During my shortened BARTG effort(I even missed XR0X, although I did hear
his pileup) I took a little time to compare RITTY with MMTTY.
Both used 16 bit SB soundboards which were each connected their own
FT1000MP which I consider identical in performance. Both radios were
connected to 3 element full size Yagis on separate towers in my back yard,
one at 65 ft and the other at 110 ft.
I listened simultaneously to a variety of European stations on 20 meter
long path as the band was going out. When the average S/N was large(
greater than about 15 db) both decoders produced roughly the same number of
errors. The print was good enough for contesting no matter how much QSB or
QRM. When the S/N grew smaller as noted on my scopes by observing the
deflections during and after transmissions both decoders produced
increasingly more errors, as would be expected. The copy was adequate for
contesting for both down to roughly S/N =10 db. At that point MMTTY started
to produce errors that made contesting print increasingly difficult and
eventually not possible for S/N below 3-4. RITTY was clearly better in the
low S/N region and it routinely yielding print adequate for contest
purposes down to below S/N=3-4. Now the error rate was high, more than
50%,but enough repeatable text could be dug out over two, if not in just
one transmission, to complete a QSO. I, of course, didn't try and make any
QSO since I was concentrating on the comparison.
I may not have had MMTTY optimized completely. I took W9ALS's advise and
used the FIR BPF detector although I tried the standard as well.
In looking at the two decoders I noticed what may be one reason for the
difference I observed. RITTY has something called a numerical flywheel that
tries to predict the timing of the received print. I doesn't work unless
there is a diddle or the text is a regular machine-generated bit
stream rather than irregular hand-typed text. Well, during contests nearly
everyone uses buffer messages that are sent in a regular machine generated
data stream that is tailor made for the flywheel.
Well that's my two bits for now. ( four bits in Aussie money!)
John VK4UC
BTW there are not a lot of VK/ZL RTTYers on these days, but if XR0X
wants to listen for us he might try 2200Z on 10 or 15 on the short path and
0600Z on 20.
|