Re: LF: F5WK EbNaut

[Markus Vester <markusvester@aol.com>]

Hi Michel,

after I realized that the time offset was now 3 seconds I got a few more decodes. Apparently SNR and condx were a bit weaker than the night before.

2016-09-28  "F5WK IN99EF"

start    symbol carrier df
 UT       EbN0   EbN0   mHz

19:00:03  12.2   14.2   -1
19:30:03  14.7   17.9   -1
20:00:03   1.1    4.3   -1
20:30    (rx interruption)
21:00:03  12.7   14.7    0
21:30:03  13.5   16.6   -1
22:00:03  10.7   13.4   -1

Yes the undefined state of a divider before the PA can cause problems for coherent EbNaut (or Opera) transmissions. With the DDS, you could try to gradually advance the phase in 64 small steps to accumulate 360°.

Best 73,
Markus



-----Ursprüngliche Mitteilung-----
Von: Michel Brunel <f5wk@mterrier.net>
An: rsgb_lf_group <rsgb_lf_group@blacksheep.org>
Verschickt: Do, 29 Sept 2016 12:14 am
Betreff: Re: LF: F5WK EbNaut

Hi Markus,

Yes I'm still transmitting with the same ERP as the other days.

The antenna is brought to resonance by a series loading coil and impedance matching is performed by a tapped transformer.
It seems to me that overcurrent applies only with an inductive mismatch, otherwise it tends toward zero. This overcurrent is not noticeable at all, except that it triggers the ACS712 Hall effect sensor and the PIC A/D routine !  
Anyway the best approach is to use a DDS. I already tried that, but as my class D PA requires 2*F, the phase offset should be 2*180° and as 360° means 0°, I thought that a immediate lower value - ie 174,375° for the AD9851 - was not adequate.

All the best,
Michel - f5wk

Le mercredi 28 septembre 2016 à 23:21:50, Markus écrivait :

Hi Michel, your signal is not weak at all, surely good enough for comfortable QRSS3 reading here.  Regarding overcurrent trip: I found that for safest operation of a class-D PA, the antenna impedance should be presented as a series resonant circuit to the transistors, because that will have high impedance for any out-of-resonance spectral components. Then during phase reversals, the current will experience a dip rather than a peak. The impedance condition depends on the total phase shift of the matching circuits and low pass filters, which should ideally be 0 or 180°. However feeding into a 50-ohm loading coil tap (or through certain PI- or T-filters) tends to have around 90°, transforming series resonance into parallel resonance, and presenting a short-circuit to the PA for out-of-band exciation. Tonight there seems to be a (somewhat weaker) trace on your QRG again, are you still sending EbNaut? Best 73, Markus (DF6NM) -----Ursprüngliche Mitteilung-----  Von: Michel Brunel <f5wk@mterrier.net> An: Domenico IZ7SLZ <rsgb_lf_group@blacksheep.org> Verschickt: Mi, 28 Sept 2016 7:13 pm Betreff: Re: LF: F5WK EbNaut Hi, Big thanks to Domenico, Markus and Stefan for their successful decoding of my EbNaut transmissions. Ebnaut decoding is not an easy task especially with weak signal. I fear my GPSDO frequency was not as accurate than expected. I had the bad idea to use a low cost GPS antenna wich performed not as good as the professional grade antenna I used for tests at home. Thus some jitter and phase jumps depending on GPS constellation position. As a "proof of concept" I made Loran-C records during the day  using the ebnaut setup and results are sometime  good sometime not so good. The transmitter overcurrent was randomly  triggered by the phase reversal. Bpsk modulation is performed synchronously by exclusive OR gates, between the D flip-flop and the MOS drivers. Phase reversal is like changing the frequency for a very short time, which leads to a current pike due to the high Q of the LF antenna. Decision was taken to bypass the overcurrent detection. The good way to do bpsk modulation is with a suitable DDS such as AD9852.  I found the antenna performs poorly at LF since the radials are burried in a sandy soil, as time goes on. No big deal for ebnaut though. A lot of things to improve for my next LF campaign,  73, Michel - f5wk/P

--