 | [Cable] The Dreaded T3 Disconnect - My Solution and Fix This post is to document what I did to fix a dreaded T3 disconnect issue. Searching google shows numerous people who have this problem but few are either resolved nor does anyone post what was ever done to fix the problem.
Very shortly after my switch to Start last week I started to notice my modem rebooting. It's a new DCM476 3.0 purchased from Start. I would watch the event logs and this was happening at random times but typically every 3 - 45 minutes. I couldn't run games, stream or do much of anything. In effect my connection was unreliable and crap. As the "red truck" left my yard the guy did comment that he was going up the road to send some "stuff" down the line. I figured, "oh well just them messing with the line." I was out that evening and didn't get back online until Saturday morning when I noticed that the problem was continuing.
As I would watch the diagnostic page of the modem I would notice that my return path numbers would slowly creep up and I would hit around 51dbvm and poof the modem would recycle. According to what I've read, this happens when it hits 55dbvm so I assume I just didn't catch a refresh on the diagnostic screen till it actually went.
Followed the advice on this forum and did multiple factory resets of the modem and ensured all the connections were tight. No improvement. Called Start tech support, very nice fellow who was helpful and apologetic about having to ask me again to factory reset when he knew I had done this already. This was 9.30pm and tech support closes at 10pm so he said if the problems continued to call back on Sunday.
Sunday morning, problems still there and bad, bad, bad. Fortunately my DSL connection was still active and I could use that while I tried to figure this out. It was baffling, you would look at the power levels and they were great. Low 40's on the return and around 5db on the incoming. Then you would see that return creep up and crash. So I started to troubleshoot.
1. Changed the coax going to the modem 2. Swapped the splitter 3. Changed outputs on the splitter 4. Turned off the TV's in the house 5. Removed the splitter completely 6. Removed the router and went straight to a laptop 7. Used an oHm meter to check all the cables to ensure nothing was shorted 8. Plugged power into a good quality power bar 9. Plugged power direct into the wall
Nothing worked! Without the splitters the numbers were PERFECT. 0-2db on the incoming and 41-43 on the return. Yet crash after crash.
Called tech support that evening. Another nice polite young man. First comment was that he was impressed by my troubleshooting He wrote a book on the file. Had me read all the power levels, frequencies and documented everything I had tried. This would all be sent to big Red for the ticket. I didn't have much faith as my reading online was that is the worst problem you can have requiring visit after visit. But the young guy kept assuring me that they would get this fixed.
While waiting for someone to fix the line I continued my research and came across this link - »homepage.ntlworld.com/robin.d.h.···nal.html It's a bit dated but one of the sections, Cable modem spontaneously re-booting, caught my attention and the comment about electromagnetic interference. Suddenly a light bulb went off in my head "oh Scott, how could you be so stupid and miss the obvious".
My cable line enters the house into the corner of the basement. There is only about 4 or 5 feet of coax into the house and into a splitter from there. I located my modem down in that corner as I did not want to add a 50ft run of coax that would degrade the signal. The modem would be where the cable entered the house and then CAT6 up to the router upstairs. Well, the table that the modem sits on is just 2 feet away from the house electrical panel, 4 feet from the water softener system, 4 feet from the vented hotwater tank, etc etc.. Can you say "an electromagnetic nightmare location" I disconnected the modem and took it upstairs and connected into one of the TV connections.
Poof, problem gone. 
It's been about 14 hours now, no disconnects and pings are solid. Power levels are holding steady at 43.5dbvm on the return and between 5-7 on the incoming.
So either the EMI in that location was bad or the power circuit where I was plugged in was bad. But either way, things are now good.
For those folks who are suffering through disconnects and modem reboots, examine the surroundings of your modem. These boxes are not built like the old days, they are very susceptible to EMI and their environment.
I called this morning and cancelled the ticket and recommend that they add this experience to their knowledge/troubleshooting database. Start - hint hint, get yourself an internal wiki or something to document these little things if you don't have one already.
I can't say enough about the courteousness of the 3 tech support folks who I dealt with over the past 3 days. 2 techs of the 3 did seem to be very knowledgeable. The 3rd I didn't get to quiz enough to be sure 
Sorry for the long book but wanted to ensure this solution was documented for others who might be having the same problems. I can't imagine how someone with little tech experience or an RF background (such as myself) would deal with this type of issue without a good tech support backing them up. This post is for them, if they are googling hopefully this solution will hit for them.
Scott |
|
 sbrookPremium,Mod join:2001-12-14 Ottawa kudos:11 | Well done!
Clearly this is something that needs to be considered by others. |
|
 | reply to scrouse
Out of curiosity, do you have a wireless hydro meter and where is it located relative to that dreaded modem location? |
|
 sbrookPremium,Mod join:2001-12-14 Ottawa kudos:11 | Another EXCELLENT point! |
|
 roccaStart.caPremium join:2008-11-16 London, ON kudos:17 | reply to sbrook
+1 Thanks for the summary and feedback on our support team.
I've seen both cases, ie strong RF interference as well as poor power sources. Where possible, I always recommend putting the modem (and other computer equipment) on a UPS if you have the means. |
|
 mlord join:2006-11-05 Nepean, ON kudos:11 | reply to scrouse
Given that the neighbourhood coax circuit is probably a couple of kilometres in length, an extra 50' of coax inside the dwelling isn't going to make any difference. Coax handles extra distance just fine, unlike DSL wiring.
Cheers |
|
 | reply to alpovs
Hmmmm good question, there is a meter about 5 feet up from that corner of the foundation but I have never made note if it's wireless or not. Will take a look when I get home. But another good thought. |
|
 | said by scrouse:Hmmmm good question, there is a meter about 5 feet up from that corner of the foundation but I have never made note if it's wireless or not. Will take a look when I get home. But another good thought.
It is indeed a Hydro One Smart Meter, I assume those are all wireless? Can't see anything on it to indicate if it is or isn't. I also have a spectrum analyzer, I might take a few shots of what that corner looks like spectral wise, out of pure curiosity  |
|
 sbrookPremium,Mod join:2001-12-14 Ottawa kudos:11 Reviews:
·WIND Mobile
·TekSavvy Cable
| reply to mlord
said by mlord:Given that the neighbourhood coax circuit is probably a couple of kilometres in length, an extra 50' of coax inside the dwelling isn't going to make any difference. Coax handles extra distance just fine, unlike DSL wiring.
Actually mlord, it doesn't. The neighbourhood coax circuit is run kind of like the power lines down a street with transformers every few homes. The power line is very high voltage, but the transformer brings that voltage down to what's needed in the home.
With coax, the trunk down the street is high power. Every tap is kind of like the transformer rather than a splitter. A tap transfers about 99% of the power from its input to its output for every drop connected. So the drop only gets 1% of the power!
That means that anything you do on your drop is moderately isolated from the other users on the street (not totally but well enough for all practical purposes. If they used a splitter, typically half the power would go to your drop and half would carry on to the next. They would run out of signal quickly. And moreover, if you shorted out the drop it really would impact your neighbours.
Therefore a little loss in the drop cable is far more significant than the little loss in the trunk cable! |
|
|
|
 | reply to scrouse
said by scrouse:It is indeed a Hydro One Smart Meter Smart Meter must be wireless. If you have time of use billing it's wireless for sure. |
|
 | reply to mlord
said by mlord:Given that the neighbourhood coax circuit is probably a couple of kilometres in length, an extra 50' of coax inside the dwelling isn't going to make any difference. Coax handles extra distance just fine, unlike DSL wiring.
Cheers
Unfortunately this isn't always the case. The forward path is in the 600MHz range which has fairly significant loss over the typical small coax cables run inside the house, such as RG-59. There are calculators online somewhere if anyone cares to find them Fortunately the return path is in the 20-30 MHz range and the loss is much lower at those frequency levels. But you can lose significant signal the longer that coax is, tie that in with a so-so connector and you can get into big trouble. Keep your coaxes short and connectors to the bare minimum when dealing with higher frequencies. Fortunately I had a good run of high quality coax cable and my loss was minimal. |
|
 | reply to alpovs
said by alpovs:said by scrouse:It is indeed a Hydro One Smart Meter Smart Meter must be wireless. If you have time of use billing it's wireless for sure. Oh yes, they are getting all their pennies from me Wireless the meter is but it's in the 900MHz range, it shouldn't interfere with the cable to much as the frequencies are not multiples. Although there could be some signal heterodyning going on, but I still think it's the electrical panel. It's very old and was probably put in shortly after twist in fuses were replaced. The house itself is over 115 years old  |
|
 mlord join:2006-11-05 Nepean, ON kudos:11 Reviews:
·Start Communicat..
·TekSavvy Cable
| reply to scrouse
said by scrouse:The forward path is in the 600MHz range which has fairly significant loss over the typical small coax cables run inside the house, such as RG-59. RG-59 is so 1970s. Rg-6 is used now, and is part of what you get for the $50 "install fee". Loss is roughly 1/2 power over 100' of coax at UHF frequencies. Negligible for a mere 50' straight run inside a house -- barely half the loss that a single splitter incurs. |
|