Gary, Marin, LF-group,
after failing to decode Gary's CMSK8 signal last night, despite is was strong,
I did some indoor sound tests this morning.
I was running the PC in the shack as "transmitter" (on the speaker) and the
notebook in the livingroom as "receiver" (via the built in micrphone).
Setting TX and RX frequency both at 1000Hz and loud volume resulted in a
audible signal in the livingroom (not so much appreciated by the other family
members). There was a strong and clear trace on the CMSK waterfall display, but
only partial decodes (estimated about 70%). Only after some fiddling with the
fine tuning I got a 100% decode.
Once the tuning was optimized I reduced the "TX power" (speaker volume in the
shack) and remained to have good decode far beyond the level of an audible
signal. Quite impressive.
Leaving the audio levels unchanged I switched to ROS1 mode. I could see some
faint traces on the waterfall, but no decodes at all. I had to increase the TX
volume a lot before I got decodes on the notebook.
After that I switched back to CMSK and produces a strong 1000Hz "tune signal"
in the shack. Using SpecLab I measured the frequency at the notebook: 999.2Hz.
It seems that the soundcard samplerates betweens both computers differ 0.8% and
the frequency difference of 0.8Hz was enough to make decodes fail, with weak
signals.
At last I tested CMSK63 the same way I tested CMSK8. Setting the frequency at
1kHz at both computers resulted in an immediate 100% decode. I could go 5Hz off
frequency before decoding failed. Then I reduced the audio level in the shack
until decoding failed (having the both computers on the correct frequency).
This time ROS1 outperformed CMSK63.
My conclusions of this tests:
1. CMSK8 can decode signals at SNR levels where ROS1 fails. It has the
potential to be a very useful mode at 500kHz (and 137kHz).
2. CMSK8 tuning is very critical (must be better than 1Hz). The lack of any AFC
function and the fact that the slightest mouse movement over the waterfall
display causes a frequency shift of several Hz makes tuning very difficult.
3. CMSK63 is far more easy to tune (as expected) and can be a very useful mode
for stronger signals (although still beyond the CW level).
The above conclusions are not meant to discourage the developers of CMSK, I am
impessed by it's possibilities and as I have written some software myself I
know that a lot of labor goes into this.
My respect and gratitude to ZL2AFP and Zl1BPU for this.
But some features such as AFC (no idea wether this is easy to implement) and
the possibility to zoom in at the waterfall display (should be easy to
implement) would make it much easier for the user to tune in at CMSK8 signals.
73, Rik ON7YD - OR7T
________________________________________
Van: [email protected] [[email protected]]
namens Martin Evans [[email protected]]
Verzonden: zaterdag 28 augustus 2010 23:01
Aan: [email protected]
Onderwerp: Re: LF: ros -MF 502 K -on now-
I managed a couple of decodes, Gary, but it's much too critical to tune at
CMSK8
Martin.
#######################
>
> I have to QRT now, I will continue tomorrow night & try faster modes.
> Thanks
> for your comments.
>
> 73,
>
> Gary - G4WGT.
>
>
> On 28 August 2010 21:40, Rik Strobbe <[email protected]> wrote:
>
>> Gary, Martin,
>>
>> I have some other things to do for the next 30 minutes, I will switch to
>> CMSK63 and see what appears on the screen.
>>
>> 73, Rik
>>
>> ________________________________________
>> Van: [email protected] [
>> [email protected]] namens Martin Evans [
>> [email protected]]
>> Verzonden: zaterdag 28 augustus 2010 22:33
>> Aan: [email protected]
>> Onderwerp: Re: LF: ros -MF 502 K -on now-
>>
>> Not just you, Rik - me too, and I guess some others, probably!
>>
>> Tell us how to do it when you find out.
>>
>> Martin GW3UCJ.
>>
>> ###########################
>>
>>
>> I had a good visual signal, but no decode. I'll try to figure out what is
>> wrong.
>>
>> 73, Rik ON7YD - OR7T
>>
>
|
