PDA

View Full Version : Correctable/Uncorrectable Codewords


pigpen
03-10-2009, 00:13
Okay, so I'm looking at my CM config pages and I'm seeing the Correctable Codewords pretty much constantly counting upwards, about 6000 a second. Is this anything to get in a twist over? Is it a sign of something that may be degrading or about to fail?

My connection does occasionally drop, it has done twice today, but it never usually happens more than twice a week requiring unplugging the modem from the mains to get it back in action. Other than that the connection is fine, 50Mbit day and night.

I think the power levels and SNR are fine from what I've read on here, but the codewords thing seems to be astronomical compared to some. These figures were reset only a couple of hours ago when I had to yank the power.

See the pic for details:

https://www.cableforum.co.uk/images/local/2009/10/81.png

Ignitionnet
03-10-2009, 09:30
Definitely should be looked at before those Correctable errors become Uncorrectable and bring your connection to a halt. Those incrementing means that they are errored, and have been fixed by Reed-Solomon error correction codes. Even though RxMER reads good what's actually hitting your modem at times is not nice.

Could be faulty modem, could be an impulse noise issue, either way needs looking at.

pigpen
04-10-2009, 12:32
Yeah, that's what I thought :-/

I'll give them a ring and try and convince them there might be something wrong then.

The-Darkside
04-10-2009, 12:58
Definitely should be looked at before those Correctable errors become Uncorrectable and bring your connection to a halt. Those incrementing means that they are errored, and have been fixed by Reed-Solomon error correction codes. Even though RxMER reads good what's actually hitting your modem at times is not nice.

Could be faulty modem, could be an impulse noise issue, either way needs looking at.

Interleaving takes care of Impulse noise. Thats what its designed to do, spread out the noise across a number of packets.
The thing that doesn't make sense to me is that some channels are showing fewer correctable errors but higher ucorrectable.

Ignitionnet
04-10-2009, 13:45
Interleaving takes care of Impulse noise. Thats what its designed to do, spread out the noise across a number of packets.

Well yes it does, these are then reported as correctable codewords. Too much noise for the error correction and they become uncorrectable. Regardless it needs looking at as it's marginal, R-S shouldn't be needing to correct anything on a clean network.

3 and 4 having slightly higher uncorrectable makes perfect sense, it may have just taken the modem slightly longer to obtain FEC synchronisation on them and while it was doing that codewords will have errored. If they aren't incrementing it doesn't matter.

The-Darkside
04-10-2009, 14:29
Well yes it does, these are then reported as correctable codewords. Too much noise for the error correction and they become uncorrectable. Regardless it needs looking at as it's marginal, R-S shouldn't be needing to correct anything on a clean network.

3 and 4 having slightly higher uncorrectable makes perfect sense, it may have just taken the modem slightly longer to obtain FEC synchronisation on them and while it was doing that codewords will have errored. If they aren't incrementing it doesn't matter.

So BB uses FEC? With cable TV we never used FEC, and if I remember correctly we used a higher interleave depth.

Ignitionnet
04-10-2009, 14:47
So BB uses FEC? With cable TV we never used FEC, and if I remember correctly we used a higher interleave depth.

Yes much higher interleave depth on TV as no need to worry about latency.

Cable TV uses FEC - it's a 'generic' term to describe any in-band error correction where extra bits are added to the stream. Reed-Solomon is a block code, takes a block of data which has been interleaved already and additional R-S error correction bits are added.

Wikipedia say it better than me.

http://en.wikipedia.org/wiki/Forward_error_correction

The-Darkside
04-10-2009, 17:53
Theres inner and outer FEC. In satelite and DTT DVB broadcasts an inner FEC method is used known as convolutional coding. This is the FEC that isn't used in cable TV broadcasts. Using it reduces the usable bitrate (as does RS).
Is the BB signal using an MPEG stream structure (188 bytes, 204 with RS added)?

Ignitionnet
04-10-2009, 22:10
Yep, downstream DOCSIS is a 188 byte MPEG frame with a fixed header (47 1F FE 1x) that identifies it as DOCSIS.

When a modem is scanning it looks for this header at each frequency to identify a DOCSIS downstream.

The amount of R-S bits per frame is configurable depending on how much throughput you are prepared to lose. It's configurable, and a compromise between coding gain, latency, and overheads.

It's why even though a raw 256QAM EuroDOCSIS downstream can push 55.616Mbit/s by the time R-S encoding and other DOCSIS protocol overheads (upstream MAPs, station maintenance) are done munching the capacity you'll get 50Mbit or so of actual usable data.

pigpen
09-10-2009, 20:35
Hmm, I've re-routed my cable from the wall to the modem, and the correctable codewords arent incrementing by several thousand a second anymore but the uncorrectables are in the tens of thousands already after about an hour :-/

What's the 50 Meg support number again?

Peter_
09-10-2009, 20:55
What's the 50 Meg support number again?
0800 052 0431 probably best to call after 0800 tomorrow morning.

REM
09-10-2009, 21:29
That's not fair, why's the OP got 4 channels and I've only got 3? :(

http://img7.imageshack.us/img7/6088/modem.gif

Peter_
09-10-2009, 21:34
That's not fair, why's the OP got 4 channels and I've only got 3? :(


Easy to answer, you are on a Cisco 10000 uBR which uses 3 Downstreams and he is on a Motorola BSR which uses 4 Downstreams, it just depends where you live and has no impact on your speed.

REM
11-10-2009, 15:38
Easy to answer, you are on a Cisco 10000 uBR which uses 3 Downstreams and he is on a Motorola BSR which uses 4 Downstreams, it just depends where you live and has no impact on your speed.

Ah, I see. Thank you. I was getting an inferiority complex. :)