Kitz Forum

Broadband Related => Telephony Wiring + Equipment => Topic started by: Pfrog on February 28, 2009, 11:54:24 AM

Title: Effects of different ADSL Filters
Post by: Pfrog on February 28, 2009, 11:54:24 AM
Hi

I am currently using a DSL-2740b (having tried many other routers but find, like others, that it holds the line well and has the added benefit of having the SNR margin controlled by DMT TOOl) and was until this morning using an adsl-nation XF-1e filter.  For no particular reason I thought I would try replacing the XF-1e with the DLink filter which came with the router.

I was very surprised to see a much broader range of tones showing on the DMT Tool display with the DLink filter, -as attached.  The DMT Tool display for the XF-1e shows only x 3 individual tones above the main block ending at around 152.  I havn't attached a copy as I am loath to switch back to the XF-1e as the DSLAM seems very keen to knock up the SNR margin (currently 12db).

Am I missing something here?  I was under the impression that the XF-1e was the "bee's knees" of filters, and yet it appears to be cutting of a large swathe of useful higher tones.

This test has been this morning at around 09.00, - I have noticed that at night the XF-1e produces no trace at all of tones at all above the main block of tones.

PS  Has anyone got RouterStats Lite working properly with the DSL-2740b?  I see it is meant to work (I am using the EU_2.89 firmware) but when I can make it start, it loses connection frequently and the readout is full of red dropouts (both via cable and wifi)......

All thoughts appreciated.....

Many thanks!

Pf



[attachment deleted by admin]
Title: Re: Effects of different ADSL Filters
Post by: roseway on February 28, 2009, 12:46:30 PM
That's rather interesting to me, as I use an ADSLNation filtered faceplate, which I believe has the same filter circuitry as the XF-1e. I don't see the same cutoff in the tone allocation as you do, so I don't think that the filter is actually cutting off those higher frequencies. But it's an active filter, so its characteristics are quite different from the normal passive filters, and your experience may just be that the XF-1e isn't so good in your particular set of circumstances.

I did use a DSL-2740b here for a while, and used Routerstats (full version) with it using the telnet method, but I found the same as you that it often failed to collect a sample from the router and displayed what looked like a lot of dropouts. There was  short discussion about this on the Routerstats forum, but it wasn't conclusive.
Title: Re: Effects of different ADSL Filters
Post by: Pfrog on February 28, 2009, 01:19:25 PM
Hi RW

Interesting.  I have in the past noted that the XF-1e introduced a clearly evident 1db (approx) of extra attenuation over simpler filters  which I have always put down to the added complications of the active circuit (is it "active" in the normally accepted sense in that it introduces some gain to compensate for the inherent circuit losses, or merely that it uses "active" transistors which could be used for gain?).  I am aware that the world of active / passive filtering is quite an advanced topic and would not attempt to debate it in any detailed way.

I shall have to see if I can replicate the results and if so I will copy both sets of traces.....  As you say it may be just my particular line which does tend to be quite noisy (in the middle of rural Worcestershire with a lot of overhead line).

The question is, shall I keep the Dlink in circuit with an undoubted lower attenuation effect (albeit very small) or put the XF-1e back in place with its probable superior filtering but more loss??

Thanks for your comments on the DSL-2740b which confirms my findings as I have no problem at all with the Telnet method.

Incidentally, the 2.89 firmware for the DSL-2740b now allows "PPP IP Extension" mode which passes the WAN IP Address to another router and thus can work effectively as a half-bridge modem only.  I have used it thus with the WRT54GL with Tomato firmware, and it appears to work well, with the extra benefit of being able to tweak the SNR Margin.....

Pf
Title: Re: Effects of different ADSL Filters
Post by: orainsear on February 28, 2009, 01:58:32 PM
merely that it uses "active" transistors

That's correct.

Passive filters use only passive components; coils, caps resistors etc and only filter when a signal is received.

Active filters will also use active compontents alongside passive; usually transistors, and filter continuously.

The question is, shall I keep the Dlink in circuit with an undoubted lower attenuation effect (albeit very small) or put the XF-1e back in place with its probable superior filtering but more loss??

The attentuation that the DLink is reporting is based on the frequencies that it can see.  If the filters are passing through slightly different frequencies this will have a bearing on the attentuation.  It's probably not something to get too hung up over unless it's causing serious issues.  You can try and experiment (gently so you don't upset your line profile) with different combinations to find out what suits you best.

Edit: Everything after 'Thats Correct'
Title: Re: Effects of different ADSL Filters
Post by: roseway on February 28, 2009, 02:03:03 PM
Yes, I don't know any details, but I think the transistors are there to sharpen up the filtering behaviour, not to provide amplification of the signal.
Title: Re: Effects of different ADSL Filters
Post by: Pfrog on March 05, 2009, 11:05:02 AM
Well, this morning the XF-1e was still showing very few higher tones (via the DSL-2740b) so I replaced it with the DLink filter and the two respective DMT Tool screenshots are below  (I refreshed the DMT Tool trace to make sure that it was fully up to date in both cases)....  Spot the difference!  The two were within 5 mins of each other so whether I have a faulty XF-1e I don't know but interesting all the same.  The good news for me is that my SNR appears to have been reduced to 9db as I didn't need to use DMT Tool to tweak it!

Anyone else able to replicate something similar I wonder?  Could it be the DSL-2740b failing to register all the available tones after a period of use?

Pf



[attachment deleted by admin]
Title: Re: Effects of different ADSL Filters
Post by: roseway on March 05, 2009, 11:29:46 AM
I certainly can't reproduce that with my XTE-2005. Apart from the odd gap, there are good levels of bit loading up to well past 800 kHz. Maybe you do have a faulty filter, or maybe it's something about the interaction between the router and the filter, I really have no idea.
Title: Re: Effects of different ADSL Filters
Post by: sevenlayermuddle on March 05, 2009, 02:00:18 PM
Forgive me for chipping inn my ha'ppence worth here, but I've sometimes experimented to try and explore the different properties of the collection of filters I've accumulated.

Whilst I certainly agree that some seem to provide a better DSL connection that others, it's also worth mentioning that I noticed, with some more than others, that physical orientation seemed to make a discernable difference (up to 1.5dB or so).  I can only imagine that the inductive components, which are there to prevent DSL tones from pollutuing the telephony voice-path, may have been acting as a directional antenna in relation to whatever noise was radiating nearby.  I'm sure the designers will have tried to mimize any such tendencies, of course.

I'm not suggesting that there's any benefit in trying to 'tune' the filter by twisting it one way or the other as the effects are too small and too spurious.  But I did find the difference was enough to make reallistic comparison of different filters more difficult, especially those that just dangle on a trailing wire from the socket.
Title: Re: Effects of different ADSL Filters
Post by: Pfrog on March 06, 2009, 08:34:22 AM
I appear to have identified the possible cause of this problem.....  This morning after having used the DLink filter over night, the DMT Tool tone trace remains at the overnight profile with very few higher tone traces.  Thus it appears to be irrespective of which filter I use with the DSL-2740b router, it seems to fail to re-recognize the increased higher tones available in the morning, following the overnight reduction.

If I reboot the router, all the higher tones reappear on DMT Tool.  I have tried to close down DMT Tool completely and then opening it again but still the much reduced higher tones are shown so it does appear to be an issue with the router, or at least with how it is reporting the tones.

Is anyone else using the DSL-2740b also experiencing this I wonder?

Pf
Title: Re: Effects of different ADSL Filters
Post by: roseway on March 06, 2009, 09:10:39 AM
I think this is going to be bitswapping then. During the night the higher tones become more prone to interference, and the data is swapped to lower tones, with the higher ones being marked as unusable.

http://www.kitz.co.uk/adsl/adsl_technology.htm#bit_swapping
Title: Re: Effects of different ADSL Filters
Post by: Pfrog on March 06, 2009, 09:14:46 AM
Hi

Yes I appreciate the bitswapping element but surely when the noise has subsided (ie now in the morning), the available higher tones would once again be used (and apparent in DMT Tool - there is no bitswapping showing in the traces at present) and the bitswapping largely ceased,- or am I misunderstanding the issue - which is very possibly!!

Pf
Title: Re: Effects of different ADSL Filters
Post by: jeffbb on March 06, 2009, 09:20:49 AM
Hi
I might be wrong but I thought that once a bin had been marked as unusable it remains so until the router is rebooted .  I thought I read this on this site ?? but cannot recall where .
regards Jeff
Title: Re: Effects of different ADSL Filters
Post by: Pfrog on March 06, 2009, 09:35:47 AM
Hi Jeff

You may well be right although this from the very helpful Kitz guidance appears to indicate otherwise:

"Bit Swapping is a way of keeping the line more stable by constantly monitoring the frequency bins (carriers) in use and reusing them if possible."

I assumed that if there was continual monitoring then bitswapping would only be applied when it was necessary - which seems to be the case as I have no indication of bitswapping during the quiet day period in DMT Tool.

Cheers

Pf
Title: Re: Effects of different ADSL Filters
Post by: roseway on March 06, 2009, 09:52:52 AM
I'm certainly no expert, but I think that routers should be able to re-use bins which have previously been marked unusable, but maybe the DSL-2740B doesn't do it so well. This is pure speculation of course.
Title: Re: Effects of different ADSL Filters
Post by: HPsauce on March 06, 2009, 10:00:55 AM
I think that would only be when SRA is active, which it generally isn't in the UK.
Title: Re: Effects of different ADSL Filters
Post by: Pfrog on March 06, 2009, 10:05:25 AM
Hi HP

Point taken about SRA but surely that is more to do with the adaptive sync speed and not how bitswapping as such is utilised?  I find it hard to understand a standard where a particularly bad burst of noise could permanently remove a great swathe of tones from ever being used - or at least until the router is rebooted? 

Pf
Title: Re: Effects of different ADSL Filters
Post by: jeffbb on March 06, 2009, 02:22:09 PM
HI
I wonder ? is this logical ?
lets assume that successful bit swapping has taken place (no loss of synch) because of noise and in that process bins X,Y,Z were marked as unusable .Then the bits that were in bins X,Y,Z are now distributed to other bins . So even though the bins may become usable again they will NOT be used until they are required . So until some noise in other channels force bits swapping to populate bins  X,Y,Z (this may not  happen if there is enough SNR in currently populated bins ) the bins will remain empty. If on the other hand you reboot or swap routers then A brand new bin allocation is created .Populating all available bins optimising  SNR. 

Regards Jeff
Title: Re: Effects of different ADSL Filters
Post by: Pfrog on March 06, 2009, 02:55:13 PM
Hi Jeff

Sounds perfectly feasible - you may well be right.  However to try and remove the DSL-2740b out of the equation as being the "problem", I have this morning replaced it with a Zyxel P660r-D1 (in half-bridge mode into a WRT54GL).  I am monitoring the Bit / Tone Allocation on OrbMT and will compare the screeeshots tomorrow morning after a night time of noise and will see if the higher tones reappear!  If they don't, the DSL-2740b will be restored......

Regards

Pf
Title: Re: Effects of different ADSL Filters
Post by: Pfrog on March 07, 2009, 10:20:37 AM
Well, they always say that practical testing is better than reading / trying to understand the theory!

The Zyxel router has behaved in exactly the same way as the DLink.  The higher tones have not been reallocated this morning and the daytime trace shows the same much reduced higher tones as the DLink......  I assume that if I were to reboot the Zyxel the higher tones whould reappear. 

I have learnt something at least!
Title: Re: Effects of different ADSL Filters
Post by: jeffbb on March 07, 2009, 11:04:35 AM
HI

So there is no problem  ;D

Yes I reckon that if you reboot then they would reappear .Not necessarily exactly the same .It would depend on how the bits are allocated .

Regards Jeff
Title: Re: Effects of different ADSL Filters
Post by: philip_l on May 20, 2009, 07:26:45 PM
Hi

The action of tones disappearing is definately bitswapping in action, this will often effect tones towards the end of the received range as these typically are the ones hardest to hear, but chunks can disappear from anywhere due to random noise that comes and goes.

Once they are completely bitswapped they will be de-activated and not used again until a resync, where the cycle continues.  This bitswapping and disabling of the tones also results in a gradual lowering of reported SNR margin until a resync.

Regards

Phil