I received an email from Todd who is the Plant Operations Manager of a fine independent Telco.
We are replacing copper with fiber but can’t do it fast enough to keep up with customer demand, with the growth of the IoT the demand for speed is now!!! We have started to bond copper to get additional speed to these customers but we have some concerns.
1. We understand we need to keep our cable fill ratio to 50% or less — is this correct? (Can we increase this when bonding?)
2. When bonding, are there specific pairs we need to use? (We understand like for like.) (What pair has the best performance: W-B, V/S, or R-S?)
3. Impulse Noise Protection may help (Calix bonding technology) ADSL2, ADSL2+, and VDSL2, are all running in PTM Mode. (Does it?)
4. We are keeping our VDSL loops to 4,000’ but are trying to bond DSL out to the 18,000’ loop length. (Any thoughts?)
5. VDSL and ADSL in the same 25pr group. Will this cause any issues?
Lots of questions but I assume others are asking the same thing!
I am going to defer some of these concerns to my bandwidth expert, Vernon May of Vernon May Solutions. Vernon’s concepts have proven to be effective for those seeking methods and procedures focused on practical network operations in the bandwidth arena.
We understand we need to keep our cable fill ratio to 50% or less — is this correct? (Can we increase this when bonding?)
Vernon’s response: I have never heard of the 50% rule on binder group working DSL circuit density. The only rule of which I am aware was one that was violated before the ink was dry. Very early in the ADSL1 days, 5 DSL circuits per binder group was supposed to be the maximum. There is no doubt that the number of working DSL circuits impacts the noise floor in the binder group. More circuits, more DSL frequency crosstalk and more bins that will not carry the full capacity of bits. However, new copper cable is very rarely installed, so we must accommodate the subscribers that need service, regardless of the amount of circuits in the binder groups. DSL bonding will double the amount of circuits and will increase the noise floor. When noise is high in the binder groups, the best thing is to reduce attenuation by using shorter loops. Also, our copper plant must be as close to perfect as possible. DSL profile tuning has also been used to reduce the noise floor (Please see my article “Be a Healer” in ISE magazine at https://www.isemag.com/2016/02/be-a-healer.)
When bonding, are there specific pairs we need to use? (We understand like for like.) (What pair has the best performance: W-B, V/S, or R-S?)
Donald’s response: Your concern of one pair’s performance over another pair in the same sub-unit of a telephone cable (pair-to-pair crosstalk) is somewhat justified. There are other concerns of pair quality but first the basics. Alien crosstalk (AXT) interference will affect any circuit. AXT is an unwanted disturbing signal from one balanced twisted pair component, channel, or permanent link to another. We are using cable pairs that were designed for analog telephone service for a multitude of services — including IoT.
Twisted cable pairs were first used on paper cables to reduce noise. This practice was derived from the use of transpositions on open wire circuits. Concentric manufacture reduced capacitive coupling from pair to pair in the telephone cables. When pulp cables were manufactured in the 1930s, unit type construction placed a lay in the cables virtually eliminating group-to-group capacitive coupling. Cables were made in 26-pair, 51-pair, and 101-pair groups, under one binder tie. Each group had 9 different pair twists from 1.5 inches to 6 inches, reducing capacitive coupling between pairs in a unit.
Our standard plastic insulated conductor cables (PIC) have 25 different twists for each 25-pair sub-unit. For example, the white/blue pair has a twist factor of 2 inches per twist. The red/slate pair has a twist factor of 4.7 inches per twist. Every other pair in the sub-unit falls somewhere in between. The pairs are fed through a 25-pair face plate with pairs 1-3 in the center; 4-12 in the middle; and 13-25 on the outside. The wheel oscillates from 180 degrees to 360 degrees in a distance of 100 to 150 feet. This separates the sub-units from a standpoint of crosstalk. A 400-pair PIC cable has 16 sub-units of 25 pairs under one sheath. No one 25-pair sub-unit interferes with any other 25-pair sub-units from a standpoint of crosstalk.
Even if you could pick the best pair for your particular service, it is not allowed. The FCC states that a telco cannot select certain pairs for their service and leave the other pairs for a CLEC or any other vendor that wants to rent your cable pairs. Therefore, to answer your question on selecting pairs for bonding, it is not allowed, so it is a moot point.
Impulse Noise Protection may help (Calix bonding technology) ADSL2, ADSL2+, and VDSL2, are all running in PTM Mode. (Does it?)
Vernon’s response: Impulse noise protection can be a double-edged sword. The times where that protection will be helpful could be rare. Setting the impulse noise protection too high will decease circuit throughput upstream and downstream. To make things worse, impulse noise varies so much in spectrum and amplitude that nothing will protect for the worst pulses. There are as many opinions on this as there are experts to opine. I believe that a setting of 2 is the best compromise between stability and throughput. BTW: ADSL2 should not be used. ADSL2+ has several performance enhancing features that help with stability.
We are keeping our VDSL loops to 4,000’ but are trying to bond DSL out to the 18,000’ loop length. (Any thoughts?)
Donald’s response: Using distance from the DSLAM to the subscriber is not at all accurate. Too many factors are involved. Wire gauge and wire temperature impacts attenuation as much as distance. Noise impacts circuit performance at least as much as attenuation does, so forecasting circuit performance by distance alone is fundamentally flawed. That said, everybody does it with mixed results.
The actual formulas that will forecast performance are wildly complicated, requiring more data than most groups have available. Therefore, distance is what is used: 4,000 feet for VDSL and 18,000 feet for ADSL2+ is as good a guess as any.
VDSL and ADSL in the same 25pr group. Will this cause any issues?
Vernon’s response: VDLS2 and ADSL2+ both use the frequencies up to 2.2 MHz. Because of this, they will both contribute to DSL crosstalk in that range. In that range, ADSL2+ circuits or a mix of ADSL2+ and VDSL2 have the same impact.
Sometimes, people set the VDSL2 SNRM higher than ADSL2+. That puts more noise into the binder group because the VDSL2 circuits are generating more power. Within the binder group, the ADSL2+ circuits are bringing a knife to a gun fight because they are set to a lower power setting.
As long as the SNRM settings are the same for both ADSL2+ and VDLS, mixing them within the binder group does not cause any special issues.
When you encounter a problem with bandwidth, take another look at the physical layer. Replace Cat3 jumper and cross-connect wire with Cat5 wire. Make sure that there are no DC type faults on the cable pair, and that the longitudinal balance tests greater than 60Db. Test the station ground to make sure that it is at the same potential as the CO or remote ground, and that it is attached to the power ground electrode.
I would recommend the DSL INT for any bandwidth service. The DSL INT instantaneously reduces over 95% of the steady-state or surge-induced voltages and currents. The DSL INT can substantially reduce voice frequency harmonic noise and impulse noise. A DSL INT has the ability to suppress induced lightning surges.
A resource for you: I just finished reading Computer Networking Breakthroughs by the noise mitigation expert, Russ Gundrum. It is a must-read for any technician or manager who is involved in provisioning and maintaining the copper infrastructure for maximum bandwidth. For more information and a PDF of the book, please visit https://www.telecomproblemsolvers.com/new-book.
I hope this column is helpful. If you have questions or comments, please reach out to me at email@example.com or contact Vernon May at firstname.lastname@example.org and