[Icom] RE: Filtering at 9 MHz vs. 455 kHz

[Siu Johnny]

Hi Aarson,

You have given a very informative message to which I agree.  Based the same
argument, I have modified the 455 IF filter of my IC756 (original version)
to get better selectively.  INRAD filters are often having better shape
factor than that of ICOM.

73

Johnny Siu VR2XMC
www.qsl.net/icom/ic756/fl44a.html

----- Original Message -----
From: "Hsu, Aaron" <aaron.hsu@unistudios.com>
To: <icom@mailman.qth.net>
Sent: Saturday, September 14, 2002 5:40 AM
Subject: [Icom] RE: Filtering at 9 MHz vs. 455 kHz


> Rick,
>
> I don't have an Icom HF rig, but analog filter chains basically work the
same on any radio.  Here's the basics...
>
> A "perfect" filter should cut off *everything* below and above the
filter's bandpass.  So, a filter with 300Hz to 2.7KHz bandpass should not
allow *any* frequency below 300Hz or above 2.7KHz through.  If plotted on
paper, it should look like rectangle with 90 degree vertical lines at 300Hz
and 2.7KHz and a perfectly flat horizontal line connecting the top of the
two vertical lines (this horizontal line represent the frequencies between
300Hz and 2.7KHz).  Now in the real world, it's impossible to have a perfect
analog filter (even digital filters aren't perfect).  The vertical lines are
more at an angle - anywhere 1 degree to 89 degrees depending on the quality
of the filter.  The shape factor is determined by taking the bandwidth of
the -6dB points divided by the bandwidth of the -60dB points.  The closer
you get to 1.0, the steeper the skirts the the closer you get to "perfect".
>
> Lower IF (e.g. 455KHz)filers have "shape factors" closer to 1.0 because
it's easier to build high "Q" filters in low frequency ranges.  However,
this often comes at a size penalty as component sizes get larger at lower
frequencies.
>
> Higher IF (e.g. 9MHz) filters have higher "shape factors" - the skirts are
further from being "90 degree" vertical.  Component sizes generally are
smaller, but it's harder to build high "Q" filters as frequency rises.
>
> So, which one is better???  It's a toss-up.  I asked this same question
many years ago and got the following two answers...
>
> By using a higher IF filter (9MHz), you prevent the overload further down
the IF chain.  In other words, filter out the crap at the door before the
crap overloads the rest of the system.  The downside is that the filter
skirts are wider, so even though you're getting rid of the crap early,
you're still letting in crap (though to a smaller extent).
>
> By using a lower IF filter (455KHz), you have much better filter skirts so
you can get rid of the crap that's near what you want to hear.  The down
side is that there's a lot of crap hitting the filter to begin with and it's
been amplified early in the IF chain.
>
> So, again, which is better?  It's personal preference.  The true engineer
would have you put the filter early in the chain.  Our ears would have you
put it later in the chain.  The optimal condition is to have filters in
both.  I originally got a 455KHz filter for my Kenwood HF rig.  I eventually
got the same filter for the higher IF (8.83MHz).  You *can* hear the
bandwidth difference if you listen for it, and if you do lots of work with
nearby signals, you'll start to notice that the higher IF filter does
prevent overload.
>
> One more thing...the "type" of filter also matters a GREAT deal.  Ceramic
filters are cheap (inexpensive) and have poorer shape factors.  Crystal
filters are more expensive, but have *much* better shape factors.  If
determining what filter to buy based on cost, defintely keep this in
mind...the ceramic filter *will* be cheaper, but at the cost of performance.
This may be one reason why the higher IF filter costs less and performs
worse than the lower IF filter.
>
> I would *HIGHLY* suggest that you get filters from InRad.  They are *VERY*
well known in the DX and Contesting world as the best filters you can get
for amateur radios.  Cost is about the same as the OEM filters, but much
better shape factors.  You can find InRad at
>
>   http://www.qth.com/inrad
>
> If you click on the links to the performance curves, you can get a general
idea of what filter shape factors look like.
>
> 73,
>
>   - Aaron Hsu, NN6O (ex-KD6DAE)
>     {nn6o}@arrl.net
>     {athsu}@unistudios.com
>     No-QRO Int'l #1,000,006
>     . -..- - .-. .-   ".... . .- ...- -.--"
>
>
> p.s.  I'm not a filter expert...just passing on what I've learned.
Comments and corrections are always welcome!
>