I actually tried that. I even went so far as to limit the component output to 480i (standard resolution). Unfortunately this didn't work either. So I don't think the copy protection circuit is "smart" enough to allow down-rezed taping. From what I've been reading, MacroVision signals (at least the earlier versions) are always present, it's just the AGC circuits that are different in recording devices vs. displays.

Another experiment I tried was to insert a "Video Stabilizer" device in the "Y" signal line. This was not a workable solution (and I didn't really expect it to be) as it introduced it's own set of distortions, but it did seem to curb the video gain problems. This little exercise really helped me narrow my list of possible culprits.

FWIW My cable provider is TimeWarner and the Scientific Atlanta DVR uses the Pioneer Passport Echo firmware.