US Telco Support > AT&T > AT&T West > [Speed Problem] Lost high frequencies over last few weeks

[Speed Problem] Lost high frequencies over last few weeks

Current DSL

From July

Those notches look very evenly spaced which could be caused by harmonic interference, especially the July chart. Have you ever tuned a portable AM radio to the center frequency of some of them to make sure they aren't local AM stations?

Is there any chance your attenuation has increased some, even only by 0.5 dB, since July?

Yes, my attenuation has gotten just BARELY worse.

Downstream attenuation:
Back then = 42.5
Now = 43.0

Unfortunately, my SpeedStream modem can't report the upstream attenuation (this is a well-known limitation in the firmware). I can dig out my old Zyxel, which can do this, but it isn't as reliable.

What's killing me is downstream noise margin:
Back then = 7.0 through 8.0
Now = 4.5 through 5.5

Obviously, this is "in the red", as it's below the 6.0 threshhold for minimum acceptable noise margin.

Anything I can do to help fix this?

My DSL wire at the house is rather nice: there's an external splitter next to the NID, and it's Cat 5 from there, all the way around to the DSL modem jack (the DSL side of the wiring has only one jack).

Testing at the NID reveals no difference in quality, between what I have at the DSL modem jack, and what I got when plugging the DSL modem directly into the NID (cutting off voice service at the house). So the post-NID wiring doesn't appear to be a problem.

reply to d_l
As for the frequencies of the holes in the graph, I looked them up, and sure enough, they do correspond to major radio stations in this area.

Bin 130 = 560 KHz = KSFO

Bin 158 = 680 KHz = KNBR (Go Giants!)

Bin 189 = 810 KHz = KGO

KGO is a real blowtorch. It's the strongest station in the neighborhood: a friend showed me a homemade crystal radio set that receives all stations simultaneously, as it has no tuner, and KGO is clearly audible above the din.

Interestingly, I live in a valley, that does a good job of blocking radio signals (it makes it hard for me to listen to the radio when I want to). I'm surprised my dropouts in the DSL correspond almost exactly to the frequencies of these major radio stations in the area. My DSL impairment matches the AM radio dial almost exactly: there's a few 50,000 watt monsters near the low end of the dial, then at the high end of the dial, there's so many little small stations that they all disintegrate into noise against each other.

The same friend of mine has a very fancy ham radio set that can take a spectrum analysis of signal strength over a range of frequencies. He lives very close to me, so the radio situation should be fairly similar. I'll ask him to scan the AM radio band, and see what kind of interference there is. It will be interesting to hold up his bandplot, and compare it to my DSL impairment!

reply to Krellan
I'm surprised that your AM station situation has stations roughly spaced out along the spectrum. That certainly creates a psuedo-harmonic interference appearance, but we can be reasonably certain that the output power levels of all those stations haven't been increased.

That's the reason I asked about the attenuation. The change in the two charts looks a lot like an attenuation experiment that I once carried out. By inserting simple resistances on my line to slightly boost the attenuation, the various interference notches I had were "magnified". I think that is what we are seeing on your most recent chart.

I was hoping the attenuation increase would be due to something accessible on your inside wiring, e.g. slight oxidation at a wire contact or plug and jack which increases the resistance slightly. If you don't see a difference of attenuation at your NID , then that probably isn't the problem. Just the same you might check all contacts/screws in your NID to be certain they are tight and clean including the ones over to the station protector.

If the attenuation change is something on your loop, then you are out of luck. This isn't a a dry loop is it? You might try ringing your phone for a long time, then power cycling the modem, and re-checking the DMT chart to look for any change. If there is any improvement, you may have temporarily burned out some oxidation/moisture at a splice.

Your CAT 5 wiring may be acting as an antenna to pull in the AM stations. What have you done with the spare pairs? Are they grounded? My experience has shown that they shouldn't be.

Yes, the AM stations here are actually rather orderly laid out.

»radio-locator.com/cgi-bin/locate···req&sid=

Interestingly, I had better DSL quality during the winter than the summer, which is the opposite of what I'd expect, if AM radio is the culprit. The wintertime has the greatest AM radio signal strength, as the MW frequencies bounce off the atmosphere during hours of darkness, and there is more darkness during winter.

That's a cool idea, just let the phone ring for a long time in order to send the maximum of 48V over the phone line for as long as possible

Most (all?) phone companies have a limit on this now, and just disconnect your call if it rings and goes unanswered for something like 30 rings or so.

As for my Cat 5 wiring, I'm pretty sure they aren't grounded. I think that both ends of the unused pairs are simply floating, connected to nothing. The NID test should bypass all of my Cat 5 wiring, regardless.

I'll try inspecting the NID and repeating the test.

reply to Krellan
Most of the causes of your problem lie beyond your control. My suggestions were aimed at those you can modify. Your wiring and connections may only add a little attenuation or add minimal amplification to the AM interference, but it is those small differences you have lost.

Probably most of the interference is injected into your line in your drop line, especially if it is older. Old drop lines are untwisted and may not be grounded and so tend to be the best antennas on the the entire loop if the IW has been replaced with CAT 5.

AM station strength depends on the time of day. Most stations have to power down or change their beam direction after sundown. So the AM interference should diminish at night until sunrise. I've tracked that very closely on my line. I found that I could maximize the maxrate simply by re-syncing after sundown and again after sunrise to allow the modem to re-train for the current AM station power levels. My line conditions aren't such that I HAVE to do this twice per day re-syncing, but it was something I could do if I ever needed the extra maxrate.

If a few ring attempts to the ring limit doesn't improve your attenuation, then ringing for a longer period probably wouldn't help. I've heard of phone techs putting even higher current on the line to burn oxidation, but that isn't something we can do.

If you can't fix this on your own, I'm not sure how much resources AT&T is willing to devote to a repair of this type of problem these days. You might need a pair swap and in some areas, the unused good pairs have been reserved for UVerse. Maybe it is only your drop line that is the problem.

Thanks for the suggestions.

I turned off the ringers on all phones in the house, and called my landline from my cellphone, just letting it ring and ring until the limit was reached (around 5 minutes or so). Repeated this a few times.

Turned off and on the DSL modem, to let it resync. Did this before and after the ringing. This was all at night. Immediately after the DSL modem synced up, I checked the sync speeds.

Before = 5440
After = 5728

A little improvement, but that's better than nothing. No difference in attenuation that I could detect (the modem rounds all readings to the nearest 0.5dB, and this change was probably much smaller than that).

What you say about AM stations at night is true, but the big lower-frequency AM stations around here are all clear channel 50,000 watt blowtorches. They broadcast 24/7.

As for the little stations on the higher frequencies, it's true that they power down, but the spectrum remains just as noisy. AM radio stations travel farther at night. So, instead of hearing 1 local station on the frequency, I'll hear about 3 or 4 very distant stations, all talking on top of each other.

Is there any way I could convince the phone company to take a look at the wiring? Or, do they have a policy of not caring about DSL quality of service, other than "it works"?

reply to Krellan
That's too bad about those stations around you all being 50 KW clear channel ones.

Also your maxrate gain could be due simply to a re-sync. The modem is most optimized for the line conditions right after it finishes training.

You could take your problem to AT&T Direct and see what they say or do. I don't know what their current repair policies are. I've just been inferring what they might be from posted repairs.

Thanks for all the advice. As for the maxrate gain being due to a re-sync, I turned the modem off for 15 seconds, then on again, for *both* of the measurements. It was immediately after a resync, for the "before" measurement, AND also for the "after" measurement. So, that rules out the quick little gain that is often seen just after a resync.

Interestingly, after letting the line stay up for a long time and not resync, my line seems to hold. It doesn't "death spiral" down over time. It appears fairly stable, even in this degraded condition.

reply to Krellan
I think ferrite chokes work best, if they work at all, on the powerline or the power cord of an interference source device right as it exits the unit and don't really work well installed on the phone line. I have lots of them in use.

reply to Krellan
I just went through this exact problem with AT&T here in the Central Valley. I could sync at 6016/768 with a SN ratio of 6 db. The line worked great at this speed since last September. Now the line is down to 5.5 db SN ratio and I can still sync at 6016/768 but the line drops once every couple of days. Sometimes twice a day.

AT&T came out and moved my DSL to the best of the 3 pairs I had coming into the house.

I am 3500' from a Uverse VRAD and DSLAM as you drive down the road. My line tests at 6500'. Obviously the line winds around a little.

AT&Ts solution. Downgrade me to 3000/512

They will not trouble shoot pairs, find a cleaner pair, or anything. I have 3 POTS lines and had their highest internet speed. That is the max I can bring into the house. If they don't care about me, I'll hit them in the wallet.

I am now currently set at 3000/512. This should hold sync. I have plenty of head room with this. But, ouch it kills to have to go down in speed. It's noticeably slower.

I talked to several of my neighbors about their phone/internet/tv services. I live on a cul-de-sac with 8 houses.

Next, I called Comcast to see if we could get the cul-de-sac hooked up with them. Comcast sent out an engineer, and said it meets their requirements.

So, about 5 of us on the block are going to drop AT&T and move over to Comcast. I am porting over my 3 phone lines also.

I tried to get AT&T to upgrade lines, change cards in the DSLAM or VRAD to push things out here further. They have cards that will push out 10,000'. Worked my way up the food chain to supervisors. No go. No help. They all play dumb and don't know what their other hand is doing.

The market is getting competitive. If you want the business work for it.

reply to d_l
I don't think I have any "interference source devices", as my interference is coming in from external AM radio stations. It's not interference from a device that's under my control (or I would have unplugged it or fixed it already).

I tried moving my chokes to the power cord. It didn't make much of a difference. I then tried moving them back to the phone line. It still doesn't have much of an effect at all. I'll try getting some more chokes and putting more on. Unfortunately, due to the small size, it's kind of hard to get many windings on the coil.

reply to Krellan
I was just saying that in my experience the chokes don't help at all on the phone lines.

How close is your power wiring to your home run? How close are various power cords around your modem to your phone line patch cord?

My power lines also seem to amplify the local radio stations. I discovered this by tuning a weak station and then moving a portable radio close to it. The signal strength always picked up as long as there wasn't some other interference also on the power line at that same frequency. I made sure I had a shielded, twisted wire phone line data cord connecting the modem to my home run to minimise AM station interference from the active power cords in the area. It helped a bit. ... a tiny bit.

Modem NID Test

Downstream Current Rate (Reported): 6016
Downstream Maxrate (Estimated): 6028
Downstream Bit Margin (Calculated): 428

Upstream Current Rate (Reported): 768
Upstream Maxrate (Estimated): 1166
Upstream Bit Margin (Reported): 96

Bit tones with notches/missing bits:

bit(s) 38:39 163.875 - 172.5 KHz
bit(s) 128:132 552 - 573.5625 KHz
bit(s) 138:144 595.125 - 625.3125 KHz
bit(s) 155:176 668.4375 - 763.3125 KHz
bit(s) 178:179 767.625 - 776.25 KHz
bit(s) 185:191 797.8125 - 828 KHz
bit(s) 196:204 845.25 - 884.0625 KHz
bit(s) 208 897 - 901.3125 KHz
bit(s) 211 909.9375 - 914.25 KHz
bit(s) 214 922.875 - 927.1875 KHz
bit(s) 222:225 957.375 - 974.625 KHz
bit(s) 228:230 983.25 - 996.1875 KHz
bit(s) 234 1009.125 - 1013.4375 KHz
bit(s) 239:255 1030.6875 - 1104 KHz

reply to Krellan
It will be interesting to see if the maxrate gain is maintained through re-syncs.

So possibly a lot of the AM interference inductance is occurring in your drop line wire because it isn't twisted.

You could experiment by grounding your spare pairs to see if it makes any difference. I've always thought that grounding them would make the AM reception worse.

The other thought is to make each spare pair a closed loop by bridging the both ends with the wire splice peanuts. I did this, but I can't really say if it made much difference.

After power cord choke

Downstream Current Rate (Reported): 6016
Downstream Maxrate (Estimated): 6528
Downstream Bit Margin (Calculated): 440

Upstream Current Rate (Reported): 768
Upstream Maxrate (Estimated): 1105
Upstream Bit Margin (Reported): 96

Bit tones with notches/missing bits:

bit(s) 38:42 163.875 - 185.4375 KHz
bit(s) 128 552 - 556.3125 KHz
bit(s) 130 560.625 - 564.9375 KHz
bit(s) 138:144 595.125 - 625.3125 KHz
bit(s) 148:149 638.25 - 646.875 KHz
bit(s) 153 659.8125 - 664.125 KHz
bit(s) 155:160 668.4375 - 694.3125 KHz
bit(s) 167 720.1875 - 724.5 KHz
bit(s) 169:176 728.8125 - 763.3125 KHz
bit(s) 181 780.5625 - 784.875 KHz
bit(s) 183:191 789.1875 - 828 KHz
bit(s) 195:209 840.9375 - 905.625 KHz
bit(s) 211 909.9375 - 914.25 KHz
bit(s) 213 918.5625 - 922.875 KHz
bit(s) 218 940.125 - 944.4375 KHz
bit(s) 220:225 948.75 - 974.625 KHz
bit(s) 227 978.9375 - 983.25 KHz
bit(s) 229 987.5625 - 991.875 KHz
bit(s) 234 1009.125 - 1013.4375 KHz
bit(s) 236 1017.75 - 1022.0625 KHz
bit(s) 250:255 1078.125 - 1104 KHz

reply to Krellan
So there really does seem to be a reduction in the re-radiation of the AM signal from your power cords and power line when you use chokes. That's good. When I wondered about the maxrate after re-syncs, I was also thinking about the long term results through the seasons. I also wonder if you could have detected the AM station signal reduction with a portable radio at all or if this is only detectable by your modem maxrate?

I pretty much only used clip on chokes with no windings of the cords to try to suppress impulse noise with mixed results. As I said, I used a Belkin shielded and twisted cord to protect my DSL patch cord from all the power cords near it and that resulted in the picked up a few bits out of the AM station notches.

I don't know how much AM RFI the modem wall wart might pass to the output (noisy DC output), but if it does allow any through the modem patch cord would lie right next to it for inductive coupling.