In my Lisa 1, the upper drive has always been a better performer than the lower drive. It was always able to read better for starters, and these days if you want to format a disk with BLU, it's pretty much the one drive that does that dependably.
I've been investigating why this might be, and I think I've found the answer now---the lower drive doesn't put as much clamping pressure on the "back" of the disk (that is, the part closest to the front of the computer; or, the part of the disk that goes in last and pops out first) as the upper drive does.
I came to this conclusion after noticing that the lower drive's performance on some of my "reject" homemade Twiggy floppies improved after I opened and reclosed their black plastic enclosures ("envelopes", whatever you want to call them). These hacked-up disks aren't precision-made: the enclosures are scotch-taped back together, and naturally they wind up not being perfectly flat after that. Not coincidentally, the part of the disk that my hack job distorts the most is... the back of the disk. And, the reads and writes that had the most trouble are also the ones that use the head that reads back there.
(As background for people unfamiliar with Twiggy drives, there are two read-write heads for each side of the disk, but they each touch the disk 180 degrees apart. For ASCII art fans, here's what that looks like from the side---imagine you are inside the CRT, looking toward the drive bay:
"Front" head | Lisa
| | | faceplate
\_/ \ / ___|
Disk: ======================================================================== ___
/ \ / \ |
| Spindle | | | |
"Back" head |
)
To test my theory, all I had to do was apply more pressure to the back of the disk (on the enclosure, near the head) while attempting to format the disk with BLU. This has the effect of pressing the media down onto the head with a bit more force. I did this with a small wooden dowel, poking it in through the front faceplate, almost like you would do with a tongue depressor. This trick appeared to work, allowing my bottom drive to format a disk successfully. As a further test, the same trick also allowed the drive to load flawlessly (i.e. with no recalibrations or apparent errors) the entire large 556K "image" provided with the Lisa Smalltalk-80 implementation on Bitsavers---something that didn't really work too well before.
(An aside: it's exciting to run Smalltalk on a Lisa at last! I'll share more about this later. I didn't know it, but Smalltalk of this era apparently loads its entire working environment into RAM in one big go? This one does, anyway, so your drive will need to be able to read much of a disk fairly reliably for this to succeed.)
As nice as all this is, this manual pressure procedure is a delicate operation! If you're careless, you can catch the dowel in the spindle, which can kick it across the top of the disk! I scratched some media that way.
I'm looking now to see if I can identify the mechanism that controls how firmly the disk is pressed against the lower head. I have some ideas. Additionally, there is some open-cell foam that presses against the back of the disk, and while this material looks okay, nobody should trust 37-year-old foam for much of anything. Anyhow: for people struggling with finicky Twiggy drives, maybe you can help things along with a little push...