Quoting Dave Blaschke <[EMAIL PROTECTED]>, on Wed 23 Apr 2008 09:33:37 PM PDT:
> At 03:08 4/24/2008, you wrote: > > . .......I'm not a CW kind of guy, so I don't recall the details of > the problem), because >> that allows you to get a rhythm. (after all, absolute delay isn't a >> huge deal, because there's already 10s of milliseconds delay in the RF >> propagation to the other station). >> >> >> Jim, W6RMK >> > > > Jim, I think your leading comment just summed it up. I AM a long time > CW operator, CW contester, and CW DX chaser. I KNOW what good QSK is, > and what it sounds like. Just to probe a bit from a different viewpoint, so maybe some light will go on in a software designers head.. when you say "sounds like" are you referring to what you (the sender) hears or what you hear from the other side? SDR-5000 + PowerSDR do not yet deliver it. > Latency is not the issue (well, maybe a little). Smooth and rapid > transition from T to R is the issue and the shortcoming. Is it the transition from T to R.. i.e. an analog radio, even with a T/R relay, might have some saturation or blocking, and it will come out of saturation non-instantaneously (affected by the one or two AGC loops typical). Not only that, but it's not the same for all frequencies in the passband. For instance, most variable gain amplifiers have a fair amount of change in phase shift with gain setting, so as the gain of the receiver chain adjusts, there's a simultaneous phase shift going on, which changes the relative phases of the audio frequencies. The current SDR implementations tend to "hard switch", with any softening probably due to downstream bandpass filtering in the (digital) audio chain. Not to mention that the noise reduction processing might have some unusual or idiosyncratic sounds. Maybe what's needed is some shaping of the receive signal during the transition, to soften the attack? If there is a > way to overcome it, I am receptive to learning how. I'll believe it > when I hear it in my QSK operating. > From a theoretical standpoint, it should be possible, so we have the classic problem of a behavior that hasn't been well described analytically, so it's tough to duplicate, but you'd know it when you heard it. Rob Sherwood has also commented on how the AGC of modern radios that use a lot of DSP in the back end doesn't provide the same subjective quality that traditional radios provide. I suspect that this is in the same class of problems. There's some psychoacoustics here that are difficult to characterize. I think that in both cases (your QSK and Rob's AGC) it's not the linearity of the receiver/transmitter chains necessarily, but the saturation and gain behavior. One should also not discount the fact that people almost always prefer the "sound" that they're used to: i.e. the comments about CDs sounding harsh compared to analog LPs, even though the CD is technically a more accurate representation. It is subjective after all, and all the skilled CW ops I know have distinct preferences that they've arrived at over years of use. And, which they would have a very tough time describing with any precision, other than by examples or the same sort of process as trying to describe subtle aspects of wines. A similar problem is faced by folks designing DSP guitar amps and trying to find mathematical representations of, for instance, a Marshall stack with 2 4 driver cabinets. This is one of the more interesting (and frustrating) aspects of software radios. You know that the creation medium is so incredibly flexible, if you could only figure out how to do it. Jim, W6RMK _______________________________________________ FlexRadio Systems Mailing List FlexRadio@flex-radio.biz http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz Archives: http://www.mail-archive.com/flexradio%40flex-radio.biz/ Knowledge Base: http://kb.flex-radio.com/ Homepage: http://www.flex-radio.com/