DMT - Re-sync and SNR Stability

I've been using DMT with a Speedtouch V5 [firmware 6.1.4.7] for some time.

I decreased the Target SNM from 15 dB to 10 dB 7 weeks ago and have been monitoring the SNR variation. Recently it typically varied rapidly and constantly by plus and minus 1 dB.

I've just re-set the Margin to its default 15 dB and the SNR is rock solid at 15.5 dB.
Is it simply the case that with the higher Margin the SNR is inherently more stable or is it the case that the re-train has increased SNR stability.

In other words does the occasional re-sync increase stability ?.

If you're saying that your SNR doesn't vary by more than 1dB during the whole 24 hours in the day, then your target SNR should be 6dB, not 12 or 15. A target SNR of 12 or 15dB is usually only set when there are jumps in the SNR which would make the connection unreliable.

As an example, my old Voyager 2091 operated around SNR 5 - 6 db, 2.5Mb data rate and used to resync every couple of days.

My new Belkin holds the connection when set at 4db giving 5Mb data speed. I have seen the variation in line quality take that down to 2db and it's still held the connection. I did try to push it to a 3db which was fine for a few hours but then the SNR dived to 0!

So you need to set the SNR to have a bit in reserve. Anecdotally, it seems some people have bursts of noise and then periods with no noise - hence they need to leave a big margin. Others seem to have a more steady noise profile so can squeeze it tight.

Richard

Different modems will work down to different SNR margins. My Speedtouch 546's will happily sync at 2dB margins, sometimes down as low as 1.5dB. I have to set higher margins than that, however, because the noise can vary by about 8 or 9dB over a 24 hour period. My approach is to have a PC which is on 24/7 (which is an old PC in my office acting as a local file server) which runs DMT, and automatically re-syncs the modem if the SNR is outside the range of 2<>7 dB. Using this technique I can squeeze the line profile up a little.

In reply to:

---

My new Belkin holds the connection when set at 4db giving 5Mb data speed. I have seen the variation in line quality take that down to 2db and it's still held the connection. I did try to push it to a 3db which was fine for a few hours but then the SNR dived to 0!

---

The SNR figures may not be reliable. My Belkin held the line at what it said was very low SNR levels, but when I changed it for a Speedtouch the SNR levels, for same sync speed, were higher and much more stable.

In reply to:

---

The SNR figures may not be reliable. My Belkin held the line at what it said was very low SNR levels, but when I changed it for a Speedtouch the SNR levels, for same sync speed, were higher and much more stable.

---

What models?

I can't remember what the Belkin was - a basic wireless with two aerials that came with my new Dell PC about 9 months ago. The Speedtouch is a 585v6 running 6.1.4.3.

The reason my Target SNM is 15 dB is because a BT fault came along but that's a separate story.

My original post relates to my having increased the Target SNM which seems to have increased the SNR stability - ie less frequent and less extreme variations in SNR.

"Is it simply the case that with the higher Margin the SNR is inherently more stable or is it the case that the re-train has increased SNR stability" - or both.

Any thoughts ?.

Edited by deleted (Sat 03-Feb-07 13:07:12)

From a common sense, rather than knowledge, point of view I'd guess that neither changing the SNR Margin nor retraining would change your SNR characteristics, but that the latter might change the way your router reported them.

Thanks Colin:

With a Target SNM of 15 dB as against 10 dB my SNR is far more stable.
My best guess is that with a higher sync rate perhaps the SNR inevitably fluctuates more.

The other possibility relates to the Speedtouch using DMT.
Perhaps when it re-syncs it takes account of errors or the SNR on the different frequencies involved and on re-syncing it may change the way different frequencies are used and may therefore give a more stable SNR.

I've no idea if anything along these lines takes place but this could account for the increased SNR stability after the re-sync. If it is the case, an occasional manual re-sync could be beneficial if the SNR fluctuation starts to increase.