Why did some things happen? Why didn’t others…and what happened instead? To what extent (if at all) were the outcomes predestined? And what outcomes beg for further study?
Ten months back, as I’m writing these words in mid-August 2026, I never would have imagined the long and winding road that awaited me when I realized last October that a water hydrant replacement across the street had inadvertently led to a Comcast outage for my wife and I, along with our next-door neighbors.
The path from there to here has admittedly had no shortage of twists and turns, not to mention frustrations. But (fingers crossed that I don’t jinx myself by typing these words) we’re now at the point where we once again have stable service, thanks in no small part to the most recent technician interaction.
I’ve learned a lot along the way. And the combined fruits (hopefully not rotten) of my experiences and education will likely manifest in further editorial content for quite some time to come, starting today, with the latest (last?) entry in my recent series, which began with these pieces:
- Debugging intermittent Comcast, part 1: Scenario-setting
- Debugging intermittent Comcast, part 2: Remediation details
As I wrapped up part 2, I was three-plus weeks into my respite of resurrected broadband and television service stability. It’s now been exactly six weeks, and my connection has remained stable through weather both hot and moderate, and dry and rain-drenched, at least as far as I can tell. Brief outages that escape attention are always a possibility, although my Google Nest-based LAN, Blink cameras and TP-Link sensors, and QNAP NASs are generally quick to log and alert me to broadband drops. And summer monsoon season is once again here with a heavy-lightning vengeance, so I’m also abundantly grateful for the newly added earth ground tether.
I’m going to subdivide this concluding (maybe) writeup into several main themed sections:
- Why did the most recent service visit’s technician tweaks seemingly do the trick?
- What’s with that mysterious Comcast-labeled box on the side of the house, not to mention the realtor’s mention of a second service feed put in by the prior owner?
- And, transforming gratitude for what is into grasping for what might be, why isn’t my broadband service even speedier than it already is?
Without further ado…
Hardware replacements, retirements, and most-likely effective candidates
As a brief reminder—read part 2 for the full details—the technician had swapped out an archaic (predating our home ownership) three-way splitter and 11-year-old (purchased and installed by me) MoCA PoE filter combo attached to the outside of our home.
The replacement hardware took the form of a filter-inclusive grounding block.
As I mentioned last time, the componentry inside either/both legacy pieces of gear could have degraded due to environmental extreme-temperature, moisture or other related exposure, lightning EMP-induced damage, and the like. More likely, the splitter was inherently a culprit due to the unavoidable signal insertion loss through it. I realized after the technician left that he’d taken it with him, so I unfortunately can’t say whether it was a balanced or unbalanced splitter. This distinction is critical, both inherently and (in the latter case) depending on how it’s hooked up.
With a balanced splitter, each output “leg” incurs a symmetrical 4.77 dB (ideally, more in real life) per-output signal loss, per the decibel power ratio equation:
10log10(P2/P1)
Where P2 is the measured power and P1 is the reference power. In this case, the ratio is 3. Keep in mind, too, that real-life SNR loss varies across the operating frequency range, as well as being affected by overall non-ideal component behavior, vendor-to-vendor and splitter-to-splitter variability, and the like. Therefore, ~5.5 dB of per-“leg” loss for a balanced three-way splitter is a more commonly quoted estimate.
An unbalanced three-way splitter, conversely, is in effect a chained combination of two two-way splitters. The “low loss” leg only passes through the first two-way splitter stage, with a ratio of 2, leading to an injected signal loss of (mathematically) 3.01 dB or (real-life) ~3.5 dB. Conversely, both of the “high loss” legs, in each case with the signal passing through both two-way splitter stages between the input and output, incur 6.02 dB (mathematical) or ~7 dB (real-life) losses.
Perhaps obviously, therefore, the signal strength received at the cable modem in the furnace room (for example) notably varied, depending not only on the Comcast-fed source strength, which we already knew had been degraded by our location at the end of the neighborhood “loop”, not to mention the late-2025 repairs and mid-2026 amplifier recalibration.
It also varied depending on whether the three-way splitter was balanced or unbalanced, and in the latter case, where each output “leg” went. And don’t forget about the final two-way input signal split within the furnace room, to the cable modem and networked CableCARD triple-tuner receiver. At least this specific situation is better than in the three-way splitter past, when a MoCA device was also in the mix!
This analysis assumes, by the way, that the coax cabling itself wasn’t egregiously lossy. I still think the greatest benefit came from eliminating the splitter itself, not from eradicating any degradation caused by the short cable runs originally connected to the splitter and feeding both downstairs bedrooms, particularly since I’d had a 75 ohm terminator installed at the end of each to prevent reflections and other signal distortions.
Regardless, I’ll need to keep all these factors in mind if/when I decide to reactivate any-to-all of the downstairs MoCA nodes in the future.
And if you’re wondering about potential signal loss through the PoE filter, per the specs for both original (Holland Electronics MPOE-TM) and new (PPC Broadband POEGB-1G70CW) devices, the passband injection loss is miniscule (~1 dB, frequency-dependent), assuming proper operation, that is! Stopband loss is a different matter—40 dB or 70 db—but that’s by design, of course.
Cracking the coax code
Next, let’s definitively sort out all the cable and other hardware mounted to and running around the outside of the house. Last time, I’d surmised that the plastic box on one wall was likely associated with the house’s legacy Comcast service, and I’d provided a glimpse of the four-way switch within it.
Here are some more views of it.
Where do its four outputs go? One heads straight up and into the attic, which baffled me until I remembered the female coax connector on an internal kitchen wall in the middle of the house’s upstairs level. I suspect (and could, I suppose, confirm by crawling around in the dusty, precarious attic) that’s where the first cable ends up.
Another cable runs both above- (presumably due to post-installation soil erosion) and under-ground to the rear of the house, where it snakes underneath the lower deck and, I’m guessing, ends up at a wall-mounted coax connector in the downstairs family room. The third traverses the outer walls, first to one side where it splits, with one output cable heading to the upstairs master bedroom. The splitter’s other output cable continues its lengthy journey to the upstairs living room on the house’s other side.
And the fourth? It’s currently disconnected and coiled up nearby the box. I suspect it was what originally fed the downstairs bedrooms, having subsequently been replaced by the newer service feed from the street mentioned earlier.
Speaking of which, what’s feeding the four-way splitter? I’d always wondered about a cut piece of coax jutting out of the ground nearby the box.
Until, that is, I learned about the earlier-mentioned tap serving our and a neighbor’s residences.
Next to it (in the background of the prior photo) is another enclosure that I’d also long wondered about.
Note the coil of coax at its base. That’s, I suspect, from the original service feed.
Remember me mentioning in part 1 that I was baffled by the two cable spans that had apparently gotten severed? I’m guessing this is why: old and new taps, next to each other, with the old cabling and hardware not removed after being taken out of service. I did definitively confirm, by the way, that there’s no ongoing service associated with the original coax cable topology, first by unscrewing the four-way-splitter input and seeing that my service didn’t go down.
I then attached the now-disconnected coax to a spare cable modem to confirm that it didn’t detect a valid service signal. Hitron Technologies makes, or maybe more accurately used to make, a portable battery-operated device that claims to more conveniently implement this valid service-or-not function. But I can’t find it anywhere at retail any longer, and the company is non-responsive to my outreach. That said, I’d still love to get my hands on one somehow!
Last mystery, at least for this section; what about the realtor’s aside at some point during the residence purchase process that the prior owner “had another line of service installed”? Comcast doesn’t (commonly, at least, more likely at all) grant customer requests for multiple from-street feeds for a single service account, specifically to preclude the otherwise-possible scenario of multiple concurrent in-use broadband modems. But keep in mind that this all happened more than a decade ago, when corporate policies may have been different (and mind you also involved a prior owner who’s now deceased and I therefore can’t ask).
What if only one of the two feeds had a modem connected to it? And what if the customer was pugnacious (as the prior owner was reputed to be)? Particularly given that the house was at the end of the “loop”, with associated degraded signal strength, did he convince Comcast to run a second line, feeding subsets of the house’s total TV service sockets from each?
From lingering past-history evidence, I’m pretty sure that the cable modem and tethered router were originally located in one of the downstairs bedrooms. But I’m aware from various neighborhood anecdotal factoids that the prior owner had active television service in multiple upstairs rooms, too. Maybe he set up the second line of service through a fictitious business listed at that same address. Or maybe he was just (impressively) pugnacious.
Regardless, although I’d initially theorized that the realtor had meant not “had another line of service installed” but “had a replacement other line of service installed”, I’m now more inclined to take her wording literally, although the newer feed is the only one now still “live”.
Next time: DOCSIS 3.1 and MoCA 2.5
Nearing 2,000 words, I realize I haven’t yet gotten (surprise! not…) to the originally planned “why didn’t my service get better” portion of this treatise. For now, I’ll concisely cut to the chase. Even though the technician had wrapped up his early-July visit with an excited prediction, “wait until you see how fast your Internet access will be now!”, I eventually realized that Comcast’s speed tier options at my location only extended to a high-end “1.2 Gbps” (your actual mileage may vary) downstream option.
I’m currently on the next-highest tier, the “Gigabit Plus” plan, with maximum documented downstream bandwidth of 1.3 Gbps and peak upstream bandwidth of 35 Mbps.
Earlier, I showed you a fairly typical bandwidth test result: ~850 Mbps down and ~42 Mbps up (the latter higher than promised, I’ll note). My router more generally does an automated speed test every few days, and scanning back over a month’s worth of data, the highest seen downstream bandwidth was 890 Mbps, along with a 43.3 Mbps upstream measurement.
To be clear, this speed is plenty fast for our needs. What I’m most happy about, of course, is its now-resurrected “dial tone” permanence. And given that both my current cable modem and router offer “only” 1 GbE inter-device connections, I’ll never see anything more than ~950 Mbps downstream (accounting for Ethernet protocol overhead) anyway.
That all said, being an engineer, I’m prone to tinkering regardless. Prior to doing sobering account research and fueled by technician enthusiasm, I did a bunch of testing rounds and learned a bunch in the process, all of which I’ll plan to pass along next time. Until then, and as always, I welcome your thoughts in the comments!
—Brian Dipert is the associate editor, as well as a contributing editor, at EDN.
Related Content
- Debugging intermittent Comcast, part 1: Scenario-setting
- Debugging intermittent Comcast, part 2: Remediation details
- Will a location change improve my MoCA?
- The whole-house LAN: Achilles-heel alternatives, tradeoffs, and plans
- A quest for faster upstream bandwidth
The post Debugging intermittent Comcast, part 3: Retrospective analysis appeared first on EDN.