Macintosh IIfx

Macintosh IIfx

Macintosh IIfx
Specifications
Manufacturer	Apple Computer, Inc.
Type	Desktop personal computer
Released	19 March 1990
Discontinued	15 April 1992
Intro price	US$8,970 – US$10,970 at introduction
CPU	Motorola 68030 at 40 MHz
Memory	4 MB standard; Apple specifies expandable to 8 MB, 128 MB achievable with eight 16 MB SIMMs
Storage	Optional internal 40, 80 or 160 MB SCSI hard drive
Display	—
Sound	Apple custom sound chip, 8-bit stereo sampling at 44.1 kHz, four-voice wavetable synthesis
Graphics	None built in — requires a NuBus video card
ROM	512 KB, 32-bit clean
FPU / coprocessor	Motorola 68882 at 40 MHz
Media	Internal 1.4 MB SuperDrive; optional second internal drive
Ports / interface	Two ADB, two RS-232/RS-422 mini DIN-8 serial, DB-25 SCSI, DB-19 floppy
Expansion	Six NuBus slots, one processor-direct slot
Power	100–240 VAC auto-configuring, 48–62 Hz, 230 W maximum
Dimensions	5.5 × 18.7 × 14.4 in (14.0 × 47.4 × 36.5 cm)
Weight	24 lb (10.9 kg)
OS / Firmware	—
Predecessor	Macintosh IIx
Successor	Macintosh Quadra 900
Codename	Blackbird, F-19, Stealth, Zone 5, Four Square, Weed Whacker
Model no.	M5525 (order no. M5510LL/A); Gestalt ID 13

The Macintosh IIfx was introduced on 19 March 1990 as Apple's performance flagship: a 40 MHz 68030 in the six-slot Macintosh II chassis, with a 32 K level-2 cache, proprietary memory and a pair of dedicated I/O processors. It remained the fastest 68030 Macintosh until the Quadra 900.

The IIfx is genuinely unusual hardware, and most of what makes it unusual is also what makes it awkward to service. The I/O processors, the 64-pin SIMMs and the SCSI termination arrangement are all specific to this machine, and each has a documented trap in it. Apple's own Technical Note HW 9 exists because of them, and is hosted on this wiki.

Everything on this page that is a number comes from Apple's Service Source for the Macintosh II, IIx and IIfx, from Apple's Technical Note HW 9, or from a named community source.

Specification

Apple's specification

Item	Figure
Processor	Motorola 68030 at 40 MHz with built-in paged memory management unit
Coprocessor	Motorola 68882
Addressing	32-bit registers, 32-bit address bus, 32-bit data bus
RAM	4 MB standard, Apple's figure expandable to 8 MB (see below)
ROM	512 K
Floppy	Internal 1.4 MB drive; optional second internal 1.4 MB drive
Hard drive	Optional internal 40, 80 or 160 MB
SCSI	5 MB/s transfer rate, eight devices maximum
NuBus	Six internal slots, full 32-bit address and data buses
Sound	Apple custom digital sound chip: 8-bit stereo sampling at 44.1 kHz, four-voice wavetable synthesis
Line voltage	100–240 VAC, automatically configured
Frequency	48–62 Hz, single phase
Maximum power	230 W, monitor excluded
Dimensions	5.5 × 18.7 × 14.4 in (14.0 × 47.4 × 36.5 cm)
Weight	24 lb (10.9 kg), more with an internal hard drive

All of that is Apple's. Corrections to figures that circulate widely:

• Height is 5.5 in, not 5.8 in, and the machine is the same size as a Macintosh II and IIx.
• 230 W is the IIfx figure; 220 W is the II and IIx figure. They get swapped.
• The model number is M5525, order number M5510LL/A, Gestalt ID 13. M5780 — which appears on some IIfx pages — is the Macintosh IIci.

Memory

The IIfx uses 64-pin SIMMs that fit no other Macintosh. They are 80 ns parts and they support parity. Eight sockets, filled in groups of four. Apple's Service Source prints an 8 MB ceiling; with eight 16 MB SIMMs the machine runs 128 MB, and unlike the IIx and IIcx it needs no MODE32 to do so, because its 512 K ROM is 32-bit clean.

The LaserWriter IINTX uses physically similar 64-pin SIMMs and they will not work here. The printer's modules are not fast enough for the IIfx. This catches people who find cheap SIMMs described as "IIfx type".

The I/O processors

This is the part of the machine that breaks software, and knowing it saves a great deal of parts-swapping. On the IIfx the main processor no longer handles floppy, ADB or serial I/O. Two Apple custom ASICs — the IOPs — do it instead:

• One IOP controls the SWIM floppy controller and contains the ADB interface that talks to the keyboard and mouse.
• The other controls the 8530 SCC, the serial communications controller.

Community sources describe both as 10 MHz cores from the 6502 family — the same lineage as the Apple II's processor. Apple's own technical note does not give the core type or its clock, and no figure for on-chip RAM appears in any Apple document; figures such as "32 bytes of local RAM" have no source behind them and are not repeated here.

Four documented consequences, all from Apple:

1. Software that pokes the ADB transceiver or its VIA registers directly does not work. There is no workaround; such programs must use the ADB Manager.
2. The SCC IOP has an "IOP Bypass" mode; the floppy/ADB IOP does not. Apple shipped the IIfx Serial Switch control panel to choose between them. In "Faster" mode the IOP handles the SCC; in "Compatibility" mode the main processor does, allowing direct access at the cost of performance. The setting lives in parameter RAM and is read at startup.
3. An application cannot tell which mode it is in. If the machine is in "Faster" mode and something makes a direct SCC call, the machine crashes. A IIfx that hangs reliably in one particular communications program is very likely a Serial Switch setting, not a hardware fault.
4. The VIA1 Wait/Request bit is gone, and the vSync bit has moved. Apple provided the `_HWPriv` trap ($A198, selector 7) to enable or disable external SCC clocking in its place — but `_HWPriv` only works in IOP Bypass mode on the IIfx.

Two more IIfx-specific errata from the same note:

• VIA2 is gone entirely. All of its functions moved to other chips. Software that reads VIA2 registers is incompatible.
• NuBus read-modify-write. If the IIfx executes a read-modify-write NuBus sequence (a `TAS`) to a card and immediately follows it with an ordinary read or write, the system hangs. Apple's fix is five `NOP` instructions between the two.

SCSI

The IIfx's SCSI arrangement is unlike any other Macintosh, and getting it wrong is a classic cause of intermittent SCSI faults. Apple's Additional Procedures chapter documents three distinct pieces of hardware:

• The SCSI Cable Terminator II — the famous black terminator — provides termination for external devices. If an external device has built-in termination, you must disable it and fit the SCSI Cable Terminator II instead. Apple's caution: never install more than one SCSI Cable Terminator II on a IIfx daisy chain, or you may damage the equipment.
• The internal SCSI filter provides the correct termination capacitance for third-party drives and for Apple internal drives shipped before 19 March 1990. It connects between the hard-drive data cable and the connector on the drive. Apple's caution: when you install a drive in a machine that shipped without one, the filter must go on the drive — connecting it directly to the logic board will cause SCSI failures.
• The SCSI termination block is fitted, together with the filter, to the logic board's SCSI connector on machines that shipped with no internal hard drive. When you add a drive, remove the block and move the filter onto the drive.

If you are fitting a modern SCSI emulator (BlueSCSI, ZuluSCSI, SCSI2SD), the same rules apply: one terminator at each end of the chain, no more, and the filter belongs on the drive end.

Apple also notes that the IIfx hardware has SCSI DMA capability that the system software of the day never used. Nothing you can do on the bench changes that.

Software versions Apple requires

Apple's technical note lists minimum versions that are not optional on this machine: Apple CD-ROM driver 3.0.1 or later, EtherTalk driver 2.0.2, and MacsBug 6.2. Earlier versions are documented as incompatible. A IIfx misbehaving with a CD-ROM or an Ethernet card is worth checking against that list before the hardware is suspected.

Power

Power supply Apple part 661-0542 (the II and IIx use 661-0375); logic board 661-0522; bottom cover 630-5806. Apple rates the machine at 230 W maximum, monitor excluded.

What Apple does and does not publish

Apple publishes no rail tolerance and no power-supply adjustment procedure for this machine. The Service Source has no acceptable-range table and no trimmer adjustment; every power fault in the symptom charts ends in "Replace power supply". The "+5 V 4.85–5.15 V / +12 V 11.9–12.7 V" table that circulates on pages of this era is Apple's Macintosh Plus procedure, copied onto machines it was never written for; it does not appear here, and no substitute has been invented. Benchmark tables, memory-bandwidth figures and "cache hit rate" percentages have gone the same way for the same reason.

Apple does ask for one rail check, in the hard-disk symptom chart: if an internal drive will not spin and its LED does not light, verify the +5 V and +12 V lines to the drive. That is a presence test, not a tolerance test.

Recapping the supply

You may open it. Understand what is live in it first: it is mains-rectified, so with the cover off and the lead plugged in the fuse, input filter, bridge rectifier and reservoir capacitors sit at mains potential and hold a charge after the plug comes out. Unplug, discharge each large reservoir through a 1 kΩ / 5 W resistor, and confirm with a meter before a tool goes near the primary side. If two reservoir capacitors sit in series across the rectified mains, replace both together with matched parts — a single capacitor in place of the pair will fail on European mains, violently.

Battery

This machine has two lithium cells, not one. Apple's battery procedures for the II, IIx and IIfx are written in the plural throughout and the illustration shows a pair of holders; the symptom charts say "check batteries". Both holders are visible in Recap-a-Mac's photograph of the board, at BT1 and BT2.

• Set a meter to the 10 V DC range.
• Positive probe on the end marked "+" on the logic board, negative probe on the other.
• Replace the battery if it reads below 3.2 V.

3.2 V is this machine's figure. It is not the 2.8 V of the IIcx, IIci and Quadra 700, nor the 3.0 V of the IIsi and the LC series. These thresholds are machine-specific and get transplanted between pages.

Apple's removal: prise the latch away from the holder with a small flat-blade screwdriver, lift the cover off and lift the cell out. Battery verification requires the cover and the drive mount to come out first.

If a cell has leaked, neutralise with bicarbonate of soda, not vinegar — these are lithium cells, and vinegar is the treatment for alkaline leakage. Bicarbonate paste in distilled water, rinse with distilled water, then isopropyl alcohol, then dry thoroughly; afterwards buzz out every track running away from the holders and check the underside of the board. See Battery Explosion, Capacitor or Corrosion Damage and Battery Refurbishment.

Take the cells out before the machine goes into storage.

Capacitors

Logic board 820-0282-B. This board is unlike any other Macintosh board of the period: it has two sets of pads at every 47 µF position, one for a tantalum and one for an aluminium electrolytic, and Apple populated boards with either type or a mixture of both.

Twenty positions in all:

Macintosh IIfx logic board 820-0282-B

Value	Working voltage	Quantity	Package
47 µF	16 V	14	Surface-mount — tantalum, electrolytic or a mix
470 µF	16 V	3	Axial
220 µF	16 V	1	Axial
10 µF	16 V	1	Axial
1000 µF	10 V	1	Axial

MacDat's list and Recap-a-Mac's annotated photograph give exactly that complement, and Console5's per-designator list for 820-0282-B agrees. Apple published no component-level capacitor data for any Macintosh of this era, so all three are readings of physical boards — but three independent readings agreeing is about as good as this gets.

Macintosh IIfx logic board 820-0282-B. The two battery holders are at right beside the ROM area; the five axial capacitors are at lower left and lower right. Source: Recap-a-Mac.

What to actually replace

Whichever mix your board has, every aluminium surface-mount electrolytic on it must go. Those are the parts that leak, and the commonest factory configuration is mostly tantalums with two electrolytics near the battery holders — which is the worst possible place for them, because that is also where a leaking cell does its damage. Look carefully: a board that appears to be all tantalum often is not.

The five axial parts are easy to overlook because they do not resemble the rest. They are electrolytic and they age.

C1: the capacitor that explodes

If C1 on your board is a factory tantalum, treat it as a fault waiting to happen. The position is a 47 µF 16 V part on a rail running at around 12 V, and 16 V is too little derating for a solid tantalum in that service. Reported failures range from a shortened life to the part detonating.

• Check whether it has already gone — a cracked, split or scorched moulded block.
• Replace it with a 47 µF 25 V part rather than a like-for-like 16 V one.

The general rule behind that: a solid tantalum wants roughly twice the working voltage of the rail it sits on. A part on a 12 V rail wants 25 V, and a part on 5 V wants 10 V or more. Tantalums fail short, so an under-rated one does not quietly lose capacitance — it pulls its rail down or burns.

Connectors and sockets

Oxidised contacts cause more intermittent faults on a machine this age than failed silicon does, and they are the cheapest thing to put right.

• SIMM sockets. Clean with isopropyl alcohol on a swab and reseat. Do not take an abrasive to a module's gold fingers — once you are through the plating, the base metal oxidises and then transfers to every socket the module touches.
• VRAM sockets. Same treatment. Video faults that come and go when the case is knocked usually start here.
• NuBus and PDS slots. Inspect for bent contacts and debris before a card goes home; check the card's edge fingers too.
• ROM SIMM socket, where one is fitted. A poorly seated ROM gives a machine that chimes and does nothing else.
• Logic-board power connector. Look for browning or a distorted housing. A discoloured power connector is a warning, not a cosmetic problem.
• ADB, serial, SCSI and floppy connectors. Reseat; check ribbon cables for cracked conductors where they fold.

A contact cleaner such as DeoxIT D5 earns its place on genuinely corroded contacts. It is not a preventive treatment and it does not belong on clean gold.

Add the ROM SIMM and the 64-pin memory SIMMs, which are the two sockets on this board most worth reseating before anything is suspected. Bank A and Bank B are silk-screened on the board; RAM access needs the drive assembly out of the way.

Cleaning

Opening the case

Release the catches at the rear and slide the top cover forward and off. Apple's take-apart sequence removes the drive mount before the logic board.

Inside

• Compressed air and an anti-static brush on the logic board, the supply and the fan. Hold the fan blade still while you blow it: spinning a fan with compressed air drives it as a generator and puts a reverse voltage back down its cable.
• Isopropyl alcohol (99 %) and a soft brush for grime, working outwards from each capacitor and from the battery holder.
• Check the underside of the board as well as the top. Both battery and capacitor electrolyte travel through vias.

Outside

• A damp microfibre cloth and mild detergent. No abrasives, no solvents — the texture on these shells does not survive either.
• ADB keyboards come apart for washing: keycaps off, warm soapy water, dry them completely, then clean the contacts with isopropyl alcohol.
• Mouse ball out, rollers scraped and wiped with isopropyl alcohol.

The platinum-grey plastics yellow with age. The discolouration is in the body of the plastic rather than on the surface, so cleaning will not shift it; see Retrobrite for what can be done and what it costs in brittleness.

Floppy drive

The 1.4 MB Apple SuperDrive is direct drive. It has no belt. Advice to fit a new belt to a sluggish or non-ejecting SuperDrive — including to the auto-inject version — is wrong: there is nothing there to replace. What these drives actually suffer is hardened grease on the head carriage and eject rails, dust in the mechanism, and dirty or worn heads. On manual-inject drives a worn eject gear is a genuine and common failure.

• Clean the heads with isopropyl alcohol on a lint-free swab, supporting the upper head arm so it is not sprung.
• If the rails are stiff, re-lubricate with a silicone grease — not white lithium grease, which gums up as it ages.
• If a disc will not eject, hold the mouse button down while switching the machine on to run a full eject cycle before taking anything apart.

See Macintosh Floppy Drive Maintenance.

Hard disk

The original SCSI mechanisms are past thirty years old and the usual failures are a stuck spindle, dried head-park grease and dead controller electronics. Solid-state replacements — BlueSCSI, ZuluSCSI, SCSI2SD — are quieter, cooler and considerably more reliable, and they let you keep the original drive on a shelf.

Whatever you fit, get the termination right: the drive at the end of the internal bus must be terminated, and the computer itself is always SCSI ID 7. See Macintosh HDD Maintenance.

Read the SCSI section above before you connect anything to a IIfx. The terminator, the filter and the termination block are not optional extras and they are not interchangeable with ordinary Macintosh SCSI practice.

Easter egg

Set the system date to 19 March 1990 — the machine's introduction date — and restart holding Command-Option-F-X. A colour photograph of the IIfx design team appears; click the mouse to continue. It is the same image the Macintosh IIci shows on its own date and key combination.

Apple's own colour-depth requirement is documented for the IIci egg, not the IIfx one; a IIfx that shows the picture without the display being set to 256 colours is not doing anything unexpected.

Things that are not faults

• A system that shuts down intermittently after long use. Apple's first cure is to check the air vents, because the thermal protection circuitry may have shut the machine down; it should be usable again after 30–40 minutes.
• Software that crashes only on this machine. See the I/O processor section — Apple documents this as the expected consequence of direct hardware access.
• SIMMs from another Macintosh not fitting. The IIfx's are 64-pin and unique to it.
• No built-in video. A NuBus video card is a requirement, not an upgrade.

Storage

• Take the battery out.
• Store dry and at room temperature — not a loft, not a garage.
• If the logic board comes out, bag it anti-statically and lay it flat.
• Nothing needs periodic running. "Power it up every few months to keep the capacitors formed" is repeated widely and supported nowhere for machines of this kind. The real hazard to a stored Macintosh is the cell in the battery holder, and removing it costs nothing.

Service documentation

Also hosted: Apple Technical Note HW 9 Macintosh IIfx The Inside Story and Macintosh IIfx Special Options and Technical Information.

Related pages

• Macintosh IIx — same chassis, one generation earlier
• Macintosh II
• Macintosh IIci — the contemporary mainstream machine
• Macintosh Quadra 900 — direct successor
• Macintosh II/IIx/IIfx Service Source
• Capacitor Failure Symptoms
• Battery Explosion, Capacitor or Corrosion Damage
• Recommended Tools