The Healing Properties of DSL Profiles
Most people would be amazed at how much control over digital subscriber line (DSL) service is based in the digital subscriber line access multiplexer (DSLAM) and modem profiles. You can mold the shape of the DSL signal with a few well-placed changes to the settings.
You can also make DSL much harder to support than it should be.
Because you can ruin service with profile changes, people have been hesitant to experiment in the field with profile improvements. Therefore, the default profiles settings often remain in some key areas and they can often be the worst choice.
I recently had the fortunate opportunity to work with a “technology-forward” service provider, Minford Telephone of Minford, Ohio (www.falcon1.net). Minford serves about 2,500 subscribers. They supply voice, TV and Internet through a combination of FTTH and FTTN network architectures. The FTTH project is ongoing. Their longest copper loops are about 23 KFT.
Looking to offer their customers the best DSL service, they found themselves facing a common DSL profile issue, and I was contacted to help them resolve the problem.
The default Signal-to-Noise Ratio Margin (SNRM) settings were pumping too much power into the binder groups, and causing high crosstalk.
Before we go further, let’s be sure to clarify some important definitions:
• Signal-to-noise ratio (SNR) is defined as the power ratio between a signal (meaningful information) and the background noise (unwanted signal).
• SNR margin (dB, a.k.a. noise margin or SNRM) is the ratio by which the signal exceeds the minimum acceptable amount (minimum SNR) to sustain a certain speed.
An example of the impact of SNRM is a crowded restaurant. You are chatting with a friend at the same table. You must talk louder than the background noise of the other conversations in the room in order to understand each other.
Now, consider what would happen if everybody in the restaurant decided that they must talk 50% louder than the background noise. The background noise will become louder and louder. Before long, everybody is talking as loud as possible and nobody can understand anybody else. Bad SNRM settings can cause this situation to happen in the binder and sheath groups.
Original Profile at Minford
(US = Upstream: DS = Downstream)
DS Target SNRM = 6 US Target SNRM = 6
DS Maximum SNRM = 31 US Maximum SNRM = 31
DS Minimum SNRM = 0 US Minimum SNRM = 0
• Target is OK. From 6-10 dB is normal.
• Maximum is telling the chipset to scream as loud as it can.
• Minimum could be turned off because 0 dB margin defies the definition of “margin”.
Minford currently has 3 Bit Rate profiles:
• 2M DS / 768K US
• 3M DS / 896K US
• 4M DS / 1M US
While upgrading all subscribers to their 4M profile, Minford experienced significant problems, so they reverted back to the old bit rates while they worked through the issues.
The old profile settings created a performance spiral so severe during the upgrade that the DSL circuits could not recover effectively.
1. Each port will pump all of the power it can into the binder group in order to reach its targets.
1A. The more power, the higher the noise floor in the binder group.
1B. The higher the noise floor, the more power each port pumps in.
2. The ports fight each other until the DSLAM reaches its maximum power output.
2A. Every subscriber served from the DSLAM site experiences noticeable deterioration in quality.
2B. The monthly site costs become approximately 50% higher than they should be.
2C. The lifespan of the electronics at the DSLAM site is greatly diminished because nothing is designed to work at peak power 100% of the time; rectifiers and power supplies take the brunt of the abuse, but everything is impacted negatively.
New Profile at Minford
Changed on September 21st, 2015
(US = Upstream: DS = Downstream)
DS Target SNRM = 6 US Target SNRM = 6
DS Maximum SNRM = 9 US Maximum SNRM = 9
DS Minimum SNRM = 3 US Minimum SNRM = 3
• The Target settings remains the same.
• The Maximum settings are changed to be above the target, but not excessively.
• The Minimum settings are changed to a reasonably stable level below the target.
Note: Since this activity, I have changed my recommended Minimum SNRM setting to 1 dB.
All subscribers maintained the same bit rates before and after the profile changes. No degradation was seen as of October 5. Techs reported less trouble seen since the change. This improvement makes sense when you consider the information depicted in Figures 1, 2, and 3.
In Figure 1, you will notice that the average DS SNRM is at about 27 dB before the profile changes. Remember, the target is only 6 dB. Now, the average is about 8 dB.
In Figure 2, you will notice that the DSLAM was pushing almost 5,000 dBm of power before the profile changes. Now, the site is down to about 3,700 dBm. This difference reduces the monthly electricity costs of the site; but more importantly, it reduces the load on the rectifiers and power supplies.
In Figure 3, you will notice that the average power per circuit was about 15 dBm per circuit before the profile changes. Now, the average is about 12 dBm. That reduction in power reduces the noise floor in every binder and sheath group. That makes higher bit rates possible for all of Minford’s subscribers.
Our goal was to maintain the current rates, but to use less power. The key was the amount of DS SNRM reduction. It is obvious that the old profiles were blasting the binder groups with power that was not needed.
The DS power tuning achieved is enough to attempt bit rate upgrades safely now. Of the 336 live circuits in this system, 319 can be upgraded to the 4M profile.
Though the improvement on the DS side was significant, the US side shows no improvement. The reason is that the customer modem controls upstream transmission and power. The modem is supposed to be instructed by the DSLAM on US settings during the negotiations phase of sync and follow those instructions. That is not happening.
Not all of the performance data fields are populated in the DSLAM port reports. This along with US control issue points to an interoperability issue between the Modem and DSLAM.
That leaves 2 options on the US side:
• Option 1. Make the handshake between DSLAM and modem follow spec.
• Option 2.Change the settings directly in the modems (possibly a broadcast?).
Further stability can be achieved by turning on multi-mode 2+ and rate adaptability.
It is important to note that the improvements generated at Minford took place without a single dispatch, and with only a few hours of work.
DSL profiles are tools that are available to every DSL provider and they can be used to heal subscriber service quickly, effectively, and with little expense.
Making the right profile decisions is critical to achieving the highest quality DSL service. Unfortunately, you can’t always trust the default settings. You must consider how the circuits within the binder groups and sheath groups impact one another.