Hey, wait a moment, I'm also having some of those issues with minor data-corruption. However, it can also be signal-noise so I'm not sure...
ok, let's get these issues hammered out. they are unacceptable.
1st, use an 80 pin cable. the extra ground wires should help with any outside noise being introduced. they shouldn't be required on something this slow, but we need to eliminate all the variables we can.
2nd, do the same test I outlined for mike. Copy some data on a known good controller and then compare it on the XT. Keep a log of some of the corrupt data, and maybe we can find there's a pattern to it (like bit 5 is always low when the data fails) and we can hopefully start to focus on where the problem is.
Here's some rules to follow:
* if there's a byte/bit or two that have failed out of a sector, it is not going to be a BIOS issue.
* if a bit is failing repeatedly, like bit 5 is always low in a data miscompare, then it is very likely to be a soldering issue.
* if entire sectors are corrupt, then it could very well be a BIOS issue.
* if it's totally random, like a byte is duplicated when it shouldn't be, or there's no pattern to failing bits, then it could be an IC timing issue between parts. This is especially true if you do the same compare test again, and it fails in a different location. Something like that will likely need a logic analyzer to figure out. (i can do this at my work)
We also need to figure out if the read is failing or the write. THat's why I'm suggesting you write the data on a modern machine and compare it on the XTIDE, and vice versa.
Please log your results, including the model # of the drive on the wiki debugging logs, unless mike b. has any other suggestions.