Kitz Forum
Broadband Related => ADSL Issues => Topic started by: noddy on May 31, 2019, 06:43:54 AM
-
Well that time of year :no: cows came out fencing goes on . Had a good stable 3.5 connection all winter then in three easy steps 1.1 with interleaving 384/8 :no: turned router off for the night but re-sync no different , oh well can cancel Netflix and Now tv :D
adsl info --stats
adsl: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 8000
Max: Upstream rate = 886 Kbps, Downstream rate = 4328 Kbps
Path: 0, Upstream rate = 886 Kbps, Downstream rate = 1150 Kbps
Link Power State: L0
Mode: ADSL2
TPS-TC: ATM Mode
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 12.3 6.5
Attn(dB): 58.5 36.9
Pwr(dBm): 17.4 12.4
ADSL2 framing
MSGc: 95 11
B: 2 110
M: 1 1
T: 8 1
R: 4 16
S: 0.0800 3.9687
L: 700 256
D: 384 8
Counters
SF: 5231240 38592
SFErr: 0 46
RS: 4226842204 3524410
RSCorr: 4498889 558568
RSUnCorr: 0 0
HEC: 0 98
OCD: 0 0
LCD: 0 0
Total Cells: 229284888 176828936
Data Cells: 19975012 1654400
Drop Cells: 0
Bit Errors: 0 14965
ES: 0 29
SES: 0 0
UAS: 29 29
AS: 84538
INP: 8.77 2.00
PER: 16.16 16.86
delay: 7.68 7.93
OR: 50.00 8.06
Bitswap: 25 3
Total time = 23 hours 29 min 38 sec
FEC: 4498889 558568
CRC: 0 46
ES: 0 29
SES: 0 0
UAS: 29 29
LOS: 0 0
LOF: 0 0
Latest 15 minutes time = 14 min 38 sec
FEC: 67527 6357
CRC: 0 0
ES: 0 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Previous 15 minutes time = 15 min 0 sec
FEC: 68578 6483
CRC: 0 0
ES: 0 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Latest 1 day time = 23 hours 29 min 38 sec
FEC: 4498889 558568
CRC: 0 46
ES: 0 29
SES: 0 0
UAS: 29 29
LOS: 0 0
LOF: 0 0
Previous 1 day time = 0 sec
FEC: 0 0
CRC: 0 0
ES: 0 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Since Link time = 23 hours 28 min 57 sec
FEC: 4498889 558568
CRC: 0 46
ES: 0 29
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
>
-
Given that the interleaving seems to be preventing any actual errors, can you over-ride noise margin with your kit to get a bit more speed? Or do you think there's a banding applied?
-
I wouldn't want to prod the DLM further, since it has intervened for a reason. Interleaving is doing its job, since we're only seeing FECs and not many CRCs (in fact, only CRCs upstream).
It's a shame it's shoved your speed down to a rate that's hardly useful for anything. The DLM will adjust the paremeters when and if it become stable again. Sadly, if the issue is interference as you mentioned then it may be there to stay until the source is removed.
I know it's a bit off the topic of your problem, but have you considered a 3G/4G connection? I understand coverage may be limited, and it could be more costly, but it may be something to consider. You could even look at satellite broadband, but pricing and usage allowances have historically not been great (might be better these days) - that ignores latency, which is usally quite high.
-
Looks banded to me.
Or just get the DLM reset and turned off if it's actually usable with whatever interference even if the DLM doesn't like it.
-
Thanks for the input , I'm going to put a connection at one end of the fence where it runs along my line so when the cattle are out of that field at least I'll be able to improve it for short stints :fingers: well that's the theory ( farmer is ok about it ) ,then when I get chance off around the whole fence ( it covers about 6 fields :(
-
A capacitor across the fence? Is there something suitable already about?
-
A capacitor across the fence? Is there something suitable already about?
sorry could you explain a bit more ? not my strong point :-[
-
I would have thought that these nasty pieces of kit could be interference-suppressed by attaching a suitable small very cheap device. One of our learned kitizens please chip in. Hardware is not my thing.
The fence has a sharp transition in voltage from off to in and back again, ‘sharp’ meaning the voltage changes very fast, the time taken to transition is very very short. That’s the source of the interference. If the voltage transition is even very slightly slowed down (https://www.illinoiscapacitor.com/pdf/Papers/spark_suppression.pdf), so that the transition takes longer and the maximum slew rate (‘instantaneous’ rate of change of e.g. voltage versus time at any particular point, or the steepness of the slope of graph at any point) is less, then the interference goes away, goes to lower and lower frequencies and reduces. Some switches also can ‘bounce (https://www.allaboutcircuits.com/textbook/digital/chpt-4/contact-bounce/)’ they sort-of go on-of-on-off-on before settling down and the output graph is all over the place. That case is very bad for interference.
-
Does DLM have a set time when it kicks in to action ? it's just I've noticed all the re-syncs have been between 6.30-7.00 ,at least this morning the interleaving dropped
adsl info --stats
adsl: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 8000
Max: Upstream rate = 886 Kbps, Downstream rate = 3544 Kbps
Path: 0, Upstream rate = 886 Kbps, Downstream rate = 2066 Kbps
Link Power State: L0
Mode: ADSL2
TPS-TC: ATM Mode
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 17.3 6.4
Attn(dB): 57.0 36.9
Pwr(dBm): 18.1 12.4
ADSL2 framing
MSGc: 59 11
B: 64 110
M: 2 1
T: 1 1
R: 10 16
S: 1.9823 3.9687
L: 565 256
D: 16 8
Counters
SF: 86809 82948
SFErr: 2 0
RS: 2821316 1408893
RSCorr: 6365 9989
RSUnCorr: 12 0
HEC: 1 0
OCD: 0 0
LCD: 0 0
Total Cells: 6814105 2917686
Data Cells: 945895 102283
Drop Cells: 0
Bit Errors: 0 0
ES: 11 153
SES: 9 0
UAS: 55 55
AS: 1399
INP: 1.13 2.00
PER: 16.10 16.86
delay: 7.92 7.93
OR: 32.28 8.06
Bitswap: 5 23
Total time = 1 days 5 min 49 sec
FEC: 10830380 1621397
CRC: 478 224
ES: 11 153
SES: 9 0
UAS: 55 55
LOS: 1 0
LOF: 9 0
Latest 15 minutes time = 5 min 49 sec
FEC: 1878 2527
CRC: 0 0
ES: 0 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Previous 15 minutes time = 15 min 0 sec
FEC: 3276 6344
CRC: 0 0
ES: 0 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Latest 1 day time = 5 min 49 sec
FEC: 1878 2527
CRC: 0 0
ES: 0 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Previous 1 day time = 24 hours 0 sec
FEC: 5248473 555819
CRC: 478 29
ES: 11 20
SES: 9 0
UAS: 26 26
LOS: 1 0
LOF: 9 0
Since Link time = 23 min 19 sec
FEC: 6365 9989
CRC: 2 0
ES: 2 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
>
-
Does DLM have a set time when it kicks in to action ?
No. Initially, for VDSL2 based services, it was observed that DLM actions would occur in the early hours of the morning but that "rule of thumb" has long since been overridden. (Not really relevant to you, of course, with an ADSL2 based service.)
-
(Apologies for off-topicness, could split if need be?)
Am I correct in thinking that Noddy’s G.992.3 link does not have PhyR ? Can’t see the associated signs of its activity in the stats’ listing.
-
Considering your line is the only 1 I've seen on these forums with PhyR active that would be a fair assumption to make. :D
If there's no mention of it in the stats then it won't be active.
-
I thought about a capacitor, it would probably have to be rated at 20kV to be safe. I'm not sure on the downside of trying it, I suppose worse case is it blows up the guy's energiser.
Just for fun I've attached a log from when I was trying to fix the fence that's killing our DSL. From left to right, the time up to around 15:10 is the normal level of interference, just under 70 CRC/min and 2800 ES per hour. At 15:10 I've switched the fence off to work on a dodgy connection, then 15:20 it's back on and looking as if I've made a worthwhile improvement. Before I got too hopeful at the end of this trace you can see the fence back off again, presumably the owner doing something, then when it's back on the errors are much worse but not quite as bad as when I started.
-
that looks worse than my one for errors :( this what it's doing just now
-
and here we go right on time this morning :)
-
How's "your" fence constructed. "My" fence is plastic rope with from memory one pair of metal strands in each strand of the rope, making 6 wire conductors in total. The work I did at that time was at a joint where the rope was knotted and the wire not making proper contact, given an audible click. Based on the initial improvement I was all set to go round the whole perimeter remaking all the joins.
-
How's "your" fence constructed. "My" fence is plastic rope with from memory one pair of metal strands in each strand of the rope, making 6 wire conductors in total. The work I did at that time was at a joint where the rope was knotted and the wire not making proper contact, given an audible click. Based on the initial improvement I was all set to go round the whole perimeter remaking all the joins.
the fence is a wooden posts multi strand barbed wire with a galvanised live running on the top . I would guess it's total run is about 3 miles but that is a total guess
-
this is looking like a daily ritual ??? todays re-sync 6.36
adsl info --stats
adsl: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 8000
Max: Upstream rate = 823 Kbps, Downstream rate = 4888 Kbps
Path: 0, Upstream rate = 829 Kbps, Downstream rate = 3067 Kbps
Link Power State: L0
Mode: ADSL2
TPS-TC: ATM Mode
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 6.2 5.7
Attn(dB): 57.5 37.0
Pwr(dBm): 19.1 12.6
ADSL2 framing
MSGc: 95 12
B: 30 94
M: 1 1
T: 2 1
R: 12 16
S: 0.3182 3.6244
L: 1081 245
D: 96 8
Counters
SF: 121206 119387
SFErr: 2 0
RS: 24483620 2147579
RSCorr: 1377728 14667
RSUnCorr: 2 0
HEC: 1 0
OCD: 0 0
LCD: 0 0
Total Cells: 14090433 3806753
Data Cells: 659749 158325
Drop Cells: 0
Bit Errors: 0 0
ES: 5167 275
SES: 27 0
UAS: 114 114
AS: 1948
INP: 4.26 2.08
PER: 16.07 16.31
delay: 7.63 7.24
OR: 50.27 8.82
Bitswap: 244 20
Total time = 1 days 4 min 16 sec
FEC: 14916243 3667504
CRC: 7758 390
ES: 5167 275
SES: 27 0
UAS: 114 114
LOS: 3 0
LOF: 27 0
Latest 15 minutes time = 4 min 16 sec
FEC: 172396 1954
CRC: 1 0
ES: 1 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Previous 15 minutes time = 15 min 0 sec
FEC: 592452 6849
CRC: 1 0
ES: 1 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Latest 1 day time = 4 min 16 sec
FEC: 172396 1954
CRC: 1 0
ES: 1 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Previous 1 day time = 24 hours 0 sec
FEC: 2203303 566479
CRC: 1901 40
ES: 1173 26
SES: 9 0
UAS: 29 29
LOS: 1 0
LOF: 9 0
Since Link time = 32 min 27 sec
FEC: 1377728 14667
CRC: 2 0
ES: 2 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
>
-
Am I missing something, because do me those stats don't look bad. Only two ES in the last 32 minutes, and last 24 hour MTBE of 73. Apart from the re-synch what is the actual impact on your service?
-
Am I missing something, because do me those stats don't look bad. Only two ES in the last 32 minutes, and last 24 hour MTBE of 73. Apart from the re-synch what is the actual impact on your service?
thats the strange thing it's has been bad then then started this almost daily re-sync that seems stable then between 6.30 to 7 each morning re-syncs again ?
-
The link below is an interesting thesis on electric fence faults.
https://mro.massey.ac.nz/bitstream/handle/10179/1161/02whole.pdf (https://mro.massey.ac.nz/bitstream/handle/10179/1161/02whole.pdf)
I was particularly impressed by the attached Figure 3-3, showing how a damaged insulator can create excessive RFI. It's the sort of effect that might clear and reappear depending on the weather.
An electric fence circuit is similar to a car ignition coil, except that the object is to generate a high voltage spike without it sparking over. Since the fence will act as a lossy transmission line, the waveform at the far end of a long fence will have fewer high frequency components than at the near end, and will also generate a smaller magnetic field from the loop current. Depends on the length of the fence, of course. But if there is a choice, place the exciter as far away as possible.
As regards the idea of adding a loading capacitor on the output of the generator, an alternative is to add a series resistor. Both of these will slow the waveform on the fence, reducing the high frequency spectrum. I suggest components rated for at least 10 kV, or 20 kV as aesmith suggests, and have them sealed in a way that prevents moisture getting at them, as they will be rated for dry conditions.
To find optimum component values, you may need to measure waveforms. In section 4.2 of the thesis, the author describes ways of sensing the pulse voltage, but misses the simplest one, which is to use a capacitive divider, as there is no need for a response to D.C.
-
thats the strange thing it's has been bad then then started this almost daily re-sync that seems stable then between 6.30 to 7 each morning re-syncs again ?
is that the daily time frame the live wire on the fence is tested to check it is still working by sending voltage down the wire, similar to the way a TAM test is run?
-
That is a nice graph Ian am just about to read that thesis, thanks for that much appreciated. (Even though I think that luckily I am completely ok, especially seeing as my missus said when I asked her ‘Well I absolutely never ever turn it on any more’.
I learned that the fence is not even turned in because our donkeys just believe in the evil power of the fence, especially seeing as it has a little bit of white tape stuck to the wire, which flaps around in the breeze. This alone is enough. They’ve been stung in the past, that’s enough.