This time in the Attic, an Apple ][+ (1979) was put together with (defective) parts from no less than five (!) different vendors: an empty chassis and case from a USA vendor, a motherboard from the Netherlands, a Power Supply Unit (PSU) from a UK vendor, a keyboard assembly also from the UK, and a keyboard encoder board from a German vendor. And that doesn't even include the sources of the multiple new, replacement parts and extra boards installed in it, so to enable a proper 21st-century experience!
This unit is not available for general sale, but I will consider serious offers from museums or private collectors who make their collection available for public viewing. If you are interested, please post a comment or send me a message on Facebook, with contact information, and include a link to the website of your museum or collection. If you represent a museum, please also include your chamber of commerce registration info. High-resolution, non-cropped versions of all the photos below can be found in this Github repository.
The serial numbers of the chassis and PSU are shown below. I am using the chassis' serial number as identifier of this machine, though the other components came from other vendors and originally belonged in other machines.
I started with the PSU, which was untested. Unsurprisingly, upon opening the unit I found a RIFA line-filter capacitor cracked and ripe for bursting. Good thing I never turn a machine on before inspecting it and performing basic restoration steps such as recapping, otherwise my workshop would have been tainted by the revolting smell of exploding caps.
Since old Apple ][ and ][+ switch-mode PSUs are tricky, unstable units that depend on delicate calibration and aren't trustworthy, I didn't even bother to test this one. Instead, I replaced it with a stealthy modern PSU made to fit into the original metal cage (if you want to continue to use an Apple ][/][+ original PSU board, the safe route is to rebuild it completely and recalibrate it properly, or it may not even start to oscillate; all this requires power electronics expertise and some special equipment).
I wanted to re-use the original DC cable harness, to keep 100% of the original look, but upon closer inspection I found that the DC connector had seen better days, particularly one sunken terminal. So I replaced the harness with a modern equivalent. I did re-use the AC socket and on-off switch, though, so the upgrade remains largely stealthy. The PSU's metal cage was cleaned and oiled, though I didn't run the risk of trying to remove the scuff marks, as doing so could make the metal more vulnerable to corrosion and remove the silkscreen markings.
The motherboard was in fairly decent condition for its age, as shown in the photos below. It was originally assembled in week 22 of 1981, though the design is from 1979, as the silkscreen markings show. Almost all ICs were present, except a 555 timer, which I replaced with a brand-new part.
On the back of the motherboard there was a factory bodge: a decoupling electrolytic capacitor was hand-soldered to the 5V power rail. As the photo to the left, below, shows, it was very poorly installed: the long and looping leads increase the inductive reactance of the part, degrading its frequency response and thereby reducing its decoupling effectiveness (not to mention the fact that one of the leads was dangerously close to causing a short-circuit). I replaced it with a properly installed, brand-new capacitor of higher capacity (10µF, instead of the original's 1µF). See photo to the right, below.
I proceeded to remove every IC from the motherboard and test them off-circuit. Surprisingly, all ICs were still working. As I would discover later, though, several were running abnormally hot and would fail shortly if not replaced.
With the motherboard freed from all its ICs, I washed it thoroughly with PH-neutral soap and distilled water, and then rinsed it with 99% IPA before drying it with an ESD-safe air-blower.
Inspecting the clean board under magnification, I found a strange patch under a socket, where the solder mask seemed to be, well, weirdly thickened and puffed up. To avoid any risk, I pulled the socket for better inspection. The solder-mask in the patch was then lifted along with socket, as it was effectively glued to it. This was clearly a manufacturing defect, and the socket may have been installed before the abnormally thick patch of solder-mask was cured. Either way, I repaired the now-missing solder mask with a fine-point green sharpie (yes, it works well!) and installed a brand-new, double-wipe socket.
Finally, I cleaned and lubricated every socket, port, slot and connector with electronics detergent and then DeoxIT D5. The result was a shinny clean motherboard that looks new, as the photos below show.
After logic tests, I visually inspected every IC before placing it back on the motherboard. Many had dirty and/or corroded pins, all of which were then cleaned with a fiberglass pen on all sides and treated with DeoxIT D5 to prevent future corrosion. The photos below show dirty pins before and after treatment.
Notice the black film of dirt on the pins: that is likely caused by air pollutants (nicotine, house dust, vehicle exhaust fumes, aerosolized kitchen fats, etc.) being electrostatically attracted to the pins during operation. Although the resulting black coat can, sometimes, actually protect the pins against corrosion, depending on its composition it can also promote corrosion. This was the case in this particular board. So it is always wise to pull at least one IC with blackened pins and check the state of the metal underneath. If it is beginning to corrode, I recommend treating all affected ICs, as I did here.
The now-shiny, tested and treated ICs were then placed back on the motherboard. From previous experience, I know that old non-LS-series ICs run too hot, and should either be replaced with modern LS-series equivalents when the latter are available, or fitted with heatsinks otherwise. I did so here, and also preemptively fitted the 6502 CPU and the E0 ROM, both of which I also know from experience to run quite hot, with heatsinks.
I preemptively replaced all MOSTEK-branded DRAM ICs because they are not the most reliable parts. In their stead, I used high-quality, high-reliability, TI-branded New-Old-Stock (NOS) 4116 chips, with late-1983 date codes. Finally, I moved the lovely Apple-branded NEC DRAM chips from the first to the second memory bank, where they will be less stressed and more visible when the case is open.
The motherboard was then successfully tested with the stealthy new PSU hiding inside the original cage. See photos below. I used this opportunity to recalibrate the video output potentiometer while monitoring the composite signal with an oscilloscope.
I also took the opportunity to check whether my 6502-replacement board, which uses a modern, currently-manufactured W65C02S CPU in lieu of the original and increasingly hard-to-find 6502, works in an Apple ][+. Because the little board was originally designed for a VIC-20, its orientation is wrong here, and makes it necessary to build a stack of three sockets so to clear the slots (see photos below). But it did work just fine! Once the test was complete, I removed my little board and put the original CPU back.
Now satisfied that the motherboard would not require any more rework, I sprayed its back with a protective conformal coating, so to eliminate any plausible risk of future corrosion. That coating, of course, also covered the ground pads that are meant to make contact with the chassis, so I carefully removed the conformal coating from those pads (and from those pads only) with a fiberglass pen.
The chassis came with a loose but original speaker. It was covered in glue and residues from the original vibration-dampening foam. I cleaned it thoroughly and re-attached it to the chassis with thick double-sided tape from 3M. This tape is not only extremely strong, it is also so thick that it provides the cushioning required to prevent the speaker's vibrations from spreading to the chassis. It also allows for easy removal: by pushing on the tape sideways with a spatula, underneath the speaker, the latter can be easily detached, as the tape's strength is on the vertical (pulling) direction. Hopefully, some future conservator, 50 years from now, will thank me for not using glue here.
The keyboard assembly looked cosmetically good, but was affected by hidden corrosion underneath due to vintage liquid spills; so much so that I had to douse the screws with WD40 before being able to remove them.
The screws were themselves quite corroded. Normally, I'd use brand-new screws here, but because these are American parts, specified in the weird Imperial system, I can't easily procure them here in Europe. So I decided to save them instead, which was perhaps the right thing to do anyway.
I first cleaned the screws in an ultrasonic bath to remove the loose corroded dust and expose the affected metal. I then marinated them in white vinegar for a couple hours, to neutralize and loosen the remaining corrosion. After that, I again cleaned them in an hour-long ultrasonic bath, and finally bathed them in WD40 to slow down future corrosion.
The keyboard mechanism itself is quite peculiar: instead of separate key switches, the matrix has a monolithic—if quite fragile—construction. I carefully cleaned each contact pad with a lint-free foam stick doused in, first, IPA, and then again, in DeoxIT D5. According to at least one source, this is a very rare, early Apple keyboard. I can't confirm this myself, as I haven't worked on many Apple ][+ units.
Notice that the traces of orange corrosion you see in parts of the keyboard switches are leakage from the iron screws. Iron corrosion is orange, while aluminum (the material used in the key switches) corrodes white. The switches are, therefore, not corroded at all; they just needed a careful cleaning and some lubrication.
The keyboard matrix PCB, with Apple part number 605-4115-0, had several corroded contact pads, particularly those of the reset key. I cleaned the corrosion with a fiberglass pen, and then cleaned the entire board with electronics detergent followed by a thin, protective film of DeoxIT D5. Because the solder joints of the socket that connects it to the encoder board looked dry, I reflowed them all with high-quality flux paste and new 60-40 solder.
A metal bar, which gives the keyboard assembly mechanical strength and prevents it from bowing during use, had several corrosion spots. After sanding them lightly, I treated them with rust converter. As a result, the corrosion was converted into an inert purple compound that prevents future corrosion. See one of the photos below.
The incandescent bulb that lights up the on-off indicator on the keyboard failed during testing. Luckily, it is powered by the 5V rail, so I could construct an LED-plus-resistor replacement, using a yellow LED for authenticity.
The keycaps were washed in soap and water and then fitted back into the keyboard, which now looks almost as good as new (there are still small scratches and scuff marks on the keycaps, which is unavoidable after almost 50 years). All keys tested good after restoration.
Now on to the keyboard encoder board. The flat cable that came with it had twisted pins and one pin missing altogether (I thus replaced it with an identical-looking modern alternative), but the board was otherwise okay. See the original condition of the board and cable in the photos below.
I proceeded to replace the sole electrolytic capacitor in it with a brand-new one, desolder and socket the discrete chips (placing them back only after testing them off-circuit and cleaning their pins, just as I did for the motherboard) for ease of maintenance, and lubricating all sockets, contacts, and the slide switch. I did replace the 555 timer with a brand-new one, just to be on the safe side (timing can be a difficult-to-troubleshoot source of headaches).
Because the encoder chip is a custom and exceedingly rare part, I added a bidirectional TVS diode to the 5V rail that feeds it, just in case one of the 12V rails were to shorten with it. I couldn't fit a heatsink on it because there isn't enough clearance above the IC once it is fitted in the case, but that wasn't necessary anyway, as the IC runs relatively cool. Finally, I sprayed a shiny conformal coating on the backside of the encoder board, for protection against corrosion. See the results below.
At this point, the machine could be put back together again. The light from the yellow LED, though weaker than the original incandescent bulb's, is clearly visible and does its job. I used a 330Ω in-series resistor with it, so to maximize the current flowing through the LED, even though 1KΩ would have been better for durability. This is not a vintage old part anyway, and can be replaced easily, so I prioritized how brightly the LED shines over current limitation.
I added three modern boards to the expansion slots: a 16KB memory expansion (for a total of 64KB), a VGA color card, and an EPROM card in lieu of a floppy disk drive. The EPROM card has several games and applications built-in, allowing the machine to be easily and promptly used upon being turned on, without any hassle.
After a few hours of stress test in the machine's final configuration, I measured the temperatures of all ICs. I realized then that seven out of the eight ROMs were running well above 50 degrees Celcius, and so were the NOS DRAMs in the first memory bank. Out of an abundance of caution, I fitted heatsinks on all of them (and also on the one ROM that wasn't running too hot, because leaving it naked would look odd). This completed the restoration of the electronics. See the results below.
The original case wasn't in bad condition. I washed it carefully, lubricated the screw holes with WD40, and that was enough. Some scratches and scuff marks are still visible, but they do not justify sanding down the original paint (thereby losing its particular texture) to re-spray the case. Restoration often entails making this kind of subjective judgment call, so to strike a balance between conservation and renovation. I trust this is the best call in this particular instance, as the case isn't e.g. drilled with non-original holes or any such drastic defacement. See photos of the final result below.
High-resolution, non-cropped versions of the photos above can be found in the Github repository of this restoration.