K-3 Roofing filter comment

[Richard Fjeld]

FYI,
I don't want to open up a 'can of worms', and Rob Sherwood could better speak to this, but this is an observation I made using a P-3.

Not long ago, there was quite a string of threads about the 5 pole 2.7 Khz roofing filter versus the 8 pole 2.8 Khz filter. Before I ordered my K-3, I had studied the filter skirts of each. I don't remember how I found the plot for the 5 pole filter, but I remember that the span of each was different. At the same span width, the 2.8 Khz filter skirts seemed to me to be considerably straighter when compared to the 2.7 Khz filter, so I ordered the 2.8 filter.

The other day, I was on a qso with a group and while watching the SSB signal on the P-3, I noticed a carrier much taller appear right on the edge of the filter bandwidth as shown by the brackets at the bottom of the P-3. The carrier was on the far end of the filter, or 2.8 Khz away from the center frequency.

I could not hear it, but expected to, so I tuned over slightly to get it within the filter bandwidth and heard it strongly. I seem to recall that a filter's width is rated at something like 6 dB down, so I thought I would have heard it within the skirts at their wider limits.  I still wonder about that, but I'm very pleased.

(Just a comment; I'd hate to do without the P-3)


Richard Fjeld, n0ce
[hidden email]
I'd rather be learning.


______________________________________________________________
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

[Don Wilhelm-4]

  Richard,

You were not hearing the effect of the 2.8 MHz roofing filter, but
instead the DSP filter.
In order to hear the effect of the 2.8 kHz filter skirt, you would have
to widen out the DSP bandwidth to max or greater.  The DSP forms the
ultimate filter, and does have nearly vertical slope.

The purpose of the roofing filter is to keep adjacent strong signals
from activating the hardware AGC and causing the resulting AGC pumping.  
It is the DSP that sets the actual filter width.

73,
Don W3FPR

On 7/12/2011 2:22 PM, Richard Fjeld wrote:

> FYI,
> I don't want to open up a 'can of worms', and Rob Sherwood could better speak to this, but this is an observation I made using a P-3.
>
> Not long ago, there was quite a string of threads about the 5 pole 2.7 Khz roofing filter versus the 8 pole 2.8 Khz filter. Before I ordered my K-3, I had studied the filter skirts of each. I don't remember how I found the plot for the 5 pole filter, but I remember that the span of each was different. At the same span width, the 2.8 Khz filter skirts seemed to me to be considerably straighter when compared to the 2.7 Khz filter, so I ordered the 2.8 filter.
>
> The other day, I was on a qso with a group and while watching the SSB signal on the P-3, I noticed a carrier much taller appear right on the edge of the filter bandwidth as shown by the brackets at the bottom of the P-3. The carrier was on the far end of the filter, or 2.8 Khz away from the center frequency.
>
> I could not hear it, but expected to, so I tuned over slightly to get it within the filter bandwidth and heard it strongly. I seem to recall that a filter's width is rated at something like 6 dB down, so I thought I would have heard it within the skirts at their wider limits.  I still wonder about that, but I'm very pleased.
>
> (Just a comment; I'd hate to do without the P-3)
>
>
> Richard Fjeld, n0ce
>

______________________________________________________________
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

[Dave Hachadorian]

> The DSP forms the
>ultimate filter, and does have nearly vertical slope.

I think one of the Elecraft folks stated on the reflector that
the -60dB BW of the DSP is (-6dB BW + 300Hz).

So at DSP BW = 2800 Hz,
BW60 = 3100 Hz.
Equivalent shape factor would be 3100/2800 = 1.1
That's pretty doggone steep.
Any reasonable amplitude signal (not triggering hardware AGC)
over 1550 Hz off center would be down 60 dB.

I think I've got that straight. Comments welcome.

Dave Hachadorian, K6LL
Big Bear Lake, CA











































.

______________________________________________________________
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

[Geoffrey Mackenzie-Kennedy]

Dave,

The sampling point for the K3's hardware AGC is before the DSP in the signal
path, so unfortunately the DSP filter does not protect the HAGC. The roofing
filters protect the HAGC.

There is another design approach to eliminating AGC pumping problems, but
that is another story :-)

73,
Geoff
GM4ESD


On July 12, 2011 at  21:21 Z,  Dave Hachadorian wrote:

> I think one of the Elecraft folks stated on the reflector that
> the -60dB BW of the DSP is (-6dB BW + 300Hz).
>
> So at DSP BW = 2800 Hz,
> BW60 = 3100 Hz.
> Equivalent shape factor would be 3100/2800 = 1.1
> That's pretty doggone steep.
> Any reasonable amplitude signal (not triggering hardware AGC)
> over 1550 Hz off center would be down 60 dB.
>
> I think I've got that straight. Comments welcome.
>
> Dave Hachadorian, K6LL
> Big Bear Lake, CA

______________________________________________________________
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

[wa9fvp]

Actually the AGC is generated by the DSP in software.  The signal VIFGAIN1 coming from a D to A converter (AD5302A) if fed back to the IF amplifier to control the receiver's gain.  After the DSP filter reduces the bandwidth, the AGC is adjusted by the DSP.  It's all done using mathematical calculations.

You can have a situation where the DSP filter is narrower then the roofing filter but if the DSP engineer does his job, the unwanted signal will be outside the DSP filter bandwidth.  The third order products will be at a minimum and the AGC will not be affected.

[Bill W4ZV]

wa9fvp wrote

Actually the AGC is generated by the DSP in software.  The signal VIFGAIN1 coming from a D to A converter (AD5302A) if fed back to the IF amplifier to control the receiver's gain.  After the DSP filter reduces the bandwidth, the AGC is adjusted by the DSP.  It's all done using mathematical calculations.

Not quite.  There are two separate AGC functions (analog and digital) in the K3.  You're describing only the second stage which is digital AGC.  It follows analog AGC which is after the roofing filter and before the DSP.  Most modern hybrid heterodyne/DSP radios use this approach (e.g. IC-7xxx, FT-5000/9000, Orion, Omni 7, Eagle TS-590S, etc) because DSPs currently only have ~100 dB of dynamic range and 140 dB is required for good BDR performance.  See Wayne's detailed explanation below.

73,  Bill

http://www.elecraft.com/K3/Roofing_Filters.htm

"In order to achieve the best K3 blocking dynamic range (desense) in the 140 dB+ range, you -must- use a narrow crystal filter (400 Hz 8-pole or 200 Hz 5-pole for closer interfering signal spacing) in front of the DSP. We use hardware AGC after the narrow crystal filter and ahead of the DSP to protect the DSP when signals inside the crystal filter exceed a 100 dB dynamic range. If you only use the 2.7 kHz 5 pole stock filter for CW or data operation you will be significantly desensed once signals within that filter's bandwidth exceed about S9+25. This is before phase noise from the transmitting station becomes a factor. Not uncommon on 40M at night, during a contest or at a multi-op station -- Or every day in major cities.  As an example, changing to a 400 Hz or 200 Hz filter reduces blocking from signals 1-5 kHz away.  I've personally confirmed this on the air with my K3 and the other commercial rigs we have here. When I've operated with the K3, or another DSP rig, on CW without using a narrow filter ahead of the DSP filtering, I frequently experienced desense (BDR) from nearby signals. Putting in the narrower crystal filter immediately cleaned it up. (While both the 400 Hz and 500 Hz filters are excellent CW filters, the 400 Hz 8-pole filter performs slightly better than the 500 Hz 5 pole filter due to its narrower shape factor. The 200 Hz 5 pole is even sharper.)"

[Geoffrey Mackenzie-Kennedy]

Jack,

As Bill W4ZV confirmed one of the AGC loops uses a sampling point which is
after the roofing filters but before the DSP and its narrow filter(s).

If the unwanted signals are greatly attenuated by the DSP's filter, but they
are either within the roofing filter's passband or in the roofing filter's
skirt/ near stopband regions, the level of their third order products could
still be of concern.  Depending on the frequencies of the fundamentals some
products could get through the DSP's filter.  Their level will of course be
determined by their spacing, and the design of the front end and its gain,
the design of the roofing filters and the quality of the crystals used in
the filters, to name just some of the areas where "Pre DSP IMD" is
generated.

73,

Geoff
GM4ESD


On July 15, 2011 at 05:43 Z, Jack WA9FVP wrote:

> Actually the AGC is generated by the DSP in software.  The signal VIFGAIN1
> coming from a D to A converter (AD5302A) if fed back to the IF amplifier
> to
> control the receiver's gain.  After the DSP filter reduces the bandwidth,
> the AGC is adjusted by the DSP.  It's all done using mathematical
> calculations.
>
> You can have a situation where the DSP filter is narrower then the roofing
> filter but if the DSP engineer does his job, the unwanted signal will be
> outside the DSP filter bandwidth.  The third order products will be at a
> minimum and the AGC will not be affected.
>
> -----
> Jack WA9FVP
> Willco Electronics

______________________________________________________________
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

[wa9fvp]

I understand that about the HAGC1.  Its purpose is to keep the level within the range of the A to D.  If this signal is too big the A to D converter U23 will over range and the DSP is unable to perform the mathematical functions.  

The HAGC1 is fixed and is not adjustable by the user.  VIFGAIN1 on the other hand is adjustable and ultimately controlled by the DSP.  The AGC loops are nested that is the HAGC1 is the main loop and the VIFGAIN1 is a sub loop.  If you look at the IF schematic you will see that the two loops are summed at the same AGC buffer amplifier U3A.   HAGC1 is routed to the DSP board and is actually an input to an A to D converter U8.  It’s then processed by the DSP.  Without knowing the DSP code, I suspect there’s a PID (Proportional Integral Derivative) function involved to maintain loop stability and data to calibrate the S meter.

The RF gain control is a potentiometer connected an analog port of the PIC controller U1 (PIC18LF8722).  It’s converted to a digital signal that’s also fed to the DSP to control the RF gain via VIFGAIN1.  

I guess where we differ is that you speak in terms of an analog signal and I like to speak in terms in the digital world.  I was involved in a DSP project in the late 80’s and early 90’s called “The Hamblaster a DSP for Ham Radio”, where I wrote two articles in QEX.