Video showing impact of loose connectors (YouTube)

This step is important, because if you are experiencing Internet connectivity issues, but the cable modem log shows NO issues, maybe the problem is with your router, Wi-Fi, or something else?

But also be practical. If you don't notice Internet issues, don't worry that much about what you find in the cable modem diagnostic web pages.

TIP: Google the brand and model number of your modem and 'manual' (eg: "Netgear CM1000V2 manual") to find the manual for your modem, which should disclose the IP address to use to get to the diagnostics webpage for your cable modem.

TIP: For unknown reasons, when I manually power cycle my Netgear CM1000 cable modem (on Comcast), after reboot, the diagnostics web page is inaccessible. However, if I disconnect the RG6 cable from the back of the modem, and reconnect it a minute later, the cable modem diagnostics web page can once again be accessed (and of course, you need to ignore some of the errors in the event log caused by disconnecting the RG6 cable).

• Started Unicast Maintenance Ranging - No Response received - T3 time-out
• No Ranging Response received - T3 time-out
• T4 time out
• Dynamic Range Window violation
• RNG-RSP CCAP Commanded Power Exceeds Value Corresponding to the Top of the DRW
• RNG-RSP CCAP Commanded Power in Excess of 6 dB Below the Value Corresponding to the Top of the DRW

Again, you are looking for frequent (every day or so) critical errors in the cable modem event log. It is generally NOT considered a problem to see one critical error a week, or regular 'notice' messages in the event log.

Downstream Power (dBmV)

If power levels are low (less than -11 dBmV), investigate why and fix. Is your modem behind too many splitters? At the cable demarc location, there should be a single 2-way splitter feeding (1) the cable modem and (2) the rest of the house via another passive splitter. See section far below on a 'powered amp' alternative.

If power levels are too high, or 'hot' (greater than +15 dBmV), contact your cable company to resolve.

Downstream SNR (dB)

SNR is an acronym for "Signal to Noise Ratio". MER is an acronym for "Modulation Error Ratio".

SNR is "how many dB above the 'noise floor' the signal is" and effectively indicating how well the signal is heard above any noise.

An analogy: If you talk normally to someone at home, they can easily hear/understand you because there is virtually no noise -- so the SNR is very high. However, if you talk normally with someone at a bar, they can barely hear/understand you because there is a lot of background talking (noise) -- so the SNR is very low. You need to speak much louder to get 'the signal' above the 'noise floor'.

Upstream Power (dBmV)

See the section far below on the expected dB signal loss of cable length and splitters. The cable modem needs to increase upstream power by the dB loss of all splitters between it and the cable company. So you can immediately and instantly improve (reduce) upstream power by finding a way to remove splitters between the modem and the cable company.

Conversely, if you see NO uncorrectable errors and effectively zero correctable codeword errors (like 1 in 100,000,000), you know that the cables, connectors, splitters, etc in the house are in great shape.

I have multiple locations where over an extended period of time (like a week), there are NO 'uncorrectable' codewords (yes, none), and the number of 'correctable' codewords is so incredibly low (one in a billion) as to effectively be zero. But also be practical. If you see codeword errors, but are not noticing any Internet issues, don't worry about the errors (something I have at other locations).

Please note this discussion only applies to QAM channels, not OFDM channels. Because with OFDM channels, seeing 'correctable' codewords is normal.

Adding a two-way splitter into the modem run will cause downstream 'power' to decrease by 3.5 dB, but cause upstream 'power' to increase by 3.5 dB.

And conversely, removing splitters in the modem run will cause downstream power to increase by 'X' dB, but upstream power to decrease by 'X' dB.

But also watch out as there may be problems with both cable company wiring AND house wiring.

Note that for the duration of this test that your Internet will work, but all other cable connected boxes will be offline.

CATV whole-house feed in the cable demarc box

Do NOT use any splitters or other RG6 cable, and do not splice smaller length cables together. You want a direct connection from your cable modem to the RG6 feed provided by the cable company using only a single brand NEW RG6 line.

Look for a gray cable box on the outside of your house where utilities (electrical / phone / cable / gas / etc) enter your house (the demarcation -- or 'demarc' -- location).

You should see a cable demarc box with an often thicker cable line from the cable company (orange cable in photo upper right) feeding into a 'ground block' (and integrated MoCA filter; commonly installed by cable companies), which then feeds a line feeding the rest of your house (white cable in photo upper right). The 'disconnect/reconnection' point is just past the 'ground block' (for an example, see red arrow in photo upper right).

Another reason for testing using only a NEW RG6 cable from the modem to the demarc location is to eliminate the possibility of a damaged/bad cable somewhere in the house/walls (like during construction a nail hitting a cable line).

100ft run directly to cable modem

The expectation (when your cable modem is directly connected to the cable company demarc location) is that downstream power levels are 'high' (above 0 dBmV) and that upstream power levels are 'low' (well below 45 dBmV). If not, is the cable company not providing proper signals to your decmarc location?!

WARNING: Finding a problem 'somewhere' with the cable company does NOT eliminate the possibility that there may ALSO be secondary issues with house CATV wiring. This is what happened at my house. There were numerous problems, both with the cable company wiring AND with the house cable wiring.

TIP: If there is a problem after bypassing house wiring, leverage that fact to get a free tech visit to fix that problem. But when the tech arrives, tell them how you tested and what you found, but emphasize that you also want the tech to verify that all house wiring is OK as well. Doing this probably would have saved me a couple of later tech visits. Because as soon as I said I bypassed all house wiring and still found a problem, the tech ignored all house wiring (and missed finding a secondary problem in the house).

This is how one person found that an improperly grounded splitter was the cause of problems. Any 'grounding' wire connected to a splitter must be properly connected to a grounding rod driving into the ground, and not connected to something else (which may not itself be properly grounded).

In the process of doing this, I found a bad splice in a run through the attic space. While there should never be splices in runs, I can only assume the run was damaged during construction, and the only fix available was to splice the line (it was too late to install a new run).

I also mapped all runs in my house and labeled all cables, because I wanted to confirm that each 'run' only ran to a single location and device. Ideally, there should be no splitters in the home feeding multiple locations/rooms/devices, but the reality is that sometimes they can not be avoided (when a home was not properly wired when built).

YouTube video showing "suckout"

Some online comments also suggest that temperature variations, causing loose outdoor cables to expand/contract, can also be a root cause of suckout.

Notice that spotting "suckout" becomes much easier with lots of bonded channels (each channel is 6 MHz wide). With 8 bonded channels covering only 48 MHz of spectrum you are not very likely to notice 'suckout'. But with 32 bonded channels covering at least 192 MHz of spectrum, spotting 'suckout' becomes much easier.

This is what I finally did in one home that had intermittent Internet issues for years that Comcast TECH's could not fix. I found some horrible F-connector workmanship behind wall plates, replaced connectors (see section below on tools), and finally fixed the problem myself.

A bad F-connector behind a wall plate in any room in the house can negatively affect the whole house (and your cable modem).

Bad: YouTube video: How to do it wrong! Good: dielectric flush with F-connector base Bad: exposed foil; by a Comcast TECH

The cable dielectric was not flush with base of F-connector -- many were significantly 'under' the base (see an example in photo right) and some were 'above' the base.

One of the prior Comcast techs, who had just replaced ALL of those F-connectors only weeks ago (in an attempt to fix the unknown problem), did it wrong, and created this problem.

So a WARNING -- not all cable techs know what they are doing. And the Comcast tech who made this mistake bragged to me that he had over 15 years of experience working for various cable companies up and down the East Coast. Even many of the videos on YouTube showing how to install F-connectors do it wrong (the dielectric is not flush)!

Instead, if the 'whitish' material is at all below (example seen in photo upper right) -- or above -- the bottom of the F-connector -- the connector was installed improperly, and may cause all sorts of problems (allow external RF signals to enter the cable and create 'noise').

• The copper center conductor should stick out around 1/16" to 1/8" above the top of the F-connector -- and any more or less is a problem to fix. For reference, the thickness of a nickel is just slightly over 1/16", which is a great target distance.
  
  
• Make sure that none of the foil, or braiding strands, are exposed inside the F-connector (example seen right). Instead, you should only see the white dielectric and the center conductor. In the photo right, you can see where the foil was pushed in and broke instead of being cut -- the TECH was likely using a cutting tool with dull blades.
  
  
• When preparing a cable for adding a F-connector, don't let the blade on your tool touch (or 'score') the center conductor. 'Nicks' in the center conductor can actually cause issues (micro-reflections).
  
  
• Avoid incredibly tight bends in RG6 cable, which can push the center conductor through the dielectric and closer to the shielding. This can inadvertently change the 75Ω electrical characteristics of the cable.
  
  
• Never rotate a splitter/post (or grounding block) onto a F-connector to tighten it. Instead rotate the loose end of the F-connector onto the post/splitter. The entire point is to avoid spinning the center copper conductor into a connector. This is because connectors were designed to have the center conductor pushed in/out, not rotated in/out.
  
  
• Gently try to twist the base of any F-connector on the end of any cable. A properly installed F-connector will be compressed against the cable and will be so tight that it can not twist. But if it twists at all, the F-connector was not properly installed -- replace the F-connector.

F-Connector Compression Kit

• The compression tool (left, red)
• A handful of compression connectors (middle)
• The cutting tool (right, gray)

TIP: Practice! Take a short cable normally used to connect a cable box to the wall jack, verify it is RG6 cable, cut off the connectors and practice putting on new connectors using the 'kit' provided F-connectors. Because frankly, you want to know you can reliably replace F-connectors on a practice cable before attempting to do so on cable behind a wall plate (with a fixed limited length; very limited retry attempts).

The result is that the copper center conductor is always the same precise height (nickel thickness) above the connector for every connector I make.

Important: The second blade must NOT touch/score the center conductor. This is very easy to accidentally do, especially you have adjusted the blades for one cable and then move to a second brand of cable! So if you are just starting out, practice (if you have enough cable). Prepare the end of the cable and instead of putting on a connector, cut back more dielectric and verify that the center conductor is not scored (you can tell by running your fingernail over the center conductor). Some tools even have a notch in the second blade to help avoid this problem.

Video: Impact of loose connectors (YouTube)

• If you screw a F-connector onto a post and it 'stops' and is 'tight' early, something is very wrong. Find out why. Instead, the connector must screw on completely (around 1/4 inch).
  
  
• Make sure that all cables (F-connectors) are screwed onto posts a strong finger tight, as any loose connectors can actually cause RF ingress problems.
  
  
• A cable connected to a wall plate, but connected to nothing -- is also bad. The exposed (loose and unconnected) F-connector can act as an antenna, allowing for RF noise to enter the system. If there is no device connected, there are inexpensive 'cable termination caps' that can be screwed onto unused posts.

Loose connector at ground block

In the example seen right (red arrow), the black RG6 cable comes down the pole from the tap, and the orange RG11 feeds a house.

However, the problem at this location was that the orange connector, while seemingly 'tight' on the ground block, was NOT actually fully screwed onto the ground block. There was a couple of grains of sand (or something) preventing the connector from screwing on fully.

And the cable modem log was full of "RNG-RSP CCAP Commanded Power in Excess of 6 dB Below the Value Corresponding to the Top of the DRW" and "Dynamic Range Window violation" warnings.

Look into any F-connector. The distance from the base of the connector to the top of the connector (around 1/4 inch) should be the distance that the connector screws onto a post.

TIP: Use a powered amp only AFTER you have solved your cable modem problems! Because the one disadvantage of using a powered amp is that any 'noise' being generated on any run (inside the house) will then only be amplified and possibly interfere even more with your cable modem.

If you are unable to solve your 'noise' problem, do NOT use a powered amp. In fact, by sticking with a non-powered splitter, you are adding 'attenuation' to 'noisy' runs, making it much harder for the noise to cause problems with other runs in the house (like your cable modem).

BUT, once you solve any noise/cable problems, using a powered amp is a great way for all devices in the home to obtain the best power levels that are possible.

A 'typical' CATV/modem installation

For most homes, this configuration will work best. But if you home is running into problems with signal strength (too many cable boxes), consider using a powered amp.

TIP: Only use the new amps when really needed, when either (1) downstream signal levels must be improved, or (2) when two-way VoD cable boxes are having trouble working.

Older 'one-way' cable boxes will often times work just fine behind a passive splitter.

Using a powered amp is a great way to ensure that all two-way cable boxes (like Comcast X1) have the best chance of working properly in a large home where all cable boxes are two-way boxes.

Many older amps (still sold today) only amplified the downstream, and left the upstream with full (lots of) power loss, which could negatively impact VoD cable boxes. These old style amps have "passive return" and should now be avoided.

So if the IPA1008D-RSV is 'good enough' for the cable company itself to provide and use, I feel confident enough to buy/use elsewhere (and I have).

Extreme Broadband IPA1008D-RSVF

It is strongly recommended that you connect your cable modem (regardless of having VoIP, or not) to the passive port on the amp. Because this port also bypasses the diplex filter in the amp (operating at 5-42 MHz) in case your cable company has expanded the MHz range used for Internet upload channels.

When the cable industry moves to DOCSIS 4.x, it is expected that the MHz range used for 'upload' will expand well past 42 MHz.

While cable modems WILL (should) continue work on any amplified port (but maybe with limited upload speeds) the noise on the un-amplified return path on the passive port 'should' be slightly better (less) than the noise on the amplified ports -- as this avoids the "10.5 dB loss + 14 dB gain" return path (see below). I say 'should' because the critical SNR value that would definitively answer this question (how much of a noise difference) is in the CMTS and is not accessible to us. In fact, in Maryland, a Comcast tech installed a PPC EVO1-9-U/U and connected the cable modem to one of the amplified ports (not the passive port) -- so yes, it works.

The bi-directional amp amplifies only "H" (high) frequencies (52-1002 MHz) in the downstream direction (cable modem download) and only "L" (low) frequencies (5-42 MHz) in the upstream/return direction (cable modem upload).

The "VoIP/Modem" port bypasses the diplex filters -- a good thing if you have a cable modem that operates outside of the 'standard' upload frequency ranges.

I ran into this at one location where sometimes Internet connectivity was horrible, but at other times, it was perfectly fine. It turns out that when Internet was horrible, the cable modem was using a specific channel that had a ton of uncorrectable errors, and when Internet was great, the cable modem was not using the problematic channel.

Ultimately, the root problem was tracked down to a single bad cable that was negatively affecting a very specific MHz range (the single channel). Once the cable was replaced, the root problem was fixed.

Here in Georgia (Comcast), for my cable boxes, I just need to press/hold '0' on the remote for around five seconds to get the diagnostic screen seen top right. In Florida (Spectrum), same thing, but with the 'info' button.

Or, for Comcast X1 boxes, hold 'Exit' for five seconds, then press the down arrow (below ok) two times, then press '2'. Seems very timing sensitive, so repeat if needed.

You then try to account for differences in power levels seen in the STB diagnoistic screens between runs/rooms by comparing the difference seen to the expected loss (see the next section) caused by differences in splitters and cable run length between the two runs/rooms.

When the power level difference is not fully explained by differences in splitters and cable length between the two runs, then you have something to investigate.

1. There was a splitter behind one of the wall plates feeding an unused second location in the same room. So the splitter was removed.
   
   
2. All F-connectors behind the wall plates in this one room had the same very serious problem -- the dielectric was WAY below the base of all connectors. So all F-connectors were replaced.

For other cable lengths not listed, pro-rate the value for any other cable length that is listed. So for a cable run of 200ft, double the number obtained from the graph above for 100ft, etc.

TIP: A great rough linear 'guestimate' for the 100 ft curve, for any MHz above 100, is 2.25 dB plus 1/2 dB for every 100 MHz over 100 MHz. The same formula, but reworked for easy 'mental math' is to take the MHz, divide by 100 (move decimal two places), subtract one, divide by 2, and add 2.25. For example: (1) for 200 MHz, think (2.00-1)/2 + 2.25 = 2.75, (2) for 600 MHz, think (6-1)/2 + 2.25 = 4.75, (3) for 750 MHz, think 6.5/2 + 2.25 = 5.50. This answer is for 100' of cable, so then take the result and pro-rate for your cable distance.

Splitter	Loss
2-way	3.5 dB
3-way (balanced)	5.5 dB
4-way	7.0 dB
6-way (balanced)	9.0 dB
8-way	10.5 dB

Also note that the cable connected to the "IN" port always has the signal coming from the cable company, and that cables connected to the "OUT" ports always sends that (attenuated) signal to other devices (STB, cable modem, another splitter, etc).

There were several prior visits over the years that failed to resolve the problem. I even went through three different cable modems (and different vendors) trying to fix the problem, and all three modems had the same problem. I finally wised up and thought 'This is crazy, why am I paying for bad Internet -- I am just going to keep reporting the problem to Comcast until they find and fix the problem'. At this point I started documenting everything that happened.


1. 2020/06/12: The TECH blamed a bad underground cable run from the pole to the house (said landscapers must have hit the underground cable). A temporary 'above ground' cable was run that day, but it was replaced days later with an underground run by a different crew. Interestingly, instead of the previous RG6 underground run to the house, a thicker RG11 (a cable with a slightly lower attenuation) was installed. The problem remained.
   
   
2. 2021/05/19: The TECH blamed a bad port on the tap up the telephone pole. The house run was moved to another port on the tap. Comcast actually charged me $100 for this visit, which I finally got refunded. The problem remained.
   
   
3. 2021/05/29: The TECH blamed splitters in the demarc box on the side of the house. Both were replaced. One was a 3-way splitter feeding (1) the modem, (2) the X1 cable box, (3) an 8-way splitter feeding all other one-way cable boxes. The problem remained.
   
   
4. 2021/08/18: The TECH proactively replaced all F-connectors in the demarc box on the side of the house. A text from the TECH said "all that was done was new fittings". The problem remained.

   But see 10/01 below -- all F-connectors were installed improperly. Also, when I looked in the cable demarc box, I found that the line feeding the entire house was 100% loose -- touching and not even screwed together at all. The TECH clearly missed putting a connection back together!

   2021/09/09: The TECH never showed up for the appointment.
   
   
5. 2021/09/11: The TECH blamed "suckout" and "low mEr" at the tap on the telephone pole (see photo upper right). He submitted a ticket to Comcast to get that fixed. The graphs far above in this paper show an improvement in signal levels. The next day, Internet seemed better and was better for weeks (but I also installed a lossless amp at the exact same time, which improved SNR), but the problem ultimately 'came back' in late September.

   Oh boy, doing it wrong: Notice the TECH (photo upper right) hooked the ladder on the actual cable lines, instead of hooking the ladder (only inches higher) on the much stronger messenger cable. Also the TECH was testing an unused port on the tap instead of the port feeding my house.

6. 2021/10/01: The TECH blamed improperly installed F-connectors in the demarc box. Major problems were found (dielectric not flush) with all F-connectors on all cables in the demarc box on the side of the house, and ALL were replaced. The TECH said it was so bad that he was surprised that the X1 cable box was even functional. We noticed a recent brand new problem of 'blocking' on TV channels that also went away after the F-connectors were replaced, confirming the impact of the bad F-connectors. Absolutely crazy since these F-connectors had been inspected and replaced (apparently improperly) only weeks earlier by another TECH on 8/18. The problem ultimately remained.

   I later found a near short (exposed foil) in one of the connectors put on by the new TECH. Not at all surprising since I watched first-hand as the TECH used the compression tool and then did NOT inspect the resulting connection for defects. If he had, he would have found his own mistake in workmanship.

   Overhead Cable Tap

   2021/10/08: The TECH never showed up for the appointment.
   
   2021/10/09: I spent the next several days tracing cable runs, inspecting ALL connectors everywhere, and replaced several bad connectors. See further below for more details.
   
   
7. 2021/10/12: FINALLY, a Comcast TECH (employee) came out (not a contractor), who said they come out when too many contractors are sent to the same location and are unable to fix an issue. The TECH focused on the telephone pole; found that the ariel connection at the tap was loose (see photo right); replaced all connectors at the pole for the drop line to the house (one high; two low); tested signal quality at each step; tightened all connections using a wrench; worked back to the demarc location; eventually ended up at the modem location inside the house; tested and said everything looked good. At first, Internet seemed better, but the problem remained.
   
   
8. 2021/10/31: The local "Technical Operations Supervisor" came out (I was not at the house) and said that one of the runs in the house was causing noise ingress. A hi-pass 'filter' was placed on that one run -- so that any ingress on this run (causing interference in the frequency range used by the cable modem) would be blocked.

   I presume what the Supervisor installed was an "in-line high pass coupler barrel adapter 54-1000 MHz filter" (example seen right) on the 'noisy' run that had a one-way cable box. The filter blocks the frequencies below 54 MHz (which are used by modems and two way cable boxes). In this way, the noise ingress from that one run is blocked from impacting the cable modem upload channels.

   2021/11/21: The problem went away for a short period of time, but I was able to remotely confirm that the problem still remained!
   
   2022/01/19: I came back to the house to find that the Supervisor had placed a 'lock' (see arrow in photo right) on the demarc box, which was a jerk move because inside that box is MY amp, and it prevented me from further testing on my own. I asked the Supervisor to remove the lock, and she agreed (in writing), but no one ever came to unlock the demarc box!
   
   2022/05/29: I finally bought the special $10 tool on Amazon that allows me to lock/unlock the demarc box myself. So once again, I connected the cable modem directly to the Comcast demarc location (bypassing ALL house wiring) and confirmed that the problem still remained, even with all house wiring disconnected. I still experienced regular cable modem T3 drop outs. A meeting with the local Supervisor was scheduled.
   
   

   tech replacing drop cable

9. 2022/06/11: The local "Technical Operations Supervisor" was supposed to come out a second time, but instead sent a regular contractor out, not even a Comcast employee.
   
   The tech arrived to find me testing and my modem connected directly to the Comcast demarc location (bypassing all house wiring) and still experiencing significant Internet drop outs.
   
   I explained a brief history of the problem and stated to the tech that even when I connected my modem to the base of the telephone pole, I still experienced problems (eliminating the underground run from the telephone pole to the house demarc location as the problem), and that this meant that the problem was somewhere at the telephone pole on up, and I stated that I wanted both the TAP (that a prior tech said had a bad port) and drop cable (the line running from the TAP to base of the telephone pole) replaced.
   
   The TECH explained that only Comcast itself (and not him, a contractor) can replace the TAP, but said that he would do the only thing that he could do, which was to replace the drop cable. So this tech: (a) replaced the drop cable from the TAP to the bottom of the telephone pole, (b) replaced the ground block at the bottom of the telephone pole, (c) replaced the connector on the underground run to the house (at the ground block), and (d) moved the drop cable to a different port on the TAP (to a port that surprisingly a prior tech had said was bad).
   
   I also watched this tech incorrectly spin a ground block onto the end of the cable, instead of turning the cable onto the end of the ground block. Argh!

The ultimate core problem at this location was either (1) a bad RG6 drop cable (the run from the TAP at the top of the telephone pole to the ground block at the base of the telephone pole), or (2) a bad ground block at the base of the telephone pole.

The F-connectors on the drop cable were replaced numerous times over the years by various techs -- without helping -- which excludes bad F-connectors as the root problem.

1. A connector at the ground block at the telephone pole for the run to the house was 'tight' but NOT fully screwed in. After fixing, cable modem upstream power stabilized for a time.
   
   
2. The TECH on 2021/10/01 had replaced a ton of connectors, but actually left with a replacement connector in a virtually "shorted out" condition (see photo right). The foil was not cut properly and wrapped around the dielectric and was a hair way from touching the center conductor.
   
   
3. I inspected ALL connectors behind wall plates in the house and replaced those connectors where the dielectric was not flush with the base of the connector.

1. Don't assume that there is just one problem to fix, because in reality, there will very likely be multiple issues to fix. And at this location in Georgia, there were problems with both the cable company and with wiring inside the house.
2. If a couple of tech visits don't solve your problem immediately, insist on a subsequent visit that ALL cables, ground blocks, etc, all the way from the TAP to your house demarc location, be replaced! If I had been more assertive and requested this, my Internet problems would have been solved years ago.

I even found a connection behind a wall plate completely loose (touching, not screwed together), and the last person to touch that wall plate was a Comcast cable tech.

During this inspection, I found (and replaced) many connectors behind wall plates where the dielectric was well above/below the base of the connector The original house builder/electrician did this many years ago and was clearly not very skilled!

"The number one thing we actually see as a 'craft error' is that the connector is not properly tightened onto the port". Rich Maroney, Product Line Manger Coaxial Cable, PPC. [Source: YouTube video]

• Evaluating Your Cable Internet Connection (by Motorola)
• F-Connector Installation Tutorial (YouTube video)
• RF Ingress - see the effects of a loose F-connector (YouTube video)
  
  
• Cable 101 - History and CATV Basics (YouTube video)
• DOCSIS 3.1 - An overview (YouTube video)
• Cable Math (PDF)
• DOCSIS basics tutorial series (by Volpe Firm)