Partial SA Sweep of ISP Cable Feed – TinySA Results

[Jan Janowski]

last week was my first attempt at this sort of thing..... But this is the area just below and above where Modem Diagnostics's say is the 26 Internet Receive channels are.

This is a partial SA sweep of my ISP cable feed as documented using a TinySA Spectrum Analyzer with 75 to 50 ohm converter with a 5.7 Db Loss... attached to the 75 Ohm RG6 coax..

At the time of this capture my Internet service was running fine, Modem Diagnostics had no complaints of connection issues, and few if any corrected or un-correctable packets, and at speeds that met my ISP's service: 400Mb service... Speed tests: 390/46.

There was little difference (couple db) if modem was off or on, looking at cable via a "T" with shortest connections, or directly connected to end of cable.... In each case - the captured display had this display's overall shape.... (display averaged across 16 scans)

 

[White Bear]

This appears reasonably healthy in terms of the physical cable connection, itself. Some rolloff at higher frequencies is normal. Spectrum analysis is good for diagnosing 'analog' issues - but not 'digital'. So are you concluding that problems are most likely with either the ISP or modem/router at your end??? Could you restate the problem you're trying to solve, as a reminder for those of us who are following your blogs?

 

[Jan Janowski]

This capture was after trying multiple different connections. And since cable is 75 and SA 50 ohms impedance, ordered a 3GHZ 75/50 ohm converter soas not to cause (or be fooled by) other SWR issues! I agree that modem diagnostics via Events list is a better way of discerning issues, especially in events of the “T” persuasion! (T3, T4, T2).
However, Modem diagnostics only work after modem is “Locked & Provisioned”… Thsts where SA gets involved! Once I know what a “Good” feed from ISP Looks like (the spectrum is nearly same with no modem). Then, when modem doesn’t lock- I can get out SA and see if Anything is being fed!
With ISP ‘Support’ 600 or more miles away, on lines that occasionally drop out randomly (

(Case in point 3 weeks ago.. I was fine and contacted support on where their amp’s RF AGC Carrier was, in frequency, and they said my line was “dropping out”.. they sent tech next day who said line was FLAT… MO PROBLEMS. From thus we both believe supports lines occasionally drop, but this also was incentive to learn more: Hence the SA)!!
The flat-ness you see was the best I could get SA to display in a week or so different length & type connections!
Thanks for the “good-enough” status of that capture. At this point it’s a reference as to what might be optimum.

 

[White Bear]

So the problem is occasional drop outs; and the suspected cause is an intermittent issue with the physical cable connection?

 

[Jan Janowski]

Oh boy! You asked!
Sometimes it’s a tree falling across line. Betcha that is flatline.
Sometimes it connects but break in between that and Provisioning.
Sometimes modem locks & Provisions, but it’s a break between Provisioning and Internet altogether.
Sometimes all works but no one in development can lock or provision modems - because the RF ACG amps in the amplifiers oscillate. This exceeds the +-10db AGC Range of modems, and no-one, Support included 600 miles away, can connect! (saw that on their own test gear and Tech was on phone with another-seeing same thing ‘somewhere else’!). That was the impetus to purchase a SA - to be able to confirm outages - but with that - I would need to understand what a good signal looked like! A ‘Reference’ as to what is good, or nominal!!

Make sense?

 

[White Bear]

Yes. Sounds like a lot of variables. Power supply degradation in aging AGC amplifiers is fairly common; and could cause oscillation. Can you correlate failures to time of day, temperature, etc.? What DOCSIS level? My EE background is mostly RF (and long ago). A friend designs cable/fiber distribution systems for rural cable companies. So we're following along... Thanks for the summary.

 

[Jan Janowski]

My modem is DOCSIS 3.1, ISP is CGNAT.

Thanks for your input!!

 

[White Bear]

Thought you might be on an older DOCSIS version. 3.1 is current. I believe CGNAT refers to a type of Network Address Translation vs. an ISP. Looks like your area is served by Spectrum or Zito? If Zito, my sympathies - they built their network on the back of several legacy cable systems; so aging and historically poorly maintained infrastructure makes for many possible points-of-failure and lowers the odds of a happy ending. Good that you are persistent.

Edit: Did a bit of research... Assuming Zito, they may be using CGNAT in your area; which could explain your previous difficulty in accessing home systems remotely (but not the outages).

 

[Jan Janowski]

Yes….Zito.

They bought out Carolina Mountain Cable, a true “Mom & Pop” ISP. Wasn’t fast, but was un-believably reliable: Because they lived here and used it!! Now I’d say we are: ‘fast & often intermittent’

One thing: you Do Learn how to troubleshoot: out of Necessity!

(Because you do it so frequently!)

That’s how I came up with how Synology Routers stop asking modem for ISP IP & Gateway address! Because I had so many instances that the connection would fail!

Our development is in a catch 22 position:
We have Internet: wired up with cable in 1994-5.. so no new ISP can apply to government for funds to cable the place with a competitor, and though Internet via RF IS AVAILABLE…. The ISP (T-Mobile) transmitting antenna is on other side of the mountain. Sig so weak it’s like DSL AT 6000’ from a DSLAM, on UL twisted pair! (Had that in Illinois, eons ago: so I know what I speak!)

So I get real good at troubleshooting and reporting it to “Support” hundreds of miles away reading from flip cards. Hoping against hope that I can educate enough folks to where system will become stable!

PS: cable feed to house: ~100’ of underground rated RG-11 from pole to garage, where it joins my 125’ of quad shielded RG6, that goes to the house via a buried 3” ID Dry Pipe, direct to modem….. So 225’ I guess explains the rolloff, ????

 

[Jan Janowski]

I Think I have found enough information to explain the roll-off I'm seeing...

From Pole to house the Cable takes this path:
~100' of underground rated RG11 cable.... from top of the Pole: to the inside of our Garage... There, it joins my Quad Shielded RG6 cable that is ~125' long from Garage to House, routed in a buried Dry Pipe... direct to the modem.... no splitters.

Looking at MFG's specs of the cables... I'm seeing that 100' of RG11 should have a 1-2 db rolloff.... Let's say 1.5 db.
100' of RG6 should have 5-6 db of rolloff..... lets say 5.5 db... 125' then = 6.875 db + 1.5 = 8.35 db,,,,, 8.35 db should then be the theoretical loss the the 225' feed should encounter on this cable route!

I'm only seeing 6 to 6.5 db of loss with ~100' of RG11 + 125' of RG6.... (This measured from carriers just Below and Above the 26 Receive Channels Indicated in Modem Diagnostics)...
So I'm saying my results are a bit better than mathematically expected... Could be shorter cables, lengths were "Guesstimates" and this accounts for the roll-off seen in the posted capture: The 'normal' attenuation ~225' of cable introduces....

This would seem to remove ME from the equation! (Which is what ISP's Service guy said 2 weeks ago... "Your OK -- FLAT" But I'm just stubborn enough to want to find out for myself!!!)

 

[White Bear]

First and second-hand opinions, here, agree - the rolloff is most likely NOT an in-house issue for you. However, there may still be coax SWR issues, given your situation. A Coax TDR (Time Domain Reflectometer) could provide a more detailed analysis; and would also help physically locate any issues (e.g. a pinch in the coax). TDRs can be rented (try google search in your area). But this level of testing is likely overkill unless you suspect the coax has been compromised.

 

[Jan Janowski]

Interesting, but now that would be like polishing a turd... (an: Up-stream Turd!)
Since the RG11 was added by previous cable company, who swept "my" cable to modem in july 2015 a couple weeks after retirement here: "We Don't usually use customer supplied cables, but your Quad Shielded RG6 and dry pipe is better than anything we would have ever used, so we're running RG11 to your garage! CMC Techs: 2015" It Passed! That type of scan might have been done then, it was bigger box than what Zito guys carry... Zito guys have come here twice in past years un-announced, saying they want to measure 'things at this elevation' in our development. Last time he was reporting via phone to another tech, who saw similar issue...so I think there might be notes on our cable...

in the 3 weeks since testing, I see SA Displays match and confirm Modem Diagnostics over the years: a Daily receive level changes of nearly 10 db... Doesn't seem to match speed tests, or Packet loss, Not certain they follow temps or humidity, but those 2 seem to follow one another.... Will be something to watch.....

Watching wider SA sweeps, and comparing to videos on internet, SWR should be addressed by the 75/50 ohm converter (good to 3GHZ, so MFG Claims SA only good to 960MHZ)... Though with it, and far wider sweeps I'm not seeing any wavy-ness in scans like what I've seen on internet, supposedly caused by SWR.....
It's a "Learning Experience"

 

[Jan Janowski]

I've been printing out modem diagnostics Status Page daily, then connecting and capturing the same SA Sweep and printout of the area of Receive channels daily.. From this have recognized the difference the ISP's AGC actions have on both SA spectrum and the Modem diagnostics's decoding of that signal...
Met with Zito Service guy and he opened up on some specifics.... He remotely, then in person at cable end to modem swept and confirmed my connection was fine...
I had been baffled by his reference to my Received Spectrum as being "Flat" on his test unit, where I was obviously seeing a roll-off.
My video background "Flat" means NO Roll-off, and I was trying things (all Un-successful) trying to flatten the display...
He clarified that his "flat" means no significant HF or LF Roll-off, nor "NOTCHES" in the RF.... Normal Roll-off from pole to Modem is acceptable... and what I was seeing was not extreme, meaning that the roll-off was within the mathematical calculation of what it would expect to be.... Semantics!!
My daily spectrum printouts have confirmed that the Roll-off is consistent within 1 db day to day... Though the RF amplitude may vary due to AGC actions of the ISP's amps....which I AM SEEING, and never knew exactly why it varied!

That explains why I could degrade the rf with different connections, but not improve upon it... The SA was showing me exactly what it saw! I now have a reference I know is 'Optimum & Correct' for my connection, that I can compare against in the future... And learned a bit in the process!!!

 

[Jan Janowski]

I submit today's SA Capture... (Via a 75/50 ohm converter with 5.7Db loss) 225' from Pole feed... It does have subtle improvements over the first one, who's modem connector wasn't perfectly crimped!
PS: the large area to left of receive channels is their Internet TV Channels, and small area to right of receive channels is "911 data area" (he wasn't clear on what that is...)

 

[White Bear]

Regarding 911 data area on the spectrum analysis... In the U.S., the FCC requires that providers of VoIP telephone services using the Public Switched Telephone Network (PSTN) meet Enhanced 911 obligations. These systems must automatically provide emergency service personnel with a caller’s callback number and, in most cases, registered service address, phone number, account identification and apartment/unit or room location within a building. The FCC vigorously enforces E911. But unfortunately, E911 does not include information about hazards (e.g. explosives, chemicals, etc.) that may exist at that location - that's maintained by local first responders (and not vigorously enforced, as we learned from a bad experience).

Did you have any luck with the test gear Zito gifted to you?

 

[Jan Janowski]

No it won’t boot at all
Could you use it?

 

[Jan Janowski]

Battery is good though

 

[White Bear]

Thanks; but I'm way past the tinkering stage - more into decluttering. Might be worth something to someone - even though it is not DOCSIS 3.1 compatible.

 

[Jan Janowski]

For all others…. ISP service guy gave me inoperative test kit:. DS2500Q
I was able to verify battery was good, (12v5A supply needed, per manual. ) it shows no signs of water or internal damage,
Does correct charge RED/Green sequence & charges correctly, but it won’t boot. Kinda heavy. I’m gonna pitch it if no one is interested. I’m in far western North Carolina.