Kitz Forum
Broadband Related => FTTC and FTTP Issues => Topic started by: Bestgear on March 06, 2013, 08:18:11 PM
-
Guys
I just thought I would share my results from playing around with three variants of the OR modem.
My Infinity was installed with the ECI b/r. I got a stable 64mb down and 14 up.
I bought a ECi B/I rev B - swapped it out and it performed very slowly - even though the upload and download MB and latency was as low as before - BUT - any browsing was deathly slow with frequent time outs on many sites - just lik eyou get when theMTU is wrong (obvioulsy not changed). I unlocked it and it was the same (no surprise) - tried it on a friend line and it was 100% perfect - go figure!
I also bought a good old HG612 - thinking want to see stats and all round more friendly.
Boy - what a shocker - I lost 11mb of download speed and upload was 1mb higher!! A couple of days later my IP profile finds its way down to 53mb or so.... AGH!!!!
I also configured it as a router (rather than having my 1841 do PPPoE) and it works very well - I can only wonder why BT did not sell/provide an all in one modem router solution (obvioulsy model LAN ports and wireless). Thinking here is that if the 612 router was ok as a basic firewall, loosing more kit saves power (have a few separate APs to cover the house).
I appreciate the the later ECI is seen as the best modem in terms of line condition handling - but what i saw was considerable to the point that it *must* be beneficial for 612 owners to go try a later ECI.
Any and all comments on the above plus your own experiences most appreciated.
BG
-
My thinking is that your FTTC Cab hardware should be matched to the same Modem hardware
IE: Huawei Cab to HG612 and ECI Cab to ECI modem, but your tests seem to confirm this myth.
-
Hi
Well... I need to go for another look, but I sure my cabinet is not an ECI one.....
It does make me wonder why the b/i was so ponderous...
BG
-
Can you tell what cabinet I am on from this ?
Status: Showtime
Retrain Reason: 2
Max: Upstream rate = 15701 Kbps, Downstream rate = 52680 Kbps
Path: 0, Upstream rate = 14999 Kbps, Downstream rate = 53308 Kbps
Discovery Phase (Initial) Band Plan
US: (0,95) (868,1207) (1972,2783)
DS: (32,859) (1216,1963) (2792,3959)
Medley Phase (Final) Band Plan
US: (0,95) (868,1207) (1972,2783)
DS: (32,859) (1216,1963) (2792,3959)
VDSL Port Details Upstream Downstream
Attainable Net Data Rate: 15701 kbps 52680 kbps
Actual Aggregate Tx Power: 6.8 dBm 11.8 dBm
============================================================================
VDSL Band Status U0 U1 U2 U3 D1 D2 D3
Line Attenuation(dB): 6.0 30.6 45.8 N/A 15.3 38.0 58.7
Signal Attenuation(dB): 6.0 29.9 44.6 N/A 15.3 38.0 58.7
SNR Margin(dB): 6.4 6.5 6.4 N/A 5.8 5.8 5.8
TX Power(dBm): -4.0 -11.9 6.4 N/A 8.5 7.8 4.0
-
Can you tell what cabinet I am on from this ?
Discovery Phase (Initial) Band Plan
US: (0,95) (868,1207) (1972,2783)
DS: (32,859) (1216,1963) (2792,3959)
Medley Phase (Final) Band Plan
US: (0,95) (868,1207) (1972,2783)
DS: (32,859) (1216,1963) (2792,3959)
From the bandplan, you're on a Huawei DSLAM. See Bald_Eagle's (Paul's) research here:
http://forum.kitz.co.uk/index.php/topic,11002.msg215116.html#msg215116
cheers, a
-
From the bandplan, you're on a Huawei DSLAM. See Bald_Eagle's (Paul's) research here:
cheers, a
Thats great, thanks...so much for ECI only working better on a ECI cab then!
-
When some kit arrives, a few tests are planned on the half dozen VDSL2 modems to hand. The idea is to discover which modem/s perform best under various adverse conditions (loop length i.e. attenuation and background noise). Others have reported that the Lantiq chipset in the ECI performs well on shorter loops but not so good otherwise. Or at least that was the finding with ADSL2+, but maybe performance is different for VDSL2. It would be nice to answer that categorically. Certainly the Lantiq isn't as sensitive as the Broadcom chipset, even on the shortest loops. [1] It doesn't obtain such a good SNR margin under the same conditions.
cheers, a
[1] http://insidehuaweima5616msan.wordpress.com/2013/02/26/contrasting-zero-loop-stats-for-eci-i-revision-and-r-revision-vdsl2-modems/
-
The idea is to discover which modem/s perform best under various adverse conditions (loop length i.e. attenuation and background noise).
Is there a quick way for me to find the loop length?
I assume the standard Bt line test is the whole path to exchange not to cabinet.
BG
-
BT has its own infrastructure records, and the LLU OLOs have access to them too.
Some DSLAMs also have a loop length estimation tool (as part of their line testing tools - SELT/DELT/MELT).
The loop length estimation function in the Huawei DSLAM firmware doesn't seem to be very accurate.
Otherwise, time domain reflectometry equipment is needed. A TDR (pulse echo sounding) is used to discover the distance to the DSLAM transceiver, whether that is in a street cabinet or the exchange.
cheers, a
-
Otherwise, time domain reflectometry equipment is needed. A TDR (pulse echo sounding) is used to discover the distance to the DSLAM transceiver, whether that is in a street cabinet or the exchange.
cheers, a
I was hoping that there would be some info buried we could read and get that!
I have a mate that works in BT, so could ask him to use his JDSU which I assume will show the length.
I assume its only the home to cabinet distance we need... the fibre run will be irrelevant.
I have moved back tot he later ECI today and wham...11mb higher throughput.
I am surprised that my reported IP profile adjusts dynamically - or very quickly (overnight it seems to discon and recon at a different rate if I have been playing around) - is this normal?!?! I am lucky it does!
I was please too that the 612 seems to be perfectly capable of sustaining throughput when setup as a router even with simple firewall and nat enabled - I feared these would flatten the box. I am back to using the ECI and an 1841.
BG
-
I'll stick to my HG612 ;) on my longer loop, though if you wanna send me an unlocked ECI modem ill be very happy to test it out :)
-
I'll stick to my HG612 ;) on my longer loop, though if you wanna send me an unlocked ECI modem ill be very happy to test it out :)
What do you class as a longer loop?
I suspect I will be circa 400m to cabinet but should get that confirmed soon.
From my experience of 612, early and late modems, the later eci outperforms all that have gone before it...YMMV of course....
-
Some more tests were done this evening.. The results are as follows:
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww2.picturepush.com%2Fphoto%2Fa%2F12368935%2Fimg%2Fmodemtests_02_13%2Fgraph.jpeg&hash=6ec68c9d26854d6b55ddfa9e91bf6ed56f11f1b7) (http://picturepush.com/public/12368935)
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww5.picturepush.com%2Fphoto%2Fa%2F12376013%2F640%2Fmodemtests_02_13%2Fgraphdata.png&hash=156f18b0192e0010676f38aad3e17648fb38358b)
(click for full size) (http://picturepush.com/public/12376013)
The raw data is attached:
cheers, a
-
absolutely fabulous asbokid and thanks for those results, I guess thats why i was given a HG612 being 725 meters from cab.
-
The results are surprising and the reverse of what was previously understood from ADSL.
The Huaweis (all Broadcom-based) outperform on the shorter loops, but lose out to the Lantiq (and the Ikanos-based Sagem F@st2764) in the longer loop scenarios. for VDSL2.
The loop lengths are indicative but are somewhat arbitrary. Ideally, the tests would encompass wider ranges of attenuation. Artificial local loops created from variable attenuator blocks would be needed to simulate this.
It's nice to see the giant Broadcom Corporation is getting a run for its money. And its competitor Lantiq, a private-equity spin-off from Infineon, is nominally European, to boot.
Although Broadcom is just about to release two new System-on-Chips (SoCs). These are the BCM63168 and the BCM63268. The Product Brief for them, is, erm, very brief, but details a PCIe architecture whereas the BCM6368 supports just PCI. Otherwise, at least on paper, everything else looks the same. Whether performance in the field is improved will be interesting to see (and to test!)
cheers, a
-
Hi
Interesting stuff...but it still does not explain to me why different modems perform differently on the same line.
I tried a ECI b/i on a friends line that had been used with an unmolested early huawei and it surprised me and sync was 15MB less! His line length was just 103m.
Unless I am reading your graph wrong, the 612 I tested should have out performed the b/r on my (confirmed) 434m line but it could not get near it (circa 11mb less).
I will get another virgin huawei tomorrow and retest my line and report back.
Weird....
BG
-
I also configured it as a router (rather than having my 1841 do PPPoE) and it works very well - I can only wonder why BT did not sell/provide an all in one modem router solution (obvioulsy model LAN ports and wireless). Thinking here is that if the 612 router was ok as a basic firewall, loosing more kit saves power (have a few separate APs to cover the house).
I don't have VDSL so cannot comment from personal experience but....
I'd imagine that having an all in one modem/router is 'pushing the envelope' a bit in terms of getting the CPU etc to manage the VDSL signal/bit swaps/margin/sync whatever, and act as a wireless access point, and manage the DCHP/routing stuff at 1G speeds, and run the firewall....not forgetting all the other housekeeping stuff running in the background.
By separating out the modem only bit the BT unit can cope with managing the line only aspect on its own without running into capacity problems.
For an all-in-one "home gateway" type application unit which could manage everything with ease at full VDSL and LAN/wireless speeds I'd guess we need to start looking at a netbook level of computing power to be available.....and methinks this sort of stuff is simply not yet available in the domestic mass market.
-
I'd imagine that having an all in one modem/router is 'pushing the envelope' a bit in terms of getting the CPU etc to manage the VDSL signal/bit swaps/margin/sync whatever, and act as a wireless access point, and manage the DCHP/routing stuff at 1G speeds, and run the firewall....not forgetting all the other housekeeping stuff running in the background.
That is in agreement with my understanding of the situation. Beattie's wizards at Grimbledon Down, having performed their experiments, decided to provide the service (for all CPs/ISPs to sell) from an Ethernet (WAN) port. Hence the definition of the modem as the active NTE for VDSL2 based services. This concept follows the precedence of the technology of the earlier Megastream and Kilostream services.
-
I'd imagine that having an all in one modem/router is 'pushing the envelope' a bit
Using the 612 as a modem/router rather than just as a modem did not exhibit any loss of throughput on my line, so should be an ok zero cost option for those that dont want a HH (for whatever reason).
What is hard work for a router is when it is managing VPN and firewall rules (ACLs not just simple NAT usually). Wireless is usually handled by a separate chipset, and are just hung off the LAN side.
It is amusing to see the quality and performance of home vdsl kit really - when you consider that even a lot of cisco kit of not that long ago would struggle to traverse the bandwidth we have in home vdsl!
Personally, I use a wee cisco router so I can have vpns and proper control of rules... and of course ditching the HH3 means that I loose the BT public wifi service without switching off my opt-in!
BG
-
Concidentially someone on the Plusnet forums has just today posted to the effect that their experiments on using the 612 as a modem/router rather than a modem only does indeed affect the download speeds - they are obviously close to the cabinet if they are getting 74Mbps download achieved.
Their conclusion is the same - the 612 simply cannot cope.
http://community.plus.net/forum/index.php/topic,112626.msg969073/boardseen.html#new (http://community.plus.net/forum/index.php/topic,112626.msg969073/boardseen.html#new)
-
Not being a VDSL2 service user, I cannot comment on usage of the Huawei HG612 as a modem/router for that FTTC based product.
However I have now used a HG612, a HG610 and a HG622 on a 'sync speed restricted by E- + D-side length' ADSL2+ service. Of the three, I rate the HG622 the best, then the HG612 and finally the HG610 in performance terms.
-
The results are surprising and the reverse of what was previously understood from ADSL.
I'm going to re-do these tests in greater depth. Resetting the linecard chipset between modems (to eliminate DLM issues), testing each modem multiple times, and best matching the line-profile used by BT, as much as we understand it.
Any suggestions for conducting new tests would be appreciated :-)
Cheers, a
-
1. You might try adding some cross talk e.g. if you have short untwisted section with an extra modem running at the same time.
2. If you have access to more than one of the same type of modem (perhaps you still have a few HG622) it would be interesting to test a few of them to see the variance in "identical" modems
I had, with help from two neighbors four DG834gv4's all running the same firmware. The sync was at about 16Mb/s and consistently varied between them. One was best - :) mine as it happens and one worst 1MB/s slower also mine :( . The other two were about 0.3 and 0.6 Mb/s slower than the best one. A rough variation of plus and minus 0.5Mb/s on the same chipset.
-
Any chance of simulating aluminium phone lines?
-
Thanks guys, keep the ideas coming!
Cross-talk can be simulated with Gaussian noise from an arbitrary waveform generator - the Hantek DDS-3x25. [1] Though it's not straightforward connecting the generator with its 50ohm coax (unbalanced) output to a balanced twisted pair with 100ohm impedance. Experiments continue ???
Certainly various levels of attenuation, including those found in aluminium pairs, can be modelled with passive attenuators (low-noise resistor-based pads). [2]
Another 'issue' is the time that a line is left to stabilise after synchronising. Five minutes, ten? Sixty? It's limited only by patience!
cheers, a
[1] http://www.hantek.com/english/news_list.asp?unid=13
[2] http://insidehuaweima5616msan.wordpress.com/2012/11/13/simulating-loop-characteristics-to-test-cpe/
-
will the SNR be taken in to consideration during tests as you know after 3 pm this seems to start to get lower in the evenings & syncing each modem at different times of the day can cause 1-2 Mbps a of difference in throughput, you need some sort of minium like 2Mbps steps in the graphs ;)
-
Just reading the above, and if I understood the point mentioned correctly, my information from my past, BT lines, unless they are loaded junction plant, which I doubt now exist, had an impedance of 1K2. Unloaded cable is just one great capacitor, which makes it a terrible transmission line. Unless we go back to the stone age and use old fashioned cadmium copper wires separated a few inches apart, having inductance and capacitance etc in a matched situation then the line would provide a 600ohm impedance. So using 100 ohm as a test line may not be totally accurate
-
Hi Benji09,
Ooh, thanks! Glad someone knows what they're talking about!
The plan is to replicate for VDSL2 modems some earlier tests for ADSL performed by Mattias Ernelli, an intern at Telenor, a Scandinavian telco.
Ernelli used an arbitrary waveform generator (AWG) to inject artificial noise into the loop under test. This from page 46 of his thesis [1]
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww5.picturepush.com%2Fphoto%2Fa%2F12393418%2Fimg%2Fmodemtests_02_13%2Fernelli-balun.png&hash=f757110c3522b65aa2713a905145e793cdb5d518) (http://picturepush.com/public/12393418)
Ernelli is using two 1:1 100ohm baluns to connect the noise generator. I must admit, it's an area about which I know next to nothing :-[ How does it work?! Any clues?!
EDIT: there's a short Application Note from Agilent on ADSL Copper Loop Measurements. [2]. The authors of the note write:
The system impedance in the POTS band is 600 Ω and in the ADSL band is 100 Ω. Meanwhile, a measurement instrument like a network analyzer typically is unbalanced and it has 50 Ω port impedance. You need to use a BALUN to match both of the system impedances. For example, a 50:100 BALUN should be chosen to measure cables in the ADSL band.
cheers, a
[1] http://www.ee.kth.se/php/modules/publications/reports/2008/XR-EE-KT_2008_003.pdf
[2] http://cp.literature.agilent.com/litweb/pdf/5968-1196E.pdf
-
Wonderful stuff asbokid.. really interesting research.
By chance do you know if the Revision of the HG612 make any difference as I've just come across a 3B, I already have a 2B version.
-
Hi Dazzlew,
Ernelli's thesis is brilliant. As for the two modems, performance seems to be identical. The 3B has physical modifications to the heatsink / BGA fabrication to try and defeat the earlier overheating problems. Seems to work. Other than that, between the 2B and 3B, there's no obvious difference in xDSL performance.
cheers, a
EDIT: heck, it's Typo Tuesday!
-
"Ernelli is using two 1:1 100ohm baluns to connect the noise generator. I must admit, it's an area about which I know next to nothing :-[ How does it work?! Any clues?! "
Balun is short for balance to unbalance transformer - (or should be it tends to also get used for devices like the transformer in the I plate, which works in a balanced line and acts as a high series inductance to an unbalanced current with negligible effect on balanced)
The simplest form is a 1:1 transformer, centre tapped on the balanced side. The centre tap and one side of the unbalanced winding are earthed. With clever transformer construction this will convert a balanced line to unbalanced, or vice versa over a considerable frequency ratio. ( For ADSL2+ the ratio is 84.5:1)
This is the simple and comprehensible form; at HF/ VHF/Microwaves MANY other variants are possible.
Regarding another point, a modem is a bit like a radio receiver, which can be designed for maximum sensitivity, e.g. for a radio telescope, or with less sensitivity to avoid overload and distortion by a strong signal - you can't have both at once and building in adaption costs money. Yes, it would be nice if someone told you what you were buying.
-
Thanks JGO...
What specifically remains beyond my understanding is how two 100:100 BALUNs placed back to back can impedance match the output of the 50ohm noise generator.
This looked like a suitable BALUN for connecting the unbalanced 50ohm output of the signal generator to the 100ohm balanced twisted pair of the cable under test. Except, the price of that BALUN ($150) renders it completely incompatible for the budget! It might be easier, and certainly more profitable to wind one by hand!
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.northhills-sp.com%2Fstore%2Fmedia%2Fimages%2Fproduct_xlarge%2F0301BB.jpg&hash=b9667a9e37e3e63bfd6261f1137b9e090cb8328a)
From: http://www.northhills-sp.com/store/cgi-bin/cp-app.cgi?rnd=9535125&pg=prod&ref=0301BB
cheers, a
-
http://www.ebay.co.uk/itm/North-Hills-NH14023-Balun-Transformer-50-to-100-Ohms-300MHz-/190808888099?pt=UK_BOI_Electrical_Test_Measurement_Equipment_ET&hash=item2c6d189323
See above -- they are cheaper on ebay but I think they really just ensure an impedance match and good transmission with no nasty reflections. Video ones are very cheap (99p) but I suspect the impedance is wrong. As a tiny lad I once made one to connect an aerial to a valve radio (before transistors were much used!) -- that did not end up working very well!!
I think the basic setup would allow "relative" testing of the modems with no balun and just the transmission line with noise injection in the middle.
-
Les - are you saying the device covers 300 MHz to 10 KHz ?!!!!!! I wouldn't have thought that is possible.
" What specifically remains beyond my understanding is how two 100:100 BALUNs placed back to back can impedance match the output of the 50ohm noise generator."
Me too !
I would be inclined to look at for unsymetrical resistive attenuators so you can separate the impedance a device looks like and the load impedance it sees. As Les says you can do a relative test that way, just allow for the attenuation.
I have used this (once) for matching a 75 ohm source and a 50 ohm receiver, but it seems more complicated with 3 devices; or I've forgotten how to do it in the intervening years !
-
Thanks guys for the advice. By coincidence, a couple of those 75ohm video baluns (50p each) were ordered from China, just to see what they did. That was over a month ago! When they eventually arrive, I'll cut one open and photograph it. ??? The North Hills BALUNs are still a bit expensive,for what is just an amateur experiment, but maybe they're the more professional answer. There is/was a course given at Univ. New Brunswick on performing these sorts of tests:
http://www.ee.unb.ca/Courses/EE3822/BC/Lab_3.pdf
cheers, a
-
First point I want to make is that ` POTS ' has not been 600 ohms since year dot. Underground cable put paid to that. Old type ASTIC telephones had to have a balance network to simulate a UG cable to cancel out sender's speech in the earphone. That balance was NOT a nice 600 ohm resister.
A 1:1 transformer will reflect the impedance that is on one side to the other side. In this case, 50 ohms in parellel with 100 from one side, to the other. As I see it the transformers, although required to deal with the generator's unbalance problem, are in this case a bit of a red herring to the understanding of the way the test set up works due to the 1:1 turns ratio. My thought is why the need for two of them. Not only do transformers EACH normally considered to introduce a 0.5 dB loss, they don't operate over an unlimited frequency range.
I personally would think at one transformer on the output of the generater, and TWO high value resistors in each leg of secondary winding to provide both generator balanced output, and a form constant current drive to the line with it's varying impedance over frequency. Would be interested in members comments to my comments....
-
My understanding is that the transceivers at each end of the link operate in differential mode. The noise generator needs to be coupled into a balanced pair such that its signal is not rejected by the differential circuit at the receiving end.
Consider the balanced transmission line stretching across the screen, just like this sentence. At the midpoint of each wire of the pair, insert a 1:1 transformer in series with the wire. For one of those transformers, the second winding is connected to the signal generator. For the second transformer, the second winding is left unused. How does that appeal to the electronics wizards?
T1
-----ooooo-----
A --------------------------------------ooooo----------------------------------- A
Twisted Pair Twisted Pair
B --------------------------------------ooooo----------------------------------- B
-----ooooo-----
| T2 |
| |
| |
- Noise Gen -
-
I personally can't see the need to inject a noise signal in series with the line, as the differential bit means that one leg of the pair is pushing, whilst the other leg is pulling a signal around it ! If the signal was injected into both legs of the pair in the same phase, then as you say, the interfering signal would cancel itself out at the modems. The fact that the differentially injected signal goes to both ends of the circuit does not stop the signal from being injected......... If my assumption was not correct, the there would never be a problem with crosstalk on cables.
I must confess that I have not had the time to study all of the postings, or the original university paper on this topic, but in what I have seen, I am not convinced that a number of the assumed facts were correct in the paper. The other point I wish to make is that if the line were a Virgin Media using coax cable to the cabinet, it would be a proper 75 ? ohm transmission line that would be terminated with the correct impedance. Therefore test equipment would be available for the correct impedance to get meaningful results.
But as I suggested in an earlier post underground cables behave more like a capacitor than a proper transmission line. Therefore the impedance will tend to drop with frequency. Due to this fact, how do you measure what power of noise signal you are actually injecting into modem, in relation to the signal being received.
In my opinion, because of BT lines being of varying characteristics, perhaps the best way of evaluating the different modems is to just to compare them in the field on various lines. In fact I am sure that BT has probably carried this out at some point already, and already know the answer !
-
Methinks that the schematic diagram (Figure 3.4, on the previous page and attached below) does not tell the truth of what was actually used. :-\
-
I took the plunge and bought a secondhand 100:50 BALUN from ebay! Ironically, the seller works for a company that performs these very tests! :o
There are a number of other "curiosities" in that master's thesis. The calculations for selecting resistor values to create specific loop attenuations do not match my own sums. At best, the calculations in the thesis seem over simplified:
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww2.picturepush.com%2Fphoto%2Fa%2F12418725%2F480%2Fmodemtests_02_13%2Fmodembackgnoise.png&hash=4e9d18aef81252540947937b7181771c05c0714e) (http://picturepush.com/public/12418725)
As for measuring noise, the $100 Hantek noise generator planned for these tests comes with Windows-only software ::) But if that handicap can be overcome, precise levels of white noise (dBm/Hz) will be injected into a zero-loop. Allowing the response / compensation from each modem to be measured.
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww1.picturepush.com%2Fphoto%2Fa%2F12418874%2F480%2Fmodemtests_02_13%2Fdds-3x25-screenshot.png&hash=6d07d3b16edf6b2e077b96a2bc9d7d93ac975c67) (http://picturepush.com/public/12418874)
cheers, a
-
I took the plunge and bought a secondhand 100:50 BALUN from ebay!
I am wondering how you propose to connect it to/into the zero-loop. Have you yet decided? ???
-
The "zero" loop will be a couple of short lengths of cat5e twisted pair, with an RJ11 crimped on one end of the first length, and the other end of the second length will be punched into the IDCs of the DSLAM patch panel. The other ends of the two lengths will be stripped of insulation, and probably soldered together. Oops, correction! A BNC to BNC will connect the noise generator to the BALUN, and the BALUN will probably be soldered to the twisted pair of the zero loop. I'll take some photos when it arrives!
cheers, a
-
I see it! Thanks for the description. :)
-
Hope you don't mind me asking a question. My connection is via a long line 800m-900m, supplied with an HG612 and have a matching DSLAM in the cabinet, would I be right in thinking reading through this thread that I might be better off using an ECI V-2FUb/r Rev B with this length of line?
-
The tests, at least the tests I performed aren't to be relied upon. They weren't conducted scientifically. For example, the linecard/port should have been reset between each modem - to flush any residual DLM behaviour from previous tests. At some point the tests can be performed again, but this time under "laboratory conditions" ! As such it's probably best not to read anything much in to them!
As an aside, they are surprisingly difficult tests to frame. A modem can be synced-up half a dozen times, with static line conditions, and yet it will perform differently on each occasion. Discovering why that happens, if not because of DLM, is crucial!
cheers, a
-
OK, I understand, thanks for explaining. I won't go and buy that ECI modem just yet then, its great to see so much work and independent research going here. I'll keep reading anyway.
-
. . . I won't go and buy that ECI modem just yet . . .
Please don't do that. :no:
As you have now probably noticed, there are two subtly different variants of the ECI B-FOCuS modem deployed by Openreach as an active CPE for its NGA GEA VDSL2 service. The V-2FUb/I Rev. B and the V-2FUb/r Rev. B. Both devices are still not able to divulge statistics in the quantity and format that is available from the Huawei HG612. :(
-
Just to add my own recent experience with this: My DSLAM is Huawei. The modem the contractor brought (first thing to hand in the van I suspect) was a /r ECI. I also 'happen' to have an HG612. The cabinet is ~150m away.
I saw no significant difference in attainable rates between either the ECI and the Huawei, but since these are (fortunately) > than my 80/20 IP profile, it really wouldn't matter much to me if they were, because I couldn't benefit from it, and as a consequence I prefer the HG612 for now. I appreciate that it will to many whose attainable is < their IP profile. :(
Having used the HG612 successfully (like many others) as a router for ADSL2+ @ ~14Mb/s, I would say it is powerful enough in those circumstances, but I agree with those that doubt its ability to do so with (say) 80/20 VDSL2.
My attainable rates are slightly lower (but not that much) than predicted by Asbokid's testing, but then in my urban environment my QLN noise floor seems to be ~20dBM higher across the spectrum at ~-120 dBM than others I've seen, which tends to explain that. What I can do about that short of BTOR implementing vectoring, I don't know. ;D
-
Just got round to connecting a noise source to the twisted pair of a (private) local loop. The noise source is a USD100 USB-based Hantek DDS 3x25 Arbitrary Waveform Generator. [1]
Two software applications for controlling the AWG were tested: the original Hantek program. see above, and the Goltek software by SOA System Engineering:
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww4.picturepush.com%2Fphoto%2Fa%2F13231617%2F640%2Farbitrary-waveform-generator%2Fgoltek-Screenshot2.png&hash=2620929ca168c3f7bc53d2f0641ab02ca0d3b40c) (http://picturepush.com/public/13231617)
To interface the 100 ohm balanced transmission line to the 50 ohm unbalanced coax output of the generator required a wideband BALUN.
In this configuration, it is possible to generate Gaussian and/or White noise in the range of 2uV to 7V peak-to-peak. Although to ensure flatness across a bandwidth of 17MHz (Profile 17a), the generator has a maximum output amplitude of 3.7Vpp.
While testing an HG622 with injected line noise, some graphs of Quiet Line Noise were created. The HG622 (and no doubt the HG612 and other BCM6368 devices) will still sync in the presence of an incredible amount of line noise. The 6368 is an extremely resilient device.
Though, unsurprisingly, bit-loading drops considerable in the presence of greater and greater levels of injected white noise*. It's actually quite hard to stop the device from syncing altogether.
Shown below is the QLN for a zero-loop. Firstly, with a natural noise floor (including some broadcast AM RF ingress) and, secondly, with the injection of 1Vpp of white noise into that bandwidth of 17MHz.
That power level of noise, equating to approximately an extra 35 dBm/Hz above the floor, brought down the attainable rate. Consequently, on resync, the actual sync rate fell from a banded 80 Mbps connection to just 22Mbps downstream.
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww5.picturepush.com%2Fphoto%2Fa%2F13231618%2F640%2Farbitrary-waveform-generator%2Fagif.gif&hash=afffe040031057fe38cf911fbe734adb84d0c694) (http://picturepush.com/public/13231618)
* increased white noise is apparently how crosstalk manifests from DSL disturbers in the same cable binder.
[1] http://www.hantek.com/english/news_list.asp?unid=13
-
Great stuff Asbo, really! :thumbs:
If increased white noise is how crosstalk manifests from DSL disturbers in the same cable binder
and the effect of increased white noise (and so FEXT) is to raise (significantly?) the QLN noise floor, then doesn't this give us a tool for detecting the (potential) presence of increased FEXT?
i.e. if there is no change in the QLN noise floor, then there can have been little change in white noise on the line, and so no observable dramatic change in FEXT, or is that a false reverse inference? I hope not. BE would be pleased to have such a (virtual) tool with which to examine his stats since last December. :thumbs:
-
Hi Colin,
I've been following the inspirational research of Mattias Ernelli [1], replicating it, where possible. Ernelli was an intern at Telenor, the incumbent Norwegian telco and studied loop qualification for VDSL2.
In the case of VDSL2 disturbers in the same cable binder, those disturbers will occupy exactly the same spectra as the "victim" loop (~25kHz to 17MHz). Such that there is a flat distribution (the flatness definition of white noise) across the (Profile 17a) bandwidth from FEXT. Ernelli writes that the noise from crosstalk should also be modelled as Gaussian - having a bell-curved distribution to its amplitude.
So maybe these humble tests need re-visiting with those observations in mind.. injecting Additive Gaussian White Noise (AGWN) into the test loop. Or, ultimately, and as a last resort, maybe eight or so modems can be linked via the same (cat6) binder, and synced up, to study the outcome of crosstalk by trial and error!
cheers, a
[1] http://www.ee.kth.se/php/modules/publications/reports/2008/XR-EE-KT_2008_003.pdf
-
Interesting findings there, asbokid.
FWIW, I have attached montages of my connection since December 2012 & the present time.
The 'stepped' speed & power reductions can be clearly seen.
As these reductions coincide with a 'stepped' deterioration in my QLN graphs & SNR (not SNRM) graphs, my/our assumption was it was a result of increased crosstalk.
Could it actually be anything else?
Strangely though, DS errored seconds increased prior to the completely unexpected removal of Interleaving & have continued ever since, going from around 100 per day in December to currently between around 600 & 800 per day.
Has the reduced power (particularly in the D2 band) caused the reduced speed or is it a consequence of the reduced speed?
Also, my connection resynced at at the most unusual times of 22:30 last night & again at 07:30 this morning.
DLM initiated resyncs usually tend to occur between 04:00 & 05:00 on my connection.
-
Just got round to connecting a noise source to the twisted pair of a (private) local loop. The noise source is a USD100 Hantek DDS 3x25 arbitrary waveform generator. [1]
<snip>
To interface the 100 ohm balanced transmission line to the 50 ohm unbalanced coax output of the generator required a wideband BALUN.
<snip>
[1] http://www.hantek.com/english/news_list.asp?unid=13
I wonder if a photograph of the loop, with a balun for noise injection, will shortly be forthcoming? Or failing that, a sketch of the circuit, please? ???
-
Sure, I'll take some photos (after disguising my pathetic soldering!)
It's not very exciting, though. Just a 'T' joint between the 50 ohm unbalanced coax output of the AWG, and the two 100 ohm balanced twisted pairs. Two pairs, with one pair going to the DSLAM, and one going to the VDSL2 CPE. An RJ11 plug on each end, as per the photo below.
As for the BALUN, it is from North Hills (NH14023):
See: http://docs.google.com/file/d/0B6wW18mYskvBeldxVDhybkFhY00/edit?usp=sharing
EDIT:
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww2.picturepush.com%2Fphoto%2Fa%2F13242120%2F640%2Farbitrary-waveform-generator%2Fawg-and-balun.small.jpg&hash=faa3c69ef980c460b78d3057cdcc53197fdb7db2) (http://picturepush.com/public/13242120)
cheers, a
-
After a couple of days toying with this, it is clear that injecting measured amounts of noise into a local loop is something of a black art!
For this experiment, enough Gaussian noise has to be injected to induce line errors (to test the modem's response). But too much noise causes an irretrievable framing loss and ultimately loss of sync. However, there's a very fine threshold between those two noise levels.
Furthermore, these noise tests using the Hantek AWG, above, exhibit something of the Boiled Frog syndrome! That is: "start with a cool pan, slowly raise the heat, and the frog won't even notice!". That seems to be the case with line noise, too!
Suddenly injecting a high power level of noise almost certainly causes a lost connection with immediate effect. One loss of framing signal is enough.
By contrast, there is 'success' when the line is synced with no noise. Only then is the noise slowly increased. Starting with just a few microvolts amplitude, the noise can be raised to several hundred millivolts over the time span of a few tens of seconds or minutes. Presumably as the CPE furiously bit-swaps in search of better carriers.
So what? Who cares?
Well, we might speculate that this behaviour has some relevance to the effects of crosstalk noise, as VDSL2 is rolled out further.
When a VDSL circuit is newly commissioned, it's possible that it will act as an immediate and severe 'disturber' on its neighbouring pairs. Inducing large amounts of Gaussian noise into them from the moment that the new port is activated at the DSLAM. There is nothing in the VDSL2 specs (sfaik) for building up transmit power on a slowly-slowly basis. So what is going to happen in those cases?
This is pure speculation, but it might be found that as a new VDSL2 circuit is activated, the crosstalk it creates causes the immediate de-syncing of (borderline) neighbours..Those lines then re-sync, knocking out the new circuit. Which then tries to resync itself. In a ping-pong effect! Just speculation though. No evidence of that actually happening, yet!
Though what has been seen is as follows: the Gaussian noise power level is incrementally raised (simulating disturbance from increased crosstalk). The actual net data rate remains static, a property of the DLM configuration. However, the attainable rate still drops. Causing the Relative Capacity Occupation (RCO) , a percentage measurement, to rise, quite possibly above 100%. Meanwhile, with the noise floor raised, the SNR margins, especially in the higher downstream bands, are seen to wither downwards, towards zero.
As those SNR margins are slashed, and with that static actual net data rate, the error rates creep up. Eventually, the DLM algorithm has to intervene. So it re-profiles the line with lower transmission rate bands, and the SNR margins rise again.
Sfaics, those stages could mark the typical effects of crosstalk at the CPE as more and more subscribers sharing the same cable binder opt for a VDSL2 service.
cheers, a
-
Sfaics, those stages could mark the typical effects of crosstalk at the CPE as more and more subscribers sharing the same cable binder opt for a VDSL2 service.
That does seem to pretty much describe what I saw when I noticed the first downward sync speed step in my connection 31st December.
The reduced SNRM persisted for a few days before the connection resynced at lower speed with SNRM back to normal.
However, this happened quite suddenly as did the increase in various error counts.
It does seem to have affected both the D1 & D2 bands to the same extent.
I can't actually make use of the D3 band due to distance from the cabinet, hence no data for it.
-
That does seem to pretty much describe what I saw when I noticed the first downward sync speed step in my connection 31st December.
The reduced SNRM persisted for a few days before the connection resynced at lower speed with SNRM back to normal.
31st Dec (2012) is such a strange date for a VDSL2 service to be activated (if that's why you measured a sudden reduction in SNRM on your line). Is it feasible that someone did actually have their connection activated on that date? It was a Monday but was Openreach open for business over that period for non-essential work?
cheers, a
-
31st Dec (2012) is such a strange date for a VDSL2 service to be activated
TBH, that date is what is making me hesitate in outright declaring this as a crosstalk issue.
The QLN graph shows a corresponding stepped deterioration since then (suggesting crosstalk), but a visiting engineer didn't confirm it & the matter (& the lack of any 'proper' explanation) has been flagged by Plusnet as requiring another visit.
-
Perhaps BlackSheep could clear up the speculation. Would Openreach have been operating on the morning of New Year's Eve on non-essential work such as activating a VDSL2 service?
The time of day (just after 8am, the start of the Openreach working day) does suggest that the slashed SNRM was caused by Openreach work rather than by some arbitrary line fault.
cheers, a