5dB at the tap enough for internet? [Text View]

View Full Version : 5dB at the tap enough for internet?

Paragons

04-14-08, 08:33 PM

GoDucksinAZ

04-14-08, 11:20 PM

Ive had TW cable out 3x now trying to get my internet service back w/o constant dropouts.

Now had new cable from the "tap" to my garage (this is about 200 ft long) and a new cable modem and new splitters and new amp.

Last tech replaced a 4 out amp with a 2 out amp and used a 3way splitter to the digital box and another 2 tvs. The outs are labeled 11db and the one going to the modem is VOIP 11bd normal/-1db backup. The amp is SVI.

In the room w/ the modem a new splitter sends the signal to a 4th tv and the modem.

The last tech said the tap shows 5db for the channel for the internet and is higher for the channels for tv (not having any problems with any tv signals).

He said they'd probably have to up the signal to the tap... only 5 db at the tap some 200 ft away even 300 ft by the time it reaches the modem, enough strength?

Im losing my mind, calling them back over and over wondering what the problem is, software cable, router etc, anyone saavy enough with some insights?

Thanks for any advice.

Pete

+5db Rx at the tap is acceptable for standard application. DOCSIS Modems have a performance range of +5db to -10db on Rx. The modem will perform to spec anywhere within that range. A tap offering of +5db allows for 15db of play so +5db is nothing for TW to be alarmed about. That said, it may be an act of congress to have them reengineer their plant to support your alt from standard household.

You have a unique situation that may be somewhat challenging to resolve due to the following numbers. I dont know TW cable systems from a hole in the ground but I imagine the HSI Rx frequency is somewhere in the 500mhz range. 500mhz loses 4.9db per 100 feet of cable. Basically, you are at losing about 10db by the time you reach the Dmark with the amp. Your signal going into that amp should be about -5db or worse on the Rx. It is already relatively poor but is now being amped - effectively, polishing a turd.

+11db on that amp should put the modem Rx back at +5db or so but then it hits a 2 way split and loses 3.5db again. If all cabling is 100% certified and without problems, your modem Rx should be about +2db. Any shoddy fittings or cabling with further decrease that number.

The Rx is just your downstream path however. You also have an upstream path to be concerned about as well. Each "standard" passive device will increase the upstream what it would loose on the downstream. For instance, a 2 way splitter drops Rx 3.5 per port. It also "increases" Tx 3.5 per port.

This is why your cable guy swapped your 4 port amp for a 2 port. The Tx has a range of 40-50 Dbmv. Each tap offers up a Tx number based on its value. Say the tap is offering 40Dbmv on Tx off the tap, plus you gain 3.5 fron the amp and 3.5 from the two way split. Your modem will have a Tx level of about 47, which is acceptable. Prior, the 4 port amp increase Tx 7 per port plus the 3.5 from the 2 way split, your modem was sitting at 50.5 - out of range, hence why he swapped it. The tap could be offering 45 however in which case you are really screwed. You would have to remove that split with the Tv to be within range.

Everyone focuses on Rx because its the most digestable but the modem will perform poorly if either level is out of range. Hopefully that makes sense to you on how to chase each level.

Good luck but it might take some serious hoop jumping to get TW to adjust their plant for a lengthy house design that they didnt not engineer to support.

kbullkar

04-15-08, 07:27 AM

The way I read your post, here is my interpretation:

Tap = +5dBmV (power levels should be in dBmV, loss/gain should be in dB)
200 ft of RG6@550 MHz (guessing here) = -13.4 dB = -8.4 dBmV
3 way dist amp = +11 dB/-1dB = +2.6 dBmV normal, -9.4 dBmV during a power failure
100 ft of RF59@550 MHz (again, a guess) = -8.2 dB; so -5.6 dBmV (-17.6 dBmV).
Another two-way, so -3.5 dB ending at -9.1 dBmV (-23.1 dBmV)

Passives lose signal power in both directions... ideally in a two-way splitter, 50% of the power goes to each output, so you have 10 log (0.5) = 10 (-0.3) = -3 (and we would say it is -3 dB). We aren't ideal and the actual loss can be closer to 3.5-4 dB depending on the quality of the splitter. When used as a 2-way combiner, half of the power goes where desired, and the other half hits isolation circuitry, so you still have a loss of 3.5 dB).

I used the loss calculator at:

http://www.timesmicrowave.com/cgi-bin/calculate.pl

During "normal" operation, the signal is hitting the cable modem at -9.1 dBmV, which is within the "legal range", and during a power outage, it will probably have enough AGC capabilities to overcome the additional loss...

Here is what I would have the tech check next time he comes out:

At the tap and BEFORE the amp have him check:
Power level (in dBmV) and carrier to noise (CNR) or signal to noise (SNR) ratio (in dB).

If you can get the MER of the DOCSIS downstream, that would be a bonus.

If you have a lot of loss at the frequency the cable modem uses, have him replace the drop.

The power level less the CNR/SNR gives you roughly the power-level of the "noise".

Now your three-way looks like it has a directional coupler in it with passive return, so I would account for it having roughly 11 dB of internal losses, meaning the "amp" part is really ~22 dB. This means your noise increases in power by ~22 dB. Noise stays relatively constant, so you want to amplify as early in the run as you can. If the CNR/SNR gets too low, you will have problems.

If you post your cable modem model, someone here may be able to give you instructions on accessing the diagnostics web-page which usually includes MER and/or SNR/CNR.

If all of this looks good - check your upstream.

If your cable modem is transmitting at max power level, chances are you are seeing too much attenuation. If it is transmitting at minimum levels, you could be over-driving the return (which will cause symbol compression and other problems).

If all else fails, check your cables inside your house. If they are RG59, consider getting some RG6. If they have crimped ends, see about having them refit with compression fittings. (If the cables have brass fittings with writing on it, you can try posting that here as well). Make sure none of the cables are chewed through or bent (a crimped cable can have some frequency specific loss added to it). If it isn't shielded cable in the house, make sure you are avoiding your electrical distribution lines.

Consider getting your own amps and splitters. Amplify first. Go to a two-way. Feed your cable modem from one leg, split the other leg to feed your three TVs.

egnlsn

04-15-08, 10:38 AM

Paragons

04-15-08, 10:39 AM

I think I remember a CSR testing a noise at 32???. Most of this jargon is way over my head.

I really appreciate the help, especially remotely.

What's so frustrating is that none of these issues occurred 3 weeks ago or for years before that.

I dont know if it was coincidence with gardeners cutting the cable or what...that's one reason I insisted TW rerun 200 ft of new cable to my house, because I was sure it had to do wth the splice, but that can't be the issue any longer.

I did have about 2 hours of use w/o any disconnects when I decided to plug the wireless router and modem into a Vista laptop and bypass the normal use of a win98se desktop. Is there any chance it's software related? Wasn't 3 weeks ago like I said, but who knows?

Pete

Paragons

04-15-08, 10:58 AM

A cable modem should always be on its own, dedicated splitter (or tap) and be the first device in the line. Sometimes, that is just not possible, but if possible, it should be so.

That said, DOCSIS calls for levels at the cable modem to be from -15 to +15dBmV at the downstream and +8 to +55dBmV at the upstream.

Numbers-wise, you should be fine. Using the Cable Attenuation Worksheet found at http://www.cencom94.com/Download.html (I always found the one at http://www.timesmicrowave.com/cgi-bin/calculate.pl to be a bit high as compared with manufacturers' specifications), you should have -2.4dBmV at your cable modem. A signal level of +5dBmV at the tap minus 9.8dB (loss through 200' of RG6) has -4.8dBmV at the groundblock. Add 11dB for the amp, lose 4.9 through the 100' to the room (RG6) and ~3.8 through the final splitter leaves -2.9dBmV at the cable modem. Is the splitter at the cable modem a splitter or a TAP? What are the numbers at your cable modem as found at http://192.168.100.1?

Keep in mind that the return path (upstream) loses signal just as the forward path (downstream) does. While the loss through passives is slightly less than the forward path, the loss through coax is significantly less. That SVI amp loses 8dB and the 2-way splitter at the outlet loses 3.5. At ~1dB/100', the RG6 loses ~3dB, so the return path loses (3 + 11.5=) 14.5dB, which is fine.

Again, is the device at the outlet a 2-way splitter or a directional coupler (TAP), and what are the numbers at the cable modem?

Thanks for your interest in helping me...I did not ask the tech what the values were that he tested at the modem. I don't know what a "tap" looks like the thing in the room is Im sure a TW supplied splitter. The tech used the term for the concrete housing across the street (200 ft away) were a bunch of their cables are and are sent to various houses.

The problem is, after leaving the amp in the garage the cables go through the floors with zero access to them, whereby it emerges in the computer room some 50 ft. away. At that point the TW 2 way splitter sends one cable to the TV and the other RG6 cable to the modem. I have no idea what type of cable is embedded in the floors.

I will find the model # of the modem later tonight...it's brand new from TW this week.

Are you saying the cable to the modem should not come out of the amp as it is now?

Pete

egnlsn

04-15-08, 11:22 AM

Thanks for your interest in helping me...I did not ask the tech what the values were that he tested at the modem. I don't know what a "tap" looks like the thing in the room is Im sure a TW supplied splitter. The tech used the term for the concrete housing across the street (200 ft away) were a bunch of their cables are and are sent to various houses.

The problem is, after leaving the amp in the garage the cables go through the floors with zero access to them, whereby it emerges in the computer room some 50 ft. away. At that point the TW 2 way splitter sends one cable to the TV and the other RG6 cable to the modem. I have no idea what type of cable is embedded in the floors.

I will find the model # of the modem later tonight...it's brand new from TW this week.

Are you saying the cable to the modem should not come out of the amp as it is now?

Pete
A tap that would be used at an outlet sort of looks like a 2-way splitter. The ports would be labeled IN/OUT/TAP and have a part number something like DCG-12FRC or something along those lines.

You should be able to type http://192.168.100.1 (or just click on this link) in your browser and get to the cable modem's diagnostic page.

A cable modem being the first device in the line gives it the purest signal and has it better isolated from the rest of the devices in your system.

Paragons

04-15-08, 11:46 AM

Well, Id say the first device is the amp that has an outlet labeled VOIP (goes to the modem) and another outlet that goes to a 3way splitter.

That link does not work for me.

I really think it's just a simple 2 way splitter, not a "tap" but Ill look more closely.

Pete

Paragons

04-15-08, 08:20 PM

kbullkar

04-16-08, 09:09 AM

re: egnlsn: I agree TM is usually in excess of manufacturer spec, and while I expect most cables to perform to manufacturer spec, I find using this calculator for the last mile provides a more real-world expectation... some loss added by violating the minimum turn-radius or through connector issues is so common and tends to scale with cable length...

Also, have to disagree with the statement that "A cable modem should always be on its own, dedicated splitter (or tap) and be the first device in the line. Sometimes, that is just not possible, but if possible, it should be so." To be clear - I am not saying you are wrong, just that I disagree with this statement when used as a hard-and-fast "rule"... I had to battle with installation contracting companies that used this "rule" with their installers to ensure that the install took the least amount of time and had the most chance to succeed... problem is most digital cable boxes have a cable modem in them (often not DOCSIS, but still a cable modem) for things such as PPV billing and VOD, and it was very common to see a working cable-box become a non-responder after a contractor installed a cable-modem using this "rule".

When I was designing systems, I used to design for an 8-way at the customer premise (200 ft RG6 drop, 100 ft RG59 cable post splitter). I think most TWC sites are designed for 4-way, with some really old systems designed for 2-way. This meant we needed more power, but we generally had shorter runs, so we accomplished this by using low-value taps... targetting end-of-line at ~3 to give some headroom. When we needed more power (for businesses), we would put in a lower value tap and pad the other drops.

re: Paragons: You are losing your lock... the "data" you shared indicates that you were refreshing while the cable modem was ranging... what is important to figure out at this time is whether you are getting/losing downstream or upstream lock... if the downstream locks and it keeps cycling on upstream, it then will go through every other possible downstream which takes a long time. Your signal level looks "OK" but not impressive. You are probably borderline. If you can capture your logs and share them that might show what it causing the CM to be unhappy.

You might also want to try putting the amp before the two-way splitter (make sure it has passive return). Actually, as a diagnostic, try connecting the cable modem directly to the feed coming in from the street (you won't have signal to your TVs during the test), eliminating all splitters and amps... see if that locks... if it doesn't lock then, you might have a frequency specific impairment... if it does lock, see if you can leave it there for several hours to make sure it doesn't lose lock. If you don't have any DVRs/recordings, overnight might be a good idea (of course, there is a chance that nightly maintenance will force the cable modem to lose lock and need to re-logon to the CMTS).

Paragons

04-16-08, 10:04 AM

egnlsn

04-16-08, 10:28 AM

re: egnlsn: I agree TM is usually in excess of manufacturer spec, and while I expect most cables to perform to manufacturer spec, I find using this calculator for the last mile provides a more real-world expectation... some loss added by violating the minimum turn-radius or through connector issues is so common and tends to scale with cable length...
It is not really a good idea to include connector issues or exceeding minimum bend radius or other negative-impact issues in the attenuation of cable. Problem with that practice is that if the manufacturer says that at 750MHz, their cable loses 5.65dB per 100', the cable loses 11.3dB over a 200' length. That is the attenuation through the cable, end of story. If you are expecting it to lose 15.9dB (according to that online calculator) and there is, in fact, a 16dB loss, you're gonna say, "Close enough to 15.9," and call it good. There could be a bad fitting or two, bad unknown splice, excessive bend radius, a staple or two that has pierced the cable or even a 2-way splitter and/or other device(s) or issues causing that 4.7dB attenuation that is in excess of what the manufacturer states, and you'd never know it because you are expecting it. An issue is an issue, and regardless of how common it may be, and it should be addressed.

Also, have to disagree with the statement that "A cable modem should always be on its own, dedicated splitter (or tap) and be the first device in the line. Sometimes, that is just not possible, but if possible, it should be so." To be clear - I am not saying you are wrong, just that I disagree with this statement when used as a hard-and-fast "rule"... I had to battle with installation contracting companies that used this "rule" with their installers to ensure that the install took the least amount of time and had the most chance to succeed... problem is most digital cable boxes have a cable modem in them (often not DOCSIS, but still a cable modem) for things such as PPV billing and VOD, and it was very common to see a working cable-box become a non-responder after a contractor installed a cable-modem using this "rule".
I don't know of one major industry engineer who advocates anything other than placing a cable modem on its own, dedicated splitter (or tap) and being the first device in the line. Yeah, DCTs have internal cable modems, but what does that have to do with the cable modem with which you connect to the internet? Absolutely nothing. A DCT becomes a non-responder if it either goes bad or loses its return path. It would lose its return path only if an either HPF-54 or a forward-only amplifier was installed between the DCT and the outside world. Installing a diplexer that only goes down to 40MHz would do the trick, also.

egnlsn

04-16-08, 10:59 AM

OK ran the diagnostics.

Downstream:
QAM 256
5360537 symbol rate
Interleave depth 32
-4dBmV power
SNR 32

Upstream:
QAM 16
2560 ksym/sec
42.0 dBmV
minislot 4

These results from running the street cable with a barrell connection to the cable modem:

-11.1 bD and SNR = 31.4 downstream...38 power upstream.

These results by using an old crappy splitter to the street cable, then one output to amp and 3way splitter to tvs and the other out directly to the modem:

They fluctuated alot during refresh but generally lots of things changed...interleave depth was 64, other and mostly 128
it was now "not locked"
SNR was about 27 dB with lots of fluctuation
Power was -18 with lots of fluctuation

Other than SNR being too low (DOCSIS calls for no less than 35dB) the first set of numbers are fine. The second set of numbers has an input a little on the low side, although nothing to worry about. Heck, until Comcast started having issues with the return path causing my cable modem to jump up to 56+ occasionally (at which time I would lose sync), I kept my input level at ~-12dBmV. Again, the concern is the SNR at 31.4dB. The third set of numbers has the input level out of spec and the SNR is way too low.

If you could print a screen shot or copy and paste that page (include times) and show it to the tech when he comes out, that would be great.

Another thing that can cause loss of sync is ingress. Ingress is when stuff floating around in the air gets into the cable system. Technically, the cable system is the system from the headend to the tap. From the tap to the TV set is called the drop system. For purposes of this discussion, the cable system goes from the headend all the way to your TV (or cable modem). The most common causes of ingress are sub-standard or loose connectors, sub-standard splitters and/or taps, and sub-standard coaxial cable. CATV needs 100% shielding all the way from the headend to the TV set (or cable modem). If it loses any of that shielding for any reason, RF in the air gets into the system and can cause issues. The return path is more susceptible to ingress than is the forward path. Jumper cables that come with VCRs are a no-no. Anything with twist-on or push-on connectors and/or copper-braid cable are huge sources of ingress.

GoDucksinAZ

04-16-08, 11:06 AM

BelowAverageJoe

04-16-08, 11:23 AM

...I don't know of one major industry engineer who advocates anything other than placing a cable modem on its own, dedicated splitter (or tap) and being the first device in the line...

I know a few :), but you may be right about many of them.

Years ago, many of our systems were fortunate enough to launch our high-speed service on the heels of fairly complete rebuilds, which also included drop system replacements/improvements. Anticipating a high rate of customer "self-installs" for the cable modems, we required every active outlet in the system would be fully two-way capable. Sure, we missed lots of them, but every new or reconnected service was brought up to spec., and that helped reduce lots of truck rolls later (I can't say we were very popular with the installation contractors. They were used to using lots of high-pass "bandaid" filters for reducing noise in the return path;)).

Every system is different, and I'm sure lots of them don't have the luxury of just handing a modem to the customer at the counter and being confident it'll work fine on any outlet in his house, but it does seem like a better goal to shoot for than the current rule of creating a single custom outlet for a modem.

egnlsn

04-16-08, 12:21 PM

That obligation calls for a fair amount of creativity depending on tap value and levels. If a D0CSIS level off the tap is 16dBmV on the downstream and 30 on the upstream, the first leg of a splitter will not place that modem in spec and gives you a fair amount of creativity to pad for other devices in the household.
That's what DCs are made for. :)

With those numbers, a DC-12 would put you at 4dBmV on the downstream, and drop loss is going to have it a few dB less than that. The upstream is determined by the CMTS, not by the way the cable modem is connected.

Paragons

04-16-08, 02:07 PM

OK Ed...how do I get that 31.4 higher or instruct the cable company to do it.

The lasttech on Sat said he would call in that the company had to up the 5 dB at the "tap" across the street...as such, it does not appear that has occurred yet. I wonder if that's another service call I have to make and meet the guy out at the concrete box?

Pete

egnlsn

04-16-08, 02:22 PM

OK Ed...how do I get that 31.4 higher or instruct the cable company to do it.

The last tech on Sat said he would call in that the company had to up the 5 dB at the "tap" across the street...as such, it does not appear that has occurred yet. I wonder if that's another service call I have to make and meet the guy out at the concrete box?

Pete
That is something that is the cable company's responsibility. Keep on them until they make it work as it should.

GoDucksinAZ

04-16-08, 02:33 PM

That's what DCs are made for. :)

With those numbers, a DC-12 would put you at 4dBmV on the downstream, and drop loss is going to have it a few dB less than that. The upstream is determined by the CMTS, not by the way the cable modem is connected.

I dont trust DC's and try not to use them if at all possible. Maybe its the extreme heat of the phoenix valley but I see at least one DC fail a month. That is far to many for my taste. Literally, ill have signal coming in, nothing coming out. Not sure why they fail but nevertheless, i try to avoid add passive equipment that I have witnessed a few too many failures.

We must be misscommunicating on the upstream because "our" upstream is determined by the node padding. The CMTS determines what the node "should be" but the node can be padded however the engineering of that neighborhood dictates. It is then further manipulated by passive devices in line. A DC-12 would increase my upstream 12dB. With an acceptable limit of 50, it is a very rare circumstance that a passive house design can tolarate a 12dB increase on the upstream. My upstream is usually around 35-40dBmV off the tap, not a lot of room to breath for passive devices or worse a DC-12.

egnlsn

04-16-08, 02:51 PM

I dont trust DC's and try not to use them if at all possible. Maybe its the extreme heat of the phoenix valley but I see at least one DC fail a month. That is far to many for my taste. Literally, ill have signal coming in, nothing coming out. Not sure why they fail but nevertheless, i try to avoid add passive equipment that I have witnessed a few too many failures.
Way too many! I wouldn't use them either if I had that many fail on me.

We must be misscommunicating on the upstream because "our" upstream is determined by the node padding. The CMTS determines what the node "should be" but the node can be padded however the engineering of that neighborhood dictates. It is then further manipulated by passive devices in line. A DC-12 would increase my upstream 12dB. With an acceptable limit of 50, it is a very rare circumstance that a passive house design can tolarate a 12dB increase on the upstream. My upstream is usually around 35-40dBmV off the tap, not a lot of room to breath for passive devices or worse a DC-12.
I probably should have worded it differently. Probably should have written "the output of the cable modem is determined by the CMTS," not the upstream of the plant itself.

Paragons

04-16-08, 10:47 PM

Here's the info after the latest disconnect...hope i formats correctly
LOGOUT | MODEM | TOOLS

CABLE MODEM• Information• Status• Downstream• Upstream• Upstream Burst• Operation Config.• Event Log

Cable Modem Event Log
First Time Last Time Counts Level ID Text
Thu Apr 17 01:22:37 2008 Thu Apr 17 01:22:37 2008 25 Critical (3) 85000100 No UCD's Received - Timeout
Thu Apr 17 01:19:09 2008 Thu Apr 17 01:19:09 2008 70 Critical (3) 84000100 SYNC Timing Synchronization failure - Failed to acquire QAM/Q...
Thu Apr 17 01:16:27 2008 Thu Apr 17 01:16:27 2008 5 Critical (3) 85000100 No UCD's Received - Timeout
Thu Apr 17 01:22:41 2008 Thu Apr 17 01:22:41 2008 31 Critical (3) 84000200 SYNC Timing Synchronization failure - Failed to acquire FEC f...
Thu Apr 17 01:15:28 2008 Thu Apr 17 01:15:28 2008 8 Critical (3) 85000100 No UCD's Received - Timeout
Thu Apr 17 01:13:57 2008 Thu Apr 17 01:13:57 2008 1 Critical (3) 82000400 Received Response to Broadcast Maintenance Request, But no Un...
Thu Apr 17 01:13:39 2008 Thu Apr 17 01:13:39 2008 1 Critical (3) 84000400 SYNC Timing Synchronization failure - Failed to receive MAC S...
Thu Apr 17 01:13:15 2008 Thu Apr 17 01:13:15 2008 48 Critical (3) 82000500 Started Unicast Maintenance Ranging - No Response received - ...

kbullkar

04-17-08, 07:01 AM

kbullkar

04-17-08, 07:29 AM

It is not really a good idea to include connector issues or exceeding minimum bend radius or other negative-impact issues in the attenuation of cable. Problem with that practice is that if the manufacturer says that at 750MHz, their cable loses 5.65dB per 100', the cable loses 11.3dB over a 200' length. That is the attenuation through the cable, end of story. If you are expecting it to lose 15.9dB (according to that online calculator) and there is, in fact, a 16dB loss, you're gonna say, "Close enough to 15.9," and call it good. There could be a bad fitting or two, bad unknown splice, excessive bend radius, a staple or two that has pierced the cable or even a 2-way splitter and/or other device(s) or issues causing that 4.7dB attenuation that is in excess of what the manufacturer states, and you'd never know it because you are expecting it. An issue is an issue, and regardless of how common it may be, and it should be addressed.

Ideally, I would agree with you, however, when running a business, this isn't practical. A good plant engineer accounts for a certain amount of "unforseen" problems (without going way overboard). The difference is between acceptable design parameters/performance and component level analysis. If you want to go component level, start adding in your loss from your connectors, ground, etc. It isn't practical.

If you are an installer/tech, then by all means, checking a cable to ensure that the loss is as close to "manufacturer design specs" is laudable. When you are troubleshooting an install you need some "threshold", I am suggesting that the time microwave values, and not manufacturer specs, should be used for the threshold... otherwise you will find yourself running around and fixing a lot of problems that didn't require fixing...

I don't know of one major industry engineer who advocates anything other than placing a cable modem on its own, dedicated splitter (or tap) and being the first device in the line. Yeah, DCTs have internal cable modems, but what does that have to do with the cable modem with which you connect to the internet? Absolutely nothing. A DCT becomes a non-responder if it either goes bad or loses its return path. It would lose its return path only if an either HPF-54 or a forward-only amplifier was installed between the DCT and the outside world. Installing a diplexer that only goes down to 40MHz would do the trick, also.

What is your definition of a "major industry engineer"? I know of at least a dozen at the operator corporate engineering level who would strongly disagree with that statement... and I know the corporate engineering guidelines of a couple of the larger operators explicitly discourage that practice (in fact, I contributed the language for one of them.)

A DCT (with Starvue RF Return), like any other cable modem, can lose its return path if additional loss is introduced that it cannot overcome. I have seen this occur a lot, and the damage it has caused has been in the millions... between un-collectable PPV revenue and costs for additional truck runs, increased churn, etc. (And if you want to talk filters, and 8-12 or 8-16 notch also causes the DCT to not be able to respond... again, your are causing the signal not to be able to overcome the additional attenuation, with a "filter" that attenuation is generally very high and very frequency specific... but it is still attenuation). Newer DCTs can use DOCSIS for their return (known as DSG) and there are other devices that use eDOCSIS... the plant should be designed such that ANY outlet can work with the cable modem...

We must be misscommunicating on the upstream because "our" upstream is determined by the node padding. The CMTS determines what the node "should be" but the node can be padded however the engineering of that neighborhood dictates. It is then further manipulated by passive devices in line. A DC-12 would increase my upstream 12dB. With an acceptable limit of 50, it is a very rare circumstance that a passive house design can tolarate a 12dB increase on the upstream. My upstream is usually around 35-40dBmV off the tap, not a lot of room to breath for passive devices or worse a DC-12.
I probably should have worded it differently. Probably should have written "the output of the cable modem is determined by the CMTS," not the upstream of the plant itself.

You are both right... kind-of. The CMTS looks for a certain level of signal and will power range the cable-modem so that the signal is at the desired level when it reaches the CMTS. The cable modem can only output so much, so if you add too much attenuation in the upstream, the CMTS will power range the cable modem, and if at Max Power it gets a low response, will leave it at max power (otherwise it is supposed to set it to min-power and disable transmission). The same thing occurs with a pad... add a 3 dB pad, and if there is enough headroom, the output level of the cable modem will increase by 3 dB... because the CMTS will tell it to do so based on the signal level it is receiving... which is based on the upstream plant.

Paragons

04-17-08, 10:25 AM

So to be clear, you had cable from the street, an F81 ("barrel looking connector"), and the first set of stats you posted were from that? You are definitely having signal issues. Are you on an underground or aerial system? Something is introducing way too much loss. If you don't mind posting what system you live in (preferably add it to your profile), someone may be able to get you in touch with a plant engineer... I would start taking readings at every connection... one of them is going to be way out of spec... (well beyond the point where Ed and I disagree about the worthiness of troubleshooting).

Well I ve tried every possible reconfiguration by swapping all sorts of connections myself and testing for eventual disconnects.

The stats I got (partially saved) with the config of connecting the brand new cable (no grounding either) with the barrell to the modem's cable was -11.1 dB down, SNR 31.4 and 38 dB upstream (post #9)...wrote nothing else down. Want me to do it again with full stats?

The current configuration is with an old 2 way splitter I found to send straight to the modem...that's the log in post #21, but even though it has the oldest crappiest part and some of the worst readings Ive seen...SNRs below 29 it seems to last longest for connections, possible coincidence though. I remember a splitter that said TW Broadband on it in a previous config that was removed by one of the techs...that's why I looked and found one of my old crappy splitters as a desparate measure.

The cable from the street is brand new, helped lay it myself...it is feed under the street for about 125 ft, then emerges at my front lawn and is feed through a pvc? tube (currently I did not rebury it because I wanted to wait until everything worked again...dont want to dig it up again) and emerges in my garage where it is barrell connected to a feed going outside to ground it and returns to the amp in the garage...after that, the cable is not accessible because it is burried directly in the floor joists with no attic space.

Here is the response I just got from Time Warner Cable:

Dear Peter,

Thank you for your inquiry.

Outside work on our taps and junctions is mostly done by a maintenance
and repair group with a typical turn around time between 3-10 business
days. We apologize for the prolonged wait and want to assure you that

we are committed to resolving the problem with low signal levels at the

tap.

Also be assured our networks are deployed and managed with methodology
and diligence similar to that expressed in the Cisco documentation you
linked.

The problem your line is experiencing however is not impacted by
utilization on Line Cards, Upstream ports, or UBRs, but by the local

tap across the street. There are only a limited number of terminals on

these taps and the likely solution is to redistribute the gain
equalization on that tap to provide you with a better
upstream/downstream signal. This could also be the result of damage
caused on the tap by the elements. Hence the large window in our
Estimated Time of Repair.

If the quality of your service does not improve by April 25th please
contact us at the number provided below. At that time we can check if
the work order has been completed and take additional steps to resolve
your cable line issue.

We thank you for your patience in this matter and encourage you to
contact our support if the problem persists.

If you have any further questions, please do not hesitate to contact us

at the number listed below. Our telephone representatives are available

24 hours a day, 7 days a week. Have a nice day!

egnlsn

04-17-08, 11:54 AM

Ideally, I would agree with you, however, when running a business, this isn't practical. A good plant engineer accounts for a certain amount of "unforseen" problems (without going way overboard). The difference is between acceptable design parameters/performance and component level analysis. If you want to go component level, start adding in your loss from your connectors, ground, etc. It isn't practical.

If you are an installer/tech, then by all means, checking a cable to ensure that the loss is as close to "manufacturer design specs" is laudable. When you are troubleshooting an install you need some "threshold", I am suggesting that the time microwave values, and not manufacturer specs, should be used for the threshold... otherwise you will find yourself running around and fixing a lot of problems that didn't require fixing...
I agree that slight variances need to be taken into consideration, such as temperature swing, connectors, barrels, and groundblocks. The connectors, barrels, and groundblocks, though, are constant and are not related to the length of the cable in question. The only one that varies with cable length is temperature variation.

A couple of compression fittings on a length of cable introduces ~0.2dB insertion loss. As in the above example, a 200' length of cable, according to manufacturer's specs (Times Fiber and CommScope), loses 11.3dB. Add another 0.2dB for the 2 fittings at either end, and it should lose 11.5dB. Say you know of an F-81 in the line and you are reading signal at a groundblock. The F-81 loses 0.3dB and there are, obviously, another couple of fittings there. The groundblock loses 0.3, which brings the expected loss through that length of cable to 12.3dB. A loss of 16dB, though, is acceptable? What happened to the other 3.7dB? Are there more barrels than was thought? Perhaps a 2-way splitter or a DC with the thru leg going to your location. If so, where does the other leg go? Perhaps a kink or otherwise too sharp of a bend in the cable? I once had a 4' piece of cable act like a 100' piece because of how an electrician installed it. Too much of a difference to not wonder why such a large discrepancy.

What is your definition of a "major industry engineer"? I know of at least a dozen at the operator corporate engineering level who would strongly disagree with that statement... and I know the corporate engineering guidelines of a couple of the larger operators explicitly discourage that practice (in fact, I contributed the language for one of them.)
I would consider a major industry engineer to be one who has been published in one or more of the trades, such as CT or CED. Perhaps speaks at chapter meetings or even major events, such as SCTE EXPO or the Cable Show. Those who participate in the SCTE-List also get a big nod.

A DCT (with Starvue RF Return), like any other ca the plant should be designed such that ANY outlet can work with the cable modem...

Agreed that the PLANT should be designed and built such that ANY outlet can work with the cable modem.

I guess we'll have to agree to disagree about excessive cable loss being acceptable and the connecting of cable modems within the residence.;)