Hi Markus,

You see the phase is varying to much, at least in this experiment, in these last 8 days. I could continue for several more nights and maybe we would be there one days. But we are already in the middle of the season and time is running.
Actually i like to spend more time below 5 kHz, better on ULF, i.e. 2970 Hz and/or 970 Hz.

A -6 dB carrier may be to weak to determine the phase when SNR is low. Regarding cheating, i think there is nothing wrong about watching the Eb/N0 building up before a decode happens on the the RX side. But of course no information must be used for getting the final decode on the RX side which is based on the known message, for example best Eb/N0 and phase of the individual phase. The only information known to the RX station is my message announcement, including the information about eventual early transmission interruptions. Also the individual noise levels on the RX site are known to the RX station. Based on these informations, the decode has to be made. Then it valid in my opinion.

73, Stefan


Am 18.12.2017 21:44, schrieb Markus Vester:

Hi Stefan,

why give up? At -2.5 dB EbN0, Mike and Paul already had about 65% of the required energy for three characters. That's about 2 out of 3, or one character's worth left to do. Now throwing that away and starting afresh, you'll have to do twice the work....     

One could also send a reference carrier to enable phase alignment before stacking, without "cheating" by using known message content, One practical way would be to unbalance the PSK by reducing the shift from +-90° to e.g. +-60°, which would spend 25% of the transmitted energy for a -6 dB reference carrier.

Best 73,
Markus
  

-----Ursprüngliche Mitteilung-----
Von: DK7FC <[email protected]>
An: rsgb_lf_group <[email protected]>
Verschickt: Mo, 18. Dez 2017 21:06
Betreff: Re: VLF: Transatlantic success East to West

Thanks Paul for the summary.

Well, 8 days already and it gets worse. I think the phase changes are not from my side, so it must be the path.
I think i will give up now with this 3 character message. But even a failed attempt gives us a set of data showing the difficulties and phase variations.

On the other side, a single character message is to short to let it be as it is. So i'm trying a new message with just 2 characters before going down to 5170 Hz.
If possible, i daily feedback is apprechated.

f = 17470.1000 Hz
Start time: 18.Dec.2017  21:30:00 UTC
Symbol period: 45 s
Characters: 2
CRC bits: 14
Coding 16K21A
Duration: 09:12 [hh:mm]
Antenna current: 1.2 A

73, Stefan



Am 18.12.2017 18:56, schrieb Paul Nicholson:

Results from eight overnight transmissions of the
3 character message:

 2017-12-10/11  Eb/N0 = -13.1 dB
 2017-12-11/12  Eb/N0 =  -6.3 dB, phase 125.3
 2017-12-12/13  Eb/N0 =  -6.8 dB, phase 143.2
 2017-12-13/14  Eb/N0 = -19.6 dB (stopped 03:40 by safety limit)
 2017-12-14/15  Eb/N0 =  -6.0 dB, phase 68.8
 2017-12-15/16  Eb/N0 = -11.0 dB
 2017-12-16/17  Eb/N0 = -14.6 dB
 2017-12-17/18  Eb/N0 =  -9.9 dB, phase 37.0 (stopped 04:19 by snow)

Stacking

 11/12 + 12/13 gives Eb/N0 = -3.6 dB.
 11/12 + 12/13 + 14/15 gives Eb/N0 = -3.0 dB.
 11/12 + 12/13 + 14/15 + 15/16 gives Eb/N0 = -2.5 dB.
 11/12 + 12/13 + 14/15 + 17/18 gives Eb/N0 = -3.4 dB.

There's some phase changing going on here which is making
stacking not so effective.

--
Paul Nicholson
--