@Robert85 wrote:
What is considered too high for CM error codewords -Correctable Codewords?
That would depend on the channel/frequency. I think that Rogers assigns a single downstream channel closer to the split point so that they can get another measurement of worst-case link performance on your line. @Datalink can probably elaborate more on this.
Generally, I'm more concerned about the number of Uncorrectable Codewords because that indicates actual packet loss on the network due to link errors.
@Robert85 can you post the whole table? To copy the whole table, top to bottom, left to right, select the left side initially, then scroll right and use "shift" + "mouse click" on the right hand corner, that should select the whole table. At that point you should be able to right click .... copy. Then start a new post,right click .... paste. The whole table should paste in and we should be able to scroll right to see all of the data. That has been done recently, so I know that it can work.
I'd like to see the frequency associated with that higher uncorrected number. Looks like channel 1, but, whats the frequency where channel 1 starts.
At the moment I don't have any info on the limit of the uncorrected words, at which point Rogers should dispatch a tech to determine what the problem is for your external cable and connectors. There's also the possibility of signal ingress into the cable system from an external transmitter, which is where one would have to know the frequency of the affected channel(s).
@Robert85 A high number of correctable codewords aren't necessarily anything to freak out about but when you see them, you need to look at the other stats, especially when the error counters are very high.
I don't have any practical experience troubleshooting DOCSIS 3.1 (so take anything that I say with a HUGE grain of salt) but I didn't think that you were supposed to see ANY uncorrectable codeword errors on the downstream OFDM stats as that might be indicative of the modem dropping to DOSSIS 3.0.
Do you see anything alarming/out of the ordinary when you look at the modem's error logs?
@Alex4161 wrote:
I was looking at my cable modem stats and noticed the following:
CM Error Codewords
Index 1
Correctable Codewords
1358115282
Uncorrectable Codewords
42
I had a ticket open for noise on the line and they said the issue was resolved. After rebooting my modem, I still get this coming up.
Is this normal or do I still have some unresolved signal issues?
I see the same as you with DOCSIS 3.1: a scary-looking number of correctable codewords but a fairly low number of uncorrectable codewords. After doing some more thinking and research into this, I believe that this is not only normal but actually a good thing.
With DOCSIS 3.0, a high number of correctable codeword errors is not good; it's either a reflection of deteriorating line conditions or could also be a precursor to some other failure. However, DOCSIS 3.1 deals with and adapts to line conditions differently so we need to interpret the error stats differently.
Forgive me for oversimplifying but with DOCSIS 3.1, you gain efficiency by being able to deal with errors better, and better error correction is what allows you to use higher modulations and achieve transmission speeds that are close to the theoretical limit. In a way, having a low number of correctable errors is actually an indicator that you could be using higher order modulations and are not maximizing the efficiency that DOCSIS 3.1 offers. In the end, what you ideally want to see are correctable codeword errors and little or no uncorrectable codeword errors.
I'll offer a different interpretation. At the end of the day, you should see zero Corrected and Uncorrectable Codewords. Have a look at the following post which is from an SB8200 Docsis 3.1 modem. Take note of the two columns on the right hand side, which are, second from the right, corrected codewords, and the right hand column, uncorrectable codewords.
https://www.dslreports.com/forum/r32626244-
The corrected codeword count increases with frequency and there is a huge leap with the OFDM channel 159. At the end of the day however, the uncorrected codeword count is nearly zero, as it should be. If you look closely at the data, the signal levels drop slightly with increasing frequency and the signal to noise ratio drop slightly as well. Not large changes, but you can see the increase in the corrected codeword count as the frequency increases.
I'd have to see your signal level data and the whole codeword table to arrive at any conclusion, but, if the remainder of the table resembles what you posted, I'd say that you still have a noise problem. Have a look at the OFDM channel on the right hand side codeword data which should be on the far right hand side. For the one active OFDM channel, if you have uncorrected codewords you definitely have a problem as the OFDM waveform, encoding and error correction is vastly improved from the DOCSIS 3.0 generation.
Now, given that you only presented a small snippet of the whole table after a reboot, consider posting more data now that the modem has been up and running for a while.
Are you using a power bar of any type? If so, disconnect the power bar completely from the wall socket and equipment. Run everything off of extension cords if necessary for test purposes. Reboot the modem and keep an occasional eye on the codewords to see if there is any difference in the amount of corrected and uncorrected codewords and the time it takes for the numbers to build up. Pay particular attention to the codeword data for the OFDM channel. You might see a high number of corrected codewords, which might be normal with the OFDM channel, but the uncorrected codewords should at zero or very close to it. If the modem is running in 4k mode with an OFDM channel width of 192 Mhz, you have 4096 sub-carriers used for the OFDM channel, give or take, depending on how many of the edge sub-carriers are used, versus 32 SC-QAM channels used for DOCSIS 3.0. So, the use of 4096 sub-carriers simply multiplies the number of codewords, both corrected and uncorrected that you might see in the OFDM channel data. Ideally they should be close to zero.
When I said that I did some research, that meant going through deployment guides from D3.1 equipment vendors as well as white papers from test and measurement companies such as Viavi.
I didn't want to get too down-in-the-weeds technical but what cable operators are aiming for is to use a modulation that provides the highest possible throughput while keeping the uncorrectable codeword errors at an acceptable level. If you can achieve that AND negligible correctable codeword errors, then that's perfect. However, that isn't always realistic and one of the goals of DOCSIS 3.1 (with OFDM and LDPC) is to allow you to get the most out of available line conditions. I'm not saying that any errors are good in and of themselves, just that the presence of correctable codeword errors is not necessarily a bad thing. Furthermore, since OFDM spreads data across multiple subcarriers, we can't look at and interpret BER stats like we did with D3.0.
So... if you are getting zero correctable errors but using a lower modulation, then you may not be (and most likely are not) maximizing efficiency. If you are using a lower modulation and getting high error rates, then that is really bad. Neither of these conditions are apparent just from looking at the error stats alone.
Given that I can't do a deep dive into my D3.1 stats, if my signal levels (Power and SNR) are okay, my uncorrectable codeword error counts remain low, and I'm getting the speeds that I am paying for, then I'm not going to get too concerned about a seemingly high number of correctable errors on the D3.1 channels.
- last edited on by
RogersCorey
@Alex4161 your signal levels and codeword data isn't too bad at all. From the posted data:
Downstream Docsis 3.0 channels (1 to 32)
Signal levels: Good. There is the usual high frequency roll off at the top end which is typical, not desired but typical on the Rogers network. That roll off will disappear over time.
Signal to Noise Ratios: Good. No real complaints. You can see a slight dip at the high end due to the high frequency roll off.
QAM: Good, running 256 for all channels, which is the highest you will see for DOCSIS 3.0
Downstream DOCSIS 3.0 Codewords.
For the most part, ok. I was concerned after your previous post that you might have had broadband noise covering the whole 5 to 1002 Mhz range, but, nope, the high Correctable Codewords and Uncorrected Codewords are isolated occurrences, at the low or high end. The low end covers the start of the first DOCSIS 3.0 channel and lower part of the DOCSIS 3.1 OFDM channel, and from what I think, probably affects the OFDM channel numbers, as seen in the Index 33 OFDM data. I suspect that there's noise ingress at the lower end of the Downstream band but, it doesn't appear to be causing an major problems. Same thing occurs at Index 30 but to a much lesser extent. All of the rest are fine.
Upstream DOCSIS 3.0
Three upstream channels running, sometimes you will see three channel, sometimes four channels, just depends on the CMTS and neighbourhood node that you're connected to. The upstream signal levels are well within spec.
Downstream DOCSIS 3.1 (Index 33)
I'm assuming that 350 Mhz is the start of the OFDM sub-carriers. It doesn't line up with the OFDM sub-carrier start point or mid-point for the CODA-4582 modems, so Rogers is probably customizing the sub-carrier frequency ranges for these modems. In any event, not a big deal, just an observation. The signal level is good, as the Signal to Noise Ratio. There is an amount of Correctable and Uncorrected Codewords. The Uncorrected Codewords at this point is only 44, basically negligible. My guess is that whatever is creating the noise at the low end of the Downstream band is causing the Correctable and Uncorrected Codewords. Not a big deal with those small uncorrected numbers.
Overall, not too shabby. I wouldn't expect any issues with posted signal data.
@Alex4161 What stands out is the number of uncorrectable codeword errors on the D3.1 OFDM channel, especially if the uncorrectable codeword error count continues to steadily increase. That number really should be close to zero. With too many uncorrectable errors, your modem will drop down to using DOCSIS 3.0.
I would call in to Rogers about this for further investigation. Do not reboot your modem. Try to get this escalated to a Level 2 tech for further investigation with the modem in its current state.
Hey @Alex4161!
Thank you for reaching out via PM! If you have any further questions or concerns definitely let us know.
-
20904 views
-
7 Likes
-
-
