Re: K3 NR (what is the equiv to F3-1 from 3.25?)

[Robin Diane Goldstein]

hi...  i'm one of the folks who is less happy with the 3.27
implementation of NR than the 3.25.  i've been trying to do a
'scientific' comparison of the two, so i can record the audio and
upload for others to hear/analyze (non trivial since band conditions
and QSO's don't seem to last between loading of the FW)

in any event, i'm listening right now to an LSB QSO on 7.182.50.
receive EQ is totally flat, noise level here is about an S5.  using
3.25, if i set the NR to F3-1, the noise drops down by about 50-75%
and the audio peaks up (sharper/more intelligable) about 25%

i then load 3.27 but can't find any settings that are similar...
even F1-1 gives me a distinct ringing/echo (echo is not quite
right... its more like the old spring reverb's of my youth!)...  i
think that F1-3 should at least match the taps of 3.25, but can i
completely replicate the filter response of 3.25's F3-1?

btw, i know there's a fair bit of multi-variable confusion and "well,
it works fine here", which is to be expected in beta testing... but
given that (even with the small sample of this thread) folks seem to
'clump' into the two camps of "works great" and "say what?" i'm
wondering if there's something else at play here?  another bit of
code that's specific to production run variation, or a change in the
underlying DSP circuitry/programming during the life of the K3?

lyle, is there, perhaps, a chart/form you could suggest for testing
with some instructions (ie, make receive bandwidth flat, for SSB 40m,
adjust bw/passband/etc., serial no of K3, etc?)...  then those of us
interested in more rigorous testing can provide some kind of unified
feedback?   i'm a beta tester for some other hw/sw consumer products
and the reporting requirements usually establish a baseline so the
engineers can look for patterns.

(oh, and THANKS for all the hard work...  if this were easy it
wouldn't be half as much fun!)

73
robin (K9RDG)

______________________________________________________________
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

[Bill K9YEQ]

I thought Lyle's email quoted below was very helpful:  

        The way the new beta NR works is:
        Fx-y selects the length of the filter.
        F1 = 121 taps, F2 = 91 taps, F3 = 61 taps, F4 = 31 taps
        (The Beta 3.25 release used FIR filters of 61 taps.)
        y selects values of Beta (gain), decay, and delay (how long the NR
algorithm waits to process a signal)
        (The Beta 3.5 release used the "x" parameter for these selections)
        73,

        Lyle KK7P


73,

Bill
K9YEQ
K2 #35; KX1 #35; K3 #1744; mini mods
ATS-3B


______________________________________________________________

______________________________________________________________
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

[Robin Diane Goldstein]

hi...  thanks for the reply...  yes, i read lyle's description of the
two filters, which is how i knew that the 3.25 filters all had an FIR
of 61 taps...

what i'm looking for (in order do to an honest A:B comparison between
3.25 and 3.27 on my rig) is what Fx-y on 3.27 would most closely
match the F3-1 configuration of 3.25

73
robin (K9RDG)

>Date: Thu, 27 Aug 2009 10:18:54 -0500
>From: "Bill" <[hidden email]>
>Subject: Re: [Elecraft] K3 NR (what is the equiv to F3-1 from 3.25?)
>To: <[hidden email]>
>
>I thought Lyle's email quoted below was very helpful:
>
> The way the new beta NR works is:
> Fx-y selects the length of the filter.
> F1 = 121 taps, F2 = 91 taps, F3 = 61 taps, F4 = 31 taps
> (The Beta 3.25 release used FIR filters of 61 taps.)
> y selects values of Beta (gain), decay, and delay (how long the NR
>algorithm waits to process a signal)
> (The Beta 3.5 release used the "x" parameter for these selections)
> 73,
>
> Lyle KK7P
>
>73,
>
>Bill
>K9YEQ
>K2 #35; KX1 #35; K3 #1744; mini mods
>ATS-3B

______________________________________________________________
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