Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[konrado5]

Could you estimate DMT tool stats?

Best regards
konrado5

[kitz]

I dont know what you mean.  DMT just isnt picking up your line rate?
AFAIK DMTtool hasnt been updated in a long time and may not work properly with some of the newer routers.


But looking at what I can see and if I didnt know anything youd previously told us about your line I would say the following.

The graphs look very good and normal aside from the gaps in tone 415-473.   


I dont know your actual uptime, but from the error counters, your line is generating some serious alarm states such as:- ,

Based on AS of 473282 =  (5 day 11hr)
ES of 23783 means that for 6hrs 35 mins your line has been in an errorred state.
UAS of 821 = 13 mins 41 when your SNRm has been too low to even have a path to the exchange.

The above implies that a 2dB target margin is too low.


Whilst you may have run some BERT tests in some other threads which showed good at the time.  BERT can only show a picture for that moment in time.  Your stats prove that there have been some periods when although your line may have remained in sync, and your bit rate errors are pretty low, there have been occasions where all is not well.


Im not going to say anything about the HECs turning into OCD

[konrado5]

ES of 23783 means that for 6hrs 35 mins your line has been in an errorred state.

Errored seconds are calculated froum 20 days (system uptime, connection uptime is 5 days).

What do you think about SNR per tone cumulations graph?

What do you think about Hlog undulations which was noticed by burakkucat? I repeat: burakkucat has noticed something else besides 415-474 gap.

[boost]

I don't understand your question either, sorry
Does that tool work with the HG612? If so, I will try it here for you if that's any help?

As an aside, having just 1 ES would irk me; it's a fault condition

It's a bit like overclocking your CPU to the max and telling yourself the occasional BSOD is 'just one of those things' but if it works for you...

[kitz]

What do you think about SNR per tone cumulations graph?

Reply =

The graphs look very good and normal aside from the gaps in tone 415-473. 

In slightly more detail, theres an odd bump at around tone 50, but since all tones up to at least tone 60 are affected very heavily by PSD masking then Im not giving it too much attention.

You have a very slight dip at around tone 200, I personally call this type of dip a bowl shaped dip and its something seen on many lines whereby recovery is in the region of tone 250.  This is perfectly normal, its the effects of crosstalk between adsl1 and adsl2/2+ lines.  In fact you seem to have got away pretty lightly compared to some.

All in all it looks damn good to me. 

What do you think about Hlog undulations which was noticed by burakkucat?

Ive said before Ive nothing more to contribute on this subject.   Theres a big difference between unusual and poor.   Your hlog isnt poor.  It just has something unusual on it...  but heres the twist, an awful lot of lines may show something unusual for which there just isnt an explanation.   In reality its a damn miricle that adsl works as well as it does.    Yes there may be some odd quirks about your line, but Im really not sure what you are looking for us to say.

[kitz]

As an aside, having just 1 ES would irk me; it's a fault condition

It's a bit like overclocking your CPU to the max and telling yourself the occasional BSOD is 'just one of those things' but if it works for you...

I and others have said many a time that running at 2dB isnt a good idea.   That line is faulting purely because the target SNRm is set too low.
Its a good job Konrado's ISP doesnt have a DLM or he'd be clobbered.  The target margin is there for a reason - to keep errors and faults down 
Yep the headline sync will look good, but whilst a line is producing errors, then the real-time throughput will be rubbish.

[loonylion]

As an aside, having just 1 ES would irk me; it's a fault condition

It's a bit like overclocking your CPU to the max and telling yourself the occasional BSOD is 'just one of those things' but if it works for you...

I and others have said many a time that running at 2dB isnt a good idea.   That line is faulting purely because the target SNRm is set too low.
Its a good job Konrado's ISP doesnt have a DLM or he'd be clobbered.  The target margin is there for a reason - to keep errors and faults down 
Yep the headline sync will look good, but whilst a line is producing errors, then the real-time throughput will be rubbish.

There is some variance though, my ADSL2+ connection ran perfectly at 1dB using G.DMT and gave me 5mbps (for some reason running ADSL2+ modulation gave me a lower sync).

[konrado5]

Yep the headline sync will look good, but whilst a line is producing errors, then the real-time throughput will be rubbish.

I don't see any real problems. I attach sample CRC errors graph.

[kitz]

From a glance that shows an average of 2 CRCs a min :

Ive only about a couple of hours ago switched over to a new version of DSLstats and I dont keep old CRC error logs....  but looking at mine since 21:51..  Ive had a grand total of just 1 crc @ 22:40.    On adsl2+ I had a big fat zero all the time unless there was something like a thunderstorm.

[konrado5]

On SNR margin 6 dB I have average 50-100 CRCs per 12 hours. However I don't see transfer rate loss on average 2 CRCs per min.

On adsl2+ I had a big fat zero all the time unless there was something like a thunderstorm.

On fast path without FEC?
Can thunderstorm cause CRC errors if line is provisioned underground?

[kitz]

On fast path without FEC?

Yes

Can thunderstorm cause CRC errors if line is provisioned underground?

Yes.  Most definitely.  FYI mine is UG.

[konrado5]

I suspect my CRC errors are caused almost only by phone line provisioned nearly the mains cables. For example this pink round box is electrical box cover.
http://img29.imageshack.us/img29/9820/gcr7.jpg
http://img197.imageshack.us/img197/4217/xvqe.jpg
I've noticed CRC error when I turn off lightning in one room but I don't see CRC error when I turn off lightning in the other room.

[kitz]

Yes..  and weve gone over that several times before, but your father wont change it because it would make pockmarks in the wall :/