Kitz Forum

Broadband Related => Telephony Wiring + Equipment => Topic started by: JGO on January 22, 2014, 09:15:22 PM

Title: A filter, " but not as we know it Jim ! " [RF3]
Post by: JGO on January 22, 2014, 09:15:22 PM
It seems that  RF3 type devices are not understood by everyone so this is an attempt to clarify and an appeal for more facts.

The device acts as a mode filter, attenuating interference. It is NOT a frequency selective filter so it can improve S/I ratio even when the frequency bands overlap.

Firstly imagine a single wire which is acting as an aerial and picking up interference. Then inserting a coil in series adds a large impedance and reduces the interference current by producing an out of phase "back emf". (fig 1)

If the "wire" is a pair then similarly, a coil in each wire has the same effect.
A 1:1 transformer in place of the coils has a similar effect to this parallel mode current(red arrows in fig 2).

However if there is a differential mode signal (shown in blue arrows),  the transformer has negligible effect on this since the back emf from one current is cancelled out by the opposing back emf by the return current in the opposite wire
fed in via the transformer.


This  simple explanation ignores losses, stray inductance and capacitance. The effect of these is a slight loss to the

wanted signal, a fraction of a dB, and restriction of the frequency bandwidth of operation. Thus a design which is OK for ADSL1 may be poor for VDSL2 due to their different max frequencies.

The I plate includes such a device as well as the bell wire choke. Can anyone confirm this is the same component as the in the RF3 ?

I think that RF3 =  RF2 but with less inductance, to work on ADSL2/2+ anyone confirm ?

The ideal no-compromise solution would be to build the device into the modem, so the requirements match and also remove any unprotected length of cable between v device and modem, but I'm not holding my breath ! Or do some modems include one
already ?
Title: Re: A filter, " but not as we know it Jim ! "
Post by: 4candles on January 22, 2014, 10:38:46 PM
The device included in the i-plate is indeed the same circuit as the RF3, as referred to here (http://www.run-it-direct.co.uk/forum/bt-junction-boxes/bt-80b-rf3-available-now-t38.html). The more authoritative link I once had seems to have disappeared from the webosphere.

The RF2 and RF3 both comprise just two inductors, so I think it must be just a case of different values.

Title: Re: A filter, " but not as we know it Jim ! "
Post by: JGO on January 23, 2014, 11:10:01 AM
Thank you for the confirmation.  Yes the more useful the link the shorter tim it lasts !
Title: Re: A filter, " but not as we know it Jim ! "
Post by: 4candles on January 23, 2014, 05:58:25 PM
Ah - how very true.   ;D
Title: Re: A filter, " but not as we know it Jim ! "
Post by: JGO on March 31, 2014, 05:09:00 PM
  P.S.

Any form of filter will have some slight loss, so if there is no interference for it to reject, it will degrade performance !  Recently I've been playing with an RF3 with the aid of SDLstats, although my line is giving close to it's theoretical performance for it's attenuation.  In fact although the number of usable tones increased, the bits per tone on some of the others decreased by 1, so the overall performance is degraded very slightly. The attenuation (at 300 kHz?) is the same and the dip due to Droitwich is removed.

If we had overhead lines and/or I hadn't isolated the phone extension ( aka interference aerial !)  it seems it could have made a considerable improvement.
Title: Re: A filter, " but not as we know it Jim ! "
Post by: JGO on August 17, 2014, 10:23:27 AM
PPS.

It seems useful to point out that the RF3 is an RF device, so unlike at 50 Hz, it's position in a cable is important.

It should ideally be inside the device it is protecting to minimise unprotected cable length. (Remember the modem is just a specialized radio receiver.)  This may not be essential, after all it's job is interference reduction to a negligible level, it can't ever eliminate it. 

RF3 devices for 'phones possibly originated for the GPO radio station at Rugby. The frequency was 16 kHz so the wavelength is 18.75 km. Thus the odd metre between device and phone was a tiny fraction of a wavelength so no great pickup.
 
With XDSL extending to 16 MHz, the wavelength is now 18.75 metres  !   so the same  unprotected metre is an effective aerial !!