But if you leave the setting at its nominal, can you hear the difference as that bandwidth is selected? I'm not sure I understand the advantage you are gaining by bringing the crystal filter in later; I'll do if there's something to be gained, it sounds an interesting approach.
David G3UNA David, Say that you have a 500 Hz (nominal) roofing filter with an (actual) 6-dB bandwidth of 550 Hz. When you choose the widest DSP bandwidth at which this roofing filter kicks in, here is what I think about the choices you have: 1. You have it kick in at 550 Hz DSP because it matches the DSP bandwidth. This may create the steepest slopes, but I would avoid it for digital modes since I would expect the group delays to vary near the flanks of the xtal filter, but not near the flanks of the (FIR) DSP filter. 2. You have it kick in at 500 Hz DSP because it is a "500 Hz filter". This logic makes no sense, since the xtal filter is actually 550 Hz. 3. You have it kick in at 500 Hz or less DSP because you want the DSP to cut away the group-delay-varying portion of the xtal filter passband. This would make sense. How much of the xtal filter response would of course depend on the group delay characteristics of the particular xtal filter. Of course if you don't use digital modes, you may want to go with approach #1. 4. You have it kick in at 600 Hz or above. Someone on the list suggested that this approach made the receiver sound more pleasant. The effect would essentially be to disable the DSP for receive purposes, except for the DSP's big improvement in ultimate rejection, and any bandwidth-unrelated DSP function that may be enabled. I have yet to hear for myself, but operating this way makes little sense to me unless there is something wrong with the DSP release. As to my original question, it seems that most respondents like to vary the DSP bandwidth as a means of switching the narrow xtal filter in and out, and to do this without lots of strong closeby signals. If the xtal filter kicks in according to #1 above, the composite bandwith of the two 550-Hz filters will be much less than 550 Hz. This will of course create an exaggerated effect of reducing the noise you hear, and here is an obvious risk of giving the narrow xtal filter way too much credit. If instead the kick-in point is set according to #3, you would reduce the effect of cascading on the overall bandwidth, and results would be more meaningful. However, as has been pointed out, it would be necessary to carefully adjust the gain for the xtal filter. Also, you would have to somehow work around the variations in effective DSP bandwidth step size, which have been stated to differ from the expected 50 Hz steps (at every 200 Hz?). Anyway, what I really wanted to know was NOT how the filters sound with just background hiss or average signals on the band. The justification for the narrow roofing filter would exist only if very strong interfering signal levels get through the standard roofing filter. I am thinking CW, and in case there is only one such signal, the AGC pumping would be easy to recognize. My understanding is that with current production K3's this would happen for an interfering signal level of somewhere around 25 to 30 dB over S9. In my mind, the interesting question is the case where there are multiple interfering signals within the roofing filter passband, including qrn etc. Even though these would not add up coherently, the peak voltage would grow with the number of signals, such that the ADC overload level would be reached without any individual signal reaching the 25 or 30 dB over S9. This suggests that the hardware AGC ideally should have an extremely fast attack time, and I assume that it does. When the hardware AGC responds under these conditions, I am guessing that the effect might be an increase in the general background noise heard. This is really what I was after. -- View this message in context: http://n2.nabble.com/Hearing-the-effect-of-narrower-roofing-filter-tp470635p795233.html Sent from the Elecraft mailing list archive at Nabble.com. _______________________________________________ Elecraft mailing list Post to: Elecraft@mailman.qth.net You must be a subscriber to post to the list. Subscriber Info (Addr. Change, sub, unsub etc.): http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/subscribers.htm Elecraft web page: http://www.elecraft.com
