I have made an unfortunate discovery. I don't know if this is just my KX3
or is more generally applicable, but: when I run WSPR with my KX3, the
frequency stability is inadequate to get good results on transmit. I
discovered this by transmitting with 1.0 watts with the KX3 and then the
K3, using the same antenna, and looking at the reports on the WSPRNet
database. I immediately suspected something, because the K3 was being heard
by 5x as many stations as the KX3.
I then listened to the K3's WSPR transmissions on the KX3 and vice versa,
using just a piece of shorted coax for an antenna to cut the signal level
down to about S9. WSPR users will know that the graphical representation of
each received signal runs horizontally across the display, with time on the
horizontal axis and frequency on the vertical. When the K3 transmits, the
line is a nice straight level stripe, as it should be. But when the KX3
transmits, it has a pronounced downward swoop of as much as 10 hz over the
first 45 seconds or so. As a result, most receiving stations are unable to
decode the KX3's transmission. Listening in the shack at a signal level of
S9, the K3 can decode less than half of the KX3's transmissions.
KX3 frequency stability on receive seems fine, as evidenced by nice level
lines on the graph. BTW I tried two different computers driving the KX3 on
transmit, but got identical results on both.
The frequency stability requirements both of WSPR and JT65 are
considerable. WSPR transmits 4-frequency FSK, with the tones separated by
only 1.5 hz. The whole bandwidth is 6 hz. The exact tx frequency is not
crucial, but the transmitter has to pick a frequency and stay there.
I got my KX3 partly to use on digital modes, and I have an important test
of a portable antenna coming up for which the KX3 would be perfect.
However, at the moment this problem is standing in the way. I'm hoping that
there's a simple fix.
The KX3's temperature stability is specified at +/- 1 ppm typical over
0-50 C. That's +/- 14 Hz on 20 meters. Normally the short-term
stability would be far greater than this. However, during long key-
down periods there is some heating of the RF board, which includes
both the PA and oscillator circuitry.
We included closed-loop temperature compensation for this effect in
hardware and firmware. Normally this compensation is entirely
sufficient, but apparently it doesn't meet WSPR/JT65 requirements.
I should be able to provide a user-settable fine compensation value in
the menu, although it would be a bit of trial and error for the user
to determine the optimal value.
Tony Estep wrote:
> ....KX3 frequency stability on receive seems fine, as evidenced by
> nice level
> lines on the graph...
> The frequency stability requirements both of WSPR and JT65 are
> considerable. WSPR transmits 4-frequency FSK, with the tones
> separated by
> only 1.5 hz. The whole bandwidth is 6 hz.