Hi all,
A number of comments and questions have appeared here recently
concerning the parameter DT displayed by the WSJT-X decoders.
K6AVP:
Only a few have posted what range of DT’s they mostly see ...
K9YC:
My memory is routinely seeing DTs in the range of +/_ 200msec, with occasional
outliers up to 2 sec or so.
AI4FU:
I would be most interested to hear if you tried pushing the limits of how far out of sync you could push things and still manage decodes.
NX6D:
I see “DT” values that range from about 0.0 to about 0.5 in my system.
... etc.
Positive DT means that a signal arrived late according to your
computer's clock. For example: if the computer clocks at both ends of
an EME QSO are bang on, the decoded EME signals will show DT = 2.5 s,
because the EME path delay is 2.5 s. Terrestrial propagation delays are
a few tens of ms or less, so the DT values most of us see in HF
operating are caused by latencies in our audio systems, context
switching delays in our operating systems, and -- most importantly --
clock errors in one or both computers.
Because people have been asking about it, I compiled histograms of the
distribution of DT values extracted from decodes in my ALL.TXT file from
two recent dates. The first was a few weeks ago, using WSJT-X v1.9.1.
The second was during the FT8 Roundup, last weekend, using WSJT-X
2.0-RC5. I limited the time ranges so that each data set had exactly
the same total number of decodes, 65010.
The distribution of DT values for these two data sets are shown by red
and purple curves in the attached plot. Most decodes have DT in the
range 0.0 to 0.3 s. The bug in RC5 that eliminated decodes with
negative DT is clearly evident in the plot: the purple line cuts off to
zero to the left of DT=0.
I operated in the FT8 Roundup with "Save all" checked, so it was easy to
reprocess all of those saved files with the soon-to-be-released
WSJT-X 2.0. This produced another 5290 good decodes -- essentially the
previously missing ones that have negative DTs.
The FT8 decoder tests for DT values from -2.5 to +2.5 s. As these
results show, most FT8 users keep their computers "on time" to within a
few tenths of a second.
-- 73, Joe, K1JT
-- 73, Joe, K1JT
_______________________________________________
wsjt-devel mailing list
wsjt-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
