Apple Microdrives used in old iPod products

[legacy]

so, I am a bit confused regarding these devices, since it seems they have a modified firmware.
Some do work on Windows, some do not, some are recognized in half size (e.g. a 4Gb recognized as if it was 2Gb), ... WTF?

I found several iPod with a Seagate ST640211CF microdrive inside. Paid 5 euro for each one, but .. a lot of troubles.

[Towger]

in 2000 I paid an obscene amount of money for a 340MB Microdrive.

[legacy]

Yes, they were expensive, but it's not the point.

What I am shocked is: the firmware of everything > 1Gbyte seems modified by Apple to only properly work with Apple.

See the above Microdrive, there is an Apple logo on the back, and this disk didn't work in a couple of PCs, it didn't' work in a camera, it didn't work with a couple of USB-CF adapters ... I had to try several combinations of tools and adapter before the one in the pic, which did work, but ... it reported 2.Gbyte instead of 4Gbyte, and it also adds something like 20 cycles to the disk_probe() function.

The same disk, the same model, without the Apple logo, not only it did work on all the tried computers, but it correctly reported 4Gbyte!


The same story to Hitachi, and again ... everything > 1Gbyte adds quirks.


So, seriously: WTF?!?!? 

[frozenfrogz]

Can not really tell you much regarding possible modified firmware, however:

Quite some time ago I updated my iPod mini 1st gen. with a 64Gb(?) CF II card. This was an easy operation. Just swapped the microdrive for the CF card and restored the device via iTunes.

I don‘t know if there is something special to the bigger microdrives. I doubt it though. If you can not see the disk on Windows maybe it has to do with Mac formatted media? You could put either a Windows FAT volume or OS X (HFS, exFAT, Ext3 ?? not sure) on them.

Edit: Out of interest, what are you planning on doing with these?

[legacy]

Code: [Select]

CPU:   IBM PowerPC 405GP Rev. E at 266.640 MHz (PLB=66 OPB=33 EBC=33)
       Internal PCI arbiter enabled, PCI async ext clock used
       16 KiB I-Cache 8 KiB D-Cache
Board: Sonoko-node-01
I2C:   ready
DRAM:  128 MiB
Flash: 512 KiB
Net:   ppc_4xx_eth0
IDE:   Bus 0: ...........OK
  Device 0: Model: ST640211CF Firm: 3.07 Ser#: 3ME785YR
            Type: Removable Hard Disk
            Capacity: 3906.0 MB = 3.8 GB (7999488 x 512)

correctly probed and identified.

Code: [Select]

IDE:   Bus 0: ...........OK

This means the microdrive took several cycles before replying to the query (can you identify yourself? ..................... yes).


But look here

Code: [Select]

# ide partitions 0

Partition Map for IDE device 0  --   Partition Type: MAC

Block Size=512, Number of Blocks=4194304, Total Capacity: 2048.0 MB = 2.0 GB
DeviceType=0x0, DeviceId=0x0

   #:                 type name                   length   base       (size)
   1:  Apple_partition_map Apple                      63 @ 1          ( 31k)
   2:      Apple_UNIX_SVR2 kernel                  16384 @ 64         (  8M)
   3:      Apple_UNIX_SVR2 rootfs                 139264 @ 16448      ( 68M)
   4:      Apple_Bootstrap bootstrap                1600 @ 155712     (800k)
   5:           Apple_Free Extra                 4036992 @ 157312     (  1G)

WHT??!?!? 1Gbye of free space?!?!? No, wrong, it's 4Gbyte, it should say 3GB, 1GB is too bad!

Number of Blocks=4194304, Total Capacity: 2048.0

what? wait what? only half of the total number of blocks are shown here, why?!?

There is the same problem even with PC-partitions, and even Linux on a common x86 PC with dd if=/dev/zero of=/dev/sdc failed at writing evrything > 2Gbyte


So this seriously means the whole LBA addressing is somehow partially working, or thay may also means you need extra commands to "unblock" something in the harddrive, since physically it's configured as "True-IDE".


Besides on Linux, this microdrive randomly fails at disk-probe(), and exactly the same beahior is shown by all the other identical microdrives i extracted from iPods.



...So,
WhAT
+ThE!
FrOG?

[legacy]

Edit: Out of interest, what are you planning on doing with these?

One is going to be used in a router, others on embedded nodes.

[legacy]

Code: [Select]

# ide partitions 0

Partition Map for IDE device 0  --   Partition Type: MAC

Block Size=512, Number of Blocks=7999488, Total Capacity: 3906.0 MB = 3.8 GB
DeviceType=0x0, DeviceId=0x0

   #:                 type name                   length   base       (size)
   1:  Apple_partition_map Apple                      63 @ 1          ( 31k)
   2:      Apple_UNIX_SVR2 kernel                  16384 @ 64         (  8M)
   3:      Apple_UNIX_SVR2 rootfs                 139264 @ 16448      ( 68M)
   4:      Apple_Bootstrap bootstrap                1600 @ 155712     (800k)
   5:      Apple_UNIX_SVR2 failsafe kernel         16384 @ 157312     (  8M)
   6:      Apple_UNIX_SVR2 failsafe rootfs        139264 @ 173696     ( 68M)
   7:      Apple_Bootstrap bootstrap env            1600 @ 312960     (800k)
   8:      Apple_UNIX_SVR2 my                    7684928 @ 314560     (  3G)


The same model, without the Apple logo on the back: it works perfectly in the same conditions, with the same firmware, etc 

[james_s]

I got some of those years ago as curiosities. I don't recall if they were Apple but they were from some kind of media player. I had to use a special formatting tool to properly utilize the full capacity, unfortunately I do not recall the details. It's possible that it is not the firmware though.

[legacy]

Formatting is performed by accessing blocks, and it's exactly here where we have a low-level problem because I cannot physically access any LBA block of 512byte for no logical address block beyond half the capacity.

Microdrives can operate in several modes { PCMCIA, MemoryBlock, TrueIDE }

You have to wire a pin to Gnd to force TrueIDE, and I have already checked: the pin is grounded!

So it must be a command issued somewhere during the initialization, in fact, the Apple version of the Seagate Microdrive doesn't correctly respond to the disk_probe function until the firmware issues several commands requiring a soft-reset, whereas the non-Apple version does respond correctly, and immediately.

Therefore, I think during these steps, the iPod does something special via software.



Anyway, I give up, I have already ordered qty=10 Hitachi Microdrive of 4Gbyte each. I know they work as expected. The only negative note: I paid much more money for them, but at least I am sure they work.

[Halcyon]

Edit: Out of interest, what are you planning on doing with these?

One is going to be used in a router, others on embedded nodes.

Can I just ask, why not grab yourself some SLC or MLC CF Cards? They are going to be far more reliable than old microdrives.

[RoGeorge]

Check if your Compact Flash reader supports bigger than 2GB cards.
Some card reader models were limited at max 2GB.

[legacy]

Check if your Compact Flash reader supports bigger than 2GB cards.
Some card reader models were limited at max 2GB.

The same CF-usb reader offers two different behaviors:
- with nonApple version of the Seagate Microdrive, it's able to handle up to 4GB
- with Apple nonApple version of the Seagate Microdrive, it's unable to handle more than 2GB

This, plus other issues, like disk_probe() failures.

[brucehoult]

in 2000 I paid an obscene amount of money for a 340MB Microdrive.

In 2002 by far the cheapest way to buy a 5 GB 1.8" hard drive was to buy it inside an iPod!

And the iPod was a perfectly good FireWire external hard drive, just as it was, case, battery, screen and all.

[legacy]

In 2002 by far the cheapest way to buy a 5 GB 1.8" hard drive was to buy it inside an iPod!

Dunno, but there must be a reason for the Apple logo on the back of the unit, and I am experimenting a lot of quirks.

I am developing my own firmware over u-boot and I can confirm the weird behavior experimented over the usb-CF adapter. It might workaround it by adding special care into the function:

Code: [Select]

int ide_preinit (void)
{
        /*
         */
        while(ide_do_some_voodoo())
        {       
                udelay (10000);
        }
        return 0;
}

But I am not willing to spend more time at reverse engineering this.

It's clear to me that at some point Apple locked down firmware to discourage people from buying the cheaper portable drive and removing the microdrive inside for camera usage, and every attempt to force the microdrive into PCMCIA-ATA mode always fails.

It means that later batches of the Microdrive were patched, and this is also the proof that there are differences between the original firmware made by Seagate, and what Apple pushed into Seagate's firmware.

Even OEM Hitachi 6GB Microdrive come with serious restrictions and does not even work in PCMCIA-ATA mode, but at least they work in True-IDE mode without any quirk.

Therefore I am not surprised to see quirks even in True-IDE mode. They must have pushed some special software for handling something special they had.


Moral of the story:

• WD60WP-32LPV0 (not OEM), 6GB: good drive, no quirks, it doesn't need ide_preinit (), and PCMCIA-ATA mode works
• HMS360404D5CF00 (not OEM), 4GB: good drive, no quirks, it doesn't need ide_preinit (), and PCMCIA-ATA mode works
• ST640211CF (OEM/Apple), 4GB: it needs ide_preinit(), but LBA only works over half the addressable blocks, and PCMCIA-ATA mode does not work
• HMS360606D5CF00 (OEM/Apple), 6GB: it needs ide_preinit(), but LBA only works over half the addressable blocks, and PCMCIA-ATA mode does not work
• HMS360606D5CF00 (OEM), 6GB: it needs ide_preinit(), LBA perfectly works, but  PCMCIA-ATA mode does not work
• HMS361008M5CE00, 8GB: cannot be used, it comes with a proprietary ZIF-ATA connector
• ST612712DE, 12GB: cannot be used, it comes with a proprietary ZIF-ATA connector

HMS360404D5CF00 (not OEM) is the best one ever over tested units! Yesterday night, I ordered qty=10 of them. Paid 40 euro each.

[Gyro]

Creative used the same drives (no Apple logo, but "For embedded application only"). I can't remember how I did it now, but I did manage to get one to format as a FAT32 drive using an external USB-IDE adapter and IDE-CF adapter board. It must have been some combination of Fdisk and computer management but at least it was possible.

EDIT: (ST650211CF 5GB)

[tooki]

I wouldn’t be surprised if the microdrive vendors made such special firmware a condition of the better pricing they gave OEMs for embedded drives, so as to protect the higher retail prices for the drives.

I’m kinda puzzled as to why you’d bother with them today, though, since flash memory is so much more reliable, and dirt cheap. A 4GB SD card costs practically nothing now.

[legacy]

I’m kinda puzzled as to why you’d bother with them today, though, since flash memory is so much more reliable, and dirt cheap.

Reliable? Of course, CF/Flash does not suffer any mechanical vibration shock, but the Microdrive offers 10^12 write cycles, while a CF/Flash offers only 10^7 in the best case. CF/Flash are cheap, but not more reliable in these terms.

[tooki]

Hard drives fail, and as you said, more susceptible to shock damage. Magnetic media offers more write cycles in theory, but that's only if the drive mechanics don't fail prior. And unless you're using Flash as swap memory due to insufficient RAM, there are very few applications where you're going to be rewriting a sector 10 million times!

Hard drives may offer (theoretically) better endurance, but definitely not better overall reliability. They're not the same thing!

Anyhow, you said they'd be in a router and some embedded nodes. Those are surely applications that don't need extended write cycles, right? (Remember also that the larger the disk, compared to the storage needs, the fewer cycles will be used. E.g. if you're writing 100MB of logs a day, on a 16GB disk it'll take 4 times as long for a block to be reused compared to a 4GB disk. So just oversize a cheap Flash card if need be. And you can get extended-endurance Flash, by the way!)

[legacy]

Hard drives fail, and as you said, more susceptible to shock damage. Magnetic media offers more write cycles in theory

In practice, not in theory.

but that's only if the drive mechanics don't fail prior. And unless you're using Flash as swap memory due to insufficient RAM, there are very few applications where you're going to be rewriting a sector 10 million times!

Dude, I have estimated the live cycle of the product, Flash is not appropriate for my application. And that's all.

[james_s]

I have a hard time believing that a vintage microdrive is going to be more reliable than high quality modern flash, although they certainly are neat devices. A real marvel of mechanical engineering and precision construction.

[Sal Ammoniac]

Anyone make a CF card that uses FRAM rather than FLASH?

[james_s]

Not that I've seen, and certainly not in capacities approaching microdrives. CF in itself is a rather ancient standard, while larger sizes did exist I never had one larger than 256MB.

[legacy]

Anyone make a CF card that uses FRAM rather than FLASH?

In avionics, we use FeRam as "non volatile ram". But the size is something like 8Mbyte (each ticket is of 128Kbyte), and it's used to log each failure at startup or during a mission. It has MILSTD spec, and it's damn expensive.

In automotive, in FOM and MotoGP, we use a large chunk (common technology, just high-premium spec) of flash-drives on mini ePCI, but this only due to vibrations, and only because hard drives are recycled (trashed) after 10-20 competitions.

They log data concerning IMUs, how each of the four tires is consumed on the runway with GPS data, fuel data, thermal data, ..... a big tensor of data, to be then downloaded and processed.

They have money, a lot of money, and these things do not have to run disk-intensive applications like the need I have for other kind of my customers, who want Erlang VMs (yes, it runs on the router), which eat up to 512Mbyte while the ram of the router is limited to 64Mbyte, hence I had to workarounded because the hardware of the router cannot be changed, and the ram is soldered and cannot be expanded.

I had a few words with engineers, and probably, the next generation of these products will use FeRam disks on SoC able to address much more ram.



Anyway, my Hitachi Microdrives arrived today, in the late morning. I spent part of the afternoon testing them, 100% tests passed, and they work as expected.

[LeonR]

I have a hard time believing that a vintage microdrive is going to be more reliable than high quality modern flash

Can you stop this? This topis is NOT about *opinons*, but rather about finding Microdrives that DO work as True-Ide.

I saw you've already managed to find thd HDDs for your project, but have you checked the microSD interface converters to mini IDE? A lot of them work in TrueIDE mode, some even have jumpers that allow mode configuration.

[Monkeh]

I have a hard time believing that a vintage microdrive is going to be more reliable than high quality modern flash

Can you stop this? This topis is NOT about *opinons*, but rather about finding Microdrives that DO work as True-Ide.

I saw you've already managed to find thd HDDs for your project, but have you checked the microSD interface converters to mini IDE? A lot of them work in TrueIDE mode, some even have jumpers that allow mode configuration.

But no flash memory is ever going to be as reliable as his second-hand mechanical data deathtraps, he's already explained that.