Effect of synch speed on upstream errors

I have just pushed my synch speed (using DMT on Speedtouch 546) up to 4704 kbps and I have noticed that the upstream errors have risen. I don't know if it will hold overnight but I wondered why I have so many upstream errors

In reply to:

---

Uptime: 0 days, 4:29:46
Bandwidth (Up/Down) [kbps/kbps]: 448 / 4,704
Output Power (Up/Down) [dBm]: 12.0 / 19.5
Line Attenuation (Up/Down) [dB]: 24.0 / 50.0
SN Margin (Up/Down) [dB]: 21.0 / 9.0
Error Seconds (Local/Remote): 66 / 0
FEC Errors (Up/Down): 52 / 207,202
CRC Errors (Up/Down): 28 / 56
HEC Errors (Up/Down): 10 / 51

---

I reckon it's a firmware problem, every version I use gives me Upstream errors except the latest, 2.6.15.5. However, in general, a higher sync will give you more errors, unless you have a very quiet line.

i think you are probably correct as my latest stats show it hanging in with an snr of 5db and no more upstream errors

In reply to:

---

Uptime: 0 days, 10:19:43
Bandwidth (Up/Down) [kbps/kbps]: 448 / 4,704
Output Power (Up/Down) [dBm]: 12.0 / 19.5
Line Attenuation (Up/Down) [dB]: 24.0 / 50.0
SN Margin (Up/Down) [dB]: 21.0 / 5.0
Error Seconds (Local/Remote): 112 / 0
FEC Errors (Up/Down): 104 / 1,363,059
CRC Errors (Up/Down): 28 / 121
HEC Errors (Up/Down): 10 / 107

---

well to start when you adjust your snr margin in dmt and your synch speed, it has no affect on your upstream snr margin as you can see both stats you posted have a 21db upstream margin.

28 crc errors is very low and I woould just ignore it, it would appear to be fixing many errors but I am sceptical about the speedtouches reporting errors but either way fec's dont affect the connection.

In reply to:

---

28 crc errors is very low and I would just ignore it, it would appear to be fixing many errors but I am sceptical about the speedtouches reporting errors but either way fec's dont affect the connection

---

I have very high error rates but what alarms me most of all is the 'error seconds' so is 6250 errored seconds (1.73 hours) from an uptime of just over 2 days bad?

http://i3.tinypic.com/2rp9r7m.jpg

is that high compared to the billion?

compared to my line yours are low.

Right now I got 6785 errored seconds for almost 11 hours uptime but I am not using error correction mechanism currently.

Tommorow when my fluctuations start I will probably get about 15000 errored seconds an hour as nearly every second is at least 1 crc error usually much more.

Dont know exactly whats bad and whats good, fec errors dont count for errored seconds so they are from proper crc errors but even 1 crc error will make the errored second go up, its quite possible to get low numbers of crc errors and the line performs fine.

My present stats are

In reply to:

---

Uptime: 1 day, 20:37:50
Bandwidth (Up/Down) [kbps/kbps]: 448 / 4,704
Data Transferred (Sent/Received) [MB/MB]: 21.47 / 195.50
Output Power (Up/Down) [dBm]: 12.0 / 19.5
Line Attenuation (Up/Down) [dB]: 24.0 / 50.0
SN Margin (Up/Down) [dB]: 21.0 / 8.5

Error Seconds (Local/Remote): 354 / 0
FEC Errors (Up/Down): 188 / 3,252,569
CRC Errors (Up/Down): 34 / 514
HEC Errors (Up/Down): 14 / 459

---

snr drops to about 5db in the evening.
I will continue monitoring but I suspect i am pushing my line too hard.
Once the BT system has caught up I will do a few speed tests using the BT speedtest and if my speeds don't go much above 3000kbps due to exchange contention I will ease it back to give a synch rate of 4100 kbps

In reply to:

---

is that high compared to the billion?

---

Yes with my Billion router, the FEC error rate was only a fraction of that figure.

In reply to:

---

compared to my line yours are low

---

Yes but don't forget that you sync higher than me (5 Mb+ I think?)

In reply to:

---

but I am not using error correction mechanism currently

---

Interesting.
What do you mean by this?

I downgraded to firmware 5.4.0.14 as its the only I can get bitswapping disabled, what I noticed between this and the 546.

on the 546 there is a big difference in error count between the 5.4 and 6.1 firmwares, FEC errors that is/

On my 585 going down to 5.4 to get rid of bitswapping crc errors are around the same and maybe even less but now the router has started resynching when the line gets a marginal snr margin again so as a result I have ended up with a very poor BRAS. I think the 6.1 firmware is buggy for the speedtouch routers and my thoery is that bitswapping was originally turned off because they couldnt get it to work right and 6.1 been the first version where they using it in a release version.

On the original subject of this thread tho I cant see how downstream speed should affect upstream errors.

In reply to:

---

On the original subject of this thread tho I cant see how downstream speed should affect upstream errors.

---

Nor me as the upstream sync is fixed at 448 Kb unless your on an Office package in which case it's 832 Kb