KX3 + JT9 -- any successes?

[Martin AA6E]

Hello, all,

I've been working with my stock KX3 and HF JT65 / JT9 via the WSJT-X program.  Brief
results:

--Mostly bad luck with JT modes until I did the "Extended VFO Temperature
Compensation Procedure".
--After doing the procedure, I can work JT65 pretty reliably for P ~ < 3 W.
--I can decode JT9 sometimes, but I can't get anyone to decode my transmissions.  (20M)

The result of the temperature calibration was that I seemed to have +/- 1 Hz
stability (compensation?) at 50 MHz from 22 to 55 C.  I thought that was pretty good,
but it still does not seem to give good JT9 results.  (JT9 is extremely narrow-band
with 9 tones spaced at 1.7 Hz across ~15 Hz of bandwidth.)

I have seen what some folks have done with larger heat spreaders and heat sinks, and
even cooling fans.  Has anyone tried liquid cooling? (Just kidding!)

My question is whether any of these strategies have resulted in reliable operation
for JT9 -- or should I just scratch that mode off my list? (I do have good experience
with this software and JT9 on the Ten-Tec Orion.)

I'm not complaining.  Nobody promised the KX3 was good for EME -- or VLF datacom!  It
is a shame that there's not better thermal isolation for the reference oscillator.  I
think Elecraft meets its own (vague - what time scale, what environmental
conditions?) +/-1 ppm "typical" spec, but the problem is that we need better
stability than that over a 1-minute transmission cycle  -- .05 ppm or even .01 ppm.  
A typical TCXO like the Orion's will do that, since the oscillator is thermally well
separated from the PA heat sink.

TIA / 73
Martin AA6E

______________________________________________________________
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
Message delivered to [hidden email]

[Jim Brown-10]

On 7/30/2014 8:58 AM, Martin AA6E wrote:

> --After doing the procedure, I can work JT65 pretty reliably for P ~ <
> 3 W.
> --I can decode JT9 sometimes, but I can't get anyone to decode my
> transmissions.  (20M)
>
> The result of the temperature calibration was that I seemed to have ±
> 1 Hz stability (compensation?) at 50 MHz from 22 to 55 C. I thought
> that was pretty good, but it still does not seem to give good JT9
> results.  (JT9 is extremely narrow-band with 9 tones spaced at 1.7 Hz
> across ~15 Hz of bandwidth.)

I've successfully made a few JT9  QSOs on 6M with my K3, and get good
copy on other JT9 signals that I know are coming from K3s. But K1JT
advises against using JT9 on VHF (and urges caution even about 10M)
because of its sensitivity to frequency drift. Quoting from his post
today on the WSJT reflector,

=   =   =   =   =

"It's hardly accidental that the docs explain that JT9 was designed for
the LF, MF, and HF bands.  Even 10 m is s stretch, with some radios and
some propagation conditions.

"Like JT65, the JT9 decoder does have an AFC as part of its bag of
tricks.  It's always engaged, solving for a linear frequency drift term
before decoding is attempted.  The algorithm can perhaps be improved.
It's a tougher job with JT9 than with JT65 -- not least, because the JT9
protocol devotes only about half as much time to a synchronizing tone,
compared with JT65."

=   =   =   =   =   =

73, Jim K9YC

 

______________________________________________________________
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
Message delivered to [hidden email]

[Don Wilhelm-4]

It is good that you are not complaining, but I just don't *get* it.
The specification is for +/- 1ppm over a temperature range of 0 to 50 degC.
You say the mode being tried needs .05 ppm or even .01 ppm.  Why would
anyone attempt to use a radio for that mode for which the specification
is worse than the needed stability by a margin of 20 to 100 times?
One might expect success with a factor of 2, or even 5, but to expect
success with an instrument that is 20 to 100 times less frequency stable
than required just makes me shake my head.

Compare it to trying to make a measurement in the low millivolt range
using a voltmeter only good to 0.1 volt.

73,
Don W3FPR

On 7/30/2014 11:58 AM, Martin AA6E wrote:
> I'm not complaining.  Nobody promised the KX3 was good for EME -- or
> VLF datacom!  It is a shame that there's not better thermal isolation
> for the reference oscillator.  I think Elecraft meets its own (vague -
> what time scale, what environmental conditions?) +/-1 ppm "typical"
> spec, but the problem is that we need better stability than that over
> a 1-minute transmission cycle  -- .05 ppm or even .01 ppm.

______________________________________________________________
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
Message delivered to [hidden email]

[Wes (N7WS)]

Maybe he's holding the temperature to 0.5 deg C :-)

On 7/30/2014 10:33 AM, Don Wilhelm wrote:
> It is good that you are not complaining, but I just don't *get* it.
> The specification is for +/- 1ppm over a temperature range of 0 to 50 degC.
> You say the mode being tried needs .05 ppm or even .01 ppm.  Why would anyone
> attempt to use a radio for that mode for which the specification is worse than
> the needed stability by a margin of 20 to 100 times?

______________________________________________________________
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
Message delivered to [hidden email]

[Martin AA6E]

That's about right. There's no reason (after some warm-up) that the oscillator
temperature should change by more that 0.5C (or whatever) over a 1 minute
transmission -- unless your oscillator is sitting close to the PA heat sink like the KX3.

The KX3, like most amateur radios, has a weasely frequency stability specification.  
It says +/- 1 ppm typical over 0-50C. What's weasely about that?

1.  "typical", to begin with.  It means something like "your chances are pretty good
it's within +/- 1 ppm if you are a normal user."  It's a weasel word, because you
would have a hard time proving that your radio did not meet this spec.  (Problems?
Well, you're just not typical.)  A more serious spec would be "worst case", but we
hardly ever see a worst-case spec in the amateur world.

2.  What does 0-50C mean?  A reasonable user might think that means ambient operating
temperature.  But it might just as well mean indicated oscillator temperature, which
is around 38C for my idle KX3 at the moment, while my ambient is ~22C.

3.  What is the timescale?  A radio that flicks around +/- 50 Hz on 6 Meters would be
in spec, but nasty even for SSB.  Most of us would probably assume that short-term
stability should be better than long-term.  That would be true if we're talking about
drift due to slow temperature changes, supply voltages, component aging, or
whatever.  The published spec does not give much insight as to what to expect in
practice.  The one minute on / one minute off that we have in JT modes is awkward for
the KX3.  The oscillator "typically" sees a large temperature swing over a minute's
transmitting -- several degrees C, depending on power setting.

For what it's worth, the Orion claims +/- 3 ppm stability (and accuracy - another
issue) over its operating range, but its short term (1 minute) stability is much
better than the KX3's.  I assume the K3 would be similar to the Orion, but sadly
there is no K3 on my desk.

This sounds like splitting hairs -- a good description of JT9 decoding!  The KX3 is a
fine radio for most applications, just not bleeding edge digimodes.

I admit to being fascinated by precise time and frequency.  I particularly like the
new Flex 6500/6700's GPS stabilized option, 5 x 10**-12 over 24 hours, for a price.  
That might support JT9 at 20 GHz.  Even there, the devil is in the details. They
don't give fluctuation vs timescale info.

73
Martin AA6E

Date: Wed, 30 Jul 2014 14:02:30 -0700
From: "Wes (N7WS)"<[hidden email]>
To:[hidden email]
Subject: Re: [Elecraft] KX3 + JT9 -- any successes?
Message-ID:<[hidden email]>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed

Maybe he's holding the temperature to 0.5 deg C

On 7/30/2014 10:33 AM, Don Wilhelm wrote:

> It is good that you are not complaining, but I just don't*get*  it.
> The specification is for ± 1ppm over a temperature range of 0 to 50 degC.
> You say the mode being tried needs .05 ppm or even .01 ppm.  Why would anyone
> attempt to use a radio for that mode for which the specification is worse than
> the needed stability by a margin of 20 to 100 times?

______________________________________________________________
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
Message delivered to [hidden email]