Notes From The Nerd Cave

Miscellaneous Electrical and Cyber Engineering Notes

Useful information which seems absent, rare, or is not as well consolidated or refined in presentation elsewhere.

16 April 2024:  More additions and modest refinements in the Restore Macintosh Mirror Door Drive AcBel API1PC36 by upgrading electrolytic capacitors and Enable Silicon Image equipped SATA cards in Classic or OS 10 PCI or higher based Mac sections.

Full privacy respect:  This site uses no cookies and acquires no visitor information.  And VPNs are welcome.

Efficiently store components of all kinds in a single set of six decade range containers.

Updated 23 February 2023:  Tiny refinement, but images still not yet added.

This is for hardware engineers who maintain stocks of small components for prototype or small quantity circuit fabrication.

We utilize a large range of numerous type of circuit components, all of which must be organized into containers for ready access or restocking.  If this is done in the most obvious manner, with a separate container for every component of every kind, the number of containers required is mammoth, rendering that method entirely impractical.  There's a far better way:

It's trivially easy to visually discern different component types irrespective of physical size - resistors, capacitors, inductors, zener diodes, fuses, crystals, ICs, varistors, and several other component types are immediately distinguishable on sight.  So since they differ in appearance, whether surface mount or leaded type, they can share a container yet cause no identification problems.

In addition, components of a specific type but different value are almost always easy to visually distinguish from each other if their values differ by six orders of magnitude.  Resistors are an exception but they're almost universally well marked so their actual values are easy to read.

This means that all these components may be stored in a single array of 72 containers, yet any component of any kind or value can be almost instantly retrieved or restocked.  Here's how to configure it:

Acquire cabinets which can be configured to provide 12 columns by 6 rows of containers.  Four column by six row container cabinets are common, three of which provide the necessary 12 x 6 matrix.  This provides 12 columns to delineate every standard 20%  component value (1.0, 1.2, 1.5, 1.8, 2.2, 2.7, 3.3, 3.9, 4.7, 5.6, 6.8, and 8.2), and six rows to delineate six decade ranges (from unity through 10 M, or 10 µto unity for example).  The decade ranges don't need to be bounded, but for most engineers 1 f (10 ^-15) through 100 G (100 x 10⁹) covers almost every discrete component value available.

The top left container holds components with values from 1.0 to but not including 1.2 with p, µ, none, or M suffixes.  The next container right holds values from 1.2 to but not including 1.5 with p, µ, none, or M suffixes.  The container just below the top left  holds values from 10 times 1.0 to but not including 1.2 with p, µ, none, or M suffixes.

So all containers hold values with six decades of difference.  For example the top left container holds 1 Ω and 1 MΩ resistors and 1 pF, 1 µF, and 1 F capacitors (the last a super capacitor), and other components whose values similarly align.  This means the container holds a broad mix of components of any kind any of which may have p, µ, none, or M suffixes.  Such differences are utterly easy to discern by eye in almost all cases, except that some resistors might require reading of their body markings.  All others are visually obvious - 1 µF capacitors of any kind are physically larger than 1 pF capacitors for example.

Delineation is by specified values with zero hedging.  So for example a small bag of resistors might have a specified value of 4.699 KΩ, and include some samples with actual values of 4.700 KΩ or a bit higher.  But they are placed in the 3.9 K container, not the 4.7 K container.  Delineation must remain absolutely strict to avoid chaos.

Images of my containers should be shown below soon.  They contain any small component of any kind which has a primary value.  Zener diodes by zener voltage, precision voltage reference ICs by reference voltage, fuses or circuit breakers by breaking current, variable resistors by maximum resistance, crystals by frequency, lamps by operating voltage, or anything else with a primary value for which the component is selected for use in a circuit, with all other parameters secondary.

An option is an array of 144 containers in a 24 x 6 matrix to delineate every standard 10% component value.  However that delineation is too fine for the vast majority of individual engineers - a 72 container array is more practical for us in my estimation.

This system works exceptionally well!  Instinct is to establish container arrays for resistors, separate container arrays for capacitors, other arrays for inductors, more arrays for zener diodes, more for fuses, and many more.  And a separate container for each individual value covering the entire range of the component (no six decade rotation).  And separate arrays of containers for every component type, such as one for tantalum capacitors, another for MLC surface mount capacitors, another for electrolytic capacitors, and another three arrays for leaded versions for all those capacitors.  That level of delineation is necessary in production environments, but is so utterly impractical for nerd cave environments that it's almost never actually achieved or maintained.  So chaos ensues, and very often engineers order new components because, although they're quite sure they already have them somewhere in their stock chaos, they can't find them.  And they rarely restock small components because they can't reasonably find a proper container to store them within.  So they're often simply discarded.  The overt waste is bad enough, but additional serious problems are long delays to component availability, which delays projects, and wasteful new order management time.

I created my first consolidated container system a bit over 40 years ago and am still happily using the same physical containers in their original matrix with original labels today.  I retrieve very nearly every small component I need almost instantly with no delay at all.  Sometimes I must check a value, but that's quite rare (maybe a few times a year) and in most cases the exercise merely confirms my initial judgment, though for certain components (usually zener diodes), when markings aren't clear I must measure to discern the precise value within the 20% container range.

Skeptical?  Well, I get it - the notion of mixing components so extensively seems guaranteed to create chaos.  But in fact the reverse occurs - this container array delineation method renders every component perfectly easy to store and extremely easy to retrieve - it works.  And it's immensely more space and time efficient than any other system - space efficient for the obvious reason, and time efficient because every component's stored or retrieved from a single array of containers right at your finger tips - there's no need to walk to a remote location to store or retrieve components in distant containers.

Try it - you'll see - it works!

Superb quality hand soldering using organic flux.

Posted 23 February 2023.

Consider water soluble organic flux instead of rosin flux.  It renders truly perfect joints and enables easy fine pitch IC soldering by simply swiping each row of pins with a ball of molten solder on your soldering iron tip.  The organic flux keeps the solder completely clean so it retains full surface tension like pure mercury, so the molten ball jumps cleanly from one pin to another leaving (usually) no shorts.  But for the last three or four pins a jump off platform is needed, such as copper braid or just a thin strip of solid copper or other solder compliant metal, or adjacent empty feed through holes often suffice.  Organic flux is available as a liquid which may be applied generously with a small art brush.  Fully immerse the working area - use plenty of it.  And use either solid solder or solder cored with the same organic flux (such as Kester 331 solder) - don't use rosin core solder.

It washes off completely with mere hot water, though I use detergent and a toothbrush as well to insure any other contaminants are also removed.  You must wash it off reasonably promptly - the flux is both corrosive and conductive so it must be completely removed.  You can work for about 90 minutes but then should wash since longer linger times result in metal surface tarnish (not serious but not as gorgeous).  You must clean your work area and all tools as well since any residue would cause corrosion and conduction trouble for anything which contacts them.

The work area cleaning overhead is a bit inconvenient, but I use organic flux for all significant soldering work because it's so superbly easy to work with, enables ordinary soldering irons to solder fine pitch easily, and the results are perfect mirror finish joints which are utterly solid and reliable.

It's available from the usual solder vendors - Kester's 2331 flux is described here for example.  AimSolder.com calls it OAJ flux.  You'll need the liquid even if you acquire organic core solder because that solder alone delivers far too little fluid for any work in my experience, especially fine pitch work.  (I doubt it was actually intended to be used alone in most cases.)

It's cheap.  And in my estimation once you experience it you'll never solder circuit boards with rosin flux again.

Storage volume target mode via SCSI ports in any Classic Macintosh OS.

17 January 2024:  Information about the MacsBug utility to cause an interrupt added.

Most well informed Mac OS 10 and higher cyber system users are familiar with storage volume target mode.  It's invoked by pressing the 'Command' and 'T' keys simultaneously immediately after power is turned on, then holding them down until a FireWire symbol appears on a display.  The system's internal volumes may then be accessed by another system by simply connecting the two systems with a FireWire cable.  The volumes in the system in target mode should mount automatically on the desktop of the second system, where they should be fully accessible in the normal manner.  Such access can be useful in a variety of situations, including bypassing file privilege restrictions which exist on the target system's volumes when booted normally.

As described here this can also be accomplished in SCSI based classic Mac systems, most of which run Mac OS 6 through 9, in a similar manner, but by interrupting the target system instead of using a 'Command' and 'T' mode boot, and using SCSI cables instead of a FireWire cable.  It's reasonably easy:  Interrupt the target system by pressing an interrupt button or using the MacsBug utility (or less elegantly intentionally cause a system crash).  Connect its external SCSI port to the external SCSI port of the second system.  Then simply boot the second system, which should mount the target system's volumes on its desktop, where they should be fully accessible in the normal manner.

However all devices on the chain must be set to unique IDs, a likely substantial inconvenience if both systems are equipped with only a single SCSI bus.  But if the internal and external SCSI buses of one of the systems is independent, such as in the PowerMac 9500, no SCSI conflict will occur (so long as any external peripheral devices are disconnected during target mode, or they have unique SCSI ID numbers).  Other independent SCSI ports such as Adaptec PCI SCSI cards provide the same advantage.

A means to interconnect the external SCSI connectors is required.  Two DB-15 to 50 pin Centronics cables and one male to male Centronics adapter might be the most common solution, but any combination of cables and adapters which enable mating of the two external SCSI ports with sufficient signal quality should be functional.

My sense is that no special sequence of operations is required, but the safest sequence might be as described above.

I suspect this method will successfully share SATA devices as well via eSATA ports, eliminating SCSI cable awkwardness, and perhaps SCSI ID conflicts, if both systems are equipped with Mac compatible eSATA PCI cards as described below.  I don't know whether the SCSI devices would be mounted as well however, but hope to explore these elements by about Summer 2024.

Enable Silicon Image equipped SATA cards in Classic or OS 10 PCI or higher based Macs.

Big kudos to "Collin aka DosDude1" (also here) and all other noble contributors for making this superb storage bus option more accessible!

16 April 2024:  Modest refinements.  But "FlashROM" and SilG3114101UpDriveR1 and SilG351210UpDriveR1 firmware installer information and the QuickSliver sleep issue remain incomplete.  Future updates are likely too as progress ensues.  Information from others highly appreciated of course.

Other than networking, eSATA is the best option for exchanging data between classic Macintosh computers with a PCI bus and any later Macintosh.  It's fastest, seems most solidly reliable, functions in all systems with an accessible PCI or later bus, and is easy to use with modern systems as well.  (For systems which have no eSATA port readily available SATA to USB bridge cases or cables may be used, though some speed might be lost.)

For SCSI systems with NuBus or no accessible system bus, SCSI Target Mode as described in the section above or sharing SD or µSD storage cards used in Blue SCSI or SCSI2SD seem to be the most practical options, though a slower and more expensive Floppy EMU is an alternative.  But for all others eSATA is clearly best.

SATA also provides strikingly superior data storage performance compared to SCSI, Blue SCSI, SCSI2SD, and PATA (aka Parallel ATA or IDE).  So in PCI equipped Macs SCSI or PATA buses can be completely abandoned.  (The internal connection to an external SCSI port can be retained so SCSI peripheral devices can be used however.)

And large data transfers between two storage devices managed by the same SATA PCI card are remarkably fast because the actual data isn't constrained by the main system bus or processor but rather flows only through the card's bridge IC, eliminating almost all system and system bus bottlenecks.  (Some command and control data flows through the system bus of course, but generally file data flows only through the SATA bridge IC, which is far faster.)  For example a backup of an internal SSD to an independent external SSD in which both are connected to the same PCI SATA card is wicked fast - far faster than an old SCSI or PATA based system in native configuration could achieve.

SATA Connectivity is provided by a variety of PCI cards but few if any are natively compatible with Macintosh systems.  However the firmware of cards with certain Silicon Image SATALink bridge ICs can be revised to render them fully functional in all PCI (and presumably higher) based Macs, except limited to Mac OS 10 or higher in many (but not all) cases.  Then internal or external, or both, SATA connected storage, including SSDs, may be utilized, including as boot volumes.

Firmware for the popular Silicon Image Sil3112 bridge IC may be installed into the SATA card's PROM as the card resides in a PCI slot in a Macintosh.  No hardware modification is necessary and the firmware installer application is very simple and intuitive so the process is superbly quick, easy, and problem free.  And the result is a SATA card which is fully functional and bootable under any Mac OS from about 7.6.1 to 10.5.8 †.

Firmware installation for the other Silicon Image bridge ICs is more complex and my information is merely tentative until I acquire personal experience with FlashROM probably running under Linux / Debian (32 bit CD ISO version suitable for G4 and lower G series Macs linked).  Also I've not confirmed the size of the firmware nor the capacities of the PROMs usually found on these PCI cards in each case.  I hope to provide more definitive information soon.

Silicon Image SilG3114101UpDriveR1 and SilG351210UpDriveR1 firmware installers include firmware, a Terminal script based installer, and instructions.  The instructions state that the installer software can only run under Mac OS 10.1 but in my experience they both function properly up to at least OS 10.5.8.  However evidently they can only upgrade an earlier version of Macintosh compatible firmware - they seem unable to overwrite alien (such as Windows compatible) firmware or empty PROMs.  I doubt Silicon Image will provide any further Macintosh firmware updates for these bridge ICs so these installers seem to be of no practical use, so I don't list them as options below.  If they were modified to eliminate all overwrite barriers they'd become very convenient and efficient for our needs.  However I'm not aware of any effort to do so.

Troubleshooting, purist's options, and other notes are described further below.

This subject is discussed at TinkerDifferent.com (and the previous and later pages, but the linked pages provide key information and file links) and at 68KMLA.org.

Firmware and firmware installers are available for these Silicon Image SATALink based PCI cards:

Sil3112:  Fully functional with all PCI bus compatible Mac OS versions (about 7.6.1 to 10.5.8 †).  The firmware installer requires Mac OS 9.2.2 or lower to possibly 7.6.1.  No hardware modifications are required.

Sil3114:  Fully functional but only in Mac OS 10.0 to 10.5.8 †.  Evidently FlashROM, perhaps best run under Linux / Debian, can install the Silicon Image Sil3114 PROM firmware, which seems to require a mere 1 Mb PROM.  Or use an external programmer such as a Model T48.

Sil3124:  Functionality and firmware installer limits not yet studied.  Evidently FlashROM, perhaps best run under Linux / Debian, can install the Silicon Image Sil3124 PROM firmware, which seems to require a 4 Mb PROM.  Or use an external programmer such as a Model T48.

Sil3512:  Fully functional but only in Mac OS 10.0 to 10.5.8 †.  Evidently FlashROM, perhaps best run under Linux / Debian, can install the Silicon Image Sil3512 PROM firmware, which seems to require a mere 1 Mb PROM.  Or use an external programmer such as a Model T48.

(HighPoint HPT372:  Functionality and firmware installer limits not known.  I doubt a means to render this IC as Macintosh compatible will be developed.)

Note:  † Mac OS 10.6 and 10.7 are likely also compatible in pre 2008 Intel processor based Mac Pro models equipped with a PCI-X system bus.

Click to download firmware installers for the following Silicon Image SATALink based PCI cards:

Sil3112:  SeriTek1S2 firmware installer 5.3.1 (or here or here)
Sil3114:  FlashROM.
Sil3124:  FlashROM.
Sil3512:  FlashROM.

Click to download the following firmware from Collin's site:

Sil3112:  Silicon Image Sil3112-1S2 patched compressed PROM firmware
Sil3114:  Silicon Image Sil3114 PROM firmware
Sil3124:  Silicon Image Sil3124 PROM firmware
Sil3512:  Silicon Image Sil3512 PROM firmware

  The easiest implementation also covers the broadest Mac OS range and requires no hardware modifications.  It applies only to a Silicon Image Sil3112 based SATA card:

1.  Download Collin's SeriTek1S2 firmware installer 5.3.1 and Sil3112-1S2 patched compressed PROM firmware.

2.  Move those files to a PCI bus equipped Macintosh running classic Mac OS from about 7.6.1 to 9.2.2 natively.  (The Mirror Door Drive with 400 MHz FireWire is the last such Macintosh.)

3.  Insert a Silicon Image Sil3112 based SATA card into a PCI slot.

4.  Boot the system in OS 9.2.2 or lower possibly down to OS 7.6.1.

5.  Launch the SeriTek1S2 firmware installer 5.3.1.  Follow the simple prompts to install the Sil3112-1S2 patched compressed PROM firmware.

6.  Test functionality.  There should be no flaws from about Mac OS 7.6.1 to Mac OS 10.5.8 †.  If not flawless consider these system limitations or the troubleshooting options further below:

Mac OS 9 HFS+ volumes must not be journaled, must have classic drivers, and if used as a boot volume must not exceed 190 GB.  Smaller volume size limits apply in some lower Mac OSs.

Mac OS 10 or higher boot volumes can be larger, can be journaled, and don't require OS 9 or lower drivers.

Evidently firmware can be installed in the other Silicon Image based cards with FlashROM.  But FlashROM's probably most reliable when run under Linux, with Linux / Debian perhaps the best option.  So unless already available you'll need to prepare a Debian or other Linux boot volume in a suitable PCI bus equipped computer for these Silicon Image cards.

Optionally firmware can be installed with an external programmer such as a Model T48, but this requires that you remove the PROM, program it in the external programmer, then solder it back onto the PCI board, a laborious task.

Sil3112 based:  Move R25 to R24
Sil3114 based:  Move R3 to R4

However a rare few PROMs require 12 V programming and thus firmware can only be installed by an external programmer, so those PROMs must be removed from the card.  They could then be programmed and soldered back onto the card but it's much better to replace them with a 5 V or 3.3 V PROM if the card provides a feasible means to utilize one or both of those voltages.  (I've not studied such feasibility.)

Troubleshooting, purist's options, and other notes:

Troubleshooting:

Mac OS 9.2.2 and lower can't boot from volumes greater that 190 GB.  Mac OS 8 and lower may be more restricted.  So partition boot volumes within the system's storage capacity constraints.  Higher limits usually apply to non-booting volumes.

Volume journaling might cause dysfunctions in Mac OS 9.2.2 and lower, so consider removing journaling from volumes used with those OS versions.

Specific PCI slot or other PCI card sensitivities might exist.  So if a dysfunction's encountered it might be necessary to utilize a different PCI slot or revise PCI card stacking order.

Inexpensive green Sil3112 and Sil3114 based SATA PCI cards from China provide no eSATA connection nor provisions to add one, a shame.  But one can be fabricated by milling an eSATA receptacle slot and drilling two round bolt holes for a panel mounted eSATA to SATA cable, then install the cable, completing an external eSATA receptacle, just as Robin-Fo illustrates here.

Target mode:  In my experience thus far interconnection of eSATA ports can't provide target mode access to volumes in a second system such as is possible with chain capable buses such as FireWire and SCSI.  All my attempts to accomplish eSATA target mode to date have failed irrespective of normal or interrupted system status.  My guess is that a system in interrupt mode can't transfer data through its PCI bus.  And if in normal mode I suspect data can't transfer out of the SATA side of a PCI-SATA bridge then directly into the SATA side of a second PCI-SATA bridge - I doubt any transfer protocol for that exists in SATA, whereas it's supported in chain capable buses.

Purist option:  Native SATA cards are often equipped with a mere 1 Mb PROM.  For Sil3112 based SATA cards there's no need to replace it because Collin's cool Sil3112-1S2 patched compressed PROM firmware is compressed to fit and it automatically self extracts when the system boots, with no known bugs.

But purists who prefer the cleanest minimal software achievable and thus prefer direct uncompressed firmware must replace the original 1 Mb PROM with a 4 Mb PROM such as for example the AM29LV040B I use, a 3.3 V component.  The voltage selection shorting bar (a 0 Ω resistor) must be placed in the correct position for the voltage of your PROM (often R25 for 5 V, and R24 for 3.3 V *).  Then the uncompressed version of Collin's Sil3112 firmware can be installed.   (I also replace the life limited electrolytic capacitors with immensely longer lived solid tantalum capacitors.)

Currently I can't find Collin's uncompressed Sil3112 firmware on his site, but this, from my archives, includes "1S2_512.rom" which might be it.  (That package also includes "FlashROM.exe", described above, and "CWSDPMI.exe", which is a mystery to me, though perhaps it's a PROM programmer application for Windows or Linux to facilitate installation of firmware via a non-Apple PCI bus.)

* Actual component designations vary, but the correct ones are often easy to find:  Look near the PROM for a three terminal regulator IC, usually in a mini power tab package under the PROM, then quite near that look for a 'resistor' marked "0" or "000", which is a 0 Ω 'resistor', meaning it's just a shorting bar.  Also look for two unoccupied pads in the same general area which are spaced for the same size 'resistor'.  To change the PROM's V+ voltage move the 0 Ω 'resistor' from its original location to the empty location, leaving the original location empty and open.  Alternately remove the 0 Ω 'resistor' completely, then short the originally empty location with a solder bridge, which is probably easier if you use a soldering iron for this work.

Bridge swapping:  If you wish to consider swapping bridge ICs be aware that even though they may have identical package profiles they aren't necessarily interchangeable.  For example the Silicon Image Sil3512 or HighPoint HPT372NLF utilize utterly different pin function layouts relative to earlier bridge ICs, so a Sil3112 replacement IC for example would render such cards completely dysfunctional in any environment.

Reports of any missing or flawed information, acknowledgment lapses, or turf encroachment very highly appreciated!

BlueSCSI storage device preparation:  A clear, elegant, and reliable method.

More information is here.  BlueSCSI device vendors are here.

I believe all this information is correct, but as of 17 May 2023 this is still rather new work which needs repeated verification to insure thorough accuracy.

This is the easiest and most reliable process I've been able to devise for preparing BlueSCSI storage media for use in classic Macintosh systems which operate in 32 bit address mode.  I present it here because I couldn't find a clearly described method which seemed functional elsewhere.  Seven to nine steps are involved but the process is conceptually rather simple and once accomplished should be easy to remember and implement.  To wit:

1.  Attach the storage device you wish to prepare, such as a µSD card, to a system running Mac OS 10 (OS 11 not explored).

2.   Launch Terminal.  Then list all storage devices and their volumes:

diskutil list

  Note the correct device identifier (diskX) for the storage device you wish to prepare:

3.   Initialize it in ExFat format with a master boot record scheme:

diskutil eraseDisk ExFat MyDevice MBRFormat /dev/diskX

Where "MyDevice" is a name for the storage device which meets BlueSCSI name conventions, for example HD10_512.
And "diskX" must be replaced with the correct device identifier (for example disk9, but be certain to select the correct one to avoid erasing other media).

(FAT32 might be more universally compatible but per BlueSCSI Wiki information ExFat is much faster, no compatibility issues are mentioned there, and I encountered none, so I recommend using ExFat.  But if a problem arises consider downgrading to FAT32.)

4.   Move the Terminal command target to "MyDevice":

cd /Volumes/MyDevice

5.   Create one to six image files in "MyDevice".  (These will be hda or iso image files rather than an img or dmg files.  The following refers to creation of hda image files.  For iso replace hda with iso.):

dd if=/dev/zero of=ImageName.hda bs=1m count=TotalCount.

Where "ImageName" is a name for the storage volume which meets BlueSCSI name conventions, for example HD10_512.  Multiple image files must have different SCSI ID numbers.  For example for three volumes HD10_512 Volume 1, HD20_512 Volume 2, and HD30_512 Volume 3.  Text between the block size and .hda is ignored by BlueSCSI, and none is required.  For example just HD10_512, HD20_512, and HD30_512 is fine.

And where "TotalCount" is the image file size desired divided by the block size (bs).  The math and notation hygiene must be correct* but here's a simple approach for bs=1m math:

The total size of all the image files created evidently must be significantly less than the maximum possible, in part because once in a BlueSCSI device the Mac OS must add data to the storage device during initialization.  (However I intend to study this further, so stand by please.)  If so set the sum of all the file sizes to a lower value than the storage device's capacity.  98% of total capacity might be reasonably safe, but 95% is evidently generally recommended, though at the cost of wasted capacity.  (I've used 97% successfully but will likely increase that after studying actual partition consumption patterns later.)

Find your storage device's total capacity in bytes in Disk Utility or the Finder's device information window.  Multiply that number by your selected trim figure, for example .98 for 98%.  Divide that result by 1,048,576 to calculate the sum of all your "TotalCount" figures, then proportion that figure among however many image files you create as you prefer.

*In the Mac OS Terminal bs=1m specifies the input and output block sizes in binary notation as 1,048,576 bytes, but evidently bs=1M specifies them in decimal notation as 1,000,000 bytes.  Be aware of such differences when calculating your total count figure.  For example if bs=1m is specified and a total image file size of 128 GiB is desired (128 x 1,024³, which equals 137,438,953,472 bytes), set "TotalCount" to 131072 (137,438,953,472 / 1,048,576).*

For large capacity media this process takes a long time since a file size of nearly the full capacity of the media must be written to the media.  (Hopefully a method which swiftly creates a large but data bereft image file will be developed later.)

When complete check the root of "MyDisk" to insure a single file named "ImageName" resides there:

ls

6.  Dismount the storage device, physically remove it, then physically insert it into a BlueSCSi device which is attached to your classic Macintosh.

7.   For just one volume, launch GFormat†.  Then initialize the image on your storage device in HFS+ or HFS format (see notes about which below).  Apply any name you wish (no BlueSCSI convention is required for this name).  When initialization is complete the image can then be mounted on the desktop, or it might mount automatically.

However GFormat won't recognize two or more images, so if you created multiple images launch FWB Hard Disk Toolkit, which does recognize multiple image files, then initialize each of them.  Then launch GFormat to reinitialize any you wish to be formatted in HFS+ format.

8.   Optional:  Launch FWB Hard Disk Toolkit, then highlight each BlueSCSi storage volume, then select "Update Driver" to install the FWB Hard Disk Toolkit driver.  When all are installed restart the Macintosh.

9.   Optional:  Check the volume with Disk First Aid, then Norton Disk Doctor (skip the media check option in Norton Disk Doctor unless media flaws are suspected).

The volume should now be ready for normal use.

In my experience this process creates flawless BlueSCSI media which functions in HFS+ format.  However my experience thus far is limited to PowerPC based Macintosh systems.  I hope to explore Mac SE, Mac Plus, and other models later then post information about them.  More about HFS+ with 680x0 systems a bit below.

† In my experience other format utilities don't seem to provide selectable format scheme options and often format in HFS, which in my opinion should never be used in any system which supports HFS+ (such as Mac OS 8.1 or higher) because it's often dysfunctional due to its serious inherent limitations.  GFormat is the only classic format tool I'm aware of which provides a clear and elegant means to select between HFS or HFS+ formatting options, which is very important since any HFS+ capable Mac should use it (otherwise immense capacity waste and some management errors will occur).  GFormat is very clear and efficient in other respects as well, and at 20,218 bytes it's of minimal size.  However I'm ignorant of the merits of the drivers it installs, though I've experienced no problems with them.  Nonetheless I know more about Hard Disk Toolkit drivers and respect them, thus my preference to install them after GFormat completes its HFS+ initialization.

As stated here:  "Mac OS 8.1 update (only from 8.0, adds HFS+ support, last to support any 680x0 Macs, and 680x0 Macs cannot boot from HFS+ volumes...)".  I've used BlueSCSI only for PowerPC Macs thus far so I have no experience with 680x0 Macs yet.  But my plan for 680x0 Macs is to try to create multiple image files on a single µSD storage card, allocating one to serve as a small HFS storage device which will hold the Mac OS and boot the system, and the rest to serve as 2 GB HFS+ storage devices for all other uses.  The hope is that 680x0 systems can manage that combination efficiently and reliably.  However ChatGPT advises that 680x0 Mac ROMs don't provide HFS+ drivers, and that third party drivers may be dysfunctional.  But I'll try at least HDT drivers and whatever drivers GFormat installs, and later a modernized ROM (MacSIMM $35 or lower, or ROMinator $45) for an SE/30 and Mac IIx.  I'll revise this section as results develop.

I posted GFormat on MacintoshRepository.org because I could find it nowhere else.  There's a GFormat reference here and it seems to be a format utility as well but the creator name's different.  If my post of GFormat violates the wishes of Allen Gainsford, the creator, my apologies and I'll remove it promptly upon request, or if he prefers, assist with contribution or payment notifications.  If anyone else has further information or another download source for GFormat please advise me or the BlueSCSI development community.

Replace corrosion time bomb clock batteries with far safer super capacitors.

Diode part number correction and minor refinements 23 February 2023.

Separate personal computer clock batteries are often devastatingly destructive time bombs.  They can be replaced with super capacitors which are far less likely to leak and presumably cause far less damage if they do.  (I've never experienced a leak failure of a super capacitor so I can't be certain how much damage would result, but feel confident it would be far less than battery leak damage.)

A super capacitor won't maintain the system clock (RTC, Real Time Clock) for nearly as long as a battery, but super capacitor values of about 2 Farads seem sufficient to maintain the clock for at least a month, which is more than sufficient in most circumstances.  In my view that's a clearly preferably trade for ridding the system of a devastating internal corrosion time bomb.

You'll need a 5.5 Vdc or higher rated super capacitor, a low leakage signal diode such as an ordinary 1N4152, and a very roughly 50 Ω to 100 Ω resistor.  The diode and resistor provide very low stress charge current for the super capacitor and prevent discharge through this path when the computer is shut down.

Remove the battery and its holder, the later a desoldering task.  Then solder the super capacitor to the circuit board pads vacated by the battery holder, insuring proper polarity.  Then solder the diode and resister in a series connection from a nearby + 5 V power location (such pads are usually quite close) to the positive pad of the super capacitor, with the cathode of the diode oriented toward the super capacitor so current can flow from the + 5 Vdc source into the super capacitor, but not out.

An option is to remove the battery but retain the battery holder, then solder the super capacitor to its terminals.  However some battery terminals may not reasonable accept solder.  (And since the super capacitor is a very long life component I consider the battery holder as useless.)

Super capacitors are low in cost and readily available from multiple distributors such as AliExpress.com.

Modern portable devices usually utilize their internal user inaccessible whole system battery to power their clock as well and thus have no clock battery to replace.  But line powered computers or portable devices with user removable main batteries usually have an internal clock battery which is a wicked corrosion time bomb, and thus a candidate for a super capacitor replacement.  I replace them in all computers I value because careful though we all try to be, perfectly disciplined maintenance of internal clock batteries is far more a nice fantasy than reality.

Restore a Macintosh Mirror Door Drive AcBel API1PC36 power converter by upgrading its electrolytic capacitors.

16 April 2024:  Results of modifications added, corrections, and more additions and refinements.

Macintosh Mirror Door Drive (MDD) computers often utilize an AcBel API1PC36 MDD power converter which essentially universally fails due to life limited electrolytic capacitors but is otherwise robust and evidently highly reliable, and thus can be upgraded to provide a quite long service life.  Some MDD computers utilize a Samsung power converter which seems prone to more complex failure modes.  In my estimation the AcBel is the superior design.

Starting from terrific work by others I developed the following approach which replaces all life limited electrolytic capacitors except the three 100 µF 450 V primary side rectified line storage capacitors with unlimited life MLC (Multi Layer Ceramic) or tantalum capacitors, or longer life PAE (Polymer Aluminum Electrolytic) capacitors, and minimizes new component ordering complexity by substantially consolidating the range of values.

(I prefer hermetically sealed electrolytic capacitors but they seem difficult to find so the PAE option seems to be the best practical alternative to ordinary aluminum electrolytic capacitors in these times.)

The consolidation of values is practical because none of these capacitors appear to be utilized in frequency or timing related circuits but rather only filtering roles (including noise suppression), where more capacitance is essentially always better (and where if capacitance degradation occurs the power converter will nonetheless remain functional for a longer time).  I can't confirm this by examination of the power converter's schematics due to lack of availability, but as of 16 April 2024 the three power converters I've upgraded thus far are fully functional with apparently optimum performance, and inspire confidence.

I purchased from the vendors linked.  Values must be selected from the arrays provided on the linked pages.

This ignores the three 100 µF 450 V primary side rectified line storage capacitors.  I judge those as much less troublesome but of course they could be replaced too for a complete electrolytic capacitor upgrade.

To wit:

Macintosh Mirror Door Drive Computer's AcBel API1PC36 REV:A Power Converter Capacitors