Kitz ADSL Broadband Information
adsl spacer  
Support this site
Home Broadband ISPs Tech Routers Wiki Forum
 
     
   Compare ISP   Rate your ISP
   Glossary   Glossary
 
Please login or register.

Login with username, password and session length
Advanced search  

News:

Pages: 1 2 3 [4] 5 6 7

Author Topic: Findings with a new connection  (Read 31921 times)

les-70

  • Kitizen
  • ****
  • Posts: 1254
Re: Findings with a new connection
« Reply #45 on: March 01, 2014, 08:07:17 AM »

   Thanks for correcting the lan2 typo.

I am fairly sure it is bad ethernet cables causing the issue. I have tried about 18 cables now -all I can find and mostly supplied with routers.  10 are within 0.0-0.1db difference when tested i.e. probably no difference, 4 are with 0.1-0,2 difference and 2 with 0.3 and just 2 with ~1 or ~2 db, I conclude 2 or 6 of the cables should be binned.   I have a second HG612 and I will try that another day but what every that shows some cables work as I would expect i.e. no impact but a minority are found to have an impact in this test. Below is some Texas Instruments text relation to ethernet noise emissions.

4 Sources of EMI in Ethernet Applications
Ideally, if differential signals are perfectly balanced, no common mode energy exists in the system. In
single ended signal systems, ideally all forward energy is contained within the signal wire or trace and
return energy is contained within a ground wire or plane in close proximity to the signal.
The source of unwanted emissions in network applications is common mode energy, radiating either from
differential signal wires, or directly from the system chassis.
This common mode energy can originate from any of three sources:
1. Imbalance in the differential signal path
2. Noise coupled to or from the system chassis or power supply system
3. Noise coupled to or from the network interface cable
4.1 Differential Signal Path Imbalance
Signal path imbalance can occur in two ways: across the differential signal pair or between the signal
source and destination.
Imbalance that occurs across a twisted pair can result from the cable medium itself being unbalanced, or
from signal termination imbalance. Imbalance across a signal pair results in one signal having a larger
magnitude than the opposite signal, which manifests itself as common mode noise.

5. Differential Signal Path Imbalance
End to end or longitudinal path imbalance can occur if the source impedance, transmission line
impedance, and destination impedance are not exactly matched in a system. This form of mismatch
causes energy reflections across the cable from end to end, which also results in common mode noise.
Recommendations for preventing differential signal path imbalance include:
• Use high quality symmetrically and tightly wound cable. ISO CAT5E or better quality cable is
recommended for 10/100 applications.
• Use equal length differential MDI signal traces with a strip line impedance of 50 ohms.
• Closely match the values and physical placement of signal termination components.
« Last Edit: March 01, 2014, 09:08:51 AM by les-70 »
Logged

waltergmw

  • Kitizen
  • ****
  • Posts: 2776
Re: Findings with a new connection
« Reply #46 on: March 01, 2014, 09:11:36 AM »

@ les-70,

I can confirm that a normal VDSL modem will run quite happily of its power supply connected to a small inverter plugged into a car's cigarette lighter socket.

Kind regards,
Walter

Logged

les-70

  • Kitizen
  • ****
  • Posts: 1254
Re: Findings with a new connection
« Reply #47 on: March 01, 2014, 09:37:47 AM »

 Thanks,   I am fairly sure the HG612 power supply is not relevant in these tests.   I have done the test with a linear power supply on a BT mains conditioner and get exactly the same changes as when run of the normal switched mode psu on the direct mains.  The mains conditioner and/or linear psu give a quite separate ~0.3db improvement when used.  The linear supply made no difference when I had adsl so I guess the vdsl bandwidth raises the risk of noise.

 In the Ethernet cable swap test the issue really has to be a couple of bad Ethernet cables. They seem to work but can't be fully OK.  I guess the bad Ethernet cables have poor connections/higher impedance on some the wires used. Most cables give No impact,  a few a small impact and a couple (one of which I first used)  a large impact.   
« Last Edit: March 01, 2014, 09:55:10 AM by les-70 »
Logged

les-70

  • Kitizen
  • ****
  • Posts: 1254
Re: Findings with a new connection
« Reply #48 on: March 01, 2014, 01:45:42 PM »

    An update::  The two "bad" cables giving drops of 1-2 db snrm are simply faulty, one shows a high impedance where there should be about an ohm and the other has a loose connection somewhere in its makeup. It took a long while and some waggling to spot the loose connection one.   The other 3 cables giving smaller snrm drops of about 0.2-0.3 db are also in the bin now but nothing was obviously wrong with them.  Below I show the impact each of these 3 poor cables had on the snrm.  The impact of the obviously "faulty" cables was similar but much bigger.  As I noted above the poor cable adds a lot more noise to the highest frequency band

   I have retested all the other cables I have and can't get any effect from them i.e. they just work and don't influence the snrm at all when used on Lan2. They work perfectly in my normal setup and also in the test set up so everything is now as it should be  :) .  Those thinking it unexpected can be reassured that if the cables are good that is definitely the case. 

   However 5 cables are in the bin (for recycling) and I assume I may not alone in having some duff cables.   
Logged

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43467
  • Penguins CAN fly
    • DSLstats
Re: Findings with a new connection
« Reply #49 on: March 01, 2014, 02:49:26 PM »

That's an interesting result, and quite surprising. It's suprising that so many cables are faulty in this way, and surprising that the fault only affects the top band.
Logged
  Eric

les-70

  • Kitizen
  • ****
  • Posts: 1254
Re: Findings with a new connection
« Reply #50 on: March 01, 2014, 03:44:44 PM »

  I seem to have had 1 in 10 cables with a serious fault and 1 in 5 of those remaining with the "mild issue" seen in the snrm above.  1 in 4 overall does indeed seem very high rate but the collection does include some quite old cables.  One of the serious faults was an "brand new" BT hub cable and the other serious one was an old cable that may have been bent once too often. 

  It is interesting that all bar one of the cables seemed to work perfectly in the sense that all  Ethernet traffic got through OK.  The one with an intermittent open circuit did fail to pass Ethernet from time to time e.g. with router stats running whilst the cable was used and waggled a few samples were lost. 

   It was by pure luck that I noticed the noise and 0.3 db drop when I swapped in new cable on Lan2 when tidying the wiring up - I was only aiming to get the Lan1 and Lan2  cable colours different!
Logged

burakkucat

  • Respected
  • Senior Kitizen
  • *
  • Posts: 38300
  • Over the Rainbow Bridge
    • The ELRepo Project
Re: Findings with a new connection
« Reply #51 on: March 01, 2014, 03:53:50 PM »

I have to agree with Eric. It is a most interesting and unexpected result. Thank you for the precise testing and clear explanation of your findings, les-70:)
Logged
:cat:  100% Linux and, previously, Unix. Co-founder of the ELRepo Project.

Please consider making a donation to support the running of this site.

burakkucat

  • Respected
  • Senior Kitizen
  • *
  • Posts: 38300
  • Over the Rainbow Bridge
    • The ELRepo Project
Re: Findings with a new connection
« Reply #52 on: March 01, 2014, 04:04:53 PM »

I can confirm that a normal VDSL modem will run quite happily of its power supply connected to a small inverter plugged into a car's cigarette lighter socket.

Deviating off topic but using Walter's suggestion, above, will add yet another noise-generating device into the circuit! The average small inverter will be a switching based, oscillatory device which generates an approximation to a sine wave. That generated stepped square-wave (~240 V AC) is then transformed back to 12 V DC by the normal switching mode PSU.

For Walter's Wheelbarrow, I would construct an appropriately fused distribution box which takes the ~12 V DC from the car battery (via the cigarette light socket -- until such devices become Verboten) and to which a flying lead can be connected, thus allowing the modem to be directly powered from the car's battery.  ;)
« Last Edit: March 01, 2014, 07:22:59 PM by burakkucat »
Logged
:cat:  100% Linux and, previously, Unix. Co-founder of the ELRepo Project.

Please consider making a donation to support the running of this site.

les-70

  • Kitizen
  • ****
  • Posts: 1254
Re: Findings with a new connection
« Reply #53 on: March 01, 2014, 05:06:21 PM »

  @burakkaucat   

  On your deviated topic I have resisted batteries as they don't seem a long term connection solution. One day I may try it simply as an experiment.  I started in the direction of mains carried noise with a BT mains conditioner which gave me +0.3db snrm gain with a fixed speed sync.

 I then tried using as well a linear PSU for the HG612.  That gave a tiny extra benefit of ~0.1db.  However leaving out the BT conditioner and just using the linear supply is a very similar benefit to that given by the BT mains conditioner.  A linear PSU should have more mains isolation than a switched one and no issue with "switching noise".  Since the linear PSU did not give much extra  benefit when used with the BT mains conditioner I guess the  main benefit of the linear PSU may be just the isolation and filtering due to the linear PSU transformer.   I find the linear PSU more attractive as it does not hum and only uses an extra 1-2 watts c.f. the switched supply.  The BT conditioner used an extra 10 watts and had a small but noticeable hum.   

 As a further deviation I have been pleased that the DLM seems to have ignored all the testing.  Either it just does not care or it detects and ignores deliberate power downs.
Logged

NewtronStar

  • Kitizen
  • ****
  • Posts: 4898
Re: Findings with a new connection
« Reply #54 on: March 01, 2014, 07:05:36 PM »


As a further deviation I have been pleased that the DLM seems to have ignored all the testing.  Either it just does not care or it detects and ignores deliberate power downs.

When reading your posts I was wondering to myself how the DLM would react to your testing and think as long as you leave a minium of 15 minutes in the powered off state between tests you should get away without any DLM intervention  :fingers:  ;)
Logged

waltergmw

  • Kitizen
  • ****
  • Posts: 2776
Re: Findings with a new connection
« Reply #55 on: March 02, 2014, 08:21:51 AM »

@ BKK,

If there are concerns about voltages and spikes etc., I would counsel some caution as car electrics can be quite variable too, depending upon engine / generator on and off, use of self starter etc.

Kind regards,
Walter
Logged

Black Sheep

  • Helpful
  • Addicted Kitizen
  • *
  • Posts: 5717
Re: Findings with a new connection
« Reply #56 on: March 02, 2014, 09:36:03 AM »

Some good fault-finding experiments from Les. The learning curve continues ...............  :)
Logged

Chrysalis

  • Content Team
  • Addicted Kitizen
  • *
  • Posts: 7382
  • VM Gig1 - AAISP L2TP
Re: Findings with a new connection
« Reply #57 on: March 02, 2014, 01:52:53 PM »

this is interesting, now I am probably going to try swapping out both lan cables between my hg and router, both are isp supplied short cables not the usual long high quality ones I use for rest of my lan.
Logged

les-70

  • Kitizen
  • ****
  • Posts: 1254
Re: Findings with a new connection
« Reply #58 on: March 02, 2014, 02:46:25 PM »

  I think that it is well worth trying the Lan cable swap test but judging by my collection there is only a 1 in 4 chance of any difference and then most probably only a change of a few tenth's of a db snrm -  but you never know unless you try.  It is a nice test as you remain in sync and and can monitor clearly what happens.
Logged

les-70

  • Kitizen
  • ****
  • Posts: 1254
Re: Findings with a new connection
« Reply #59 on: March 17, 2014, 05:18:24 PM »

 An update.  I have now gained about 0.8 db snrm or 3mb/s of attainable by:-

   1. Replacing a dodgy lan cable  ~0.1-0.2 db
   2. Replacing the flat dsl cable with a homemade dsl cable using twisted pair  ~0.1db
   3  Running the HG612 on Bt mains conditioner ~0.2-0.3db
   4  Using a linear psu for the HG612  ~0.1
   5  Earthing the disconnected pair on the incoming drop wire and internal cable to the master socket ~0.2db

  Note that the last mod is not legal or recommended and has been done so that it can be easily and quickly removed.
  I checked carefully that the unused pair had no connection to earth and was open circuit before adding the earth connection.

  With mods 1-4 all in place the average error rate was reduced by about 30% and odd big crc spikes were reduced in size but still gave about 5 ses/day

  With mod 5 the error rate has halved and over a period of one week no ses or crc spikes have occurred - I believe this is the biggest useful change. I think I have obtained the benefit from the spare pair giving partial screening of the used pair. It may also be significant that the run to my house is underground until it goes up the pole that drop wire comes from. 

   
« Last Edit: March 18, 2014, 08:29:05 AM by les-70 »
Logged
Pages: 1 2 3 [4] 5 6 7