Fine tweaking of the software timing has reduced any long term timing drift to well within MEPT_JT parameters. In reality its stability outstrips my shack computer when it is free running, (no time sync from the Web).

I have changed the name of my coding to avoid confusion. Whilst I am happy to have bits of binary code and bits of chewing gum in the same sentence, I can see that it could cause problems. Equally the system is 5 tone, I have gone for RTTY terminology of units.

It is now 5 Unit Multiomde + WSPR

All of the coding is based on MEPT_JT elements 682.66mS long. A QRSS3 Dot in theory 3 Seconds long is in fact 4*0.68266 Seconds = 2.73064 Seconds a QRSS3 Dash, 9 seconds in theory is 13*0.68266 = 8.87458 Seconds.

S/MT Hell Characters are produced by elements of MEPT_JT code. A 5th element is produce by an extra 3rd Binary bit.

All spaces are multiples of 0.68266 seconds apart from one filler, for accurate 2 Minute slot timing, just prior to the start of the MEPT_JT data and after the 2 step waveforms. The Steps are produced by the 5 element tones spaced precisely ;) 1.46484375Hz apart and 4 elements long. Originally I had a 5-4-3-2-1 countdown with 5 of these. Now just 2-1, as they convey no real information. The top of step 1 is twice the length of step 2.

My WSPR Data starts at 1.36532 Seconds past the even minute, 2*682.66mS, as near to 2 seconds as the elements allow, although the 2 Minute filler can place this where ever it likes.

Paolo IZ1KXQ produces some fine screen shots of my signal.