Unfortunately I see little improvement in the CW auto-spot feature,
broken in the last release. It is still much less accurate than the
original algorithm, and less consistent in it's behavior.
Try this - tune in a good steady (S6 or S7) cw signal. Set tuning to
FINE, enable CWT and then press SPOT repeatedly, observing the
frequency display. The beta version (like 1.88) settles on a
considerably different frequency almost every time - moving 5 Hz or
more commonly, and sometimes as much as 20 Hz or more. There is a
tendency to tune low. The original version, on the other hand, rarely
moves more than 1 or 2 Hz with each auto-spot invocation, bearing out
the fact that it is more accurate in general, or less likely to get
confused (by noise, varying morse code speed etc.).
Since I have two K3 units here, I set one up with 1.78/1.58 and the
other with the beta firmware, and then compared their auto-spot
behavior side by side on the same received signal using an antenna
switch. The difference is quite pronounced. Here are some series of
freqs (last three digits) resulting from 10 successive invocations of
auto-spot on the same steady S7 signal on 40 meters (at 13 wpm or so,
fortunately a long winded transmission):
v178/158 (TEXT DEC threshold = AUTO)
125, 125, 125, 129, 130, 130, 127, 128, 129, 129
beta (TEXT DEC threshold = AUTO)
115, 118, 124, 079 (!), 106, 106, 124, 107, 112, 100
beta (TEXT DEC threshold = 5)
124, 123, 109, 116, 118, 122, 117, 116, 120, 123
The newer firmware version tends to "bounce around" more with a
tendency to tune low, often requiring multiple invocations of
auto-spot to hit the right pitch, whereas the original algorithm
generally tunes right the first time. With threshold set to 5 in the
beta version auto-spot is a bit better, but all too often it takes
multiple tries to hit the right pitch, as it still tends low. If I set
the threshold much higher, it doesn't want to tune at all for signals
that the original algorithm handled easily.
I'm not sure what the "improvements" to auto-spot introduced in 1.88
were supposed to do, but IMO the feature is still negatively affected
by the attempt. A side by side comparison of the two versions, on the
same signal, is very revealing in this case - can someone else try
this?
Too bad, since I use the feature a lot it's back to 1.78/1.58 for now.
It "just works".
Bob NW8L
On Mon, May 12, 2008 at 7:08 PM, wayne burdick <[EMAIL PROTECTED]> wrote:
Thanks to your input on the last firmware release (1.88), we've made a
number of important improvements.
For the first time, we're posting new firmware as "beta." Upgrading is
optional: if you want to minimize your risk, you might want to wait until
we've blessed it as a normal production firmware release. (Or until all
your
friends have it and you're a little jealous.)
We'd love to have as many thrill-seekers as possible take the plunge and
load the new firmware now. We think it's greatly improved, and it will
certainly clear up many of the issues noted over the past week. Even
though
we're headed to Dayton in a couple of days, we'll be monitoring closely
for
any additional issues.
For release notes and instructions on how to load beta firmware, please
see:
http://www.elecraft.com/K3/k3_software.htm
The instructions (linked from within the release notes) explain how to
create a separate folder for the beta revision. This will allow you to
easily revert to an older revision if you find it necessary. Hopefully
not!
For your convenience I've also provide the release notes below.
73,
Wayne
N6KR
* * *
MCU 1.94 / DSP 1.73, 5-12-08
* VOX: Increased sensitivity. Typical VOX GN settings now ~30-50.
* VOICE MONITOR: Eliminated "scratchy" artifacts with compression set to
high levels.
* FM DEVIATION ADJUSTABLE: Use CONFIG:FM DEV menu entry (default: 5.0
kHz).
* Flash memory driver improved to prevent occasional receive dropouts,
"ERR
PL1", and
"FPF LOAD PENDING" messages.
* XMIT SWITCH IN MENU: Tapping XMIT while in menu correctly starts/ends
transmit.
* PASSBAND SHIFT BELOW "0.00" CORRECTED: The SHIFT control no longer
allows
continued rotation below 0.00. (This was occurring only with sidetone
pitch < 500 Hz.)
* SQUELCH/RF GAIN CONTROL CHANGES: By default, the RF/SQL concentric
pots
control main and subreceiver RF gain, and squelch is controlled by two
CONFIG menu
entries (SQ MAIN and SQ SUB). But setting SQ MAIN to "=SUB POT"
assigns
SUB RF/SQL pot to both main and sub squelch, and in this case, the
main
RF/SQL
pot controls both main and sub RF gain. Setting SQ MAIN back to a
numeric
value
restores the default assignments. Note: Squelch applies only to FM
mode
at present.
---
http://www.elecraft.com
_______________________________________________
Elecraft mailing list
Post to: Elecraft@mailman.qth.net
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com
_______________________________________________
Elecraft mailing list
Post to: Elecraft@mailman.qth.net
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com
