Hello Chen san
Thanks for your comment.
By MMTTY-ML of JA,
I received the lecture of DSP from Mako-san and friends.
but I did not understand well.
Your comment gives useful hints.
I will try again.
73, Hisami 7L4IOU
>On Feb 16, 2005, at 6:28 AM, 7L4IOU wrote:
>
>> I discussed with Hal JF1PJK and Terry JA7IC
>> about 756PRO Twin Peak Filter.
>
>By the way, I wasn't sure if Don got his "turn the RF gain down" hint
>from Hisami-san, but that is a very good general advice for everybody
>using internal DSP on _any_ rig.
>
>The reason is that when the A/D converter of a DSP unit is saturated,
>you will no longer get the frequency response of the filter that you
>had originally designed it for.
>
>The same thing occurs when using an external A/D such as a sound card.
>But with sound card software, you can include a "VU meter" that gives
>you a clue that the A/D is clipping.
>
>Another comment on dual-peak input filters. If you are already using a
>"matched filter" in the demodulator, using a dual peak filter at the
>input could degrade copy. RITTY for example, uses matched filters.
>The better modems that appeared on the analysis which Alex VE3NEA
>posted earlier are also probably using matched filters.
>
>Using a double peak filter is not the same as using a matched filter.
>It kinda gets you half-way there, i.e., it is better than a flat
>filter.
>
>With a real matched filter, the input of the slicer sees is triangular
>waveform. It is the peak of the triangles that gives the slicer that
>fractional better dB of SNR which gives you copy on the marginal
>signals. I suspect that if you were to look at the slicer of a
>non-matched filter modem, but using a twin peak filter ahead of it, you
>will see some preemphasis at the peaks too, just that it is not a pure
>triangular waveform.
>
>73
>Chen, W7AY
>
>P.S. If you are using PSK31, it is also a good idea to turn the RF
>gain down (or apply attenuation) when there is a loud station in the
>passband. You will be able to notice that the louder stations suddenly
>have better IMD and are clobbering the weaker stations less.
>Basically, IMD from the front-end of your rig. This is also a problem
>with RTTY if there are more than one loud station under the roofing
>filter of the rig -- so the comment applies even if you're using a
>narrow IF filter.
>
>Since the mark and space do not overlap in an RTTY station, there can
>be no IMD from a single station. But during WPX, I had captured the
>spectrum of a loud station that had a small IMD component precisely 170
>Hz away from his mark at a little over 20 dB down. It is possible that
>multipath was causing a mark/space overlap (I have no other explanation
>- a front end filter that has some group delay would also do that but
>mine should, I think, not). Unfortunately, I only saw this spectrum
>afterwards, or I would have recorded the signal itself for a better
>look at the phenomenon.
>
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|