Print

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

[burakkucat]

As Wolfy (Howlingwolf) has released unlocked variants of Beattie's latest firmware image for the Huawei HG612 [1], the time was ripe to take the webgui image for a "test run".

My testing HG612 was flashed with the bcm96368MVWG_fs_kernel_HG612V100R001C01B030SP08_Webgui firmware image. It was then connected via a 1.5 metre "local loop" to a Planet VC-230N, the latter device being configured to operate in CO mode.

The VC-230N was powered up and allowed to stabilise. After two minutes had elapsed, power was then applied to the HG612. A further two minutes were allowed to elapse, so that the latter could also stabilise.

A sequence of screenshots were then captured. Two, showing the device configurations, are attached below.


[1] https://mega.co.nz/#F!LdJFDIJL!e_E1twsIg2kTet8mPjrb4w

[burakkucat]

Two further screenshots, showing both devices' status, are attached below.

[burakkucat]

Finally, a telnet connection was established to the HG612 and the usual data was harvested. Bald_Eagle1 then took the harvested data and produced his usual collection of graphical data.

Both the raw harvested data (as a ZIP format file) and the portrait montage are attached below.

[NewtronStar]

and waiting for your conslusion   

I think the new 2014 firmware has zero impact in performance unlike kernel_HG612V100R001C01B030SP06 which did improve FTTC somewhat

Until B*Cat can test his model in the real world on FTTC your 1.5 meter local loop test is  unrealistic 

[burakkucat]

Eleven months ago I attempted to gather statistics for the then available HG612 firmware images. [1] All was good for the ultra-short local loop (1.5 m) but the results for a longer loop were seriously degraded by a spool of copper coated steel cable. I made a note to source some new all-copper cable but never got around to doing so.

Last month I collected and shared the data relating to the latest firmware image from Beattie, once it had been unlocked and given back its GUI by Howlingwolf. N*Star's comment (the previous post) was still in my mind when I saw an offer of two 100 m spools, 0.5 mm copper conductors, three pair, CW1308 specification cable on a well-known auction web-site. The price? £11-99

The two spools of cable were delivered on Thursday afternoon. The first test, with a magnet, was passed. Not copper coated steel. The second test, a good scrape of a conductor with a sharp blade, was passed. Not copper coated aluminium.

Batteries were fitted into the Tester 301C, the green pair of one cable was shorted at the far end and the near-end was connected to the LINE sockets of the 301C by 1.6 m test leads. The near-end of the blue pair was connected to the REF sockets of the 301C by an identical set of 1.6 m test leads. With the BAL control set off, a very satisfying trace for a short-circuit was seen. Look closely at the second image (attached below) and you can see that the indicated length is slightly more than 104 m. The next increment, to 105 m, positioned the cursor to the right of the deviation from horizontal by as much as it was to the left, when set at 104 m. So the measured length of that spool of cable is 104 m "plus a bit". Remembering to subtract the length of the test leads, 1.6 m (i.e. 1 m "plus a bit") from the measured length, gave me a result of 103 m.

Next the Ohmmeter 18C was used to check the insulation resistance and conductor resistance of that spool of cable. Each pair (green, blue and orange) showed a very healthy insulation resistance greater than 1000 MOhms. Each pair showed an identical 18.4 Ohms value when shorted at the far end (over 206 m of conductor). Each individual conductor (of each pair) showed 9.5 Ohms (over 103 m of conductor). It was very satisfying to obtain consistent results for all the tests, for they showed the first spool of cable to be in excellent condition. I can therefore say that each of the six conductors has a resistance of 0.09 Ohm/metre.

Obviously I still need to check the second spool of cable, to determine its electrical length and condition. Once that has been done I should then be able to perform six separate experiments, with local loop lengths of 103 m, 206 m, 309 m, (309 + L) m, (309 + 2L) m and (309 + 3L) m, where L is the electrical length of the second spool of cable. I am assuming, of course, that fellow kitizens will be interested to see the results? 
[1] http://forum.kitz.co.uk/index.php?topic=13160.0

[NewtronStar]

I am assuming, of course, that fellow kitizens will be interested to see the results? 

I am very interested to see the results 

[Ronski]

Wouldn't using the three pairs joined in the single cable and still coiled up cause interference?

What about trying 103 + L? Then it can be compared to 103 + 103 in the same reel. Hopefully reel 2 is also 103 meters.

[roseway]

I'm certainly interested in the results. Measurements under rigorously controlled conditions are very much needed, to balance the anecdotal evidence which we so often depend on.

[Bald_Eagle1]

I too would be interested in seeing the results.



Just thinking aloud.......................................

Another 'interesting' experiment at some stage might be this:-

Run a generated VDSL2 signal down one pair & a live ADSL signal down another pair, sort of simulating differingent DSL modes carried within a D-side cable bundle.

Connecting & disconnecting the ADSL & VDSL2 signals in turn, logging the effects accordingly might/might not provide some indication of how severe crosstalk really can be?

[burakkucat]

Thank you all for the comments and suggestions. 

Ronski -- The self-interference by local coupling is something that Asbokid was uncertain about and he & I concluded that it could skew the results obtained but in a consistent fashion. I like your proposed experiment in using one pair from each spool in series and comparing the result with that of using two pairs from one spool in series.

Eric -- I share your view. Some results from a well-defined experimental set-up will be better than none at all. I can control the experimental environment quite carefully but I am limited with the equipment I have to paw.

BE1 -- A good proposal. Unfortunately that will be something I can not perform. The Planet VC230N only operates in VDSL2 mode and only has one "line port". To perform your suggested experiment would require usage of a DSLAM with two different line cards installed. Again, if I am remembering correctly, that was on Asbokid's ToDo list -- using his Huawei SmartAX MA5616 DSLAM.


I have now quantified the second spool of CW1308 cable. Just like the first spool, it is marked as 100 m of 0.5 mm diameter solid copper wire, 3 pair cable.

This time I connected one open-circuited pair to the LINE terminals and the short-circuited pair to the REF terminals of the Tester 301C. With the BAL control set off, a very satisfying trace for an open-circuited pair was seen. Look closely at the second image (attached below) and you can see that the indicated length is slightly less than 110 m. The previous increment, to 109 m, positioned the cursor to the left of the deviation from horizontal by as much as it was to the right, when set at 110 m. So the measured length of that spool of cable is 109 m "plus a bit". Remembering to subtract the length of the test leads, 1.6 m (i.e. 1 m "plus a bit") from the measured length, gave me a result of 108 m.

Once again the Ohmmeter 18C was used to check the insulation resistance and conductor resistance of that spool of cable. Each pair (green, blue and orange) showed a very healthy insulation resistance greater than 1000 MOhms. Each pair showed an identical 18.6 Ohms value when shorted at the far end (over 216 m of conductor). Each individual conductor (of each pair) showed 9.5 Ohms (over 108 m of conductor). As previously, I can therefore say that each of the six conductors has a resistance of 0.09 Ohm/metre.

[Bald_Eagle1]

BE1 -- A good proposal. Unfortunately that will be something I can not perform. The Planet VC230N only operates in VDSL2 mode and only has one "line port". To perform your suggested experiment would require usage of a DSLAM with two different line cards installed. Again, if I am remembering correctly, that was on Asbokid's ToDo list -- using his Huawei SmartAX MA5616 DSLAM.

TBH, my thinking aloud included a somewhat veiled 'hint' that someone with maybe a Planet VC230N for the VDSL2 element & a real live ADSL connection (possibly even provided via TalkTalk) may be willing to use their ADSL connection as the temporary donor for the other element of any such experimentation 


BTW, the word "differingent" as used in my previous message was an accidental Eagleism - an example of multi-tasking at its worst 

[burakkucat]

I am gradually compiling a list of experiments to perform:

• With the default CO configuration of the VC-230N and a 1.5 m local loop. (Already done. Result above.)
• With the default CO configuration of the VC-230N and a (103 + 108) == 211 m local loop. (The Ronski experiment, part 1.)
• With the default CO configuration of the VC-230N and a (103 + 103) == 206 m local loop. (The Ronski experiment, part 2.)
• With the default CO configuration of the VC-230N and a (108 + 108) == 216 m local loop. (The Ronski experiment, part 3.)
• With the VC-230N set to fastpath, the target SNRM set as small as possible and a 1.5 m local loop. (A Konrado5 experiment.)
• Reverting back to the default CO configuration of the VC-230N, then setting a rate-limit of 80 Mbps US and 20 Mbps DS. (Note: I have not got my US & DS mixed up, we are considering the CO end, not the CPE end, of the circuit.) Using a:
  
  
  • 103 m local loop.
  • 206 m local loop.
  • 309 m local loop.
  • 417 m local loop.
  • 525 m local loop.
  • 633 m local loop.

For each experiment I propose to attach the raw harvested ASCII text as a ZIP format file and, hopefully Mr Eagle will be able to provide the corresponding snap-shot graph montages.

I am hoping that the six different local loop lengths of experiment no. 6 will show a linear relationship and, perhaps, provide something that could then be extrapolated to "real" VDSL2 circuit parameters.

As I do not own a pair of crimping pliers and a tub of gel-crimps, I propose to make all the joints via "chocolate block" screw terminals. Each end of the loop will be fitted with an LJU2/3A (using IDCs) and a pair of RJ11 to 431A cables will connect the VC-230N & the HG612 devices to the circuit.

[burakkucat]

Adding some further information, before I begin the testing, so that those interested can comment, etc.

I have now defined three separate configurations for the VC-230N:

• Basic configuration
• Konrado configuration
• Openreach mimic configuration

(Three screen-scrapes are attached, below.)

According to a specification sheet for CW1308 cable, each conductor should not exceed 97.8 Ohms/km. My tests show that the spool 1 cable conductors have a resistance of 90.8 Ohm/km and that the spool 2 cable conductors have a resistance of 87.1 Ohms/km.

My next consideration is how to serially connect the pairs for the experiments using the Openreach mimic configuration. I have thought about two different methods. For the following descriptions, I need to define the abbreviations used:

NE -- near end (the outer end of the cable on the spool).
FE -- far end (the inner end of the cable on the spool).
S1 -- spool one.
S2 -- spool two.
B -- blue pair (blue/white stripe & white/blue stripe).
O -- orange pair (orange/white stripe & white/orange stripe).
G -- green pair (green/white stripe & white/green stripe).
LJU2/3A -- Line jack unit with IDCs (secondary).

The first method of connection:

LJU2/3A to NE,S1,B
FE,S1,B to NE,S2,B
FE,S2,B to FE,S2,O
NE,S2,O to FE,S1,O
NE,S1,O to NE,S1,G
FE,S1,G to NE,S2,G
FE,S2,G to LJU2/3A

The second method of connection:

LJU2/3A to NE,S1,B
FE,S1,B TO FE,S1,O
NE,S1,O to NE,S1,G
FE,S1,G to NE,S2,G
FE,S2,G to FE,S2,O
NE,S2,O to NE,S2,B
FE,S2,B to LJU2/3A

For simplicity of connecting the pairs, I plan to use the first option. Any comments or observations, please?

[konrado5]

I'm curious of SNR margin graph and errors on fast path.

Best regards
konrado5

[burakkucat]

I'm curious of SNR margin graph and errors on fast path.

I assumed that you would be interested to see how far I can "push" the test circuit. Hence I named the second configuration (fast path, zero target SNRM, no DS or US rate limit) in your honour!