Link Search Menu Expand Document

Apple Macintosh Classic

Computers in Collection

General Info

Replacement Components

Case Fan

Existing fan is 12V DC and 60 to 65 mA – Sanyo “DC Pico Ace 25” 109R0612M405 or Elina HDF6025L-12MBB-0, which are almost the same models as my SE/30s. The fan blows upward – drawing air from under the case. Mounting holes are spaced 50 mm apart and are approximately 4 mm in diameter. Case dimensions are 60 mm x 60 mm x 25.4 mm. Cable is two-conductor (power and ground) and I eyeball it as less than 4” long, and terminates in a two-pin 0.1” female header socket. I found a datasheet with model HDF6025L-12MB listed as 10.2 to 13.8 V, 70 mA, 0.84 W, 3000 RPM, 12.7 ft^3/min (0.36 m^3/min), 26 dB(A)/1m, 85 g.

Macintosh Classic, serial E0500BRM0435LL/A

Apple Macintosh Classic suffering from "Sad Mac 0x0000000f 0x0000000a"

Inventory

“Acquired” from Jenny.

Case

Attribute Value
Name Apple Macintosh Classic
Manufacturer Apple
Model M0420
Serial E0500BRM0435LL/A
Manufacture Date December 1990
Source 2004, Jenny

Logic Board

Apple Macintosh Classic logic board after cleaning and recapping

Attribute Value
Name Apple Macintosh Classic logic board
Manufacturer Apple
Part TODO:
Bar Code SG0508WZ03F5
Solder Mask bright green
Assembled In Singapore
Markings “820-0390-03”, “630-0390-__”, serial “bar code” SG0508WZ03F5, “APPLE COMPUTER INC. (C) 1990”
Notes Slight corrosion near surface-mount electrolytic capacitors. Electrolytic capacitors replaced.
Source 2004, Jenny

Memory Expansion Board

Attribute Value
Name Apple Macintosh Classic memory expansion board
Manufacturer Apple
Part TODO:
Bar Code (none)
Solder Mask dark green
Assembled In Singapore
Markings “APPLE COMPUTER INC 820-0405-01 (C) 1990” “630-0405-S2” “470S” “G9049” “OC AHENG”
Source 2004, Jenny

Battery

Apple Macintosh Classic logic board original lithium battery

Attribute Value
Name Inorganic Lithium Battery
Manufacturer Sonnenschein Lithium
Model SL-150
Size 1/2 AA
Voltage 3.6
Date Code WX/10/90
Source 2004, Jenny

Power Supply Board

Attribute Value
Name Apple Macintosh Classic Analog Board
Manufacturer Hitachi?
Model  
Serial *SG0494KS03K4*
Markings ””
PCB Markings “MAC CLASSIC ANALOG BOARD”, “(Hitachi) HCS-106C 94V-0”, “ASSEMBLED IN S’PORE”, “820-0395-C”, “630-0395-8”, “450S”
Date Code L 9049

Internal Cathode Ray Tube

Attribute Value
Name Apple Macintosh Classic internal monochrome cathode ray tube
Manufacturer TODO:
Model TODO
Serial TODO
Markings TODO: “EIA 1420” “*CL0453814801FB*” “4590”
Date Code TODO:
Source 2004, Jenny

Internal Floppy Drive

Attribute Value
Name Apple Macintosh Classic internal floppy disk drive
Manufacturer Sony
Model MP-F75W-01G
Serial *SD04440600P2*
Made In Malaysia
Source 2004, Jenny

Internal Hard Drive

Attribute Value
Name Apple Hard Disc 40SC
Manufacturer Quantum
Product ProDrive LPS
Model TODO:
Serial TODO:
Made In Japan
Case Markings “440011150937” “42S 940-40-9404” “1059”
PCB Markings “Quantum (C) 1989” “800-09-94 20 0” “MK 0 46 6RG 037A”
EPROM Markings “J2801 (C) 1998 QNTM CORP DisCache”
Notes Backed up via SCSI2SD, contents do not match source
Source 2004, Jenny

External Modem

Attribute Value
Name Incomm Midget 24E Modem
Manufacturer Incomm
Model Name Midget 24E
Model Number 0200010
Serial 0024494
Source 2004, Jenny

Miscellaneous

Attribute Value
Description DB-25 male to mini-DIN (8 pins) cable, 6 feet
Source 2004, Jenny

Maintenance Log

Not functional, sad Mac.

2020/Aug

Replaced SMT electrolytic capacitors (TODO: Component list. I believe this is before I got the aluminum solid polymer capacitors.). Touched up pins that appeared corroded. No improvement in “sad Mac” behavior.

2020/Mar

Tons of logic analyzer traces and ROM code reverse engineering eventually led me to suspect a VIA pin connected to the ADB chip. I compared the signals between the VIA and ADB chip at power-up, and noticed that the interrupt signal from the ADB chip was asserted earlier on this Mac than on the one that was working. Then I noticed one of the two state signals going from the VIA to the ADB chip wasn’t changing – it was stuck high. I hacked up a quick 1K pull-down on that signal, and lo and behold, the system error was bypassed and I got a blinking “?” floppy icon! Then I hooked up a SCSI hard drive and the machine booted fully! While I didn’t test it, I have to assume the mouse and keyboard (ADB peripherals) would not work with that VIA “state” signal being permanently pulled low.

In all the ROM reverse-engineering I did, I eventually traced a divergence in the ROM code executed by the working and non-working system. The first hint something was wrong is when the ADB interrupt signal is prematurely asserted on the non-working system. Tracing the code, this causes the zeroing out of a pointer in an ADB structure in RAM. Later, this pointer is used to jump to a code handler of some sort. Since the pointer was zeroed due to the early interrupt, the jump goes to the start of RAM (0x000000) and executes whatever is there. The processor encounters valid instructions until 0x000100 or so, when it comes across an Line F instruction, which sends us into the Line F exception handler, and then to the sad Mac 0x0000000F 0x0000000A display.

A replacement 65C22 was ordered. Western Design Center is still making them, and Mouser had several PLCC variants to choose from. I chose the “NMOS” version, which is closest, it seems, to the chip I’m replacing.

I desoldered the bad VIA from the board using a board pre-heater (set to about 150C) and my Quick hot air wand (set to 275C and “50” air flow). I applied some flux once the board was heated (probably better to do while cold, so the flux isn’t as much of a thermal shock). I set the tweezers to gently pry at a corner, and then I waited until the chip was loose all the way around. (I have a terrible habit of using too much force and lifting or even tearing off pads.) It came off cleanly! Strangely, a few of the pads look a little dry after the fact, which maybe indicates I was still pulling a bit too hard and/or not using enough heat.

I got the replacement WDC 65C22 part, placed it on the board, and preheated it for a couple of minutes before using the hot air wand to reflow the area around the new chip. However, I connected the wrong temperature probe, and the board was heated quickly, and beyond the 150C temperature I set. Fortunately, I noticed something was amiss and cut the power to the preheater. Using the hot air wand, I did get the 65C22 seated into the existing solder on the board. Once the board was cooled, I inspected my work. The joints on the new chip looked good enough, though I went through and touched them up with my soldering iron, just to get a more obvious solder fillet. I also retouched a couple of nearby chips that had small solder balls that had formed on their pins due to the reflow process. I think the more corroded pins seem to wick solder from the joints and form balls on the more oxidized portions of the pins. Using some flux, I was able to collect many of these balls, either on the iron tip, or back into the solder joint. I noticed the bottom of the board had a lot of joints, almost all vias, that had started to “drip” solder, almost certainly due to my overheating the board with the preheater. I used my soldering iron to reflow the solder back into the vias. Many of the “drips” were actually bubbles! I think maybe the hot air wand air rate was too high and I was blowing molten solder through the vias? Or I suppose the heavy solder was drooping from the bottom of the via and creating some negative pressure, pulling air into the via and into the bubble? I’m not sure. At some point I discovered a little black plastic disk on the table, which I quickly realized was the button plunger piece of the reset switch. It looks like the programmer interrupt switch was also a bit melted, but hasn’t fallen out just yet. So I guess I’ll need to replace those too…

I probed the connections from the new VIA chip to all the connected components, and everything checked out fine. I tested the continuity and capacitance between all the supplies and ground, and they were sensible.

I got impatient and hooked the logic board up to my bench power supply. (I didn’t have a Mac case or power supply available.) I set the +5, +12, and -12 voltages and current limits extremely low, then ramped them up to a reasonable current. The processor was executing code, so my preheater accident didn’t (immediately) destroy any of the chips or via interconnections. WHEW The video signals were bouncing around correctly – the board was producing video! After a bit of oscilloscope trigger fooling, I get a single horizontal line of video on the display and could see indications in the signal that there was a flashing floppy disk icon on the screen. Observing the audio output, I get what looks to be the “bong” sound at the right time after a power-on or reset.

2023 Nov 19

Powered up and got checkerboard pattern. Checked 5V at hard disk power connector, and it was extremely low (4.4V, IIRC). Inspected PCB and found surface corrosion under the low-voltage filter capacitor area (e.g. CP2, 6, 7, 8, 9, 10, 12, 36).

Partial Inventory of Analog Board Capacitors

According to the Bomarc schematics, it appears the switching circuit operates at 100 kHz. So capacitor specs for 100 kHz should be of primary concern.

Ref Des BOMARC Visual r H Pin Tested Replace With
CP2 ? Nichicon 1000uF/35V PL(M) 105°C H9030 12.7 mm 31.8 mm 5.0 mm 930 uF, D 0.35 @ 1 kHz Panasonic EEU-FC1V102 1000uF 35V 12.5x12-5
CP6 2200/10 Nichicon 2200uF/10V PL(M) 105°C H9032 12.7 mm 26.0 mm 5.0 mm 2400 uF, D 0.177 @ 120 Hz Kemet A750MW228M1CAAE010 2200uF 16V 10x18-5
CP7 1K/10 Nichicon 1000uF/10V PL(M) 105°C H9026 10.1 mm 22.1 mm 5.0 mm 910 uF, D 0.50 @ 1 kHz Kemet A750MS108M1CAAE013 1000uF 16V 10x13-5
CP8 2200/16 Nichicon 2200uF/16V PL(M) 105°C H9032 12.7 mm 32.0 mm 5.0 mm 2210 uF, D 0.062 @ 120 Hz Kemet A750MW228M1CAAE010 2200uF 16V 10x18-5
CP9 470/25 Nichicon 470uF/25V VX(M) 85°C M9037 10.2 mm 15.6 mm 5.0 mm 407 uF, D 0.365 @ 1 kHz Kemet A750MV477M1VAAE018 470uF 35V 10x16-5
CP10 470/25 Nichicon 470uF/25V VX(M) 85°C M9037 10.2 mm 15.6 mm 5.0 mm 411 uF, D 0.377 @ 1 kHz Kemet A750MV477M1VAAE018 470uF 35V 10x16-5
CP12 1K/16 Nichicon 1000uF/16V PL(M) 105°C H9026 10.2 mm 30.3 mm 5.0 mm 938 uF, D 0.460 @ 1 kHz Kemet A750MS108M1CAAE013 1000uF 16V 10x13-5
CP36 2200/10 Nichicon 2200uF/10V PL(M) 105°C H9032 12.7 mm 26.0 mm 5.0 mm 2380 uF, D 0.139 @ 120 Hz Kemet A750MW228M1CAAE010 2200uF 16V 10x18-5

The Nichicon “PL” series is their obsolete “Extremely Low Impedance, High Reliability” line, replaced by the PM line. The Nichicon “VX” series is their obsolete “Standard type”, replaced by VR series. The “(M)” suffix apparently means +/-20% tolerance.

Nichicon PM (substitute series) ratings for values of interest:

Value R @ 20°C 100 kHz R @ -10°C 100 kHz mA Ripple @ 105°C 10-200 kHz ma Ripple @ 105°C 120 Hz
470uF/25V 0.065 0.130 1060 810
1000uF/10V 0.060 0.120 1060 915
1000uF/16V 0.047 0.094 1410 1210
1000uF/35V 0.029 0.058 1980 1710
2200uF/10V 0.034 0.068 1710 1530
2200uF/16V 0.028 0.056 2010 1800

All the solid polymer replacements vastly outperform in ESR and ripple current. And they shouldn’t leak. Hopefully they don’t unsettle the switcher.

Macintosh Classic, serial E0442TDM0435LL/A

Inventory

Acquired from Glenn via Sean@CCC on 2020/Dec/06.

Case

Attribute Value
Name Apple Macintosh Classic
Manufacturer Apple
Model M0420
Serial E0442TDM0435LL/A
Manufacture Date November 1990
Source 2020/Dec/06, Glenn via Sean@CCC

Logic Board

Attribute Value
Manufacturer Apple
Part TODO:
Bar Code SG043WLW03F3
Solder Mask dark olive green
Assembled In Singapore
Markings silk: “820-0390-03” “630-0390-S5” “APPLE COMPUTER INC. (C) 1990”, copper: “820-0390-03 (C) 1990”, “S9043”
IC UT4 BBU: VLSI “9032AV” “VGT7910-6207” “141743” sticker: “343S0107-A (M) (C) 1986 Apple”
Notes Extensive corrosion around surface-mount electrolytic capacitors. Electrolytic capacitors replaced.
Source 2020/Dec/06, Glenn via Sean@CCC

Power Supply Board

TODO:

Internal Floppy Drive

Attribute Value
Manufacturer Sony
Model MP-F75W-01G
Manufactured In Malaysia
Markings “2MB” “4-614-735-03” “SD0403KW00P2” “MFD-75W-01G 72185504”
Source 2020/Dec/06, Glenn via Sean@CCC

Internal Hard Drive

Attribute Value
Manufacturer Quantum
Model ProDrive LPS
Apple Name Hard Disk 40SC
Case Markings “40010084399” “42S 940-40-9404” “1059” “MK0421MR037A
PCB Markings “Quantum (C) 1989” “800-09-94 200 Made in Japan” “EMMM1”
EPROM Markings “J2801 (C)1988 QNTM CORP DisCache”
Notes Backed up via SCSI2SD
Source 2020/Dec/06, Glenn via Sean@CCC

Maintenance Log

2020/Dec/06

Not functional, checkboard pattern on screen. Extensive corrosion around surface-mount electrolytic capacitors.

Apple Macintosh Classic surface-mount electrolytic capacitor corrosion

Apple Macintosh Classic surface-mount electrolytic capacitor corrosion

Capacitors were removed, board was cleaned, new capacitors installed.

After the capacitors were replaced, I powered the machine up, and this time got thin horizontal lines. I eventually identified that the 12V supply wasn’t making it to all the places on the board. Now the system boots!

While using the chassis (power supply, CRT, cabling) to test the other Mac Classic, the CRT started doing odd things. I believe the image first got really dim, so dim I thought there was no picture at all. I adjusted some of the CRT settings, and was able to brighten things up, however the image was now squashed – too narrow in width. And later, when I booted the system off the hard drive, there were image abberations whenever there was hard drive activity. It seems like the 12V supply is getting weak? TODO: Was this also observed with the other Mac Classic’s logic board?

2023/Nov/19

Powered on to see if it was behaving the same as my other Mac Classic, and sure enough it was. This one didn’t even display a picture. I checked the supply voltages and the 5V rail was extremely low, about 4V if I’m remembering correctly. And the low-voltage ripple-smoothing capacitors on the analog board looked to have some oily residue in the vicinity. So I suspect this board also needs a recapping.