Elecraft

K3S DSP filter plots

Classic

List

Threaded

7 messages Options

charles-2

K3S DSP filter plots

Thanks for the reply Guy.

I understand your rationale for aligning IF and DSP filters for
additional out-of-passband rejection in crowded contest conditions.

My concern stems from using IF shift to move the DSP filter within the
roofing filter passband. The Elecraft rep I spoke to last week told me
that the 8 Mhz IF roofing filter and the DSP filters do not track. The
roofing filter is fixed and will present a fixed bandwidth in which the
DSP filter can operate. Therefore, there will be an interaction between
the two filters, the fixed roofing filter, and the DSP filter which can
be moved higher or lower with the IF Shift control. If the roofing
filter and the DSP filter have the same bandwidth, the interaction
between the filter skirts creates a high cut or low cut condition
depending upon the direction of the IF shift. If the IF can be shifted
enough, a stop-band condition occurs and all signals disappear.

I am primarily a "search and pounce" contester, but I can appreciate
your desire to have the sharpest combined filter skirts to keep the
adjacent run station from getting into your bandpass. However, my
limited experience with "run" contesting revealed that extremely sharp
filters are a double-edged sword because of the inability to hear
off-frequency callers. In fact, I think there is an option in the K3S
menu system that provides for higher shape factor DSP filters in order
to allow off-frequency caller to be heard. In the S&P contest mode, I
also want very sharp DSP filters, but I want them to exist within a
wider roofing filter bandwidth so that I can use the IF shift to
discriminate between signals in that passband. If my DSP filter and my
roofing filter are about the same bandwidth, I can't use the IF shift
effectively because its filter skirts will interact with the roofing
filter skirts and cause an undesirable high cut or low cut resultant
filter response, or worse, the dreaded stop-band condition.

Another consequence of cascading a roofing filter and a DSP filter of
the same bandwidth in a run contest mode is that the only recourse to
hear an off-frequency caller is to engage the RIT (since engaging IF
shift with result in the above stated undesirable resultant filter
response curves). When I am in "run" contest mode, especially in very
crowded conditions, using RIT changes the pitch of every station that is
audible, and I just don't like to lose my awareness of where each
station is. I much prefer to use the IF shift to "highlight" the caller
I want to contact without changing the pitch of every single station.
Therefore, I hope you can see why I would not want to have a roofing
filter that is too narrow for the DSP filter that I plan two use in
combination with IF shift. Since I like to use the 100 hz DSP filter
bandwidth, I would not chose the 200hz roofing filter as an option
because it would severely limit the amount of IF shift that I could use
with a 100hz DSP filter. If I were not a fan of IF shift, I would do
what you are doing, cascade the 200hz roofing filter and use the 100 or
200 hz DSP filter.

In summary, I am hoping that the DSP filters have very steep skirts to
maximize the IF shift that can be used before the DSP filter skirts
interact with the roofing filter skirts. If you have any data on the
narrow DSP filter plots, please let me know.

Charles K5UA

______________________________________________________________
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
Message delivered to [hidden email]

Don Wilhelm-4

Re: K3S DSP filter plots

Charles,

I believe most operators use the RIT control rather than IF Shift to
tune for off frequency stations.
The IF shift is normally used to reduce interference from an "offending
station". That along with the Width control can isolate a single
station with ease and with the 50 Hz minimum DSP bandwidth, operation in
a crowded band segment can actually be a pleasure - unless an extra
strong signal gets past the roofing filter passband and activates the
hardware AGC.

When you get your K3S, give it a try and I think you will find the use
of RIT to be more effective.

Of course, if you ordered the KRX3, you can always go to SPLIT and
eliminate the need for using RIT. Split will cover a much greater range
than RIT.
The KRX3 is not required to operate SPLIT, but if you want to
dynamically listen to your transmit frequency and the desired signal at
the same time, the KRX3 is the only way to do that. Without the KRX3,
you can use the REV button to temporarily listen to your transmit
frequency, but you will not hear both frequencies at the same time with
the REV button.

Yes, I realize this is a bit different than the discussion of filter
shapes and the interaction of the roofing filter with the DSP
filtering. If it is not relevant, then please ignore my comment.

73,
Don W3FPR

On 11/16/2015 12:35 PM, [hidden email] wrote:

>
>
> Thanks for the reply Guy.
>
> I understand your rationale for aligning IF and DSP filters for
> additional out-of-passband rejection in crowded contest conditions.
>
> My concern stems from using IF shift to move the DSP filter within the
> roofing filter passband. The Elecraft rep I spoke to last week told me
> that the 8 Mhz IF roofing filter and the DSP filters do not track. The
> roofing filter is fixed and will present a fixed bandwidth in which the
> DSP filter can operate. Therefore, there will be an interaction between
> the two filters, the fixed roofing filter, and the DSP filter which can
> be moved higher or lower with the IF Shift control. If the roofing
> filter and the DSP filter have the same bandwidth, the interaction
> between the filter skirts creates a high cut or low cut condition
> depending upon the direction of the IF shift. If the IF can be shifted
> enough, a stop-band condition occurs and all signals disappear.
>
>

Bob McGraw - K4TAX

Re: K3S DSP filter plots

Are we interested in the shape of the DSP filters inside the roofing
filters or are we interested in the performance of the receiver using
recovered audio?

I have done measurements using a broadband noise for the source and
measured spectrum of the recovered audio. Of course as the DSP BW is
changed the recovered audio changes accordingly. The definition at 6 dB
and 60 dB can be attained. I doubt the S meter method is more than an
approximation of values unless all values are related to S units @ 6 dB
per S unit or the value in dBv which can be obtained from the detector
via the Config Menu..

Perhaps Wayne is better qualified to define the performance of the DSP
filter system, usually limited by the number of poles and is done
mathematically. This of course would be somewhat different than the
recovered audio method. Yet, my preference is recovered audio in as
much as that's what we hear.

73
Bob, K4TAX
K3S s/n 10163

On 11/16/2015 6:00 PM, Don Wilhelm wrote:

> Charles,
>
> I believe most operators use the RIT control rather than IF Shift to
> tune for off frequency stations.
> The IF shift is normally used to reduce interference from an
> "offending station". That along with the Width control can isolate a
> single station with ease and with the 50 Hz minimum DSP bandwidth,
> operation in a crowded band segment can actually be a pleasure -
> unless an extra strong signal gets past the roofing filter passband
> and activates the hardware AGC.
>
> When you get your K3S, give it a try and I think you will find the use
> of RIT to be more effective.
>
> Of course, if you ordered the KRX3, you can always go to SPLIT and
> eliminate the need for using RIT. Split will cover a much greater
> range than RIT.
> The KRX3 is not required to operate SPLIT, but if you want to
> dynamically listen to your transmit frequency and the desired signal
> at the same time, the KRX3 is the only way to do that. Without the
> KRX3, you can use the REV button to temporarily listen to your
> transmit frequency, but you will not hear both frequencies at the same
> time with the REV button.
>
> Yes, I realize this is a bit different than the discussion of filter
> shapes and the interaction of the roofing filter with the DSP
> filtering. If it is not relevant, then please ignore my comment.
>
> 73,
> Don W3FPR

k6dgw

Re: K3S DSP filter plots

No doubt Wayne is qualified, however he has deferred on several
occasions to Lyle who I believe actually writes the code.

And a teeny point from a math major ... inasmuch as arithmetic is a part
of math, it is clearly "done mathematically." The "real" math however
is in figuring out what arithmetic to do. At the implementation level,
it's all arithmetic, generally in base-2.

I'll bet my next paycheck, if I got one anymore, that both Lyle and
Wayne understand both.

73,

Fred K6DGW
- Northern California Contest Club
- CU in the Cal QSO Party 1-2 Oct 2016
- www.cqp.org

On 11/16/2015 4:18 PM, Bob McGraw - K4TAX wrote:

> Perhaps Wayne is better qualified to define the performance of the DSP
> filter system, usually limited by the number of poles and is done
> mathematically.
______________________________________________________________
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
Message delivered to [hidden email]

Elecraft mailing list

Re: K3S DSP filter plots

And I will add a filter is not a filter. No, I am not crazy. There are many types of filters with various trade-offs. In band ripple.....loss......phase shift.......skirt shape....are just some of the trade offs. DSP software can emulate any or combos

George, W6GF

Sent from my iPad

> On Nov 16, 2015, at 5:03 PM, Fred Jensen <[hidden email]> wrote:
>
> No doubt Wayne is qualified, however he has deferred on several occasions to Lyle who I believe actually writes the code.
>
> And a teeny point from a math major ... inasmuch as arithmetic is a part of math, it is clearly "done mathematically." The "real" math however is in figuring out what arithmetic to do. At the implementation level, it's all arithmetic, generally in base-2.
>
> I'll bet my next paycheck, if I got one anymore, that both Lyle and Wayne understand both.
>
> 73,
>
> Fred K6DGW
> - Northern California Contest Club
> - CU in the Cal QSO Party 1-2 Oct 2016
> - www.cqp.org
>
>> On 11/16/2015 4:18 PM, Bob McGraw - K4TAX wrote:
>>
>> Perhaps Wayne is better qualified to define the performance of the DSP
>> filter system, usually limited by the number of poles and is done
>> mathematically.
> ______________________________________________________________
> 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
> Message delivered to [hidden email]

Jeff Blaine

Re: K3S DSP filter plots

It's easy to measure this. Here's one way to do it - assuming you already
have a functioning AFSK/PSK interface working with a rig/PC.

Download and install the WD6CNF Audio Spectrum Analzyer. Set it up to
listen to your rig interface AF feed.

On the rig, turn of the AGC and the rig's preamps. Turn of NR, DNR -
whatever the noise reduction and noise blanking stuff is called.

Set the mode to SSB or CW (I forget if there are min DSP BW restrictions in
SSB). Set the roof selection manually to something wide relative to the DSP
(so you don't comingle the two transition bands). The 2.7 Khz filter

Feed a stable S8 (-80 dBm) into the antenna and tune it so you have roughly
a 700 hz signal.

73/jeff/ac0c
www.ac0c.com
alpha-charlie-zero-charlie

-----Original Message-----
From: George Fritkin via Elecraft
Sent: Monday, November 16, 2015 7:49 PM
To: [hidden email]
Cc: [hidden email]
Subject: Re: [Elecraft] K3S DSP filter plots

And I will add a filter is not a filter. No, I am not crazy. There are
many types of filters with various trade-offs. In band
ripple.....loss......phase shift.......skirt shape....are just some of the
trade offs. DSP software can emulate any or combos

George, W6GF

Sent from my iPad

> On Nov 16, 2015, at 5:03 PM, Fred Jensen <[hidden email]> wrote:
>
> No doubt Wayne is qualified, however he has deferred on several occasions
> to Lyle who I believe actually writes the code.
>
> And a teeny point from a math major ... inasmuch as arithmetic is a part
> of math, it is clearly "done mathematically." The "real" math however is
> in figuring out what arithmetic to do. At the implementation level, it's
> all arithmetic, generally in base-2.
>
> I'll bet my next paycheck, if I got one anymore, that both Lyle and Wayne
> understand both.
>
> 73,
>
> Fred K6DGW
> - Northern California Contest Club
> - CU in the Cal QSO Party 1-2 Oct 2016
> - www.cqp.org
>
>> On 11/16/2015 4:18 PM, Bob McGraw - K4TAX wrote:
>>
>> Perhaps Wayne is better qualified to define the performance of the DSP
>> filter system, usually limited by the number of poles and is done
>> mathematically.
> ______________________________________________________________
> 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
> Message delivered to [hidden email]

______________________________________________________________
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
Message delivered to [hidden email]

______________________________________________________________
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
Message delivered to [hidden email]

Jeff Blaine

Re: K3S DSP filter plots

In reply to this post by charles-2

Oops. Hit the go button too fast.

73/jeff/ac0c
www.ac0c.com
alpha-charlie-zero-charlie

-----Original Message-----
From: Jeff AC0C
Sent: Monday, November 16, 2015 8:25 PM
To: George Fritkin ; [hidden email]
Cc: [hidden email]
Subject: Re: [Elecraft] K3S DSP filter plots

It's easy to measure this. Here's one way to do it - assuming you already
have a functioning AFSK/PSK interface working with a rig/PC.

Download and install the WD6CNF Audio Spectrum Analzyer. Set it up to
listen to your rig interface AF feed.

On the rig, turn of the AGC and the rig's preamps. Turn of NR, DNR -
whatever the noise reduction and noise blanking stuff is called.

Set the mode to SSB or CW (I forget if there are min DSP BW restrictions in
SSB). Set the roof selection manually to something wide relative to the DSP
(so you don't comingle the two transition bands). The 2.7 Khz filter

Feed a stable S8 (-80 dBm) into the antenna and tune it so you have roughly
a 700 hz signal.

Fiddle around with the AF spectrum analyzer to show a clear picture of the
bandwidth and the top 40-60 db of the signal.

Turn on the PEAK HOLD setting of the AF spectrum anal.

Set the average setting to "2"

Swing the VFO back and forth to pain the passband on the peak hold display.
Depending on how you have the AF SA setup, you may need to crawl along.

Voila, there's your passband. Hard data and no MSEE required. Have fun!

73/jeff/ac0c
www.ac0c.com
alpha-charlie-zero-charlie

-----Original Message-----
From: George Fritkin via Elecraft
Sent: Monday, November 16, 2015 7:49 PM
To: [hidden email]
Cc: [hidden email]
Subject: Re: [Elecraft] K3S DSP filter plots

And I will add a filter is not a filter. No, I am not crazy. There are
many types of filters with various trade-offs. In band
ripple.....loss......phase shift.......skirt shape....are just some of the
trade offs. DSP software can emulate any or combos

George, W6GF

Sent from my iPad