VDSL Performance & Tuning

Forums › New Zealand Broadband › VDSL Performance & Tuning

1 | ... | 94 | 95 | 96 | 97 | 98 | 99 | 100 | 101 | 102 | 103 | 104 | ... | 285

40 posts

Geek

#830823 5-Jun-2013 09:18

The good news: a truckload of rain had no effect on CRC errors. :D

The bad news: I can't get SNRM tweaking to do anything. It continues to aim for 12dB SNRM despite best efforts.

So, has anyone been able to get downstream SNRM tweaking working on the newer firmware? I saw a mention by quakeguy back on page 77 of someone they know that upgraded their firmware and lost their SNRM tweak.

I tried applying the tweak as documented on page #1, adding "NoChecks=yes", modifying "DownstreamMarginOffset=-50", uploading the config file to the Fritz and rebooting it. I get the "changes not supported by manufacturer" and "[Modified Settings]" notices on the home and DSL settings pages respectively. However, downstream SNRM is still at 12dB, and "Request Extended Port Status" reports:

Downstream Training Margin: 12.0 dB

Have I missed something, or is it time to locate and install and older firmware? (Perhaps tomorrow to avoid upsetting the DLM-1 gods...)

SamF

1578 posts

Uber Geek

Trusted

#830945 5-Jun-2013 13:39

@seatil, which modem do you have & which firmware did you upgrade from / to?

seatil

40 posts

Geek

#830946 5-Jun-2013 13:42

It's a Fritz!Box 7390 and came with 84.05.51 on it. I haven't yet changed the firmware, just followed the SNRM tweak instructions.

SamF

1578 posts

Uber Geek

Trusted

#830968 5-Jun-2013 14:03

Hmm, yes that does correlate to the post on page 77. You could try PMing the user who posted it and see if he has any further info on this. I'd be interested in finding out what the deal is with that!

Has anyone upgraded a 7340 to v99.05.51 yet?

I did a bit of Googling and found this:

http://forums.whirlpool.net.au/forum-replies.cfm?t=1991683&p=35&#r684

sidefx

3711 posts

Uber Geek

Trusted

#830981 5-Jun-2013 14:27

@SamF: Love that graph, really cool man, nice to see it graphically like that. :)

(I wonder if it would be more meanful across a range of "base" sync rates with a percentage increase across the bottom axis instead of straight Mbits?)

"I was born not knowing and have had only a little time to change that here and there." | Octopus Energy | Sharesies
- Richard Feynman

SamF

1578 posts

Uber Geek

Trusted

#830982 5-Jun-2013 14:31

Hehe, cheers! :)

What do you mean exactly re base sync rates?

sidefx

3711 posts

Uber Geek

Trusted

#830983 5-Jun-2013 14:40

So base sync rate = sync rate at 12 db Target SNRM and then along the X axis show the percentage increase compared to that base sync rate. The rationale being that if your base sync rate is 20Mb and you increase through SNR tweaking by 5Mb (20% increase) that's different to someone with a base sync of 50Mb increasing by 5Mb?

"I was born not knowing and have had only a little time to change that here and there." | Octopus Energy | Sharesies
- Richard Feynman

Equinox IT

Cloud spending continues to surge globally, but most organisations haven’t made the changes necessary to maximise the value and cost-efficiency benefits of their cloud investments. Download the whitepaper From Overspend to Advantage now.

SamF

1578 posts

Uber Geek

Trusted

#831115 5-Jun-2013 16:29

Oh yeah, I can do that! Will whip it up tonight.

sidefx

3711 posts

Uber Geek

Trusted

#831154 5-Jun-2013 17:37

Well, regarding my massive error rates, they've replaced the failing bulb and there hasn't been a single error since the lights came on today. Touch wood it looks like that was the cause; I'll check again in the morning and if there aren't any errors ring snap to reset my DLM.

Moral of the story: If you get a massive increase in errors during the night (I was hitting 10000+ per hour) and barely any during the day, look at your street lighting for flickering.

"I was born not knowing and have had only a little time to change that here and there." | Octopus Energy | Sharesies
- Richard Feynman

SamF

1578 posts

Uber Geek

Trusted

#831279 5-Jun-2013 20:54

Good news on the errors!

Here's some graphs I created based on your suggestion, plus another couple I thought of:

Obviously the data set is a little on the small side, but there's some interesting trends here:
1) 'x'dB of SNRM decrease yields almost exactly the same amount of sync speed increase in megabits between our lines
2) 'x'Mbit of sync increase on your line yields far fewer errors than on my line. But this makes sense as you're running at a lower speed than me.

stevehodge

393 posts

Ultimate Geek

#831309 5-Jun-2013 21:30

SamF:
Here's some graphs I created based on your suggestion, plus another couple I thought of:

Obviously the data set is a little on the small side, but there's some interesting trends here:
1) 'x'dB of SNRM decrease yields almost exactly the same amount of sync speed increase in megabits between our lines
2) 'x'Mbit of sync increase on your line yields far fewer errors than on my line. But this makes sense as you're running at a lower speed than me.

That's really interesting. In hindsight it's obvious that SNRM decrease vs sync rate increase is so similar and so linear. 1db of SNR represents a certain number of bits of signal that can be sent and this is close to a fixed relationship. It'd be great to get some more examples, but so far I'm pretty confident saying that 1db of SNRM reduction will give you a bit over 1 Mbit/s of sync speed.

stevehodge

393 posts

Ultimate Geek

#831322 5-Jun-2013 21:38

SamF:
Here's some graphs I created based on your suggestion, plus another couple I thought of:

Can you do one more? Absolute SNRM vs CRC/min. Or perhaps SNRM vs CRC/min % increase. I'm thinking that reduced margin should result in a fairly predictable increase in errors - a db of buffer should stop a failry constant number (or %) of errors.

SamF

1578 posts

Uber Geek

Trusted

#831339 5-Jun-2013 22:08

Here ya go:

I did actual CRCPM vs % as both sideFX and I were close to 0 CRCPM @ 12db SNRM so that graph would be virtually identical to this one.

As you suspected the trends are pretty close for this. This graph is probably a better indicator of expected errors than my original CRCPM vs Sync Speed Mbit Increase one.

SamF

1578 posts

Uber Geek

Trusted

#831351 5-Jun-2013 22:10

I did this one out of interest as well:

The trends are almost identical!

insane

3236 posts

Uber Geek

ID Verified

Trusted

#831868 6-Jun-2013 21:38

Got another line test done to see what the line looks like after finding a better socket in the house to use which is much closer the the demark point.

Have gained tones in the upload frequencies, but lost a little at the higher end. Really need to get my wiring properly sorted as the Cabinet is literally outside my neighbors driveway.

Before changing phone socket:

Sync 55/7mbps

After changing phone socket:

Sync: 49 / 10 mbps

1 | ... | 94 | 95 | 96 | 97 | 98 | 99 | 100 | 101 | 102 | 103 | 104 | ... | 285