Hi,
I've recently set up a C887VA-W-E-K9 router to connect to Vodafone fibre 80/20. It's my first time setting up a xDSL circuit from scratch on these routers so I don't know if the config is 100% right - but it is working anyway (much to my delight).
After an initial teething period of unexplained CRC & header errors, and chronically low download speeds (~6Mbps download, 19Mbps upload) the line seems to have "settled".
Despite being told we would likely be only able to get around 50Mbps due to our distance from the cabinet, the router is reporting that attainable speeds in the 80s, and syncing consistently over 73Mbps.
Although no one has complained of any issues in the office that I'm aware of, I am seeing consistently high Reed-Solomon EC values in the statistics for the VDSL controller. The fact the rest of the numbers there are zero makes me think it shouldn't be this high?
Here is the output from sh controller vdsl 0, taken about 30 minutes after I cleared the counters on it:
Controller VDSL 0 is UP
Daemon Status: Up
XTU-R (DS) XTU-C (US)
Chip Vendor ID: 'BDCM' 'IFTN'
Chip Vendor Specific: 0x0000 0xB203
Chip Vendor Country: 0xB500 0xB500
Modem Vendor ID: 'CSCO' ' '
Modem Vendor Specific: 0x4602 0x0000
Modem Vendor Country: 0xB500 0x0000
Serial Number Near: FCZ17****** C887VA-W 15.3(3)M3
Serial Number Far: 55023*****
Modem Version Near: 15.3(3)M3
Modem Version Far: 0xb203
Modem Status: TC Sync (Showtime!)
DSL Config Mode: AUTO
Trained Mode: G.993.2 (VDSL2) Profile 17a
TC Mode: PTM
Selftest Result: 0x00
DELT configuration: disabled
DELT state: not running
Trellis: ON ON
SRA: disabled disabled
SRA count: 0 0
Bit swap: enabled enabled
Bit swap count: 0 0
Line Attenuation: 0.0 dB 0.0 dB
Signal Attenuation: 0.0 dB 0.0 dB
Noise Margin: 6.3 dB 9.8 dB
Attainable Rate: 86392 kbits/s 28273 kbits/s
Actual Power: 12.7 dBm 4.5 dBm
Per Band Status: D1 D2 D3 U0 U1 U2 U3
Line Attenuation(dB): 10.9 25.7 39.9 0.6 20.4 31.5 N/A
Signal Attenuation(dB): 10.9 25.7 39.9 0.6 20.2 31.4 N/A
Noise Margin(dB): 6.3 6.3 6.3 10.2 10.0 9.6 N/A
Total FECC: 107824 69
Total ES: 1 3
Total SES: 0 0
Total LOSS: 0 0
Total UAS: 0 0
Total LPRS: 0 0
Total LOFS: 0 0
Total LOLS: 0 0
Full inits: 1
Failed full inits: 0
Short inits: 0
Failed short inits: 0
Firmware Source File Name (version)
-------- ------ -------------------
VDSL embedded VDSL_LINUX_DEV_01212008 (1)
Modem FW Version: 130205_1433-4.02L.03.A2pv6C035j.d23j
Modem PHY Version: A2pv6C035j.d23j
Vendor Version: Ap6v35j.23j 68
DS Channel1 DS Channel0 US Channel1 US Channel0
Speed (kbps): 0 73775 0 20000
SRA Previous Speed: 0 0 0 0
Previous Speed: 0 0 0 0
Reed-Solomon EC: 0 107824 0 69
CRC Errors: 0 1 0 3
Header Errors: 0 0 0 0
Interleave (ms): 0.00 8.00 0.00 0.00
Actual INP: 0.00 3.00 0.00 0.00
Training Log : Stopped
Training Log Filename : flash:vdsllog.bin
As you can see - the Reed-Solomon EC count is already at 107,824 and it carries on rising into the millions. There doesn't seem to be any particular consistently to it.
I've been in contact with our broadband providers who haven't really able to help, and insist that we have to use the BT Openreach modem in front of our router otherwise it's not a supported configuration. Vodafone are also telling them the same thing when it was escalated. I've also been told - which I don't believe - that the 887VA can't support 80/20 VDSL, but the CPU utilisation on it barely goes above 15% even with CBAC and IPSEC active, so it seems meaty enough.
Vodafone also gave me a load of settings that don't (I think?) apply to VDSL at all - e.g. VCI, VPI and VC-Mux/VC-based multiplexing? I can't see any way of setting these on the Dialer0 interface, and the ATM0 interface is administratively down as it's unused.
Can anyone shed any light on this?
Is our "Noise margin" too low? Could a mismatch of MTU cause this kind of problem? (currently I have no MTU settings explicitly set, and sh int shows it as 1500 - which Vodafone also said we should aim for).
I should point out that speed wise the speed tests I've run have consistently shown download speeds of ~70Mbps, and upload of ~18Mbps, and there hasn't been any desyncs that I've seen - so whatever is causing those Reed-Solomon EC numbers doesn't seem to be adversely affecting the line or the quality of service.
