[Top] [All Lists]

Re: [RFI] RFI U-verse

To: "J.Gordon Beattie, Jr., W2TTT" <w2ttt@att.net>
Subject: Re: [RFI] RFI U-verse
From: "Christopher E. Brown" <cbrown@woods.net>
Date: Sun, 27 Feb 2011 19:14:31 -0600 (CST)
List-post: <rfi@contesting.com">mailto:rfi@contesting.com>

I think there is a terms issue here.  I did refer to a common mode choke 
as a filter once, but the important part is _common mode_.  And it does 
have band-stop, high-pass or low-pass behavior depending on the core and 
windings.  32turns on a 2.4" #31 toroid is a common mode only high pass. 
High common mode impedance on 160, decent on 80M and dropping from there. 
Very little impact on say 20M given the environ it is normally installed 

A lower turn count can provide say a "band stop" type behavior on centered 
on 20M, but doing very little on say 160M.  Again, common not 

So yes, I do consider a common mode choke to be a form of common mode 
filter because it does show frequency dependant behavior.

The desired signal is differential.

One can use low pass, high pass, etc multi-element filters to provide 
_differential mode_ filtration.

Rolling you own here is generally a bad idea, too many standards, etc.

Proper use of differential mode filters is by the provider.

In a whole house DSL + POTS setup "DSL blockers" are approp on every POTS 
drop.  These are a simple low pass filter designed to pass the audio to 
the POTS device without shunting higher freqs or reflecting them.

A better setup is a "whole house splitter" installed in the NID.  This is 
a high pass and a low pass filter in the same case.  The low pass leg goes 
to the POTS, the high pass to the DSL modem, and all properly designed for 
proper split, no reflections, etc.

These should come from the service provider, and are very specific to the 
technology in use.  Don't roll your own, to easy to cause issues. 
Pass-band is not the issue, phase shift and waveform distortion _is_.

The other being common mode.

A proper common mode choke is as close to non-existant as possible in the 
differential mode.  I do slip into saying "common mode filter" at times, 
as it does act as a very wide "band stop" filter, but only in the common 

The 2 biggest issues I see in the field are

1) Common mode becoming differential mode due to issues in the cable or 
devices connected.  Block the common mode before it becomes diff mode and 
improve the line.

2) Very high levels of common mode acting to "de-sense" the diff mode 
receiver.  This one is much more rare, but happens.


Diff mode filtering, stick with the service provider supplied filters 
selected for the specific tech, it is not as simple as is seems to roll 
your own.  Better to work on improving the cabling...cheap cabling and 
poor install methods degrading balance is very, very common.

A proper _common mode_ choke, A.K.A common mode filter that does not 
interfere with the diff mode signal is generally a win, and in my expr 
never a loss.

On Mon, 28 Feb 2011, J.Gordon Beattie, Jr., W2TTT wrote:

> Chris,
> I think we have more of a terminology issue here.  Common mode
> isolators/chokes, or as you have called them filters, are of value.  I just
> don't think of them as filters as they do not have bandpass or reject
> qualities.  Small point of difference, but I think we agree.  As you
> suggested, common mode isolators do all the good things that you suggested.
> We have seen cases where people do put bandwidth-limiting devices ahead of
> the DSL modem in an effort to keep specific frequencies out of their modem.
> The result may be more permissive for Amateur Radio operation, but can
> reduce the overall bandwidth available for the DSL service, which can make
> the service more vulnerable to other noise or line performance-driven
> impairments.
> On your other points, you are "spot on" as well, as those cleanup measures
> do make a significant operational difference when seeking to improve the
> bandwidth and the robustness of any DSL service.
> Thanks & 73,
> Gordon Beattie, W2TTT
> 201.314.6964
> -----Original Message-----
> From: Christopher E. Brown [mailto:cbrown@woods.net]
> Sent: Sunday, February 27, 2011 11:54 PM
> To: J.Gordon Beattie, Jr., W2TTT
> Cc: jim@audiosystemsgroup.com; 'Gordon Beattie'; rfi@contesting.com
> Subject: Re: [RFI] RFI U-verse
> On Sun, 27 Feb 2011, J.Gordon Beattie, Jr., W2TTT wrote:
>> Jim,
>> A low frequency cutoff strategy can negatively impact overall service
>> stability.  The best strategy is to keep the transmitter RF from getting
>> into the cabling and as you said, good cable, properly installed is a key
>> starting point.  Bonding and grounding of the cabling is another good
> step,
>> as is reducing the proximity of the antenna system to the service drop and
>> other cabling.  Line isolators are another good tool for the service line
>> and other cabling to the DSL RG/modem.  Chokes only seem to be effective
>> when applied in front of STBs and not on the service line to the DSL
>> RG/modem.
>> Thanks & 73,
>> Gordon Beattie, W2TTT
>> 201.314.6964
> While I agree with most of this, I have to strongly disagree with you
> comments about a common mode filter on the DSL side of the line.
> When you are saying is the _exact opposite_ of what I have seen many
> times.
> I have _never_ seen a proper common mode choke reduce the performance of
> any DSL variant, symmetric or asymm, legacy through VHDSL.
> By proper, I mean (as one example) a high twist rate 22 - 28ga pair,
> wrapped through a 2.4" Type 31 toroid, turns count approp to the
> interfering signal.
> Someone doing a scramble wrap of untwisted, or worse yet a split
> winding will cause issues but this is not a proper setup.
> I have never seen this cause issues or reduce sync rate.
> It often increases sync rate, due to enforcing balance and blocking
> "common mode" RF pickup from the environment that then becomes
> differential somewhere between the pre and the CO.
> This is my personal experience, where in my last three homes (all <
> 6kft from CO) initial ADSL2+ sync rates were < 5Mbit and any HF use would
> cause loss of sync or errors depending on band.  Fixing the house cabling
> (ADSL2+ rated splitter at the NID, general cleanup and dedicated run from
> the splitter to the modem) brought sync rate up to 10+ Mbit, with only
> minor error rate increases on some bands.  Placing properly wound chokes
> at both ends of the NID to modem run increased sync rates to 15+ Mbit and
> eliminated any change in error rates when on HF.
> This is also my experience at work.  I have been a packet data networking
> guy for 17 years now, near 8 years in the engineering dept of of one of
> the 3 largest communications providers in the region.  My normal workflow
> centers around things much larger than a single customer drop, but the
> department is the final level of technical escalation for issues.  We use
> alot of ADSL and various multipair symmetric Ethernet over Copper
> platforms.
> The first thing to check is proper cabling, then equipment grounding and
> bonding.  After that...   I have had chokes installed in a number of
> cases, on simple ADSL lines and one per pair on multipair EoC drops in
> high noise environs (commercial locations with lots of large VFDs, etc,
> repeater sites, broadcast toper locations, etc).
> It does not always help, but I have never seen a circuit degraded by
> properly built and applied chokes.  About half the time, we see a increase
> in sync rate and lower error rates...
> On the other front (degrading the line), I have seen chokes like
> 2 separate 28ga wires, wound in 2 "lumps" about an inch apart on a 2.4"
> #77 core
> one wire loose/messy wound over one third of the toroid, the other
> loose/messy over the top of the first with ~ 50% overlap
> A very very clean winding, but split on opposite sides of the core, and
> with one side reversed during install
> All of these pretty much took the circuit offline.
> I normally take 24ga insulated, twist to about 4 turns per inch (4 full
> 360, not 4 half turns) and wrap anywhere from 10 - 32 turns on a 2.4" #31
> toroid.  There are 2 inline with the ADSL2+ annex M line I am using right
> now.  Line is 5.5kft, set 9db/9db for channel threshold levels and syncs
> about 2Mbit faster in the downstream with these in place.
> And yes, I have tested at work with "perfect" circuits, 1kft of quality
> cable in the CO for ADSL1, ADSL2+, eGSHDSL and several VHDSL platforms
> with 2 inline found no negative impact on a line already running at max
> sync.
RFI mailing list

<Prev in Thread] Current Thread [Next in Thread>