Chen:
Where does cocoaModem stand among these 2 demodulators in relation to this
discussion?
73, John N1JM
On Feb 16, 2013, at 4:08 PM, Kok Chen <chen@mac.com> wrote:
> On Feb 16, 2013, at 1:37 PM, Bill Turner wrote:
>
>> Can someone explain why slower is better for 2Tone? Is this true only for
>> 2Tone or for other decoders as well?
>
> It is true under selective fading, for all demodulators that implements a
> good ATC circuit.
>
> For an overview of ATC, see these two articles:
>
> http://w7ay.net/site/Technical/RTTY%20Demodulators/Contents/am.html
> http://www.w7ay.net/site/Technical/ATC/index.html
>
> I don't know what MMTTY uses for ATC (or even if it has such a circuit), so I
> can't speak for it, but both fldigi and 2Tone use pretty much state of the
> art ATC.
>
> Both 2Tone and fldigi have some form of squaring circuits before the slicer.
> As the second of the two above articles show, the squaring can either be
> applied to the signal itself (what Leonard Kahn had used for diversity
> reception) or to the ATC control voltage, as I had done. I think fldigi lets
> you select between both methods, and 2Tone squares the Mark and Space signal.
>
> The reason Kahn's work (for diversity reception) applies to RTTY is because
> when you use an AM type detector, the Mark and Space tones can be considered
> as independent (but complementary) tones which operates under frequency
> diversity.
>
> Both 2Tone and fldigi use AM detectors instead of FM detectors. For a
> distinction between AM and FM RTTY detectors, see here:
>
> http://w7ay.net/site/Technical/RTTY%20Demodulators/Contents/am.html
>
> The Allnatt, Jones and Laws (researchers with the British Post Office
> Research Station) paper that I mentioned in the web article mentions that
> they should have used a squarer, but did not because of the complexity (the
> original 1957 paper actually references Kahn's 1954 paper). Today, in
> software, that is half a line of code :-).
>
> I mention all this because none of this squaring stuff to do ATC is really
> new -- merely forgotten by hams, or they could not justify the additional
> hardware cost.
>
> A good ATC depends on precise knowledge of the signal envelope, and this is
> where I believe 2Tone will prove superior to the current fldigi.
>
> You definitely don't want the receivers' AGC to interfere with the modem's
> ATC when a good ATC is better at adjusting the threshold than some arbitrary
> AGC circuit that was not designed for RTTY reception. Just like you don't
> want a receiver's filter to interfere with the modem's filter, if the modem's
> filter is already optimal.
>
> I usually run with no AGC, and just use the RF gain control manually to back
> off the receiver gain when an extra large signal arrives. Beyond that, I
> just use a high dynamic range sound card to send a very high dynamic range
> audio signal to the modem.
>
> The only time you will want some receiver AGC is when the signal undergoes
> periodic, deep "flat fading" where both Mark and Space fades at the same
> time, and the fading depth exceeds the dynamic range of your receiver. But
> even there, you would still want to use very slow AGC (as G3YYD recommends)
> so that the AGC activity does not affect the ATC circuit. If the AGC is slow
> enough, the ATC will adapt to it. If the AGC is a typical
> fast-charge-slow-discharge AGC, or worse still, a hang AGC that suddenly
> drops, the ATC envelope tracker will have more problems.
>
> Not to be flippant, but if fast AGC helps copy, the modem needs a better ATC
> circuit.
>
> Oh, you do not, do not, do not, want to use fast AGC with a fluttered signal.
> Turn off AGC (and if there is a software button, even turn off ATC) when you
> hear flutter and set the RF gain control to keep the receiver and sound card
> from overloading. And if you really must use AGC, choose the slowest AGC and
> without hang, for flutter conditions.
>
> AGC is also useful before a limiter that has insufficient dynamic range when
> you use an FM type detector. But that is a different story since 2Tone does
> not use FM. But even there, you really don't want to use AGC that is much
> faster than a hundred milliseconds.
>
> 73
> Chen, W7AY
>
>
>
>
>
> _______________________________________________
> RTTY mailing list
> RTTY@contesting.com
> http://lists.contesting.com/mailman/listinfo/rtty
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|