[Elecraft] KNB2

[Jim Wiley]

Joe et al:


I experienced the same thing with my KNB2.  A "sort of" fix, for me, was 
to increase the size of both timing capacitors so as to lengthen the 
blanking pulse.  I increased C11 from .001 uF to .0047 uF, and C12 from 
.0068 uF to 0.033 uF.  In each case, this represented an approximately 
5X increase.  This seemed to help quite a bit, particularly on power 
line buzz,  but  I am sure there are further improvements that could be 
made.  I believe that the basic design of the KNB2 needs to be changed, 
but I haven't the time  to play with it now.


What got me started was when I had just completed the KNB2, and I set 
about to test it's operation.   I found that the blanker did not respond 
to my traditional "noise source" (more about that in a minute) but did 
seem to work when scratching the top of Q22.     So, the KNB2 works, in 
it's own way, but it seems to be effective  for only a limited range of 
noise types. Immediately after completing the KNB2,  I tried my 
"traditional" tests (below) and found that the KNB2 was totally 
ineffective.  I thought it was broken.  However, after communicating 
with Gary Surrency, he put me onto the "scratch Q22" method, and of 
course, it was just fine. 


So, what was the difference?  Why had my traditional noise blanker test 
failed?  Some background follows:


Several years ago (before I owned a K2) I was searching for a 
controllable noise source to use for testing and aligning noise blanker 
systems.  I thought about what a noise pulse was, and decided that the 
most important characteristics of noise were (1) very short duration, 
high amplitude signals, and (2) very wide bandwidth. 
 

After a few tries involving use of Model-T ignition coils, door buzzers, 
some other ways to create "artificial noise",  I hit upon the idea of  
using a sweep generator.  If you think about how a sweeper works, it 
sends out a signal that covers a very wide band, and the dwell time on 
any particular frequency is very short.  If the sweeper is set so that 
it's output signal covers the entire range of the receiver being tested, 
then depending on the sweep rate,  the  time that the swept signal is  
present in the receiver's IF passband is very small - it appears 
basically as a noise pulse.  At the same time,  because some blankers 
work on the idea of "out of band" noise,  the swept signal also meets 
that requirement.


 For example, for a HF receiver, I might set the sweeper to start at 
100KHz and sweep up to maybe 50 MHz. For a VHF or UHF set, I might set 
the sweeper to run from 20 MHz below the receiver tuning range to 20 MHz 
above. In any case, you can see that when using rapid sweep rates the 
signal is on any particular frequency for only a very short time 
(microseconds to milliseconds, depending on sweep rate).  As a bonus, 
the output level controls of the sweep generator allowed me to adjust 
the "noise" pulses from very weak (less than 1 microvolt) to extremely 
strong (+10 dBm) at the receiver antenna jack.  I was using a old HP  
8601 that I found on eBay for less than $50,  but most any sweeper 
covering the range will do, and home-made units are  not that hard to 
make, particularly if  all you want is "swept noise" and precise control 
over frequency and output level are not required.  By this I mean you 
don't need "frequency counter" type readouts, and the output level need 
not be precisely calibrated.


What I was actually trying to do was simulate 160 meter LORAN pulses and 
"Woodpecker" over the horizon radar signals, which a few years ago were 
among the most egregious  "QRN" ever heard.   By adjusting the sweep 
rate, I could simulate signals "ticking over" once or twice a second (an 
idling engine type noise), to  "power line buzz" which occurs at 120 Hz. 


This worked like a charm, and was great for adjusting noise blankers, 
such as the ones in my Drake "C" line.  As years passed, I found the 
setup worked with pretty much any rig having a noise blanker, such as 
the Kenwood, ICOM and Yaesu sets, as well as most everything else. In 
fact, this setup has worked on anything I have tried so far, except for 
the KNB2!   To really give a "real world" test,  I could also mix in 
signals from conventional signal generators to see how much the blankers 
"hashed up" the wanted signal while doing their thing.  I used a scope 
to look at either / or the receiver IF signal or the audio output.   It 
was all very interesting and informative, and I learned quite a bit 
about blankers in the process.


So, what's with the KNB2?  Possibly one reason the Elecraft KNB2 fails 
this test is because it was designed after the LORAN A stations and the 
"Woodpecker" were history.  I don't know this for a fact, it is just 
supposition.   However, I maintain that a good noise blanker should be 
able to handle this type of interference, because who knows when some 
administration or military group somewhere will decide to use the HF 
bands for  pulse (LORAN) or swept signal (Woodpecker) type transmissions 
again.  To me, a good blanker should be able to handle ignition noise, 
power line leaks, electric appliances, lightning, static discharges from 
raindrops or snowflakes hitting the antenna, the "Woodpecker" and more. 
No blanker can completely clean up all noise, and some will distort the 
intended signal to a greater or lesser degree while operating, but all 
should produce a significant reduction in noise.  Natural lightning is 
perhaps the most difficult, because the pulse duration can be quite 
long,  but I believe putting up with a "hole" in the desired signal for 
several milliseconds (or tens of milliseconds?) is preferable to having 
to listen to those very loud crashes.


My present rig, other than the K2, is a Yaesu FT-1000 Mark 5.  It has 2 
blanker settings, each of which is adjustable over a wide range of  
levels.  It is interesting to hear even raucous power line noise or TV 
set horizontal oscillator signals disappear as adjustments are made.  
However, I will also say that at the more aggressive settings, the Yaesu 
blanker can "hash up" a signal pretty badly.  It becomes  a balancing 
act to use enough blanker to get rid of the noise while at the same time 
maintaining enough intelligibility to copy.  


Just some idle ramblings.  I will go home now.  I hope some of you find 
this interesting.


- Jim, KL7CC