I took a slightly different approach. I set two of the filters for acceptable passband ripple and attenuation and let the bandwidth fall where it may. I did this for CW, but a similar approach would work for data.
Since the actual bandwidth and the indicated bandwidth of the variable filter can differ quite a bit, I did not see the point of picking an arbitrary value for the bandwidth. Rather I set the widest filter where the ripple started to look fairly flat and the narrowest filter where filter loss became more noticeable. My filters are centred on 589 Hz which is the pitch of the sidetone when set to 580 Hz. My sidetone circuit has been modified considerably, and that may account for the 9 Hz difference between the setting and the measured result. An audio spectrum analyzer and wideband noise source are required to see the passband of the filters. So here's the result: XFIL 1 - OP1 XFIL 2 - Widest bandwidth where ripple begins to look acceptable. This is subjective as to what you are willing to tolerate with respect to ripple. My filter ended up at 970 Hz bandwidth indicated and 512 Hz measured at the -6 dB points XFIL 3 - Set about half way between XFIL 2 and XFIL 4. Nice smooth passband and no attenuation. Indicated bandwidth was 600 Hz, with actual -6 dB bandwidth of 307 Hz. XFIL 4 - Set at the point where the attenuation began to be quite noticeable. Indicated bandwidth of 200 Hz and actual measured bandwidth of 178 Hz. Of course there will be a bit of variation in the actual bandwidth vs the indicated bandwidth between individual K2s due to component tolerances and construction differences. 73, Darrell VA7TO K2 #5093 On Sunday 25 November 2007 17:03, Don Wilhelm wrote: > There is a lot of ripple in the variable filter passband when it is set > to a wide filter value. Even with the ripple, the filter is quite > usable on SSB voice, but for data modes, it does have its shortcomings. > > That is one good reason that I set up the RTTY filter set for use on > data modes. I typically set the FL1 position the same as SSB FL1 (with > the OP1 SSB filter), but I set relatively narrow widths for RTTY FL2, 3 > and 4 - I like to use 1.00 for FL2, 0.70 for FL3 and 0.40 for FL4. I > center these filters at 1000 Hz (or 800 if the BFO range is not adequate). > > 73, > Don W3FPR > > Stephen Prior wrote: > > Running spectrogram with random noise at the antenna (not much else on > > 12m, four hours after dark!), I can see a distinct ripple in the passband > > of all filter widths, in every mode. Except, that the OP1 filter in the > > ssb board gives a flat response in all modes. So it's something to do I > > guess with the setup of the variable crystal filter. I am wondering how > > far back I will need to go in the setting up of the K2 to find the > > adjustment which will perhaps flatten the shape. > > > > Using spectrogram (latest) with the xtal filter in usb at 2.1kHz, there > > is about 10dB between peak and trough of the ripple. The peaks are > > roughly 600Hz apart. I'd never noticed this in casual listening, I doubt > > that I could have done. Most of the time I am on cw and the 700Hz filter > > setting shows no obvious ripple anyway. It does however certainly not > > look right on the waterfall on psk31! -- Darrell Bellerive Amateur Radio Stations VA7TO and VE7CLA Grand Forks, British Columbia, Canada _______________________________________________ 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