Re: LF: Signal bandwidth in Op32 LF

[Markus Vester <markusvester@aol.com>]

PS: Why does the displayed bandwidth in opds vary so much?

To maximise sensitivity, opds is using synchroneous AM demodulation, and an estimate for the carrier is extracted from the central peak in the spectrum. To cope with spread from propagation and TX/RX instability, the peak is evaluated down to it's -10 dB points, which is the width indicated in the output. For random phase symbols the spectrum won't contain a discrete central peak but will look like a ragged noise hump, approiximately as wide as the symbol rate. Depending on unpredictable details of the shape of the spectrum, the point where it meets the -10 dB criterion may vary arbitrarily.

Opds sensitivity should be about 3 to 4 dB better  for coherent signals. For incoherent signals, the carrier estimate uses wider bandwidth and becomes fast enough to partially track the random symbol phases, approximating standard AM demodulation by rectification. 

Best 73,
Markus


-----Ursprüngliche Mitteilung-----
Von: Markus Vester <markusvester@aol.com>
An: rsgb_lf_group <rsgb_lf_group@blacksheep.org>
Verschickt: Sa, 29. Sept 2018 11:27
Betreff: Re: LF: Signal bandwidth in Op32 LF

Yes. As Andy says, if you are using a divider it should run continuously to preserve carrier phase. All modulation (either XOR for EbNaut-PSK, or on-off-gating for Opera) should be applied after the divider stage.

Best 73,
Markus


-----Ursprüngliche Mitteilung-----
Von: Andy Talbot <andy.g4jnt@gmail.com>
An: LineOne <rsgb_lf_group@blacksheep.org>
Verschickt: Sa, 29. Sept 2018 10:44
Betreff: Re: LF: Signal bandwidth in Op32 LF

If you stop and start a divider, the phase jumps about randomly.  Each time you start the divider, its output signal starts up in a differnet phase to what it would have been if left running

To achieve phase coherent, you must keep the divider running and gate the OUTPUT separately

This issue is a problem with beacons in general that employ FSK keying for their CW ident.   If they adopted on-off keying, then long term integration cold take place though the keying cycle.  But unfortunately, IARU (or whatever organisation that tries to mandate these things), appear to require FSK keying.  SO most beacon keepers comply.   (I don't, the Bell Hill microwave beacons, GB3SCx all use on-off keying)

Andy

www.g4jnt.com

On Sat, 29 Sep 2018 at 09:10, VIGILANT Luis Fernández <luis@vigilant.es> wrote:

Hi LF

I'm puzzled with the readings of bandwidth at the Op32 correlation detection grabbers

As source I'm using a GPS generated signal, which is coherent in phase and works for EbNaut

So I tested this at 20x frequency to drive an small D class PA and modulated in Opera32

The OOK was achieved by enabling/disabling a divider by 10 just after GPS

Sourprisingly, my decodes at DF6NM and DL0AO varied in width, but always showing width of a non coherent signal

2018-09-24 16:55:09 EA5DOM  1500km 137485.051Hz  71mHz -99.9dBOp  78% 17.6dB
2018-09-24 15:35:10 EA5DOM  1500km 137485.026Hz 114mHz -99.9dBOp 100% 17.3dB 

2018-09-24 14:13:18 EA5DOM  1500km 137485.039Hz 164mHz -37.2dBOp 100% 20.7dB

2018-09-27 19:08:06 EA5DOM  1536km 137485.013Hz 110mHz -38.3dBOp  94% 18.8dB

Last night I tried without PA.  Using just a mosfet driver this time getting

about 1W output. The GPS generates the 10x frequency and the OOK is controlled

by enabling/disabling the divider by 10. Got a couple of decodes at DF6NM

2018-09-29 02:17:39 EA5DOM  1500km 137485.007Hz   5mHz -46.4dBOp  85% 15.4dB
2018-09-29 00:37:50 EA5DOM  1500km 137484.986Hz 112mHz -43.2dBOp 100% 19.6dB

So, 5 and 112mHz, for the same signal. I'm thinking that the problem is caused

by glitches when the signal goes off. A synchronous off would be needed to

cut signal at zero crossing ?

Reports to other stations, like SM2DJK, show just 2 to 3mHz

Any ideas ?

PS: Stefan. Your Op32 correlaton detector is not working actually

73 de Luis

EA5DOM