[600MRG] Long callsigns in WSPR

Hans Summers hans.summers at gmail.com
Wed Aug 27 21:07:41 CDT 2014

Hi all

I read the recent correspondence on long callsigns and I have useful info
to impart!

I have just completed work on firmware version v3.05 of my Ultimate3 kit
http://www..qrp-labs.com and this new version upgrade can send the long
callsigns and 6-character locators. So because of intense work on this over
the last week or so, getting the programming done, I can tell you what is
hopefully quite close to facts on the issue.

Firstly, ordinary WSPR messages contain callsign, 4-character locator, and
power. The callsign must be 3-6 characters long and there are some rules
about whether letters or numbers or spaces go in various positions etc.,
but that's not important right now for this discussion. The important point
is that the callsign cannot contain a prefix or suffix, no / character is
allowed in there. It doesn't matter how short your callsign is, you still
cannot have a pre/suffix. E.g. just because your call is A1A don't think
you can A1A/P, you cannot. This is the original WSPR protocol.

Later on it was extended to include support for compound callsigns, I mean,
callsigns including prefix or suffix (but not both). I call this "extended
mode" and in the U3 (new v3.05 firmware) it is enabled by setting a
configuration parameter "Ext. WSPR" to 1. In this mode, 6-character locator
is sent, rather than the usual first 4 characters only.

The long callsign form must be ONE of the following forms (not a

1) single letter or number suffix e.g. /P
2) double number suffix e.g. /26
3) 1-3 letters/numbers prefix e.g. MM/

Again: it doesn't matter how long your ordinary callsign is. A 6-character
call is fine, as long as the callsign obeys the original WSPR callsign
rules and the prefix/suffix style is one of the above three.

The extended WSPR mode sends the information in two 2-minute message
transmissions, alternating between the two message parts. So transferring
your information takes 4 minutes (2 slots) instead of 2 minutes (1 slot).
It's therefore harder to get your information through.

The first half of the message contains the long callsign and the power. But
NOT the locator. This message shows up something like
G0UPL/P 23
at the receiving end.

The second part of the message contains the 6-character locator, the power,
and a 15-bit hash code. But NOT the callsign. The hash code is used by the
WSPR decoder to tie together the two message parts. The hash code can be
thought of as an encoded, compressed version of the callsign. Kind of. But
it is is a lossy compression, you can't work backwards from hash code to
call sign. If the decoding station receives a part 2 extended message but
has never received a part 1 message then it shows something like:
< . . . > IO91ab 23
To show that it does not know the callsign. Once it has received the past 1
message, it knows what callsign relates to your hash code, and so it would
show something like:
<G0UPL/P> IO91ab 23

So, a station that had received the first part, not the second, would know
your callsign but not your location. A station that had received the second
part, not the first, would know your location but not your callsign.

It's worth mentioning that as far as I know, David VK2DDI's recent spot of
'XIQ was a single spot only, ever. One single 2-minute transmission. If XIQ
had been using the extended format, this would not have been enough,
because both pieces of the extended message are needed, two transmission
slots. If David had received the first part he would have at least known
the callsign (not the location). But if David had only received the second
part message, he would not have been able to know the callsign. Then an
awesome feat would not have been accomplished so beautifully!

So using compound callsigns in WSPR's extended mode makes it less likely
that your signal will be copied. In weak signal conditions this is not
good. 600m DX is surely weak signal!

So in fact there are two reasons not to use the extended mode:

1) it requires 2 cycles to transmit, so it makes it less likely to get a
reception report. It's like doubling the bandwidth, half the signal to
noise ratio, this kind of thing.

2) the hash code is 15 binary bits, which means 32,768 permutations, which
means there isn't a unique one to one correspondence between your callsign
and the hash code it produces. There is a small probability of false
decodes, if two stations happen to produce the same hash code and are
received. The more people use the extended mode, the more the probability
of false decodes increases.

So, support for this is there in the WSPR PC software, and now it is
implemented in my U3 kit as well, but really I don't recommend people to
use it unless they *really* need to do so. As people have said here, it
isn't a license requirement to encode your full extended callsign into the
WSPR message. So if possible, stick with your usual 4-6 character callsign,
don't put in any / character.

I hope this clarifies the situation and is useful to someone.

73 Hans G0UPL

More information about the 600MRG mailing list