New Zyxel Router Disconnections and Sync Speed

[#1]

I recently changed my router from a 3COM 3CRWDR100A-72 to a Zyxel P-660HWP-D1. The change was forced due to wireless hardware incompatibility.
It appears that both routers use the TI AR7 chipset yet the 3Com suffered from less disconnections and managed to achieve a higher data throughput than the Zyxel (up to 2200kbps vs 1800kbps via the BBC iPlayer diagnostics page).
My router displays the following information.
ras> wan adsl perfdata
near-end FEC error fast:  0
near-end FEC error interleaved:22230
near-end CRC error fast:  0
near-end CRC error interleaved: 333
near-end HEC error fast:  0
near-end HEC error interleaved:  0
far-end FEC error fast:  0
far-end FEC error interleaved:2856
far-end CRC error fast:  0
far-end CRC error interleaved:  12
far-end HEC error fast:  0
far-end HEC error interleaved:  0
ras> wan adsl version
ADSL Chipset Vendor: TI AR7 07.02.03.00
ras> wan adsl chandata
DSL standard: ADSL_G.dmt
near-end interleaved channel bit rate: 2336 kbps
near-end fast channel bit rate: 0 kbps
far-end interleaved channel bit rate: 448 kbps
far-end fast channel bit rate: 0 kbps
ras> wan adsl coinfo
ATUC Information
country code:15  0
vender:ALCB
ras> wan adsl targetnoise
Current target noise margin value:0
ras> wan adsl linedata far
noise margin upstream: 13 db
output power downstream: 16 db
attenuation upstream: 31 db
ras> wan adsl linedata near
noise margin downstream: 9 db
output power upstream: 11 db
attenuation downstream: 55 db
According to one calculator I found online, my connection speed should be around 2550Kbps for ADSL and if using ADSL2+ 2350Kbps to 4250Kbps. Regardless of the calculator, the ADSL connection is not performing as it was with the old router.
Can anyone offer an opinion based on the above data?

[#2]

I assume the increased number of disconnections with the Zyxel router router has probably caused the DLM to increase your target noise margin from 6dB to 9dB. This will have reduced sync speed by around 500kbps so your IP Profile is now 2000kbps whereas previously it was 2500kbps. Throughput will have correspondingly reduced as you have found.
So the key to this is probably to establish why the Zyxel disconnected more frequently. Is it still doing that even with the reduced sync speed? Does noise margin vary markedly between daytime and night-time periods?

[#3]

I have an open support request with PN to obtain details of disconnections from the RADIUS logs. Unfortunately the Zyxel router logs are not very helpful. The router only stores 128 log entries and I'm having difficulty configuring to only show connection related events. When the log fills, old events are lost. The old 3Com router could store log events for many days showing this information.
With regard to noise margin, I am not at home during the day Monday to Friday and hence would need to look at the weekend. If I'm up late one night I could look at the logs but currently I'm restricted to evenings and weekends. Is there any particular time at the weekend that it is worth checking the router stats (i.e. a particularly quiet time)?

[#4]

A further thought. My Zyxel router provides the functionality to enable or disabled the ADSL rate adaption mechanism. IIRC (it was late last night when I looked) this is turned off by default.
Should this feature be enabled or disabled? A quick search appears to show that this could help combat the target noise margin issue. Is this correct, should I enable it if it isn't already?

[#5]

I imagine if rate adaptation is enabled the connection could drop more frequently to try to track the target noise margin within whatever tolerance it uses (perhaps user selectable). It's the equivalent of you seeing noise margin was 11dB (say) with 9dB target, so restarting the router; it will reconnect with 9dB margin, but higher sync speed. Then in the evening you might notice noise margin is down to 7dB, restart again and it goes back to 9dB but with reduced sync speed. With the DLM overseeing things that will, at best, be totally ineffective since the IP Profile will not have time to improve; at worst it could be counter-productive with the DLM raising target noise margin because it sees disconnections.
For checking noise margin variation between day and night, all that is required if to record the value a couple of times during daylight (say 10am to 2pm) and similarly after dark when atmospheric conditions, street lights, CH, and other equipment in the home are likely to be generating RF noise at ADSL tone frequencies.

[#6]

When I first received the router I was unable to configure the router fully using the latest firmware and had to resort to an older version. With the router correctly configured I've discovered that I can upgrade the firmware and retain the previous configuration. The lastest Zyxel firmware does not upgrade the TI AR7 firmware but does contain other changes.
Is it possible to reset the target noise margin to so that I can see if the issue re-occurs?
From what I've read, it may take a while for the target noise margin to decrease if the later firmware cures the problem.

[#7]

PN have very kindly provided this timeline of my disconnections. Prior to the new router, the connection had been up for six days. Unfortunately after upgrading the firmware last night at 6pm, there was a further disconnection or two a couple of hours later.
Do the recent number of disconnections look reasonable or excessive?

[#8]

I'd say the recent number of disconnections is excessive, and almost certainly would cause the DLM to take action to reduce/eliminate them. Since interleaving is already on, for a 20CN connection that will mean in increase in target noise margin.

[#9]

Hi MKICampaign,
I certainly think you should raise a fault and have it investigated properly by the faults team.
Jojo

[#10]

For what it's worth, I used to connect using a ZyXEL router - albeit an old one, a P660HW-61, bought 2005. The variation in noise margin between daytime and nighttime on my line was much greater than it now is with my Plusnet-supplied Thomson TG585 v8. The margin with the ZyXEL used to drop from 6 to around 1-2 at night, whereas with the Thomson the drop is from 6 to around 4. I had more disconnections with the ZyXEL, which led to speed problems before I went on the 'no DLM' trial.
Apart from that (admittedly not trivial) problem I thought the ZyXEL was a pretty good router.
Graham

[#11]

Jojo - Thanks for the suggestion regarding raising a ticket. I'm a little loathed to do this as the issue occurs with the new router but did not with the old router as seen in the disconnections chart. The probable outcome would be to change the router.
grahamt - Thanks for the input regarding your experience with Zyxel routers. It appears this sort of behaviour is not uncommon with Zyxel routers although in my case it does appear that there are more disconnections than normal even for a Zyxel router. I know that the connection can be stable with a TI AR7 chipset, therefore my suspicion is that the firmware or other hardware in the router is the culprit.
I have read references to no DLM/static target noise margins, is this an option that could be considered here?

[#12]

Fixing the target noise margin is not an option on 20CN. However even if it were you would still suffer problems with disconnections and probably slowdowns until the connection dropped.
The viable solution has to be to replace the router with one that doesn't have this problem. Anything else is an attempt to paper over the cracks.

[#13]

Is there a no DLM option on 21CN now and if so would this be an option in my case?

[#14]

I have run a quiet line test (17070 -> option 2) with a single phone connected to a filter on the master socket (nothing else connected in any socket). Is this the correct procedure?
I experience the following:
- Low level constant background "static".
- Intermittent higher levels of static.
Should the line be completely quiet? What is likely to be causing the noise, is it a line problem or interference?

[#15]

Ideally the filter and phone should be plugged into the test socket so that all internal wiring is disconnected. During the quiet line test there might be a slight hum, but shouldn't be any crackles (static).
When your router is plugged in do you notice significant variation of noise margin second-by-second, especially during the daytime?