toggle quoted messageShow quoted text
I’ll crack open my copy of Oppenheim and Schafer, and have a read. I think the next step is to test the hypothesis – whether subjectively (e.g., listening) or objectively. The trouble with objective measurement is how to get the test signal to test with. Seeing what the filters do to an impulse would be a good way to compare the filters, but where do you get an RF impulse? Bob
From: Elecraft-K3@groups.io [mailto:Elecraft-K3@groups.io] On Behalf Of K9MA
Sent: Monday, February 15, 2021 1:18 PM
Subject: Re: [Elecraft-K3] K3S Crystal Filtering
All filters have phase shift. The important thing is how that phase shift varies across the passband. Filters with nearly linear phase shift produce minimal distortion of the signal, because linear phase shift is equivalent to a constant delay. If I recall correctly, these are Bessel filters, possibly among others. The trade-off is poorer skirt selectivity for a given order, hence the requirement for higher order filters with more latency.
As Jim points out, DSP filters are just mathematical implementations of hardware filters, though they can do things that would be impractical with hardware filters.
On 2/14/2021 10:49 PM, Jim Brown wrote:
It isn't only hardware filters that have the phase shift -- DSP filters use the math that describes the performance of hardware filters. There is a class of DSP filters that don't have the phase shift problem, but they have greater latency and, I believe, require more processing power.
And because of the interrelationship of phase, frequency, and time, the effect of the phase shift is that the different frequencies that make up a waveform take more or less time get through the filter, so the signal is "smeared" in time.
The same thing happens in transmission lines at low frequencies -- indeed, equalization was developed in the earliest days of telephony (and telegraphy) because both the loss AND the velocity of propagation were changing dramatically at audio frequencies; without equalization, on long distance circuits, the high audio frequencies would arrive long before the lows, so speech would be unintelligible!
73, Jim K9YC
On 2/13/2021 7:24 AM, Bob Novas via groups.io wrote:
After hearing Eric talk about the K4 filtering (https://www.youtube.com/watch?v=OoMoJddbFyI&t=1221s <https://www.youtube.com/watch?v=OoMoJddbFyI&t=1221s>) and the phase shift problems with narrow crystal filters, I wondered if you could run the K3S with wide crystal filters and narrow DSP filters at times when you don’t need tight roofing filtering. And, if you did that, would that improve the situation with respect to muddy audio when listening to highly filtered speech.