CRC error rates

Hey there,

Just a question, I am wondering if anyone has any guidance on what a good, average and poor range would be for CRC errors?

Any help appreciated!

It might be worth doing a search through the forums for error rates (e.g. search for posts with CRC, HEC, FEC). Quite often you'll get a complete set of data from a router including sync rates and router up time. The post context will usually indicate whether the line is (subjectively) thought to be good, average, or bad.

was only a very recent post here on the subject, there is no figures to give you as the answer depends on the router you are using, what your line condition is and ultimately how your connection performs.

zero would be good.

Nice idea, I will put that together tonight!

Cheers,
Aslan
DSL broadband, dsl isps, and DSL modem advice
www.adslgeek.com

- I agree that the rate should be as low as possible and to be perfectly honest, anything over 0 is typically an indication of an issue and they can typically be seen from the other modem router settings (eg SNR, Attenuation/distance etc) but it is a good indicative tool I think.

Possibly it is a 0 = good, 0-100 in 24 hours indicates a problem and higher than 100 generally indicating a physical layer issue (or sometimes the modem is having difficulty with a protocol eg

- from what I have seen is generally the rate will be very low, and then thousands in a short space of time. And typically for a good reason.

Cheers,
Aslan
www.adslgeek.com (DSL, dsl modem, ISPs and all things broadband)

well from what I have seen here and other dsl sites, users with the best of lines ie. sub 20db attenuation still have more than 0 crc errors but very litle maybe 10-100 crc errors over the course of a week.

Users on long lines typically will get much more even with interleaving and as such its just accepted as been normal providing the connection isnt suffering packet loss and loss of synch as a result of it.

Good call.

I will go through the forum and get the consensus on what actual results I can see, and I will try to see if there are any trends between distance, SNR, or modems and then try to work out a formula to take that into account.

So the vote so far from my side would be

CRC errors per week:
0 - 10 - Excellent
10 - 100 Good
>100 Poor (though often an indication/result of distance, noise or a poor modem)

Does that sound about right?

It may be right. It doesn't feel right.

As a first stab for a week I'd guess

Good <1000
Fair <=50,000
Poor >50,000

Depends a lot on when they happen, averaging out over a week could avoid the fact that 10,000 happened in a one hour window which would be really bad.

There is no fixed figures, you get a feel for these in combination with all the other data for a line.

again it depends.

under 10 crc errors a week I have not seen a single line achieve that so I would say more like the following.

sub 30db attenuation line on adsl1.

crc a week

0-200 excellent
200-1000 ok
1000+ needs looking at

45+ db adsl1

0-1000 excellent
1000-10000 good
10000-100000 ok but may cause problems
100000+ needs looking at

the best of lines on adsl2+ are likely to get significantly more errors than they do on adsl1, all these figures are far from set in stone as so many variables and different routers handle crc errors better.

In reply to:

---

There is no fixed figures, you get a feel for these in combination with all the other data for a line.

---

Agreed.

My CRC error rate of around 63,000 is good for an uptime of nearly 34 days!


Uptime: 34 days, 6:44:55
Modulation: G.992.1 Annex A
Bandwidth (Up/Down) [kbps/kbps]: 448 / 3,520
Data Transferred (Sent/Received) [MB/GB]: 834.66 / 12.21
Output Power (Up/Down) [dBm]: 11.5 / 18.0
Line Attenuation (Up/Down) [dB]: 31.5 / 55.0
SN Margin (Up/Down) [dB]: 18.0 / 8.5
Vendor ID (Local/Remote): TMMB / 
Loss of Framing (Local/Remote): 0 / 0
Loss of Signal (Local/Remote): 0 / 0
Loss of Power (Local/Remote): 0 / 0
Loss of Link (Remote): 0
Error Seconds (Local/Remote): 20,341 / 0
FEC Errors (Up/Down): 5,288 / 7,031,983
CRC Errors (Up/Down): 3,188 / 63,768
HEC Errors (Up/Down): 2,218 / 56,264


p.s. the FEC error rate is quite low as well.

In reply to:

---

My CRC error rate of around 63,000 is good for an uptime of nearly 34 days

---

That's a CRC count. The rate per day is that divided by 34.

Swapping words that mean different things isn't a good idea . You and I both know what you mean, but does everybody?

Does the OP even, seeing as in another thread he clearly didn't know the difference between SNR and SNRM?

"There is no fixed figures, you get a feel for these in combination with all the other data for a line. "

I know what you mean but the DLM doesn't have feelings so it must work to some sort of algorithm. Is it really so secret that BT won't release it or is your target SNR based on some sort of random pot luck generator.

John.

They won't release it, and I wasn't realy referring to the DLM, but CRC and error counts in general

fwiw I'm on adsl2 synching at 24574/1402 with an attenuation of 18db and snr of 4.9db and usually get around 40 crc errors per hour.

I agree with that completely - if the distance is far, or there is noise on the line, I would expect that there to be a much higher CRC error rate and that to me would be a natural correlation.

If we assume that the line was within 1.8 km with a good SNR, and with a good modem, then we could work out some sort of baseline. And then I can work out what the other factors are (eg if Attenuation is X then that is likely to increase the CRC errors by y)

If it gets a higher CRC error then those other factors could be relevant. (does that make sense?)

Do you think sampling it over an hour would be better? That is certainly something that I have traditionally done I have seen some routers do thousands within 3 mins, so my feel for it is that it can generally sit quite low all things being equal, and it can spike up dramatically once the fault condition is on the line. And then LCP typically has a paddy so drops the line.

I have heard that a modem is relevant and that certainly sounds correct to me, and I may try to work out what brands of modems have this as an issue and I have seen some firmwares just appear to suck.

But it often as a combination of say firmware, interleaving, DSL2+, distance, noise, CPE etc, but I think it would be possible to work out some sort of baseline and then compare modem types against that to see if it is over or above and then I could work out the "Less stable" modems.

You could try to factor in the Errored Seconds and Severely Errored Seconds to try to cater for the bursty nature of noise. A problem is I'm not sure if those numbers have a consistent meaning as reported by the various routers.

In reply to:

---

My CRC error rate of around 63,000 is good for an uptime of nearly 34 days

-------------------------------------------------------------------------------------------------------

That's a CRC count. The rate per day is that divided by 34.

Swapping words that mean different things isn't a good idea . You and I both know what you mean, but does everybody?

---

Sorry I didn't know that (above)

No problem

Sorry if I came across a bit stroppy. Just that after a lifetime in IT inaccurate descriptions are anathema, as very few correct diagnoses come from incorrect information.

In reply to:

---

Sorry if I came across a bit stroppy

---

No problem.

I learned something anyway