[Elecraft] Price Comparison and Comment

[Don Wilhelm]

Bill,

All of what you say is true *EXCEPT* for one item - and that is the AGC.

The hardware AGC is developed before any DSP processing, so when there 
are strong signals inside the roofing filter passband, they will 
activate the AGC and reduce the receiver sensitivity.  If you have the 
DSP cranked down to a narrow bandwidth, you may not even hear the 
offending strong signal because it is outside the DSP bandwidth but 
still inside the roofing filter bandwidth.

BTW, this can happen on the K2 (and many other receivers) if a wide 
bandwidth is selected and the DSP is used to reduce the signal to only 
the desired one.

That is not really tough to deal with if you are already listening to an 
S-9 signal even though an S-9 +30 signal comes on - the S-9 signal will 
be reduced to the equivalent of an S-4 or S-5 signal and that can easily 
be solid copy (because other signals in the DSP passband are similarly 
reduced).

Now, if the signal you are trying to copy is an S-1 or S-2 level and an 
S-9 signal comes into the roofing filter passband (again it may be out 
of the DSP passband and will not be heard), the receiver gain will be 
reduced and you will no longer hear the S-1/S-2 signal - just like QSB, 
but it is QSB induced in the receiver, not due to propagation effects.

I will have at least one narrow filter available in my K3.  Bill, you 
were correct that a filter is needed for each major mode, but the 
filters that any one person needs will be determined by their operating 
habits, and not based on the 'average'.  For ragchewing and  casual 
contacts, even the wide 2.7 kHz filter may be all that is needed, but 
for serious CW, QRP, contesting or DX chasing, an array of filters will 
be needed - just how many will depend on the operator's desires to 
reduce QRM and the amount available in the ham radio budget.

73,
Don W3FPR

Bill Tippett wrote:
> W5EWA:
>  >but now someone says that we DON'T need them all.
> 
>         Definitely not...unless we have more money than
> sense.  You need only ONE per major mode, which will
> cover the widest bandwidth you ever expect to use in
> that mode.  For example:
> 
> CW - 500 or 400.  250 and 200 are definite overkill
> IMHO and not necessary in addition to narrow DSP BWs.
> The reason is that IMD/BDR spurious issues at <500 Hz
> will be far overridden by other issues such as phase noise,
> key clicks, etc that come from the transmitted signals.
> There is no point in making a receiver many dB better
> than the environment in which it must live!  No matter
> how good a receiver is it cannot eliminate transmitter
> defects (maybe Flex's SDR-X can but that remains to be
> proven).
> 
> SSB - 2.7 kHz unless you want ESSB (not sure how high
> that goes these days but IMHO we should be trying to
> minimize communications bandwidths rather than playing
> broadcast disc jockey (my personal opinion!).  Those
> folks should go to 29 MHz FM if they want to do that.
> 
> AM - 6 kHz
> 
> FM - 15 kHz
> 
>         In fact you may not need ANY filters beyond
> the 2.7 kHz stock filter unless you expect to have lots
> of very close spaced S9+30 dB signals on CW (e.g. huge
> low-band pileups or in contests).  The 2.7k should be
> perfectly adequate for nearly any situation on SSB,
> although it wouldn't surprise me to see someone offer
> 1.8 kHz, mainly for contesters.
> 
>         A roofing filter's purpose is simply to minimize
> spurious artifacts from being created *inside* the RX
> by external signals.  It does NOT create the ultimate
> selectivity (which is determined by the DSP filter at
> the 2nd IF).  Please read the last sentence again!
> Here's a simple summary:
> 
> 
> Antenna > roofing filter > DSP filter > ear.
> 
> ...roofing filter prevents very strong signals from
> creating spurious (i.e. *NOT REAL*) signals which
> will then appear inside the DSP.
> 
> ...DSP filters provide the ultimate selectivity (i.e.
> separating the *REAL* signals from each other).
> 
> 
>         The number of poles (i.e. shape factor of the
> filter) can affect IMD/BDR issues inside the RX.  For
> this reason hopefully Elecraft will provide actual
> IMD/BDR measurements with each option to help answer
> the following questions:
> 
> 1.  Is the optional 8-pole 2.8 kHz better than the
> stock 5-pole 2.7 kHz (by "better" I mean resulting
> in better IMD/BDR performance which is the primary
> purpose of a roofing filter).
> 
> 2.  Is the 5-pole 500 Hz better than the 8-pole 400 Hz?
> 
> 3.  Is the 8-pole 250 Hz better than the 5-pole 200 Hz?
> (academic to me since I don't feel either is necessary
> ...see comment above re TX key clicks, phase noise, etc).
> 
>         The variable BW filters are interesting but
> I would not consider them until we have some actual
> results for IMD/BDR.  They are NOT providing ultimate
> selectivity as they do in the case of the K2, which
> seems to be confusing many folks.
> 
>                                 73,  Bill  W4ZV