Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

I have a Raspberry Pi and set it up to record the SNR margin each second.

Unless you are doing active testing then I would change the sample rate. Querying the modem every second can be quite intensive especially if the modem is also having to correct any errors. 
G.INP adds a slight standard processing overhead, then more depending on if it can itself correct errors plus more for interleaving and RS correction. We usually suggest a sample rate of every minute... or 30 secs at most.

Reasons
~ Running both DSLstats & HG612 modem stats too close together every min causes collision issues. To avoid this Eric added a small routine to check if HG612 stats had finished sampling before starting a DSLstats sample.
~ Watch how long the DSLstats "Sampling" message is displayed for

I have a daily burst of SHINE usually lasting a split second, but occasionally it gets stuck regenerating errors until I refresh the connection. 
 I have an alert set up in DSLstats to send me an email notification if I had a certain amount of errors.  This system worked fine for many years and I would manually reboot the modem to clear the errors.  However g.inp doesnt work too well on all types of noise.  Since g.inp has been applied to my line the errors seem to get backed up and no useful data seems to get processed..  meaning the  DSLstat alerts never reach me. DSLstats log shows that its tried to send mail but failed.

That's a weird setup, I mean why didn't they just put the master socket where that junction is and call it a day?

Problem is if the extension is officially the master socket now, then technically you aren't allowed to touch any of this wiring (at least not the active bits).

Is that socket below not physically connected to anything above?  I'd be tempted to completely remove it if so, or at least the cable so its clearer to see what is going on.

Seems theres a lot of loose cable in here not actually doing anything that could be cut away to make this tidier (the unused pairs from the extension cable), you just don't want to mess with anything that is the active pair.

The important thing is to not have anything connected to the active line then leading to nowhere, although from what I can see there is only two cables going into crimp, suggesting it does only connect to the new master socket and 99% of what is in here is redundant cabling.

That's not even the most practical way to do it:
https://www.aliexpress.com/item/1005003533018920.html

Although I own one of these and had problems getting it to work in the appliance I got it for, but it did work in a desktop PC.

Found something you may like:
https://www.candelatech.com/adtran_sdg_8733_dev.php

Is the extension lead twisted pair and are you sure it has been wired to use an actual pair as any real length of non-twisted cable will be prone to interference?

I don't think it is twisted pair. There are various unused 'strands' of the extension cable and it looks like they've just been cut, creating a spaghetti of wires.

Below are two photos. The top plate covers where the (underground) phone line comes into the house. Behind this is where the extension lead (off to the left) is connected and runs about 10 metres around the room to a socket on a wall which looks like a BT master socket.

Also connected is a short cable going to the bottom socket, but it terminates in the socket and isn't connected, so the socket is dead.

The connections behind the top plate seem to be made using some sort of transparent connector. None of it looks good to me!

Is the extension lead twisted pair and are you sure it has been wired to use an actual pair as any real length of non-twisted cable will be prone to interference?

Hi

I've got FTTC broadband with BT. Download sync speed is around 42 to 48Mb. SNR margin usually around 5db, but this fluctuates quickly (e.g. it could be 5 one second and then 5.8 just a second or two later).

An extension lead is wired into the original master socket (which itself is disconnected) and the lead runs around the edge of the living room into a new BT master socket on the wall, which the Billion router's connected to. (This setup was done by the previous homeowner!)

Since I moved in in 2018, I've had the occasional dropout but nothing too bad, with syncs often lasting for several weeks. At the end of November, after being synced for a month, the router resynced with a lower SNR margin (around 2 or 3db) and the sync speed was much higher than usual (52Mb). I assume this was DLM intervention.

This caused a lot of errored seconds and I assume the DLM intervened again as it resynced again at a higher SNR margin, lower speed and then the errors went. Since then there have been random resyncs once a week or so at random times of day. I have a Raspberry Pi and set it up to record the SNR margin each second.

I've found that every few days the SNR margin drops to below 2 for about 5-10 seconds but it doesn't cause a resync. Last night though it twice dropped down like this and went so low it caused two resyncs (I went into the router stats and it registered -4 db just before dropping). I listened to the phone line when the SNR margin dropped and there was no noise at all.

I've checked against when my heating kicks in and I don't think there's a link. I can't think of anything else that could cause it.

However, I've noticed the router can be sensitive to what I assume is electrical noise. For example I once switched a fan on upstairs and it caused a resync. I tried turning the fan off/on again and it caused another resync so it was definitely the fan. I rebooted the router, tried the fan again and it was then fine.

The router's near my Raspberry Pi and a couple of other power adapters for the TV etc., which I know isn't ideal, and neither is the odd master socket setup. However I've had it like this for years and it's not been a problem.

Is it possible that something like the fridge or freezer (in the next room) has started to cause interference on the mains? Or perhaps a nearby power adapter has started doing it?

I'm considering buying a Tacima mains conditioner to see if this will help, but I'd appreciate any advice as I'd like to ensure this problem doesn't worsen.

Attached are the current router stats since last night's resync.

Thanks.

There's a guy designing his own high end unit on Youtube etc Tomaz Zaman. Doing the full SFP route.
You might want to take a look!

I feel they missed the mark here.  What's the point of having only ONE 2.5Gbit ethernet port?

Good point, maybe the idea is that most people don’t need 2.5 Gb on the WAN side and it keeps the cost down? I dunno.

It does seem a bit odd not to just make both 2.5 Gb, VanTechCorner test seems to suggest its not a hardware / cpu issue.

OpenWRT One

Really like this idea, a low power, open source wifi / router at a reasonable price, I note it is also bravely described as 'unbrickable'. It is manufactured by Banana Pi ..

C

Filogic 820 is a very well supported platform:
https://downloads.openwrt.org/releases/24.10.0-rc5/targets/mediatek/filogic/
The WiFi6, 2.5GB WAN and 1GB LAN do not rely on proprietary drivers, and the Mediatek Filogic 820 Eval Board supports this as standard.
the unbrickable hardware switch between NOR or NAND is a nice touch for a developer, same with the serial USB C console:
https://docs.banana-pi.org/en/OpenWRT-One/BananaPi_OpenWRT-One

I feel they missed the mark here.  What's the point of having only ONE 2.5Gbit ethernet port?

Same chipset as the GL.iNet Beryl AX MT3000 and Brume 2 MT2500