MODEM Question -- Can a factory reset & re-register clear up intermittent drops?

[NTL2009]

Have had very reliable internet in the ~ 2.5 years we have been in this home, using my purchased Modem/router. A few months back, we started getting intermittent drop outs (~ 10/day?), lasting a couple minutes each. Mostly would come back on their own, sometimes I'd have to reboot the modem, sometimes the modem seemed to reboot on its own (or did Comcast/Xfinity trigger this?).

At this same time, on the Xfinity status page, they had a string of announcements about scheduled maintenance to correct noted network problems, so I thought maybe this was all on their end (and they did find some bad cables/connections along the way, and my signal levels improved from marginal to good), but the problem persisted. After unproductive runarounds with Xfinity, and a tech deciding we should have our buried cable replaced ( ~ 15 YO if original to the home- but it made no difference, as I expected, as my signal levels and S/N ratio were at a good level before/after). I have some scripts running to check the connection every minute and log any drops, so I've got a good record and measurment (Ethernet connection to modem/router, to eliminate WiFi as the culprit) . There's more to this, but to cut to the chase...

I got frustrated with the Xfinity run-around, and decided to rent their modem @ $15/mo to avoid some finger pointing. Well, to my surprise, the rented modem worked perfectly. I let it go a week (my old one was perfect for 3 days once over this trouble period, so I still can't be sure which end is having the problem). Then I wondered whether simply doing a factory reset, and the re-registering that old modem with Xfinity would magically clear up something. So I did that, and my old modem seems to be working near-perfectly for four days now. I did have one reboot and later, one error log "Started Unicast Maintenance Ranging - No Response received - T3 time-out;", but zero drop outs (other then when the reboot occurred). The 'corrected errors' are ~ 230 total now, after two days of uptime, and zero uncorrected, I thinks those numbers are not considered a problem?

Does it make any sense that a factory reset and re-register could clear up intermittent drops? I rebooted that old modem many, many times previously. FYI, the 'old' modem is on their 'approved' list, a Motorola MG7550 DOCSIS 3.0, which might be newer than the one the provided (an ARRIS TG1682G DOCSIS 3.0).

Hopefully, it keeps working and I'll return their modem, but I'm curious if there is a any sound technical reason the reset/re-register would help (over and above a regular reboot)?

 

[Tom.G]

Your description sounds much like a power problem, either short spikes or dropouts on the power line. (it could also be the cable network dropping out)

Try getting an Uninterruptable Power Supply (UPS) for it. Modems do not draw much power, so a small UPS would be fine.

If that cures the problem, you might want to get a larger UPS for your expensive computer and monitor as 'cheap insurance' for when the problem gets worse.

Cheers,
Tom

 

[NTL2009]

In the back of my mind, I was wondering if it could be as simple as a power problem. I was thinking about setting up something to monitor it (or put it on a UPS as you suggest). But then again, it was good for about 3 days, and has been good for 2 days since I reset and re-registered it. I'll keep an eye on it.

 

[Tom.G]

I was thinking about setting up something to monitor

Recording Power Monitors cost a LOT more than a small UPS. Many UPS models can report events to your computer on a coarse level (5-second sampling). You could rent a Monitor but a small UPS may still be cheaper.

If you do get a UPS, shut the system down for a real outage, otherwise they will run until the battery dies. That includes shutting off the UPS. My 750W UPS is driving a 210W system (typical sizing, 3 to 4 times load), a new battery will keep it running for 10 to 15 minutes. I replace the battery when the self-test reports 75% to 80% of the run time for a new battery.

Actual available power is roughly 20% to 25% of rated battery Watts.

On request, the larger Electric companies will often attach a Recording Powerline Monitor to your powerline on a temporary basis for troubleshooting (or to prove innocence ). If they do, try to get a look at the results rather than believe the installer when they come to remove it.

Cheers,
Tom

 

[NTL2009]

Recording Power Monitors cost a LOT more than a small UPS.

Oh, I'd just monitor the 12V supply, or I could grab the 12V battery out the grandkids electric trike and a trickle charger for a stable supply. To monitor, I could whip up a bit of code for one of my NodeMCU boards - the 10 bit ADC can make ~ 10,000 reads per second. Even with a 7.5:1 divider to get to the middle of the 3.3V input of a NodeMCU, that would give me 30~40mV resolution, allowing for a few bits of dither/noise, it shouldn't be hard to catch a few tenths of volt droop/spike, and the thing ought to be able to handle a few tenths. Of course, the internal regulators might be getting flaky.

But it's been up for 5 days, and except for one reboot, no drops in connectivity at all, so I'm thinking the factory refresh/re-register did something, or they fixed it on their side. Probably never know.

 

[NTL2009]

UPDATE: I learned (hovered over the 'info' icon) that what I thought was my purchased modem spontaneously re-booting, is not (my scripts would log a drop, and I'd see that the "System Uptime" on the modem admin page was reset at that time). But their "System Uptime" is not time since a reboot, but the time since that connection was made. So, as far as I know, it has not spontaneously re-booted, I had cause-effect reversed apparently.

I've had my purchased modem up and running since 08AUG, so ~13 days, and I have not performed a reboot, and things are running well (maybe not perfect though). My scripts have recorded a few drops, but just a one or two events at a time (some scripts run every 60 sec, some every 30 seconds) - so I'd call those just temporary 'glitches' - less than one minute, not enough to interfere with streaming or browsing, and we haven't really noted any problems. I do get some corrected/uncorrected errors on a few of the 16 channels, but I don't think this is a significant issue? The connection reset this AM, and I've had only a single corrected error in those 3 hours. Earlier, over a 20 hour period, I had Total corrected: 56,113 Uncorrected 1,748. So ~ 2,805 and 87 per hour...

I'm going to keep it up another week, and if still reasonably good, return the rented modem on my next trip out that way.

 

[Rive]

Have had very reliable internet in the ~ 2.5 years we have been in this home

That's around the time I take those kind of things apart for a preventive re-capping (capacitor check and replacement).
Old story, but still prevalent
Can you check the interior of the device?

 

[NTL2009]

Well, it seems to be working OK now, and now that I've learned it does not appear to be spontaneously re-booting, I kinda doubt it is a power supply problem. But I'll keep this in mind. It's only 2.5 years old, seems early for a re-capping, except for that bad patch produced ~ 2000?

https://en.wikipedia.org/wiki/Capacitor_plague

The capacitor plague was a problem related to a higher-than-expected failure rate of non-solid aluminium electrolytic capacitors between 1999 and 2007, especially those from some Taiwanese manufacturers,[1][2] due to faulty electrolyte composition that caused corrosion accompanied by gas generation; this often resulted in rupturing of the case of the capacitor from the build-up of pressure.

 

[Rive]

Wiki is wrong about that, it did not end in 2007. I still regularly finding affected electronics: though affected motherboards are now indeed rare.

ps.: now these are usually cheap (not solid-polymer cap worthy) electronics in small enclosure.