Roofing Filter Overkill

[Bill W4ZV]

N2DTS wrote:

 >The K3 2.7 KHz filter is just the first filter, unless
you have real strong local signals close, you likely wont have
any problems.

         This is very good advice.  Some may forget that
the DSP filters in the 2nd IF are the primary means for
achieving selectivity.  The 1st IF roofing filter is simply
preventing issues caused by very strong signals from reaching
the DSP stage.

 >Now if you often find yourself working a signal and having
a real strong signal very close, you might think about
something narrower...

         I would suggest not adding filters unless you
ACTUALLY experience problems.  Since many of us tend
to be closet engineers, we tend to overkill potential
problems that will most likely never exist.  If you
actually do experience a problem, ordering a filter
is easy and you'll have it in 2-3 days via Priority Mail.

         I'll repeat why a 200-250 Hz filter is most likely
a good example of overkill in our real world of transmitter
phase noise, key clicks, etc.  Let's assume you have a
S9+40 dB signal spaced 200 Hz from your operating frequency
that is causing desensing, either BDR or analog AGC
activation (i.e. pumping).


         First let's try Inrad's 400 Hz 8-pole filter
below (identical for K3):

http://www.qth.com/inrad/graphs/701.gif

The graph is not quite centered but the 6 dB bandwidth
is 435 Hz (i.e. +/- 218 Hz from center) so the rejection
at 200 Hz from the center is just under 6 dB.  We now
have a S9+34 dB signal at 200 Hz away from us.  Possibly
some help if it prevents the signal from activating
analog AGC (assuming the DSP ADC limit is around S9+35).


         Next let's try a 250 Hz 4-pole filter like
Inrad's for Orion II (we don't yet have a plot of
K3's 200 Hz 5-pole but I expect it may not differ
significantly in shape factor from Inrad's below).

http://www.qth.com/inrad/graphs/353.gif

Because the 6 dB bandwidth is actually 562 Hz, it will
actually have *less* rejection at 200 Hz than the 8-pole
400 Hz above.


         OK let's try another.  Inrad's 250 Hz 8-pole
(again identical to that in the K3).

http://www.qth.com/inrad/graphs/708.gif

The 6 dB bandwidth is actually 370 Hz, so it has the
best rejection at 200 Hz spacings (but only marginally
better than the 8-pole 400 Hz above which has a 6 dB
bandwidth only 65 Hz wider (435-370).


         In the real world, will ANY of these filters
really allow us to operate 200 Hz from an S9+40 dB
signal?  The simple answer is NO.  Why?  Because the
transmitted phase noise (for only a steady state
carrier) far overrides desense issues.  Furthermore,
when the carrier is keyed, we must then add the
effects of key clicks.  We actually are pressing
our luck if we try to operate 500 Hz from such a
strong signal, and I doubt very few on this list
have ever done that unless they are very serious
contesters.

         Bottom line...we can add the world's best filter
but it will NOT solve the fundamental problem of
transmitter phase noise and key clicks, which become
the limiting factors at very close spacings.  Today
Orion has the best close-in phase noise performance
and CW waveform shaping of any commercial rig, but its
sidebands are down only about 70 dB at 1 kHz spacings.
Move closer and it is more like 30 dB down at 200
Hz spacings.  What good does it do if we could achieve
IMD of 100 dB and BDR of 140 dB at 200 Hz spacings when
the interfering transmitter signal itself overrides
the receiver's performance by at least 60-70 dB?
That's pure overkill and a waste of money IMHO.

         200-250 Hz roofing filters might make us feel
good but, in the real world of transmitted phase noise
and key clicks, they will not allow us to operate
significantly closer than we could with 400-500 Hz
filters.  And, as we saw above with the 4-pole "250"
Hz, we may want to see all filter plots before we
decide which is actually better.

         I ordered my K3 without any optional filters,
and I may even swap out the 2.7 kHz 5-pole once I see
complete IMD/BDR performance data and the actual
filter plots.

                                 73,  Bill  W4ZV






_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com