After watching spectrogram on a noise signal and noting a much sharper shape on 50 Hz DS as an audio peak filter, I began to wonder about some of the posts I was seeing asking for APF. What was going on? After some measurements...
1) the 50 Hz DSP bandwidth is *narrower* at the top than any APF I have used. The 100 or 150 Hz DSP is more like it. 2) the 50/100 Hz DSP has *far* steeper skirts than the APF. 3) the APF *sounds to the ear* to drop off a signal more, but that is because the signal as it moves outside the APF is still controlling the IF AGC off the APF center, holding the input to the APF constant. In the 50/100 Hz DSP the AGC opens up the gain when one tunes off the signal, giving the *appearance* of no skirts. Watching the S meter will show what is going on. This effect dissappears entirely with the AGC *off*, where the signal falls off the table on the DSP skirts as one tunes away. If one simply measures the amount of discrimination between two close signals, the 50 Hz DSP can beat the APF by 10, 20 db or more. Listening to two S9 signals only 150 Hz apart, the 50 Hz DSP can completely isolate them. In run situation, the shift and 50 Hz DSP would allow me to pick up just one and then the other by remembering their respective tones, without changing my transmit or RX reference. The SOUND of tuning across a signal with APF in a 500 Hz IF bandwidth is illusory in close signal discrimination, even if it is one's favorite analog sound. Isn't this just dumbing down a superior DSP function to "sound like" an analog result? And reversing myself in one of my earlier posts, the modulation "pops" produced by the sharp skirt drops on 10 Hz shift increments off a strong signal are annoying, and perhaps the shift can be damped just a tiny bit to remove them. -- 73, Guy K2AV ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
