It's flabbergasting how good Woz's designs were. Almost on a whim, he with the Disk II did something no one anywhere in Silicon Valley—anywhere in the world—was doing. Forget about IBM, HP, Shugart, Tandon. Just within Commodore and Tandy, Apple's direct 1977 competitors, there were abundant human and engineering resources to come up with a fast, inexpensive, and reliable floppy drive and controller; Chuck Peddle at Commodore was certainly no average engineer. And yet, Commodore was still unable to do this in 1984.
Whether one believes in the reality of the existence of the "10X developer", it's hard not to see what Woz did between 1976 and 1978—Integer BASIC, Apple II color graphics, and Disk II—as proof that such a being can exist, even if (as I have written elsewhere) that brilliance straddled the line between optimized and overoptimized. <https://news.ycombinator.com/item?id=41685888>
I disagree that the approaches are similar. The 4040 <https://en.wikipedia.org/wiki/Commodore_4040> is a monstrosity; even the later single-drive models, such as the 1541, are massive. Apple's 1978 floppy drive + Disk II card takes up less space than 1985's 1571 drive (and still significantly faster).
>The only significant difference is that the 6502 (actually two of them) is in a separate enclosure from the Pet , communicating via IEEE-488.
Many things are possible when another 6502 is used just for the drives! That Commodore takes this approach is, as I said, no credit to its army of engineers versus one Berkeley dropout.
>Since Commodore manufactured the 6502 presumably it was ok to use them liberally.
I acknowledge that, had Apple been the owner of MOS and manufactured 6502s, it might also have been tempted to take the easy way out designwise and built Commodore-style drives, or implement the Disk II with a 6502 on it. But I'd like to think that Woz would have done the "right" thing regardless of available resources.
Even though the Disk ][ was significantly faster than the 4040, the 4040 was significantly faster than the 1541. Apparently the difference was largely due to a bug in the controller used for the Commodore serial bus.
Commodore's 8-bit micros also used IEEE-488 (or a serialized form of IEEE-488) for their floppy drives, rather than a dedicated expansion card that was connected to a bus with direct access to the CPU and RAM. While the expansion bus on the Apple II was fast enough to control the drive directly, the most Commodore could do was send and receive a stream of data to the drive. The drive had to have the smarts to interpret that stream of data. The drive electronics was going to be more complex than Apple's even if Commodore assigned amazing engineers to the task.
Finally, Commodore's approach had its own advantages. Since the drive accepted and handled commands, the CPU could be freed up for other tasks. I don't know how useful that was in general, but the classic example involved two drives autonomously copying floppies. (You could literally remove the cable between the computer and the drives after the operation was setup.) I also recall hearing about classroom setups where multiple computers would share a drive (PET era). It also allowed more drives to be connected to the same bus. Apparently 8 drives were supported. In contrast, the Apple II would support two drives per expansion slot.
... which is still significantly slower than the Disk II.
The 1978 Disk II is about 30 times faster than the unaccelerated 1982 Commodore 1540/1541, as well as more reliable,[1] cooler-running,[2] and just more elegantly designed. The same comparison holds for the 4040 and other pre-1541 Commodore drives, albeit with the speed disadvantage vis-a-vis Apple being less.[3]
[1] The 1541's drive head often goes out alignment despite both it and Disk II using similar "ratatatat" methods to return to the first track. Also, the first couple of years of 1541 production were notorious for failures in general.
[2] As the 1541 is a standalone 6502-based computer with an integrated power supply, overheating is constantly an issue.
[3] Yes, the 1541 is slower than earlier Commodore drives, because of the move from parallel to serial IEEE-488 and a hardware bug (that both necessitated and allows for Fast Load and its counterparts)
But the Disk II card was 8 chips you could get from Radio Shack, where the 8050 was a monster with a whole CPU/memory/bus (you can see the board appear in this video at 1:15 https://www.youtube.com/watch?v=S3d2cNSAB9A&t=69s).
And Woz's was faster. And hit market almost two years earlier. There's an aesthetic judgement to be made here too, and... it's not remotely close.
https://josephoswald.nfshost.com/circuit-sim/woz-machine-hw-...
Also Magic-1, another homebrew CPU, also made out of 74LS TTL chips from the 1970s.
Fantastic trend, really.
https://www.bigmessowires.com/bmow1/
(Of course it's a lot easier [but still quite fun] to make your own homebrew CPU/system on an FPGA.)
It's also a great early example of the massive win you can get by replacing hardware with software (and "software" -- in the form of a state transition table encoded in a small ROM).
It's also one of the reasons there were so many fascinating and weird copy protections for Apple II software: since so much of the behavior was in software on the computer, it was malleable. (Since it uses the CPU for tight timing loops, the Apple II couldn't really do much else while using the disk.) The write-ups by 4am on IA are fun reading if you're into this kind of thing: https://archive.org/details/apple_ii_library_4am
There are some fun projects to record disks at the level of magnetic flux transitions. I'm mostly familiar with https://applesaucefdc.com by the amazing John Keoni Morris, which came with a new file format too, and some lovely UI software.
The copy-protection stuff was completely puzzling back when it mattered, but of course makes complete sense now.
I hadn't seen the applesaucefdc.com stuff, that's great.