I found it quite noticeable last night too, I'm 150 miles or so North of Ron's QTH. In my experience, the marker on ARISSatTLM was moving down rapidly as the satellite approached. After TCA (which was at about 68 degrees elevation, I believe) it reversed back to "normal". I was using the latest keps that I obtained from Space Track yesterday afternoon. I did find my PC clock to be off (ahead) by just less than one second, too. For what it's worth. My thoughts given what you all have written, would be that errors in all three (my location, my clock, and the keps) as Alan mentioned, probably contributed to it. On a typical linear satellite pass with a 2 meter downlink, I have never noticed it. I imagine that the effect is amplified in a perceptive sense, by sitting and watching the signal on a graph with pretty good resolution and not being distracted by copying the CW or working a voice QSO.
73, Jerry N0JY
Ron,
Alan's points are spot on. That being said---I see something similar to what you are seeing and "blame" the Keps :)
In the first few days, the manual freq. correction at TCA was over 1,000Hz (seems like I had to shift it down, so -1,000) as compared to AOS and LOS...which were "on frequency."
It has gotten much better over time. It was off -500Hz, and now it's under -100Hz at TCA--so my thought is that the Kep elements are to blame, not a drifting satellite transmitter, etc.
73!
Mark N8MH