wondering if extra overhead brings the occupancy up driving the error rate up (but correctable). |
|
planiwa
join:2009-02-19
Toronto M5S
| That's a very interesting observation.
So, perhaps Interleaved reduces capacity as it increases SNRM, assuming low SNR is due to short noise bursts rather than weak signal or steady noise?
»www.kitz.co.uk/adsl/interleaving.htm |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| Hummm...
I did document my transition from 5056/800 Kbps FastPath to
6016/512 Kbps Interleaved with some numbers, see what i got:
%20.PNG/thumb.jpg)
Tweaks, Bicephale, 2007-Oct-23
%20.PNG/thumb.jpg)
Tweaks, Bicephale, 2007-Nov-30
That's not much, really, but it seems my RCO improved despite
the higher DownStream rate (thanks to the lower UpStream one,
we might say). If we can called this Bell's roughly approximated
version of what's tagged "Spectral Shaping" then that's why i've
been so curious about the D-Link DSL-2320B/2640B units lately:
»Re: DMT Tool for D-LINK DSL 2320B
»Re: Velcom Average Speed?
Those captures were less conclusive about SNR Margins, though!
 |
|
dslrocker3
join:2002-05-26
Toronto
3 edits | reply to planiwa
said by planiwa :That's a very interesting observation. So, perhaps Interleaved reduces capacity as it increases SNRM, assuming low SNR is due to short noise bursts rather than weak signal or steady noise? » www.kitz.co.uk/adsl/interleaving.htm According to that website:
"It should also be pointed out that whilst BTw state that applying interleaving shouldn't reduce your line speed, it does reduce the maximum line rate achievable from 8128kbps to 7616kbps due to the additional overhead required for check bytes."
So is it not possible that the interleaving is what can actually cause more errors due to the fact that it could end up pushing you the the edge of the usable signal?
For example, I am at 87% on 2496 interleaved but there are errors on the line. Is it not reasonable to assue that the error correction is what can actually cause some of the errors in the first place (due to more of the line being used)? |
|
Bicephale
join:2005-09-24 | reply to Bicephale
Oups! But i've just noticed that this was with "tweaking" in action...
 |
|
Bicephale
join:2005-09-24 | reply to dslrocker3
Very relevant questioning but i can think of only one way to find out!
|
|
dslrocker3
join:2002-05-26
Toronto
3 edits | said by Bicephale :Very relevant questioning but i can think of only one way to find out! I have had my line set to interleaved before and to fast path before. I can't say for sure because my 2496/800 fast path experience was a week ago but there was never more than 100 or 200 (in multiples of 68) for FEC in a 15 minute period. The CRC errors were sometimes at rate of about 20 per 15 minutes.
Now on 2496/800 interleaved, I sometimes get FEC values of 1000 per 15 minutes (but less retransmission due to uncorrectable errors).
I can't say for sure because this is spread out over 2 different weeks but I'm pretty sure that the extra overhead in interleaving is playing a part in the increased error rate.
Either way, if it's like RAM error correction (1 bit for error correction for every 8 bits of data), that would add another ~12% to your line rate usage.
While there is no doubt that interleaving can increase reliability, I have heard some people say that interleaving will improve your noise margin. To me, that part doesn't make sense because the extra error correction overhead would force your modem to use "dirtier" parts of the signal. |
|
dslrocker3
join:2002-05-26
Toronto
2 edits | reply to Bicephale
said by Bicephale :Hummm... I did document my transition from 5056/800 Kbps FastPath to 6016/512 Kbps Interleaved with some numbers, see what i got: I have a feeling that 3008 down would have been fine if they tried 512 or 640 Kbps up. But I'll never know because they never tried. I actually find it kind of strange that they tried 2496/800 instead of first playing with the 640Kbps upload speed since it was already set to 640.
just a thought: I have seen modems where you can use software to limit the downstream speed but is there anything out that let's you limit the upstream speed? It would be interesting to be able to easily compared attainable downstream rates vs the current upstream rate. |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| If this had depended on me i would have given him 4032/512 Kbps
Interleaved knowing he can tweak using his Thomson SpeedTouch!
I've had quite a few profiles myself but i didn't post 'DMT' captures
in the begining so those numbers got flooded in a see of characters
and searching for them would cause excrutiating pain to my optical
nerves! I won't be able to find them and i need to rest a bit anyway.
All i can add is: Good luck! |
|
Bicephale
join:2005-09-24 | reply to dslrocker3
Euh...
"Limit the upstream speed"?
The last sample posted before your eyes was taken while doing just that!...
 |
|
dslrocker3
join:2002-05-26
Toronto
| said by Bicephale :Euh... "Limit the upstream speed"? The last sample posted before your eyes was taken while doing just that!... oh... I didn't notice that part. but for me it really doesn't matter because Bell would have you believe that I'm comitting murder by asking for profile adjustments. |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| Well, the title of that thread was "Tweaks", right?!
I can only advise guys with so little luck to gather
whatever proof they need to support a request as
i've done myself... I must admit i wasn't in such a
bad situation but i bet an intelligent phone worker
can see when the customer's assumption is based
on detailed evidence. Show you can deal with the
HardWare and one of them might start listening.
Hummm...
But one would need to have an opportunity to talk
to Bell's people 1st - i did get such a chance and i
used it to the best of my abilities: i was prepared!
So, get prepared too!!! The trouble-shooting tools
have never been so good!... |
|
dslrocker3
join:2002-05-26
Toronto
2 edits | reply to dslrocker3
something strange going on with my line today...
-was connecting at 2496/800 rx snr at 9.5 or 10 settling off at 9.0 to 9.5 after I start using it
-rx snr spikes to 12.0 dB out of nowhere but starts bouncing all over the place... down to 0.5 dB, back up to 12 dB, back down to 8dB, etc... showing characteristics of 3 or 4 different lines when it is the same line, almost like someone is connecting and reconnecting stuff... I didn't open another ticket and was not informed of any pending dispatch.... Also, attenuation has gone up a dB or 2 since I noticed the Bell van sitting there.
I look outside and there's a Bell van sitting in front of my next door neighboor's house. Coincidence? I am wondering if something on my neighbor's line is what has been causing my problems. I never actually lost sync (even when i went down to snr 0.5dB).
maybe I can quietly ask the technician if i bump into him/her |
|
Bicephale
join:2005-09-24 | As i wrote, the trouble-shooting tools have never been so good... |
|
%20.PNG)
%20.PNG)