dpz@pilot.njin.net (David P. Zimmerman) (09/05/88)
Let me tell you a story about a PC clone. My Ultra-Comp 10 MHz PC clone with 3.5" drive kept doing reseeks, but not consistently. It would only happen on reads, and then really weirdly, like only during the last 40 tracks of a backup, or somewhat into a disk sector verify. Well, after much theory and thought, and a swap of this and that, I found that replacing my floppy controller with a true blue IBM floppy controller fixed my problems. Now, this can mean one of two things: either my machine is too fast for the floppy controller, or the floppy controller is too fast for my PC. I suspect my PC was throwing requests too quickly for the controller to handle, for two reason. My controller has a Zilog chip at its heart, while the true blue has a NEC. NEC chips are generally faster than "standard" chips like the Zilog. Also, the true blue controller had a faster clock crystal, for whatever that would mean. I believe my controller has a 10MHz crystal, while the true blue one has a 16Mhz one. If the true blue one isn't divided to a lower effective clock than mine, there's a second answer. The moral of this story is BE CAREFUL if you are designing an IBM clone, PLEASE! Something like this isn't immediately obvious when you would be speed testing the various parts. David -- David P. Zimmerman, the Rutgers Dorm Networking Project, the UUCP Project, etc dpz@dorm.rutgers.edu rutgers!dpz zimmerman@zodiac.bitnet