Hello Roger,
as JT9-2 and JT9-5 signals are resampled to JT9-1 to be decoded by WSJT decoder engine it all depends on the drift tolerance of that decoder.
For JT9-2 the frequency stability must be 2.222 times better than for JT9-1.
For JT9-5 it must even be 5.926 be better.
As I couldn't find any information about the required JT9-1 stability I did some brief tests with JT9-5:
- At -5 dB level a drift up to 0.35 Hz/min seems acceptable.
- At -15 dB level it is 0.26 Hz/min.
- At -25 dB level it is 0.24 Hz/min
- At -30 dB level it is 0.22 Hz/min
- At -34 dB level (more or less the JT95 limit) it is 0.20 Hz/min
But as said this was just a brief test, these numbers should be handled with some caution.
Also keep in mind that this numbers represent the combined TX and RX drift (according to Murphy TX and RX drift will always be opposite).
If using a transverter the TX drift is the combined drift of the driver TX and the transverter.
Furthermore I assumed that the drift was linear over time, for frequency jumps it might be different.
Another frequency instability source will be the ionosphere, but there is little we can do about that (but the more stable TX and RX the more room for ionospheric instabilities).
Based on the above I would suggest to aim for a LO stability of no worse than 0.1 Hz/min, 0.05 Hz/min would be even better for weak signal DX (and that is what JT9-5 is all about).
Tests over the past weeks have shown that JT9-5 works very well for 136 kHz DX (N1BUG was copied by numberous EU stations).
On 472 kHz K3MF has been reported in Europe (as far as I know by G0LUJ and myself).
73, Rik ON7YD - OR7T