Kitz Forum
Broadband Related => ADSL Issues => Topic started by: badassgeek on February 16, 2014, 01:42:43 PM
-
I know there's a lot out there about sync speeds and chipsets etc, but I haven't seen much on copper overhead lines in particular. That's what I'm on and I'm getting constant drop outs, and some unexpected latency, inconsistencies over resyncs etc. The line is adsl2+ and both my gear and the isp issue adsl kit are dropping out. Whilst we're on this topic, anyone know if there's a theoretical or rough max speed for this type of line?
I just moved in november and the only thing that has changed is I'm now using overhead lines (according to the engineer who fitted the phoneline it's a thick copper line from the 80's) and obviously a different exchange. My ISP uses talktalk at the exchange, and I know they usually use infineon chipsets but according to my isp issue router the other end is a broadcom in my case. I don't know what chip was used for the exchange at the previous house but my line there was brilliant.
I have a buffalo WBMR-HP-G300H (lantiq/infineon chipset) using openwrt firmware which worked perfectly on my previous line and gave good sync speeds. It still does give higher sync speeds by about 0.5-1mbit actually, but it drops out several times a day. There could be issues with this router as another openwrt user has told me it's unstable for him on adsl2+, but like I said adsl2+ was fine for me before, I just wish I'd asked the bloke for more details now.
The isp issued Technicolor (thompson) TG582n is using a broadcom chip (6328 according to http://wiki.kitz.co.uk/index.php?title=BCM_routers) and holds the line for 3-4 days before dropping out. One of the things with this router is that I don't think it's really intended for a household full of gamers (who are gaming and voice streaming at the same time etc), so in any case I need something a bit better than this.
According to sam knows the exchange is less than a mile away. My downstream attenuation is lower than 24db with both routers so that seems right.
As the isp and openreach are so far unable to solve the problem I want to try replacing the kit as a last ditch attempt.
I know that's a lot of back story, but basically I'm wondering if anyone has any recommendations for chipsets that ought to be solid on this type of line. Instead of getting an all in one router/modem, I want to get a bridge modem that is solid so I can just put in my wndr3700 I have from a couple years back when I was using cable and be sure of wireless performance. So long as the modem/router can be put in to bridged mode using pppoe bridging or else pppoe to pppoa and doesn't give me two lots of NAT or something else to interfere with routing/firewalling it will be fine for me. I have a short contract, so will be happy to just unplug a bridged modem later on and use my WAN router on a cable service if it really has to come to that.
I read trendchip are 'good for short lines' but that's vague and nobody is claiming they're a rock as far as I can tell. I see a couple of tp-link modems with these chips that come highly recommended when used solely in bridge mode, but I think the line needs to be really good in the first place for trendchip to be any use. Can anyone verify that?
I know broadcom keep getting a lot of praise, and my own broadcom is doing well in comparison and I've seen a d-link 320b but as far as I can tell there is no good rep for this modem like some routers have.
I'm considering getting a draytek 120 v2, which is lantiq/infineon as a lot people are saying this router syncs high and is stable. Though, I'm seeing my buffalo (also lantiq) torn to shreds by isp issue kit, so obviously I have some doubts. If I had a good idea it would hold on a line like mine, this is what I would prefer to get because of the pppoa bridge option.
Anyway, thanks for reading and suggestions welcome.
-
Two points
There were some inconclusive postings in this forum suggesting that the TG582n was under- powered in the CPU for fast ADSL/VDSL when the child filter is also in use.
You don't mention any anti interference precautions you have taken.
Both these might contribute to your problems.
-
Thanks for the reply,
The TG582n router is using firmware version 10.2.0.B, and I have not setup the content filter. Most of the feautures I can disable, I have already and the content filter seems to be disabled from the web interface, at least it says there is no address based filtering in use and unknown sites are allowed. I am not familiar with the commands via telnet for this router yet, so if that's something that needs disabling via telnet I haven't done that yet and would appreciate any instructions (or a link) to disable that or other unwanted junk as I'm not a fan of 'features' anyway.
I use a shielded adsl cable from belkin which is up to cat5e standard, I've removed all energy saving bulbs and any possible interfering equipment from the vicinity of the line all around the house. There is perhaps noise on the line as I do hear a low buzz on it, but according to the SFI engineer that came here there was no noise so I assume that's within 'acceptable levels'. I have two faceplates, austin taylor and xte-2005 clone from run-it and I've tried both with and without the shielded cable and notice no improvements in stability.
-
I think before you do any more modem swapping it would be helpful to get some performance stats, which might point towards the reason for the dropouts. If you've still got the TG582N connected, you could install DSLstats to monitor the connection. A day's worth of graphs and other data will help a great deal.
http://dslstats.plainroad.me.uk
-
I think before you do any more modem swapping it would be helpful to get some performance stats, which might point towards the reason for the dropouts. If you've still got the TG582N connected, you could install DSLstats to monitor the connection. A day's worth of graphs and other data will help a great deal.
http://dslstats.plainroad.me.uk
Thanks, I didn't realise there was something like this, apart from in fancy routers or I'd have used it. I've set it monitoring and will see what happens, and now I think of it the openwrt might have something similar so I will try the buffalo when this drops out.
-
Yes I agree with roseway, some data could be very useful - wait and see.
The only thing which occurs to me is that "overhead copper" = aerial !
i.e. you might be picking up interference from some distance not just your own appliances. In which case I would be inclined to put an RF3 in the line, close to the modem input. BUT wait and see the log first .
-
Yes I agree with roseway, some data could be very useful - wait and see.
The only thing which occurs to me is that "overhead copper" = aerial !
i.e. you might be picking up interference from some distance not just your own appliances. In which case I would be inclined to put an RF3 in the line, close to the modem input. BUT wait and see the log first .
Yes, I thought the line might be slightly worse when we were moving in, but I assumed as I have good kit it would come down to just using a slightly slower speed and I was prepared to put up with an unusual number provided it was 12Mbit+. Cable from vm is available, but I wasn't feeling desperate enough at the time :p
Speaking of aerials, our TV aerial runs parallel to the drop down cable for the phone line (I say phone line, but we only use the adsl) as they are both running into this room (there is no tv or other stuff like that).
Another source of interference could be the industrial units over the road which can be no more than 18 meters away. The phone pole itself is about 12-15 meteres away from a halfords garage and other units a little further than that.
Other than that, a tree is close to the line with a few branches that could touch it if it was windy (nearest one just under a foot away maybe), I'll have a closer look tomorrow.
There has been bad weather, including a storm but the line has been bad since November when it was fitted. The weather hasn't always been this bad, and it's settling down but the line is not.
I have a pair of ferrites on the adsl cable at either end too, which seems to help a weird latency problem I have. The issue with the latency is about 4 out of 5 times when I sync the line has an extra 14ms of latency added to the line which I don't think is congestion. That is to say, if it syncs 'right' for that 1 out of five I would get a more usual RTT when I ping say a public nameserver that I know is okay - usual being 28ms since the ISP put on interleaving to try to solve the problem (previously 9.5ms non-interleaved). Adding the ferrites to the adsl cable improved that, and more often now it will sync with normal 27ms round trip times rather than 40ms.
What is an RF3?
I did get some data, I'm sifting through it so I'll add that in another post. So far I see the downstream SNR sometimes dropping by more than 1db and there are a lot of FEC errors and some CRC errors, but much more of them when the SNR is dropping. Using the xte-2005 and the adsl cable with the ferrites so far it's better but I've tested that in a different time frame.
I'm curious about this faceplate http://www.run-it-direct.co.uk/btvdslfaceplate.html, as it includes an RF filter that is seperate from the other mode chokes and I don't see any other faceplates like that. Does anyone know if these work well?
Thanks for the replies so far, appreciate the constructive input :-)
-
I couldn't figure out how to link in images properly (new here and forum noobish), so I attached some of the results from dslstats.
I had a look into the possible power issue with the technicolor tg582n, and found this http://npr.me.uk/telnet.html#cpu, and tried turning off the power saving features as suggested but the router became unstable for me so it didn't improve it but I don't think there's any power issues for me anyway. I'm not sure which set of tests reflect when I tried that (it was late) and it became unstable, but I'll post one set from when I think the line lost sync (test1). I just included basics for this, put in some more in test2 as I took snapshots of important changes this time.
The average errors:
Average error rates over a 24 hour period
24 hours starting 16 Feb 2014 14:20:23
Per second Per minute Per hour Per day
CRC Up 0 0 0 0
Down 0 0.01 0.44 10.5
FEC Up 0 0 0 0
Down 0.02 1.19 71.6 1718
HEC Up 0 0 0 0
Down 0 0 0 0
ES Up 0 0 0 0
Down 0 0 0 0
Basically a lot of errors but when the SNR went down, the amount of errors went up.
In the second test I put in the xte-2005 and the shielded adsl cable with two ferrites on it. I don't have the latency problem like before, I'm sure it will come back at some point but the averages are lower (test2):
24 hours starting 17 Feb 2014 14:20:42
Per second Per minute Per hour Per day
CRC Up 0 0 0 0
Down 0 0.01 0.41 9.79
FEC Up 0 0 0 0
Down 0 0.26 15.3 367
HEC Up 0 0 0 0
Down 0 0 0 0
ES Up 0 0 0 0
Down 0 0 0.20 4.90
This is over a longer time period than the last test, but I still think the errors are less.
Posting a fair few hours later but these are the stats now from the web interface:
Uptime: 0 days, 18:09:38
DSL Type: ITU-T G.992.5
Bandwidth (Up/Down) [kbps/kbps]: 973 / 12,189
Data Transferred (Sent/Received) [MB/GB]: 234.03 / 2.66
Output Power (Up/Down) [dBm]: 12.3 / 19.8
Line Attenuation (Up/Down) [dB]: 13.3 / 23.5
SN Margin (Up/Down) [dB]: 14.9 / 15.0
System Vendor ID (Local/Remote): TMMB / ----
Chipset Vendor ID (Local/Remote): BDCM / BDCM
Loss of Framing (Local/Remote): 0 / 0
Loss of Signal (Local/Remote): 0 / 0
Loss of Power (Local/Remote): 0 / 0
Loss of Link (Remote): -
Error Seconds (Local/Remote): 3 / 0
FEC Errors (Up/Down): 0 / 677
CRC Errors (Up/Down): 0 / 5
HEC Errors (Up/Down): 0 / 9
I've had a listen on the quiet line test again, but with another handset and it's quieter with this handset but still with a low hum. So, maybe the line isn't that capable or could it still be noise/interference?
-
RF3 - see http://www.run-it-direct.co.uk/BT80BRF3.html
For how it works see earlier postings.
Filtered faceplate is good ! saves external microfilters from damage for one thing but doesn't include an RF3, only in the I-plate which doesn't have microfilters !
-
Do you have a snapshot of the SNR margin graph covering a time period of a couple of hours or so? This would give a lot more useful information. Your stats indicate an upstream and downstream SNRM of 15 dB, but if those values actually fluctuate significantly over time, that would indicate varying interference levels and may suggest a possible source of the interference.
-
Filtered faceplate is good ! saves external microfilters from damage for one thing but doesn't include an RF3, only in the I-plate which doesn't have microfilters !
I thought it was disclosed that the Mk 2 SSFP does, indeed, contain the RF3's components? :-\ Or am I imagining it? ???
-
It was I that put that info up I think, B*Cat ??
I haven't the time to go trawling through my historical posts though.
-
"I thought it was disclosed that the Mk 2 SSFP does, indeed, contain the RF3's components? :-\ Or am I imagining it? ??? "
I thought the point about the Mk2 faceplate was that it was suitable for VDSL i.e. to 17.7 MHz ? and though the RF3 had more bandwidth than the RF2 it wouldn't cover that. In fact the bandwidth ratio needed for VDSL is extremely wide for any such device and the RF3 is a low frequency core design i.e. not sectionalised or distributed ?
As we are talking ADSL it isn't vital for this thread but I'd like to be sure.
-
Don't know if will help you, JGO ........... but here's a link to a link :)
http://forum.kitz.co.uk/index.php?topic=13092.0
-
I have a BT iplate. Since you mentioned it, I put it in today to see what would happen. It's in the graphs, but it's easy for me to see less errors occured with the xte-2005, than with iplate + adsl filter.
Per second Per minute Per hour Per day
CRC Up 0 0 0 0
Down 0 0.13 8.03 193
FEC Up 0 0 0 0
Down 0.10 6.04 362 8697
The line can drop any time of day, and the best I've seen is 5 days uptime
So is the consensus the MK2 plate is better than a run-it xte-2005 for keeping out RF on adsl2+? Or would I maybe be better to do xte-2005 + RF3?
-
I thought the point about the Mk2 faceplate was that it was suitable for VDSL i.e. to 17.7 MHz ? and though the RF3 had more bandwidth than the RF2 it wouldn't cover that. In fact the bandwidth ratio needed for VDSL is extremely wide for any such device and the RF3 is a low frequency core design i.e. not sectionalised or distributed ?
Again, I think a fellow Kitizen recently posted a description of the mode of operation of an RF3 . . . implying that it does not operate as a band pass filter. :-\
Ah ha! I've just found the thread (http://forum.kitz.co.uk/index.php?topic=13473.0). And the originator of that thread was (drum roll) . . . JGO!
-
" So is the consensus the MK2 plate is better than a run-it xte-2005 for keeping out RF on adsl2+? Or would I maybe be better to do xte-2005 + RF3? "
From your point of view you probably need something like that which attenuates even mode pickup on the overhead wires. The modem should have some rejection of it's own but manufacturers are coy about giving figures !
As to which is better, suggest you need at least week long test with each to decide which is best. ( In my case if the speed is going to drop it is nearly always on a sunday evening, but on average only 1 week in 3. I just check on monday and if necessarily switch the modem off for 5 secs and swear - most important !)
Now off topic :=
The RF3 etc aren't supposed to be bandpass filters, they are common mode rejectors, BUT because of the limitations of magnetic cores etc each wire has some leakage inductance which doesn't couple with the other winding. The effect of this is a low-pass filter which limits the bandwidth ( REAL bandwidth in Hz!) So the design inductance has to be limited, which curtails the low end useful frequency in order to get a desired high frequency performance. (I understand the RF2, aimed at telephones ? had to be scaled down for ADSL as the RF3)
VDSL has an enormous frequency range , about 608:1 as against 38 for ADSL1 & 2 and 76 for ADSL2. 200:1 is a more comfortable design range.
So I suspect a VDSL common mode filter is a compromise - either sacrificing high end performance "The S/N is usually poor anyway at that end so no loss" OR low end " The high end is more susceptible pick up and low frequency interference comes in via power supply anyway". Measured performance would be useful.
-
P.S.
The I plate doesn't isolate the master socket as far as ADSL nor does it include micro filters so you need to add them at each socket - fragile ! . Thus a filtered faceplate is better for pickup on the extensions.
Reading the blurb on a Mk2 there is something about working up to 30MHz (?!) so it may be inferior to a MK1 for ADSL as opposed to VDSL. As you say you can always add a RF3 which at least is wall mounting.
Frankly without test gear it ends up as suck it and see since interference can enter by several routes and you need to clobber them all. (If it is half by one route you could eliminate one but the level will only drop substantially when you deal with both.)
Good Luck
-
I had a significant improvement using the MK II faceplate (the vigor 120 I got as well which seems to hold well), there are still some small issues, but i also have an rf3 at hand, i was wondering if anyone had instructions on usage? I have some cat5e to play with so I want to give it a try and see what happens. Thanks :-)
-
The RF3 goes in the line, before any splitting for phone and Broadband but is made with terminal screws. Internally it is screw connections for input and V connectors for output.
Important point; if you do filter the combined phone/BB line - don't put an extra twist as this will invert the polarity of the DC and upset the phone !
I found it convenient to cut a modem cable into two and insert the RF3 in it so I could plug it in and replace the plain Mk1-faceplate-to-modem cable with a RF3'ed one quickly. This is just my idea; if you want to screw it to the wall and hard wire it do so.
Incidentally, confirmation from another thread, the RF2 slowed someone's ADSL speed. It was designed to keep RF out of phone full stop.( Broadband what's that ?) Probably N times as many turns on the coils as the RF3 since there was no need to compromise.
-
The RF3 goes in the line, before any splitting for phone and Broadband but is made with terminal screws. Internally it is screw connections for input and V connectors for output.
Important point; if you do filter the combined phone/BB line - don't put an extra twist as this will invert the polarity of the DC and upset the phone !
I found it convenient to cut a modem cable into two and insert the RF3 in it so I could plug it in and replace the plain Mk1-faceplate-to-modem cable with a RF3'ed one quickly. This is just my idea; if you want to screw it to the wall and hard wire it do so.
Incidentally, confirmation from another thread, the RF2 slowed someone's ADSL speed. It was designed to keep RF out of phone full stop.( Broadband what's that ?) Probably N times as many turns on the coils as the RF3 since there was no need to compromise.
So A-B as normal on the screw connectors, I guess the unmarked one is A, as I see only B and E that is marked there. Extra twist?
-
"Extra Twist ?"
Yes - if you reverse the polarity of the 50v you'll screw up ! One reason I put it in the BB only wiring where no 50v.
Suggest you use something like DSLstats to look at the bits/tone to get an idea what happens to the response over the bandwidth before and after. If the Faceplate Mk 2 has a RF4 the results might be "interesting" !
-
"Extra Twist ?"
Yes - if you reverse the polarity of the 50v you'll screw up ! One reason I put it in the BB only wiring where no 50v.
??? b*cat has to comment. In reverse order:
- The nominal 50 VDC is still present on the pair leading to the modem.
- The polarity of the pair is actually, nowadays, irrelevant as far as any connected equipment is concerned. (Although I always ensure that the B-wire is more negative than the A-wire.)
-
" The nominal 50 V DC is still present on the pair leading to the modem."
Quick check with ohmeter and a Microfilter agrees with you ! BUT WHY ?
That branch is supposed to be a High Pass filter which blocks DC in my language ? What is the point of extra components, which all cost money ?
"The polarity of the pair is actually, nowadays, irrelevant as far as any connected equipment is concerned. (Although I always ensure that the B-wire is more negative than the A-wire.) "
OK so it doesn't matter, thanks for the information.
-
" The nominal 50 V DC is still present on the pair leading to the modem."
Quick check with ohmeter and a Microfilter agrees with you ! BUT WHY ?
That branch is supposed to be a High Pass filter which blocks DC in my language ? What is the point of extra components, which all cost money ?
The only High Pass filter present in the circuit will be inside the actual modem. There is just a Low Pass filter within either a micro-filter or a centralised filter.
Attempted ASCII art:
---- Incoming pair ---- Low Pass filter ---- Telephony Circuitry
|
|---- Broadband Modem Circuitry
-
Corr !
Thank you very much for the clarification.
-
I love B*Cat ...... he knows everything and makes it easy to understand. :)
-
I love B*Cat ...... he knows everything and makes it easy to understand. :)
:blush: (Thank you for the complement.)
-
:hug:
-
How would I avoid reversing the polarity?
-
How would I avoid reversing the polarity?
The polarity does not matter. Hence do not worry about it.
However, if you are a perfectionist --
- Connect the positive terminal of your DVM to a good earth.
- With a probe connected to the negative terminal of your DVM, test the wires in the pair.
- The wire with the larger voltage reading is the B-wire and the wire with the smaller voltage reading is the A-wire.
- The B-wire should always "appear" on terminal 2 and the A-wire should always "appear" on terminal 5.