Writing Mac 800kB disks

[Eudimorphodon]

My gut suspicion is the flux reader issues people complain about are more likely to be the software that has to interpret the transitions getting confused by the inconsistent/non-standard flux intervals than anything in the drive's hardware. (I mean, sure, I guess it's possible some drives have some kind of clocked data separator in them to "amplify" standard MFM flux transitions that fails with non-standard width/interval pulses, but I've never seen a datasheet saying that.) But I don't have a flux reader or any experience with them so, yeah, no idea. Apparently some people believe it's a big enough problem that they need "special" flux readers that interface to Apple drives.

 

[SomeGuy]

It's definitely something in the drive. I've never gotten down to the signal level, but reading the same Macintosh disk with a Kryolfux or similar device will have vastly different results depending on the drive. I even have one that lets me override the density notch - in one density mode outer tracks are unreadable and inner tracks are readable. in the other, inner tracks are unreadable and outer tracks are readable.

Then a drive that works, works fine regardless if a Kryoflux, SCP (with PCE tools since SCP has no decoder), or DOB/Tanscopy card is used.

Since there seems to be some confusion, the Central Point Deluxe Option Board/Transcopy board is a flux-level device. In that respect it is similar to the Kryoflux, Super Card, FluxEngine, Greaseweazle, and so on. That means it is in the same boat regarding floppy drive compatiblity.

All of these are significantly different in operation than the Macintosh's Internal Wozniak Machine chip. Mainly because they do not change the rotation speed of the drive, instead changing the communication data rate they they use. Also because this is mostly done using software instead of a dedicated microcontroller chip.

So, yea, that means at a very low level what is read/written with a PC floppy drive is not 100% the same as it would be if it were read/written in a real Apple variable bitrate drive. Usually it is close enough to work, but not always.

 

[njroadfan]

Central Point called out Alps, Mitsubishi, Mitsumi, and Panasonic 1.44MB drives for being particularly troublesome with Macintosh 800k GCR disks.

ftp://ftp.oldskool.org/pub/OptionBo...kage Contents/Documentation/OB_Drive_Note.pdf

Citizen, TEAC, and Toshiba drives were recommended.

Ironically Apple started using Mitsubishi drives in late model PowerPC machines......and even they had problems with GCR formatted disks!

For flux reading and writing of 400/800k drives, there is a new option of using the AppleSauce device and a real Apple 3.5" drive. Sadly its not available at the moment.

 

[Chuck(G)]

It's a matter of the bandpass filters on the drive, mostly. That's why, for example, you can't read a HD floppy that was written in a DD drive without covering the density indicator hole. And why you can't read a DD floppy written in HD (some old PS/2s could do this) in a regular drive that pays attention to the indicator hole without making a hole in the jacket. The rotational speed is the same in either case--and the drive doesn't tell the FDC which type of media is inserted.

Also note that Amigas handled HD media by dropping the spindle speed--pretty much akin to the old 5.25" Welltec drives that could handle HD media on an XT by dropping the spindle speed to 180 RPM on an otherwise standard 96 tpi (720K) drive).

You can vary the spindle speed or the datarate interchangeably, provided you pay attention to the bandpass/frequency response of the read/write channel. The big difference in the two approaches is that the read signal level is proportional to the linear speed of the media, so a fixed-speed drive may actually be superior to a SuperDrive in that respect.

On he old F85, we used a ~~300KHz clock to write GCR to a DD 5.25" floppy. This put the top end slightly above the filter on current 100 tpi drives, so the drives were ordered with a "tweak" to the electronics, which usually involved little more than a capacitor change.

Why Apple decided to stick with the IWM for handling floppies is a mystery to me. It made the drives more expensive and the media non-interchangeable with established industry standards. Traditional vendors were already putting 800KB on 3.5" DS2D media using plain old 250KHz-clocked MFM. Must have been a vanity thing.

 

[Al Kossow]

Must have been a vanity thing.

Backwards compatibility of media.
The IWM was really painful to program, SWIM got rid of the worst and they finally added DMA
in SWIM III because I told them soft modems wouldn't work if they didn't add it. Not that it
mattered, GeoPort was a big dud in the market.
GCR should have been dropped with the change to PowerPC. No new software was being
sold on 800K disks by then, but people still wanted a way to read/write their old disks.
You could still do this well into the G3 era. It only died with OS X

The IIfx actually added separate 6502s for floppy and scc support, but no platform other than
the Quadra 950 ever used that again and it broke too much software that expected to be able to
bang the swim and scc directly.

What should have been done was an external 400/800K GCR floppy for people who cared.

The other thing I thought of is SWIM was integrated into the I/O ASICs early on, so it cost them
nothing to keep using what they already had. The Quadra AV's had an expensive NEC custom
floppy controller that got ripped out in the cost reductions done for all later machines.

 

[Eudimorphodon]

Apple had a massive, raging case of NIH Syndrome in the early 80's. Remember, the 3.5" Microfloppy drive only ended up in the Mac at all because of Apple finally admitting at the last minute that Fileware, aka, Twiggy, had been a total waste of time and they were never going to get it to work acceptably, but even that realization didn't stop them from insisting that some of the technical dead-ends from that project (mainly the multi-speed drive motor idea) be incorporated into the Sony mechanism.

(Apple shuttered their catastrophically ill-fated hard disk division at about the same time, the one that, so far as I know, resulted in the "Widget" drive some Lisa models are cursed with as the only fully-shipped product.)

Maybe at the time the whole affair looked like it made sense, but in the rear-view mirror it comes across an awful lot like all the over-the-top praise Apple got for the "low parts count" Disk II controller went to someone's head and stuck there a little too hard.

(I have to admit I'm young enough that at the time I never really got it why the Disk II was such a big deal. Around the time it came out fully discrete "board full of chips" disk controllers were being replaced by much more integrated devices like the WD1771/1791, and Apple's solution *always* came with the trade-off of requiring the CPU to do a lot more work anyway, so, yeah, but apparently it made Apple management decide they were destined to be the One True God of Magnetic Storage.)

Since there seems to be some confusion, the Central Point Deluxe Option Board/Transcopy board is a flux-level device. In that respect it is similar to the Kryoflux, Super Card, FluxEngine, Greaseweazle, and so on. That means it is in the same boat regarding floppy drive compatiblity.

To be clear I knew the DOB was a flux device, or at least I assumed it was, but I never owned one and on the forums today where people complain it's about modern devices. I didn't know that there were specific recommendations at the time regarding floppy compatibility for the DOB, if there were that certainly supports the idea that there could be a real issue there.

The original Apple 400/800k drives actually spun at between 400-ish and 600-ish RPM, almost twice as fast as a standard 300RPM fixed speed drive, so perhaps that has some bearing on making disks written in them harder to read. (Per Chuck's comment regarding response being proportional to media speed.)

 

[krebizfan]

The original Sony Microfloppy had its own variable speed motor so incorporating that in Apple systems cost nothing.

The Disk II margins were amazing. A drive without the expensive logic board controlled by the CPU; can't get much cheaper than that. Alas, fewer components was no match for the power of cost reduction through mass production and in a few years, Apple was left in an expensive cul-de-sac.

 

[Eudimorphodon]

The original Sony Microfloppy had its own variable speed motor so incorporating that in Apple systems cost nothing.

According to this datasheet for an early Sony mechanism it did have a 600 RPM motor (they intended it to be used with 500kbs controllers, not the 5.25 standard 250kbs) but... maybe I'm missing it, but I don't see anything in there about the drive either automatically changing speeds based on track (which is what the 800k Mac drives did) nor do I see a PWM signal on the connector pinout (how the original 400k drive worked.) I don't see anything about PWM or speed control in the older 1982 manual on Bitsavers either(*). Citation needed?

(* Actually, if I'm reading the '82 manual correctly that version didn't have a motor-on signal at all, it spun all the time when a disk was inserted like some 8" drives do?)

And, yeah, technically almost any drive with a direct-drive motor uses PWM for speed control so you could argue that *any* drive is "variable speed" if you just hack it, but the point that Apple insisted on customizing the drive in a non-standard way remains. In addition to the speed control thing the Apple drive uses a *completely* alien way of sending commands to the drive in place of the normal standards, a method directly derived from the Twiggy instead of the direct stepper-motor control of the Disk II.

 

[rayarachelian]

Just for the record: I can write 800k DiskCopy images to a 1.44MB disk using WinImage and a USB floppy drive and my Mac Classic will boot from them, too. Doesn't say much about whether or not a Mac with a real 800k drive can boot from them. Don't own one for testing, sadly.

So there's two things at play here:

1. the physical disk format for 1.44M floppies is MFM - i.e. for PCs. On Macs this is normally GCR.
2. the file system itself is either MFS or HFS, or HFS+ for much newer macos systems.

So what you've done there is to place either an MFS or HFS file system onto a MFM formatted media. The file system itself fits within the 1.44M media because it's smaller and appears to be 800k, but this is because the file system itself contains data that says "this is 800k" or "this is 400k" in size.

However, if you place such a disk in a classic mac which doesn't have a superdrive, it won't be able to use it even though the file system is on the media, because those drives require GCR encoding on the physical format.

Similarly:

a. You could also boot an ancient Linux or one of the BSDs on a 68030 mac and use mkfs to place a file system on an 800k GCR formatted floppy. However, if you boot up MacOS, it won't understand the file system although it could read the physical sectors from the media, and so it will offer to erase it. This is equivalent to what you've done with WinImage.

b. You could have potentially placed a FAT32 or FAT12 or FAT16 file system on an 800k GCR formatted floppy sector by sector, and if the appropriate extensions are installed on MacOS, it will be able to use it. But no PC will be able to make use of it without additional specialized hardware and drivers (such as kryoflux, etc.)

These last 2 things are analogous to using WinImage to write an MFS/HFS disk image to an MFM formatted floppy.

If your classic mac has a superdrive it *might* be possible for MacOS to boot off it, if the ROM Toolbox supports MFM, but that's a big IF.

 

[tingo]

People might be interested to know that I've just merged in successful write support for those horrible 800kB disks to FluxEngine, my USB FDD interface project (tl;dr: it's essentially a very cheap and open source DIY Kryoflux or SuperCardPlus replacement). I can download a System 6 DiskCopy 4.2 image from MacintoshGarden, write it to a floppy, and then boot from it.

Nice!

If anyone's got one of the 800kB-only machines like a Mac Plus, I'd be really interested to know if it works there. I have very limited hardware for testing (one beaten-up PowerBook 150).

I've only got a Macintosh SE/30 - no 800kB drive unfortunately.