Print

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

[g3uiss]

Total disaster tonight on my VDSL2 circuit. Loss of sync first tehn many resyncs. US sync gone to minimum 512kbps, low SNRM on US and HLOG looks like many U1 tones have gone. Will contact Zen tomorrow any steers appreciated metallic pathway or noise ?

Stats uploaded to www.west-lodge.com/vdsl/index.htm

[burakkucat]

Hmm . . . Does the underlying telephony service still operate? Do you receive a "dial" tone? Have you performed a quiet line test?

In the QLN plot, I see evidence of significant cross-talk. In the Hlog plot, the significant absense in the US0 and US1 bands is ominous.

[g3uiss]

Phone service seems normal. Quiet line is perfect.

[kitz]

Upstream only can indicate telephony or HR fault.   
Note that its across both bands so something causing noise over wide range of frequencies and sufficient to cause DSErrs, possibly indicating short sharp burst type noise as DS SNR graph looks pretty normal aside from when loss of sync.     QLN shows very obvious crosstalk on D2.   Noticed the gradual US SNR slide.

I would be tempted to give the line a port rest for an hour if you can, or at very least 30mins   Fully power down modem and leave it switched off.   That 3dB range of jaggedness after a loss of sync is something that occurs from time to time even on the best of lines.   It sometimes doesn't seem to resolve without a full power down of the modem and leaving a sufficient period of time to register an unforced retrain.

[g3uiss]

Thanks @kitz, meanwhile DLM applied Interleave US, but effectively nothing changed apart from reduced CRC and ping times. Never had much on D2 ever

www.west-lodge.com/vdsl/index.htm

[g3uiss]

Spoke with Zen. fast tests OK. Want modem swapped out and wait 1 day to see result. Seems pointless to me, but done it. Stats updated. Any advice on taking forward please

[kitz]

I would definitely try powering down the modem for a while, if you havent already done so.  We've seen that jaggedness occur many times after a sudden retrain, the SNRM goes mental. Has happened to me personally several times, and I've seen it many times on other lines.  It seems to happen more often on ECI cabs, but its not exclusive to ECI's - for example adslmax is on a Huawei can and seen it twice on his.  There are many more.     Obviously it's not going to fix or help understand what caused the line to lose sync in the first place, but it is well worth ruling it out.   

No idea why it should happen, sometimes the bit-swap goes crazy too.  Ideally a fresh sync should create a new bit allocation table, but sometimes its almost like it doesn't so I'm not sure if its perhaps its something to do with the modems, which is why I always say fully power down the modem and leave it off for at least 30mins to ensure that there is a complete 15min empty bin period.  This also has the benefit of signalling to the dslam that it is an unforced retrain.   Also, I think it was b*cat who came up with additional and plausible theory as to why letting the port rest may help.    If it was b*cat, then perhaps he may refresh my memory what it was.   

[g3uiss]

@kitz. Not been able to do that yet as my daughter is working from home on her university lectures , and while the connection is flaky she is able to manage at the moment.

It would seem from looking at the FEC that there are a lot of US errors being corrected and the sync is still low presumably from the absence of tones. So not sure if there are two effective faults, noise and missing tones

OR due PM tomorrow now.

[g3uiss]

Have had the modem off for 35 minutes, and certainly not much improvement. DS sync slightly lower (lost D2) US slightly better, interesting its syncing double the max attainable rate ?

www.west-lodge.com/vdsl/index.htm

[burakkucat]

Also, I think it was b*cat who came up with additional and plausible theory as to why letting the port rest may help.    If it was b*cat, then perhaps he may refresh my memory what it was.   

I wish I could! 

It was many years ago, now, that I first suggested the "power-down, disconnect and leave in that state overnight" technique to N*Star. I can remember using the phrase "allow the line-card port time to relax". Back then, his circuit statistics were showing the demented oscillations in the SNRm plot.

[burakkucat]

Having just reviewed the latest statistics, I now have "tingles in the whiskers" that we are looking at the result of a joint showing HR or semi-conductive tendencies. Such a fault would effectively destroy the AC balance of the twisted pair and, thus, account for the obvious cross-talk that can be seen in the DS2 & US2 bands of the QLN plot.

[g3uiss]

@bcat. Thanks for that suggested diagnosis! Do such faults no show on the Fast Test that the ISP runs ? I do know the OR check the balance however on their kit.

[burakkucat]

Do such faults no show on the Fast Test that the ISP runs ?

Sorry, I have no idea.

If I was testing for it, I would connect a time domain reflectometer directly to the pair. Then, with the trace displayed, adjust the gain so that any major peak is no more than about half of the height of the display. Finally, from a mobile telephone, make a call to the telephone number associated with the line. Watch the TDR trace closely. If there is a blip or peak that moves in synchronism with the applied ringing voltage, that will be the likely culprit.

[g3uiss]

Um my electronics degree was in 1968, so that method is not totally beyond my comprehension, however I’m not sure if our OR engineering at the first level would. However hopefully they might have some kit that produces a diagnostic of the balance.

My lines are on AL and very flaky, however at them moment the ADSL circuit is stable.

[Black Sheep]

Sorry, I have no idea.

If I was testing for it, I would connect a time domain reflectometer directly to the pair. Then, with the trace displayed, adjust the gain so that any major peak is no more than about half of the height of the display. Finally, from a mobile telephone, make a call to the telephone number associated with the line. Watch the TDR trace closely. If there is a blip or peak that moves in synchronism with the applied ringing voltage, that will be the likely culprit.

Speaking for the engineers with the JDSU (no idea about the EXFO) ... but the in-built TDR recognises the incoming AC ringing voltage as 'High voltage', and flashes a warning all over the screen making it impossible to carry out any form of interrogation.

Can't beat the old-school stand alone TDR meters, for this kind of faulting technique.