Slight FTTC speed drop

[#1]

And you really notice that missing 2Mb/s, right?

[#2]

So the latest is... Following a recent break in connection, likely when I was out during the day, what do I find?

I now get speeds testing up to 37/38 Mbps again. (Though very variable, for some reason)

[#3]

...But not today! 17 August:

02:17:02, 17 Aug. DSL Link Down: duration was 817320 seconds
02:17:03, 17 Aug. WAN connection WAN2_INTERNET_PTM disconnected.[ERROR_NO_CARRIER]
02:17:03, 17 Aug. WAN Sensing Auto sensing Running
02:17:03, 17 Aug. WAN Sensing Auto sensing Complete, interface selected
02:17:32, 17 Aug. WAN Auto-sensing detected port DSL WAN
02:17:40, 17 Aug. DSL Link Up: Down Rate=40000kbps, Up Rate=9999kbps; SNR Margin Down=17.0dB, Up=22.1dB

Yeah, that's definitely a higher SNR margin than previously reported, after a half hour break overnight. Something must have happened, somewhere.

Nobody knows! 😀

Another curiosity, as before slight speed reduction, Hub 2 now showing;

Downstream sync speed: 40 Mbps

Upstream sync speed: 9.999 Mbps

Instead of what was showing after slight drop in speed:

Downstream sync speed: 39.999 Mbps

Upstream sync speed: 10 Mbps

[#4]

The only thing which is going to explain your experience here is massive electrical interference giving rise to a high data retransmission rate.

This line is capable of far more than the provisioned 40/10 line rate, which is what is it synching at.  If the DATA speed test is not consistently close to that, then you need to look at the possible causes of slow data transfer OTHER THAN the line sync rate.  This considerations include...

• Local Area Network connection
  • WiFi
  • Ethernet over Power adapters
  • Host PC NIC configuration
  • Concurrency of use (other devices uploading / streaming)
• Device
  • Processor performance
  • Disk read / write performance
• WAN - data retransmission over heads due to (transient) line noise (is the phone line always quiet?)
• Exchange back haul (VLAN capacity)
• Remote data speed test service performance 

[#5]

There was never any doubt in my mind about the cause of the initial disconnection issue. It was due to the inevitable crackles when I moved the router from the normal extension socket to using it on the master socket - due to my testing a Wi-Fi 5Ghz link from the router at the master socket to the PC. Instead of the usual, router on extension socket connected to PC via Ethernet cable.

What then surprised me was, when everything settled back down to normal, the link was stuck at a slightly lower speed than the usual 37 Mbps. This persisted for several days until, following a disconnection overnight (I think), it reconnected at the usual 37 Mbps. Which it has stayed at ever since. Probably should have tried rebooting the router myself at the time (I though I had).

[#6]

That sound like normal DLM behaviour, in that it recognised that the line interruptions had stopped happening after several days (often around 10 days), and correspondingly increased your sync speed - typically done between 1am and 2am on my connection.

.

[#7]

...Could well be. I didn't though realise the DLM was that sensitive to minor interruptions - ought it to be?

Plus, my experience with the DLM on ADSL 2+ was that being marked down was a 'thing'. But you could NEVER get marked up again, ever!

From my above posts it appears to have been 3 weeks rather that 10 days.

[#8]

Yes, my ADSL was the same, ANY brief loss would case a drop that then NEVER recovered until getting a DLM reset.

When I was on Plusnet FTTC, DLM dropped my speeds typically twice a year, every year, usually due to brief mains glitches.

My stand alone VDSL modem is now protected by two uninterruptible power supplies (in series), and don't have any issues.

The only time DLM sees my line drop, is when I've done a modem firmware update and the modem needs a reboot.

.

[#9]

...I eventually gave up on ADSL 2+ when even a DLM reset stopped working. At least it did work until, around 5am the next morning, when the DLM unfailingly reasserted its authority - for no discernable reason, other than it could.

[#10]

---

@pvmb wrote:

...Could well be. I didn't though realise the DLM was that sensitive to minor interruptions - ought it to be?

From my above posts it appears to have been 3 weeks rather that 10 days.

---

Mistake! A misreading of the posting dates.  From 4 August to 17 August - so around ten days.