|
|
Hello everyone
I was wondering if someone could explain please - whenever we use the landline, our SNR drops from 6db to 5.5db or lower. Now, we used to suffer from frequent disconnections - they're less frequent now (thanks to a Broadcom chipset based router), but they've not gone completely. Whenever our SNR drops to 4db or around, we get errors in the router stats. What would this point to? Is it normal for SNR to drop whenever the tel. is used? It's a Siemens c610a tel. and I'm using an ADSL nation filter.
Thanks for your time and help!
Edited by deleted (Sat 28-Jan-12 21:15:33)
|
|
|
There really should not be any noticeable drop in the SNRM when the telephone is used.
Please borrow and then test test with;
(1) a different filter
(2) a different telephone.
-----------------------------------------------------
100% Linux and, previously, Unix.
|
|
|
Thanks very much.
I've tried with another phone - an old one with a dial (love them!) and the same thing occurs - a drop by 0.4db in both down and upstream SNR. Maybe it's not a lot - I don't know, but earlier this afternoon it was down from 6db to 4.4db and it caused some errors, as shown in the router stats, unless it was something else. I'll try a different filter (and from the test socket) and report back.
Also, there was a noticeable hiss, louder on the phone with a dial than on the remote one, on both phones. I think I've asked about it before and I think it was down to the wireless signal from the router, but shouldn't be heard on the wired phone.
Anyway, thanks - you're all great, very helpful and knowledgeable!
Edited by deleted (Sat 28-Jan-12 23:52:06)
|
|
Register (or login) on our website and you will not see this ad.
|
|
|
I'll try a different filter (and from the test socket) and report back. That is a good idea and -- oops -- something I should have mentioned previously. Also, there was a noticeable hiss, louder on the phone with a dial than on the remote one, on both phones. I think I've asked about it before and I think it was down to the wireless signal from the router, but shouldn't be heard on the wired phone. The general rule of thumb is to keep the base station for a cordless phone and a modem/router as far apart as is practical.
It is possible to notice inductive coupling between a wired telephone and a modem/router (even one which has its WiFi disabled) if they are too close to each other.
A drop in SNRM of less than 0.5 dB (if possible to detect and display such fractional values) when a telephone "loops the pair" would be perfectly understandable and acceptable. A greater than 1 dB drop most definitely indicates that "something is not right".
-----------------------------------------------------
100% Linux and, previously, Unix.
|
|
|
I will disagree with that. There will be a change to either SNR, received power level, or attenuation. When the the phone is taken off hook a low impedance is placed across the filter outputs. All filters are single stage and changing the load impedance may change the inputs complex impedance and the frequency characteristics of the filter.
Predominately the impedance will change and even for the DSL frequencies it will be slightly lower and that will add additional load to the DSL signal and thus result in a slightly lower level being received by the modem. That will be reflected as a lower SNR.
Going on "gut feel" and remembering doing a few checks years back, I would think that up to 1dB difference would be normal and maybe on some line 2dB could be seen.
You will know that a decent modem can cope with an SNR down to about 3dB from a 6d sync and excellent modems such as the 2700 can cope with slightly negative SNRs.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
M H C
taurus excreta cerebrum vincit
|
|
|
FWIW
Whenever I've experienced lower SNRM due to a phone being used it's always turned out to be a line or equipment fault!
|
|
|
I shall respect your opinion on that point. My own personal observations are that the reduction in SNRM is very minimal -- to the point of being virtually unnoticeable. Whenever I have noticed a reduction in SNRM, it has been the result of a line fault -- more often that not a HR or semi-conducting joint.
Being an enthusiast for the 2Wire devices, your last point is well known. 
-----------------------------------------------------
100% Linux and, previously, Unix.
|
|
|
ACK. That agrees with my observations. 
-----------------------------------------------------
100% Linux and, previously, Unix.
|
|
|
I come from an electronics design background and having designed very some very complex telephony switches, line impedances and attenuation were something we had to pay a lot of detail. A lot of detailed analysis was done on differing lines, handsets at each end of tolerance, effects of filters and more. On some, the effect was minimal and almost immeasurable whilst on others it was significant.
I will have a look at one first thing tomorrow morning and see what happens. I know it is a good stable line ...
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
M H C
taurus excreta cerebrum vincit
|
|
|
I come from an electronics design background and having designed very some very complex telephony switches That much I have deduced from reading your postings in these fora. line impedances and attenuation were something we had to pay a lot of detail. A lot of detailed analysis was done on differing lines, handsets at each end of tolerance, effects of filters and more. On some, the effect was minimal and almost immeasurable whilst on others it was significant. Interesting background information. I will have a look at one first thing tomorrow morning and see what happens. I know it is a good stable line ... Thank you. Your experimental observations are duly awaited.
-----------------------------------------------------
100% Linux and, previously, Unix.
|
|
|
Some results:
The line is fairly long and gives around 1.6 Mbps. It has not failed in 3 years and has been extremely consistent in performance with a 2wire 2700 business hub installed. The Netgear that was originally there have under 1Mb and would drop out most nights. Voice quality is clear/excellent - and the only interruption was a recent theft of underground cable. The drop wire terminates on the outside wall and comes straight through to a filtered face plate from which two extensions are taken.
I tried various combinations or 1, 2 or 3 phones off hook several times.
Taking a single phone off hook resulted mainly in differences of 0dB, -0.1 and -0.2 and one of -0.5dB.
With two off hook, again there were varying results, mainly in the -0.2 to -0.6dB range. Going to all three off the highest was -0.9dB.
Obviously, these are quick and dirty readings and changes in background SNR could be making the reading better or worse but it does show that you can get a minor change when a phone goes off hook.
I also looked at what happens when an incoming call occurs - 0dB each time.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
M H C
taurus excreta cerebrum vincit
|
|
|
I'm using an ADSL nation filter.
I'm going to be controversial here, and suggest that you replace this.
Reason?
I went through a couple of these (faceplate) filters in about 3 years. For some reason they didn't last particularly well on my line.
Can I suggest you try an Openreach-branded faceplate?
I'd also go with the suggestion of a 2wire 2700 as a cheap, effective bit of kit.
|
|
|
Thank you for performing the tests, although obviously not as rigorous as in a laboratory environment.
Purely for interest, how were you determining the SNRM values in that experiment?
-----------------------------------------------------
100% Linux and, previously, Unix.
|
|
|
Taken from the 2700 page ...
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
M H C
taurus excreta cerebrum vincit
|
|
|
Right. So we would both say that a particular value was x.y +/- 0.1 dB.
-----------------------------------------------------
100% Linux and, previously, Unix.
|
|
|
|
Thanks guys for a very insightful discussion.
I've not managed to do my tests - I'll do them on Wed.
The router I'm using is a Broadcom based router - TP-Link 8960 (couldn't afford the Billion 7800).
I'm on a relatively short line, close to the exchange:
Downstream Upstream
Line Coding(Trellis): On On
SNR Margin (0.1 dB): 59 96
Attenuation (0.1 dB): 120 41
Output Power (0.1 dBm): 0 115
Attainable Rate (Kbps): 23964 888
the way I saw fluctuating was from the router's stats.
I'll report back as soon as I've done the tests and I'll post screen shots.
Thanks!
|
|
|
Lots of factors come into play, but it has to be assumed that the router is consistent in its measurement over the hour or so and thus +/- 0.05dB. So, what was registered as 0, could be -0.099 or +0.099dB and -0.9 could be anywhere from -0.801 to -0.999
It was getting light, so the overnight increase in atmospheric noise would be receding resulting in an increase in SNR, could there be additional domestic created noise possibly ... But to alleviate those, as much as possible, I used the SNR just before taking a receiver off hook, using an F5 page refresh for the new value.
It definitely showed that a handset can cause a drop in SNR. On a shorter line, it may be a little worse, a different handset could be better or worse - an older device maybe worse and a new DECT better ...
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
M H C
taurus excreta cerebrum vincit
|
|
|
|
mine has an auto refresh function, so it was showing almost immediately.
there was a drop from 59 = (5.9db) to 57 = (5.7db) during the day.
when there was additional domestic noise (in the eve), the figures were much lower, approx. from 54 (5.4db) to 48 (4.8db) or slightly less.
it just happened i was looking at my router stats when my partner was on the phone!
|
|
|
I will disagree with that. There will be a change to either SNR, received power level, or attenuation.
Seconded.
|
|
|
The SNR figure is the easiest to use in most case. But it is actually the power seen by the modem input that is lower with noise staying constant and hence a lower SNR.
Glad to see that someone else has seen the same effect.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
M H C
taurus excreta cerebrum vincit
Edited by MHC (Mon 30-Jan-12 23:39:29)
|
|
|
Interesting at the academic level but, considering that end-user equipment is being used in non-laboratory conditions, I would not consider results at that level of precision. You have, however, made your point clear to me.
My overall feeling is that the average end-user, using the average end-user equipment, will only notice significant changes in the SNRM -- when a telephone loops the pair -- under fault conditions.
-----------------------------------------------------
100% Linux and, previously, Unix.
Edited by burakkucat (Wed 01-Feb-12 12:49:58)
|
|
|
Those drops (0.2 dB and 0.6 dB) I would class as insignificant and nothing to be concerned about. If the drop was more that 1 dB, then I would wonder . . .
-----------------------------------------------------
100% Linux and, previously, Unix.
|
|
|
Was thinking similar, any variation of under 1dB I usually ignore in peoples stats.
The noise environment even ignoring on/off hook phones is not constant, if it were then algorithms would be used to filter it out and thus use even lower signal levels and reduce cross-talk
|
|
The author of the above post is a thinkbroadband staff member. It may not constitute an official statement on behalf of thinkbroadband.
|
|
|
I have performed my own experiment. I have:
* a SSFP (an ADSL V1.0) fitted at the NTE5/A
* a total of three telephones (two 20CN, one 21CN)
* a 2Wire 2700HG
With all telephones "on-hook", the 2700HG shows --
| Text | 1
23
4 | Current Connection:
Current Noise Margin: 5.8 dB 7.3 dB Current Attenuation: 45.7 dB 27.3 dB
Current Output Power: 19.6 dB 12.5 dB |
With one, two or all three telephones "off-hook" there is absolutely no changes to any of the above reported values. Just to be sure, I refreshed the screen three times for each test.
For the record, one of the 20CN telephones is a 300-series, dated 1956. I was surprised to see that one had no effect.
-----------------------------------------------------
100% Linux and, previously, Unix.
Edited by burakkucat (Wed 01-Feb-12 13:06:31)
|
|
|
|
How do you define 20cn abd 21cn telephones?
|
|
|
The century in which they were manufactured.
-----------------------------------------------------
100% Linux and, previously, Unix.
|
|
|
|
Ah, that makes sense - nothing to do with the normal meaning of 20/21cn in the BT/ broadband sense then
|
Pages in this thread: 
Print Thread
