[Elecraft] 8 pole vs 5 pole..I knew this would happen

[Geoffrey Mackenzie-Kennedy]

With respect Larry I must disagree about the term "roofing filter" being 
misleading.  I completely agree that a narrow filter at the first IF is 
desireable if not essential, and it could be identified as a roofing filter 
in some instances - see below. This approach has been the norm in the design 
of certain classes of high performance receiver for some time, and obviously 
this filter's bandwidth must be compatible with the receiver's operating 
mode.

I think that the confusion in understanding the meaning of the term "roofing 
filter" started in the amateur community sometime after commercial double 
conversion receivers began to appear on the amateur market, and appears to 
be increasing. Because these receivers used very wide roofing filters, and 
many still do, the myth arose that roofing filters were always wide and were 
only used in double conversion receivers. In turn this gave birth to other 
myths about the poor performance of double conversion receivers vs single 
conversion receivers, which often can be traced back to poor design and poor 
electro-mechanical construction. The term "roofing filter" was intended, and 
has since when correctly used, to identify the first narrow bandwidth IF 
filter appearing in a receiver's signal path after the first signal mixer, 
but *only* in those cases where additional IF filtering was introduced 
further down the IF chain for the purpose of establishing the overall RF / 
IF selectivity - as found in many amateur double conversion receivers and 
early single conversion ISB receivers for example. In the ISB receivers with 
which I was involved in the 1950s, the typical bandwidth of the roofing 
filter was slightly greater than twice the required traffic bandwidth of 
each of the following USB and LSB filters i.e.roughly speaking 7 kHz for a 
basic two channel at baseband receiver, not tens of kHz. In later years a 
variety of roofing filters, some wide some very narrow have crossed my path. 
The term does *not* and was *not* intended to imply that that the receiver's 
architecture is double conversion nor that the bandwidth of the roofing 
filter is by default wide, and is not used to identify any filter outside of 
the IF cascade. Although it is tempting to identify the roofing filter as 
the 1st IF filter, this could imply that there were other IFs used elsewhere 
in the receiver in question e,g dual conversion or triple conversion, and is 
usually avoided.

In the case of a straightforward single conversion receiver using a single 
set of filters (or variable bandwidth in the case of the K2) the IF filter 
should not be identified as a roofing filter.The small filter prior to a 
product detector to attenuate unwanted sideband IF generated noise does not 
count as a second filter, because according to the "rules" the same result 
can be achieved by using an image reject mixer as a product detector  On the 
other hand if for some strange reason a single conversion comms receiver did 
employ a widish bandwidth IF filter close after the mixer and narrow 
bandwidth IF filters further down the IF chain, at the risk of questionable 
IMD performance if the cascade between the filters is weak, it would be 
correct to identify the first filter as a roofing filter.

In my opinion if a filter is performing the role of a roofing filter its 
identity should not be changed from "roofing filter", which is a well 
established term both inside and outside of the amateur community.

73,
Geoff
GM4ESD


On Friday, May 04, 2007, at 2:11 AM, Larry Phipps wrote:

>I think the term "roofing filter" is misleading. A narrow filter at the 
>first IF protects a receiver even better than a "roofing filter", so there 
>is nothing inherently distortion reducing in using a wider filter at the 
>first IF and then a narrower one later. The ideal situation for IMD would 
>be a pair of matched narrow filters at both IFs. The real reason for a 
>"roofing filter" it seems to me, is to allow passband or slope tuning. This 
>compromises IMD and AGC performance for the sake of a feature... which may 
>or may not be valuable to the user.
>
> Therefore, the "roofing filter" should be termed the "passband tuning 
> enabling filter", or "PBTE" filter ;-)
>
> Thankfully, I think Elecraft has done a brilliant job of giving us the 
> options we want without compromises. By tying the DSP bandwidths and PBT 
> functions to the "roofing filters", we have the ability to have the 
> combination of 1st and 2nd IF BW we want,,, and with the variable "roofing 
> filters", I think we will be able to almost set the relative BWs between 
> the two... allowing a window for PBT or not as we choose. This is an 
> exciting development, and will be copied by many companies over the next 
> year. Kudos to the design team on this.
>
> 73,
> Larry N8LP