VLF: Fwd: question about MFSK-37

[Markus Vester <markusvester@aol.com>]

Hi Steinar, LF,

 

for an explanation, please see my mail to Alex RN3AUS.

 

Best 73,

Markus (DF6NM)

 

 

Von: Steinar Aanesland <saanes@broadpark.no>
An: rsgb_lf_group <rsgb_lf_group@blacksheep.org>
Verschickt: Fr, 6 Apr 2012 4:22 pm
Betreff: Re: VLF: DF6NM 8970.00x

Hi
>> I am not familiar with this MFSK-37 mode . Where can I find some info about it?
LA5VNA Steinar

-----Ursprüngliche Mitteilung-----
Von: Markus Vester <markusvester@aol.com>
An: Alex K <rn3aus@mail.ru>
Cc: Stefan Schäfer <Stefan.Schaefer@iup.uni-heidelberg.de>; lubomir.bobalik <lubomir.bobalik@rtv5.cz>
Verschickt: Sa, 24 Mrz 2012 8:28 am
Betreff: Re: question about MFSK-37

Hi Alex,

 

my version of MFSK is no real "digimode", it's just a simple allocation of TX frequencies to the numbers and letters of the alphabet. I am now using frequencies #0 to #9 for the numbers, #10 as idle or space, and #11 to #36 for the letters A-Z. Please note that since my first experiment I have exchanged idle and zero, and now have the lowest frequency as "0".

 

I have tried this at three different speeds, targeted at "6000", "600" and "60" spectrograms. For easy readout I have chosen decimal frequency steps (1 mHz, 10 mHz, 100 mHz), and transmit times associated with whole UTC minutes (30, 5 and 1 minutes).

 

I have pasted my two earlier emails on the subject beneath yours. Also attached is a calculation of character error rate vs SNR, which shows the benefit of  multifrequency MFSK (eg. "64 ch", pink curve) versus DFCW ("2 ch", red). It is even slightly better than conventional PSK (dashed black).

 

Best 73,

Markus (DF6NM)

 

From: Alex K

Sent: Monday, March 19, 2012 4:39 PM

To: Markus Vester

Subject: question about MFSK-37

Hello dr Markus!
I now develop the control program for my new DDS-synthesizer. I would like to realize MFSK-37 mode for use on VLF and, probably, on LF (faster variant). Could you inform me about the specification of this mode?
73 de rn3aus/Alex

 

_______________________________________________
Von: Markus Vester <markusvester@aol.com>
An: rsgb_lf_group <rsgb_lf_group@blacksheep.org>
Verschickt: Fr, 16 Mrz 2012 10:48 pm
Betreff: VLF: MFSK-37 tests above 8970 today

 

To explore the possibility of further two-way VLF contacts from my home antenna, I have conducted a test in MFSK-37 mode this morning. I was hoping to reach 0.42 mHz ("6000") grabbers using UT-synchronized half-hour dashes. Similar to an earlier kite experiment, I used absolute frequency encoding of characters in 1 mHz steps (0 = 8970.000, 1 = 8970.001, ... 9 = 8970.009, idle = 8970.010, A = 8970.011, .. Z = 8970.036 Hz).

 

The plan was to transmit my complete callsign in 2.5 hours between 8:00 and 10:30. Unfortunately DHO locking failed during the first two characters, which came out some 9 mHz low. Attached image shows what was actually sent according to my "monitor", which is a high harmonic of the transmitted VLF signal accidentally aliased into the LF TA window.

 

I was pleased to find all the dashes showing up clearly as bright dots on the OK2BVG "6000" grabber, and even left visible traces in his "600" window. As I had also received Lubos' transmission last weekend with a good margin in 0.48 mHz, a two-way QSO between us would probably be feasible within a few hours.

 

Paul Nicholson's 0.278 mHz super-sensitive spectrogram also shows the transmission at 120 degrees azimuth, even though the dashes were really too short to reach full SNR in this bandwidth.

 

Results on the DK7C grabber were not quite as good, with only a single clear dot on .006 Hz. This may partly have been due to the previously experienced daytime minimum at our distance (180 km). Nothing distictive was visible at OE3GHB, who is at similar distance as OK2BVG, but seems to have suffered from a bit of local QRM at the time. 

 

Again we find that a great deal is possible with very modest means...

 

Best 73,

Markus (DF6NM)

 

Von: Markus Vester <markusvester@aol.com>
An: rsgb_lf_group <rsgb_lf_group@blacksheep.org>
Cc: Paul <paul@abelian.netcom.co.uk>
Verschickt: Mi, 30 Nov 2011 11:11 pm
Betreff: VLF: MFSK-37 kite transmission on Nov 27

 

This is a somewhat belated report of my Sunday afternoon VLF kite activity, Nov 27th, 15 to 17 UT. The site and configuration was the same as on June 19 and July 30 this year (see emails below). This time I had mostly good kite elevation and an antenna current up to 0.85 A, radiating about 2 to 3 mW EMRP. I had originally hoped to engage in a two-way contact with DK7FC, but unfortunately sufficient wind came up only much later than expected.

 

Stimulated by discussions with Paul Nicholson and Stefan about the advantages of visual decoding and the usefulness of DFCW, I decided to experiment with a simplistic multifrequency coding approach, with one frequency channel assigned to each letter in the alphabet, plus the numbers. The concept is somewhat similar to Alberto's Jason mode, but simpler by using absolute frequencies (instead of differences), only one dash (not two) per character, and bit timing based on UT alone.

 

For a given speed, such MFSK dashes can be about 5 times longer compared to straight DFCW, indicating a 7 dB advantage. There are other subtle factors involved in the sensitivity, like the higher probability of a false hit in the much larger number of channels, versus the necessity to have several dashes correct to decode one complete DFCW character. But a rigid calculation based on gaussian noise statistics confirmed that at the same speed and error rate, 32-channel MFSK is actually 6 dB more sensitive than 5-bit DFCW. On the other hand, trading bandwidth efficiency for sensitivity is nice on VLF, but may not be appropriate for densely populated bands.

 

The first MFSK transmission from 15:00 to 15:25 was targeted at Stefan's  4.5 mHz "600" window; and I sent "DK7FC" using 5 minute dashes with 10 mHz spacing. One of 37 tones were generated by manually editing the frequency of SpecLab's signal generator every five minutes, according to the following scheme:
- base frequency, space: 8970.00 Hz,
- numbers 1 - 9: 8970.01 to 8970.09, zero: 8970.10 Hz,
- letters A - Z: 8970.11 to 8970.36 Hz.

 

Then Stefan called on the phone and informed me that I had good SNR in his "60" window. I switched to 60 s conventional DFCW sending "FC DF6NM" from 15:33 to about 16:05. After that, I repeated my callsign in 37-MFSK, this time using 1 minute dashes spaced 0.1 Hz. The last two characters were "73" on 8970.7 and 8970.3 (see attached pic, revealing my typing error).

 

From 16:24 on, I finally sent a straight carrier again on 8970.002, until I finally had to shut down at 16:54.

 

As an aside result, I found that it is quite easy to receive on the kite antenna without additional T-R switching hardware, by moving the little 23.4 kHz samplerate-lock ferrite antenna towards the loading coil, and shorting the TX transformer to ground.

 

Many thanks to Stefan, Chris, Mal, Lubos, Eddie, Victor and Jacek for the valuable reports, and to Paul for running his very sensitive spectrogram.

 

Best wishes,
Markus (DF6NM)

MFSK_60.jpg
Description: JPEG image

MFSK_vs_PSK_log.png
Description: PNG image