Real floppy drives typically vary +/-10us, and that is a lot to me. I have several 3.5" drives that I use that vary only 2-3us on every read. I deal with things on a 25ns interval, so perhaps I am picky.
This may depend of the drive and probably on the floppy disk inserted too. Generally manufacturer specify something around 1.5%/3%.http://www.millennium.ab.ca/products/fl ... _Drive.pdf
Anyway the speed variation of this SCP is not due to the drive, but more to the STX format itself. This format is clearly not designed to regenerate a flux stream or a floppy disk…
I got some difficulties some years ago when i have done the first reverse engineering of this format. Some tracks in these file contains only the sector data (all the track must be create), while others contains the data coming from a READTRACK command to the WD1772 (data NOT at the MFM level of course…). Unfortunately all information (data, timing…) in these files are related to the WD1772. And half of the disk data are missing : the clock stream. This make the disk recreation somewhat tricky. And i think that some STX with track-based disks are almost impossible to recreate properly : the missing clocks information to resync the track in the real world are … missing from the file
. So lot’s of thing are done by guessing from this format… And the timings may vary depending of the tracks…
Another SCP example generated from ST files :http://hxc2001.free.fr/vrac/Odd_Stuff_2 ... One_st.zip
This time the timings are a lot better