[Elecraft] DSP - 4 questions & a LONG answer

[Dale Boresz]

Vic Rosenthal wrote:

> Lyle Johnson wrote:
>
>> Many of the testers are CW ops as well.  Some of them report really 
>> liking
>> how the noise reduction lets them open up the bandwidth to hear 
>> what's going
>> on around them, without increasing the noise.  And the CW audio 
>> filters are
>> useful, too, with both "hard" and "soft" slope characteristics.
>
>
>
> I would REALLY like to read some 'reviews' from the field testers, 
> particularly regarding CW performance.  I have never found a DSP unit 
> that has the ability to change an unreadable weak signal in noise to a 
> readable one, but considering the source of this one -- maybe this is 
> it.  How about some impressions?


Vic,

I am very fortunate to have been one of the field testers of the KDSP2, 
and as a CW op myself (for the past 38 years), I'd like to offer my 
impressions of the KDSP2.

First, I should say that I have been a huge fan of the KAF2 audio 
filter, and if you review the archives, you'll find a number of postings 
that I have made in response to folks questioning whether or not the 
KAF2 was a worthwhile addition. I've found it to be extremely helpful in 
extracting weak CW signals from QRM, QRN or both. It does so by passing 
the audio signal through a relatively high-Q filter with a peaking 
response, which tends to discriminate against all other signals except 
for those within it's very narrow passband. The result is that noise is 
reduced, and of course QRM is also reduced.

The KDSP2 achieves much the same effect but in a different way, and it 
does so with, in my opinion, a distinct advantage over the KAF2: the 
noise reduction is completely independent of the bandwidth. In other 
words, if you happen to like to use wider bandwidths (as I do) in order 
to more easily hear other signals on the band, the KDSP2 permits you to 
switch in noise reduction (NR) while maintaining your current bandwidth 
- whether it be wide or narrow. This also means of course that 
absolutely NO additional ringing is introduced when the NR is engaged, 
since it does not restrict bandwidth.

You mentioned that you've "never found a DSP unit that has the ability 
to change an unreadable weak signal in noise, to a readable one", and I 
would say that I pretty much agree. However, while working with the 
KDSP2, I've noticed that I find myself invariably engaging the NR 
whenever I encounter an extremely weak signal, even if the band is not 
particularly noisy at the time, and it definitely helps. The greater the 
background noise, the more effective the NR seems to be. I still think 
that the human ear/brain does the best job of pulling signals out of 
noise, but as anyone knows who has strained to make out the callsign of 
a weak station sitting right at the noise level, it can be a very 
fatiguing experience. This is where the NR excels - by removing 
irritating static pops and ticks, and thereby providing the ear/brain 
with a much less distracting environment in which to work. In other 
words, it provides a major reduction in listening fatigue.  Throughout 
the testing period, I had the interesting opportunity to perform 
side-by-side comparisons between the K2/KDSP2 and my ICOM 746PRO's DSP 
NR. I had a coaxial switch set up, which enabled me to switch the 
antenna instantaneously between rigs, and during the many hours spent 
testing, there was never a situation in which the 746PRO's NR exceeded 
the performance of the K2/KDSP2's. Of course, in my opinion the K2, even 
without the KDSP2, is the best sounding CW receiver I've heard (and, 
with the QTH being about 2 miles from the Ohio AES store, I've heard 
pretty much all of them), and now the KDSP2 extends that margin even more.

I think it's important to recognize what DSP can and cannot do. It 
cannot, as far as I've been able to determine, pull a signal which is S6 
out of an S7 noise level. I am aware of only two ways in which signals 
which are actually below the noise level can be extracted. One, is via 
the Noise Blanker - if the noise is of the right type. Fortunately for 
us K2 users, the KNB2 is an excellent noise blanker, and if the noise 
has the right spectral content, the KNB2 can virtually eliminate it, 
thereby enabling the reception of signals which were below the noise 
level. Another device which also can actually eliminate noise, is a 
noise nulling device such as the MFJ-1025/1026, which can completely 
null out interference - provided that the noise source is from a single 
direction. Neither of these systems however, is very effective against 
static crashes due to electrical storms. However, a well designed DSP 
noise reduction system, such as the KDSP2 is very effective at reducing 
static crashes, as well as any other type of continuous, 
white-noise-like interference, and greatly reduces listening fatigue, 
making for much more enjoyable operating. The KDSP2 also permits the 
user to fine-tune the noise-reduction system, by providing adjustments 
for 'Beta' and 'Decay' (loosely analogous to an AGC system's attack and 
decay parameters), as well as the NR Gain. The excellent Technical Note 
and Glossary which Lyle wrote,
really helps to explain these parameters.

Regarding the bandpass filters of the KDSP2, for each of the modes - CW, 
SSB, DATA, there are four filters, which are selected in the same manner 
as the AF1 and AF2 filters of the KAF2 are selected - except that for 
the KDSP2, we have four filter settings rather than two. Each of these 
filters have parameters which can be changed as desired, including 
filter center frequency and bandwidth (for CW), low-pass and high-pass 
corner frequencies for SSB, and filter gain for all modes. Additionally, 
each of the four possible CW filter positions have a 'Normal' and 'Soft' 
position which is easily set via the menus. The selection of these 
filters is completely independent of the I.F. filter settings. The point 
here is that although the KDSP2 comes with some carefully-determined 
default settings, they are all adjustable, for those of us who are 
inclined to experiment a bit.

In summary, now that I've been using the KDSP2, I wouldn't want to be 
without it. I had a great time during testing, trying to find weak CW 
signals buried among QRN and QRM, and then using the KDSP2's NR and 
filters to pull them through.

I hope this helps, Vic.

73, Dale - WA8SRA