Re: LF: Coherent BPSK on LF using EbNaut

[Markus Vester <markusvester@aol.com>]

Hi Stefan, Domenico,

 

to get an idea of the tolerances, you can convert the spectrum from
http://df6nm.bplaced.net/VLF/fec_tests/ebnaut_data_09200300.zip

into a wav and play around with the time offset and frequency offsets for the EbNaut decoder. The apparently best decode of this file is shown in the screenshot, using 1114.49 s and 2.1382 Hz. 

 

Changing the frequency by 0.25 mHz up or down introduces a phase slope of ~ 90° along the 1040 s sequence (end to end, i.e. +-45° deviation), which produced only negligible degradation (about 0.5 dB or a couple of extra symbol errors). Doubling the frequency offset to 0.5 mHz already affects the decode significantly, with about 3 to 5 dB degradation. This is also obvious in the constellation diagram (top right in show_rawsyms3a), where the two separate blobs are being smeared around and can no longer be distinguished visually.

 

A linear frequency drift results in a quadratic phase, and a drift rate up to about 1 mHz in 1000 s seems ok. This is within easy reach of a good OCXO. But the difficulty with that approach is that the receiver will have to re-determine the correct frequency for each subsequent transmission. 

 

EbNaut-RX can actually cope with some degree of (propagational) phase variation by repeating the decode attempt using different phase permutations between quarters. Having the frequency right will give constant phase, and be likely to decode soon during one of the first attempts.

 

Regarding the timing, +- 0.1 symbol durations seems to be ok, but 0.2 s already introduced a significant SNR penalty.

 

Best 73,

Markus (DF6NM)

 

-----Ursprüngliche Mitteilung----- 
Von: IZ7SLZ <iz7slz.domenico@gmail.com>
An: rsgb_lf_group <rsgb_lf_group@blacksheep.org>
Verschickt: Do, 1 Okt 2015 7:01 pm
Betreff: Re: LF: Coherent BPSK on LF using EbNaut

Hallo Stefan,

thank you very much for your message.

On 10/1/2015 4:29 PM, DK7FC wrote:

Hi Domenico,

Thanks for your report.
I see you on my OPDS32 grabber, as you say.

2015-09-28 21:00:47 IZ7SLZ  1159km 137502.120Hz   1mHz -46.7dBOp  93% 17.9dB
2015-09-28 20:00:47 IZ7SLZ  1159km 137502.120Hz   2mHz -41.1dBOp  90% 20.4dB
2015-09-28 19:00:45 IZ7SLZ  1159km 137502.120Hz   5mHz -44.1dBOp  93% 18.8dB
2015-09-28 18:24:51 IZ7SLZ  1159km 137502.121Hz   2mHz -45.0dBOp  75% 15.2dB

Is 137502.120 Hz the frequency of your choice?  

YES. CONFIRM.

If your TX and my RX is GPS locked (or tracked) then where does the 1 mHz offset come from?

No. My Tx is not yet fully GPS-Locked but it is steered by an OCXO. That's why i need to check the frequency and make a small correction as done in the above OP32 sequence . After correction QRG remains quite steady for some hours.

Maybe a phase variation due to changing propagation?

The tests carried out so far show that the phase stability of the LF path is sufficient for  transmissions of at least 20-30 mins. Of course, it is  necessary to do many other tests with differents conditions (message and symbol lenght, QRB, hours, seasons,...)  in order to better understand the channel capability.
There is an interesting result achieved on Sept 14, 2015 with some my transmission of 43 characters at 0.2 s (duration = 7 min 47 s)  succesfully received by DF6NM. See:
https://www.dropbox.com/s/kimklrv25sbmwiu/ebnaut_150914_results.zip?dl=0

Which offset is acceptable to avoid SNR losses in EbNaut (at a symbol rate that allows a QSO (in one evening! ;-) )) ?

Here i quote a Markus' observation regarding this question:

On LF, the frequency stability of the TX and RX local oscillators comes in as an additional variable. It is very important that these are known and stable to better than 1/message_duration (about a mHz for a 20 minute sequence), which requires either GPS locked oscillators, or a very good OCXO or Rubidium. To circumvent possible soundcard glitches on TX, it may actually be better to use a hardware oscillator (eg. a DDS on the final RF frequency).

73, Stefan   PS: Unfortunately not QRV on LF at the moment. I need to find a new loading coil body because my glass coil fully broke during maintenance attempts. Now i need something to come out with a stronger signal because each modification on the system has to be an improvement, by definition. So maybe i can come closer to 1W ERP then...

OK. We wait the restoration of your coil hoping that you want to try coherent bpsk. Your signal will be very useful for TA EbNaut test.

All the best

Domenico / iz7slz

09200300_best_result.png
Description: PNG image