The Myth of Lying to Your K3

[Ed Muns, W0YK]

Setting the crystal filter engagement bandwidth is sometimes described as
"lying to your K3" when the number is different than the number on the
filter.  While this phrasing is cute and descriptive, it can be very
misleading.  There is no lie, but in fact a very precise setting that
determines the DSP frequency at which the crystal filter engages.  If
anything, the user is lying to themselves.  Some facts:

1.  The bandwidth in the filter's marketing name is not the actual bandwidth
of the filter.  Some filters are significantly different than their name.
For example, the "KFL3A-250  250 Hz, narrow 8-pole  filter" is nominally
370Hz wide at the -6dB points.  The "KFL3A-400  400 Hz, narrow 8-pole
filter" is 435Hz.

2.  The actual IF bandwidth of the K3 is determined by the cascade effect of
the DSP and crystal filter bandwidths.  When those bandwidths are close to
one another, the actual IF bandwidth is significantly narrower.  So, if you
configure your K3 to engage the 250Hz filter (actually 370Hz) at 350Hz or
400Hz, the resultant bandwidth will be closer to 300Hz.

3.  If you engage a crystal filter at a DSP bandwidth much greater than its
actual bandwidth, then the K3 IF bandwidth will decrease immediately to the
crystal filter bandwidth and the passband shape will be governed by the
"rounder" shape of the crystal filter.  Some users do this because they
"like the sound" of the crystal filter passband shape compared to the
sharper DSP passband.  Keep in mind, though, that the bandwidth step at the
crystal filter engagement point is much larger than the normal DSP steps.
For example, if you have the 1.8kHz filter engage at 2.2kHz, you give up any
IF bandwidth between 2.3kHz and 1.8kHz.  The K3 WIDTH control shows the DSP
bandwidth at values between 2.3-1.8kHz, but the actual K3 IF bandwidth is
governed by the narrower 1.8kHz crystal filter and fixed at that bandwidth
for these intermediate DSP bandwidths.  Failure to recognize this fact is
lying to yourself.

4.  The crystal filter engagement point can alternatively be set at a DSP
bandwidth much less than the actual crystal filter.  In this case, the DSP
bandwidth steps are consistent and represent the true IF bandwidth of the K3
at any time.  The crystal filter is always wider than the DSP bandwidth and
therefore does not cause the IF bandwidth to be narrower due to cascading.
In this case, the K3 IF passband shape is governed by the sharper DSP shape
and the crystal filter is only providing roofing filter protection for the
DSP, which happens to be the primary function of the crystal filter.

Any value you choose for a crystal filter to engage is valid.  There is no
right or wrong.  There is no lie.  The engagement point you choose is a
simple fact.  The important thing is to fully understand the implication of
your choice so that you accomplish what you intend to accomplish.

Ed - W0YK
-----------------------------------------------
Ed Muns
Muns Vineyard - www.munsvineyard.com
FaceBook - www.facebook.com/munsvineyard

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Guy, K2AV]

With you all the way until you said this:

"The crystal filter is always wider than the DSP bandwidth and
therefore does not cause the IF bandwidth to be narrower due to cascading."

To construe the sentence correctly requires the assumption that ONLY
the shape INSIDE the rated 3 dB or 6 dB passband matters. It is true
that this "flat" part of the curve is governed most by the
characteristics of the narrower curve. HOWEVER, everything cascades.
PARTICULARLY so on the skirts.

When one is trying to listen to a signal at S1 just up a ways from a
signal that is S9+30, it is the skirts that count the most. The losses
on the two filtering curves in dB at a given frequency are ADDING
together to create the 98+ dB needed to make the S1 signal comfortably
louder than the unwanted signal.  Note that merely down 78 db on the
skirts makes them EQUAL. Another 20 dB is needed for the "comfortable"
part.  Neither the DSP nor the roofer will accomplish this on their
own. It is the sum of the skirt losses.

There is a point where the skirts on a roofing filter are the
steepest. If the steepest part of the DSP curve is set to coincide,
this will result in the most sudden and abrupt filtering out of the
unwanted signal as one tunes across and will result in the least
interfered open bandwidth around a run frequency.

That is why I "lie" about the 8 pole 400 and 250 and call them 450 and
350. I WANT those super-sharp COMPOSITE skirts you get whenever the
steep parts coincide.

73, Guy.

On Sat, Mar 13, 2010 at 10:15 AM, Ed Muns <[hidden email]> wrote:

> Setting the crystal filter engagement bandwidth is sometimes described as
> "lying to your K3" when the number is different than the number on the
> filter.  While this phrasing is cute and descriptive, it can be very
> misleading.  There is no lie, but in fact a very precise setting that
> determines the DSP frequency at which the crystal filter engages.  If
> anything, the user is lying to themselves.  Some facts:
>
> 1.  The bandwidth in the filter's marketing name is not the actual bandwidth
> of the filter.  Some filters are significantly different than their name.
> For example, the "KFL3A-250  250 Hz, narrow 8-pole  filter" is nominally
> 370Hz wide at the -6dB points.  The "KFL3A-400  400 Hz, narrow 8-pole
> filter" is 435Hz.
>
> 2.  The actual IF bandwidth of the K3 is determined by the cascade effect of
> the DSP and crystal filter bandwidths.  When those bandwidths are close to
> one another, the actual IF bandwidth is significantly narrower.  So, if you
> configure your K3 to engage the 250Hz filter (actually 370Hz) at 350Hz or
> 400Hz, the resultant bandwidth will be closer to 300Hz.
>
> 3.  If you engage a crystal filter at a DSP bandwidth much greater than its
> actual bandwidth, then the K3 IF bandwidth will decrease immediately to the
> crystal filter bandwidth and the passband shape will be governed by the
> "rounder" shape of the crystal filter.  Some users do this because they
> "like the sound" of the crystal filter passband shape compared to the
> sharper DSP passband.  Keep in mind, though, that the bandwidth step at the
> crystal filter engagement point is much larger than the normal DSP steps.
> For example, if you have the 1.8kHz filter engage at 2.2kHz, you give up any
> IF bandwidth between 2.3kHz and 1.8kHz.  The K3 WIDTH control shows the DSP
> bandwidth at values between 2.3-1.8kHz, but the actual K3 IF bandwidth is
> governed by the narrower 1.8kHz crystal filter and fixed at that bandwidth
> for these intermediate DSP bandwidths.  Failure to recognize this fact is
> lying to yourself.
>
> 4.  The crystal filter engagement point can alternatively be set at a DSP
> bandwidth much less than the actual crystal filter.  In this case, the DSP
> bandwidth steps are consistent and represent the true IF bandwidth of the K3
> at any time.  The crystal filter is always wider than the DSP bandwidth and
> therefore does not cause the IF bandwidth to be narrower due to cascading.
> In this case, the K3 IF passband shape is governed by the sharper DSP shape
> and the crystal filter is only providing roofing filter protection for the
> DSP, which happens to be the primary function of the crystal filter.
>
> Any value you choose for a crystal filter to engage is valid.  There is no
> right or wrong.  There is no lie.  The engagement point you choose is a
> simple fact.  The important thing is to fully understand the implication of
> your choice so that you accomplish what you intend to accomplish.
>
> Ed - W0YK
> -----------------------------------------------
> Ed Muns
> Muns Vineyard - www.munsvineyard.com
> FaceBook - www.facebook.com/munsvineyard
>
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:[hidden email]
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
>

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[NZ0T]

Umm, do you guys ever just have fun operating?
Just wondering....................

[Dunc Carter - W5DC]

NZ0T wrote:
> Umm, do you guys ever just have fun operating?
> Just wondering....................
>  
YES!
______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Ed Muns, W0YK]

K2AV noted:

> "The crystal filter is always wider than the DSP bandwidth
> and therefore does not cause the IF bandwidth to be narrower
> due to cascading."
>
> To construe the sentence correctly requires the assumption
> that ONLY the shape INSIDE the rated 3 dB or 6 dB passband
> matters. It is true that this "flat" part of the curve is
> governed most by the characteristics of the narrower curve.
> HOWEVER, everything cascades.
> PARTICULARLY so on the skirts.

That's right.  I chose to simplify the point by not getting into the skirt
issues.  It's clear from the posts on this reflector that the basics of this
topic are not well understood, let alone the subtleties of the cascaded
skirts.

Ed - W0YK
-----------------------------------------------
Ed Muns
Muns Vineyard - www.munsvineyard.com
FaceBook - www.facebook.com/munsvineyard

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Guy, K2AV]

I'm not sure the flat passband is the better pick if you have to pick
one. One narrows the passband to NOT hear things, or no one would ever
buy anything but the wide body roofer.  The FT1000MP would do.

The rejection of signals before the analog to digital converter, the
vertical range of the converter, and defensive hardware AGC in front
of the converter are central concepts of the radio.  Might as well
take them head on.

73, Guy.

On Sun, Mar 14, 2010 at 12:34 AM, Ed Muns <[hidden email]> wrote:

> K2AV noted:
>> "The crystal filter is always wider than the DSP bandwidth
>> and therefore does not cause the IF bandwidth to be narrower
>> due to cascading."
>>
>> To construe the sentence correctly requires the assumption
>> that ONLY the shape INSIDE the rated 3 dB or 6 dB passband
>> matters. It is true that this "flat" part of the curve is
>> governed most by the characteristics of the narrower curve.
>> HOWEVER, everything cascades.
>> PARTICULARLY so on the skirts.
>
> That's right.  I chose to simplify the point by not getting into the skirt
> issues.  It's clear from the posts on this reflector that the basics of this
> topic are not well understood, let alone the subtleties of the cascaded
> skirts.
>
> Ed - W0YK
> -----------------------------------------------
> Ed Muns
> Muns Vineyard - www.munsvineyard.com
> FaceBook - www.facebook.com/munsvineyard
>
>

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Ed Muns, W0YK]

> I'm not sure the flat passband is the better pick if you have
> to pick one. One narrows the passband to NOT hear things, or
> no one would ever buy anything but the wide body roofer.  The
> FT1000MP would do.
>
> The rejection of signals before the analog to digital
> converter, the vertical range of the converter, and defensive
> hardware AGC in front of the converter are central concepts
> of the radio.  Might as well take them head on.

Yes, as I said to your first comment, I agree.  My intent was to get folks
to first base before we go to second base.  Choosing the crystal filter
engagement point as well as setting the DSP bandwidth affects several
things, not just the skirt issues that you are focused on.  There is not a
single best answer for every situation, so it is useful to understand each
of the impacts affected by these filter choices.

Ed - W0YK
-----------------------------------------------
Ed Muns
Muns Vineyard - www.munsvineyard.com
FaceBook - www.facebook.com/munsvineyard

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html