Repairing Apollo 3040

[hese]

After repairing a bunch of Amiga motherboards lately, it was a pleasant change to attempt to repair something else for a change.
This time I had a faulty Apollo 3040 68040@33MHz to repair. One of the 74ACT16543 transceivers had taken a hike and took ten tiny pads with it.



Also one of the electrolytic capacitors had leaked on the board and corroded some pins of the Boot ROM socket.



Before attempting to repair the board, the first thing was to test if the MACH chips were okay.
If the MACHs were faulty, there was no much point to spend more time on this one.
I tested the chips on another, working Apollo board. The working Apollo booted with the MACH chips,
so now I knew the faulty board had some hope. I also tested the 68040 CPU and the Boot ROM
at the same and they were also okay, nice.

So back to repairing the board. First I replaced the electrolytic capacitors and the Boot ROM socket.
After that I began working on the missing pads of the 74ACT16543 chip.
I used a small copper wire about the same thickness as the pads as a substitute.
The plan was to keep the changes as indistinguishable as possible. I soldered the wires to vias when possible,
otherwise I scraped some lacquer off from the broken traces and soldered the wires there.



A few photos after fixing the pads and soldering a new 74ACT16543 in place.



I also replaced the original ghetto 68040/68060 CPU socket with a new one.



It was time to test the board, would it work? Well, it didn't. The Amiga only booted to a dark grey screen.

Probably a bad connection somewhere. I cleaned the CPU slot connector pins with isopropyl alcohol but that didn't have any effect.
The connector pins had some oxidation that was difficult to scrape off. Sockets of the MACH chips were also oxidized and
I couldn't get them cleaned satisfyingly enough, so I desoldered the sockets and soldered the chips directly to the board.



Another try to boot the Amiga with soldered MACHs still gave only a dark grey screen.
Hmm, the remaining problem was probably with the CPU slot connector or the other transceiver chips.

I replaced the 74ACT16543s and 74LS245s with faster 74FCT16543s and 74F245s keeping a 68060 upgrade in mind.



Visually the solder joints of the CPU slot connector pins were okay, but I refluxed the pins just in case.
Only after refluxing the pins the board woke almost alive, this time the Amiga booted to a light grey screen.
Now the Amiga could execute the boot up sequence up to some point before the computer hung.

So the CPU slot connector was the culprit. I desoldered the old connector and compared it with a brand new one.
The new connector is in the picture on top and the old one is below it.
The old connector had some oxidation that most likely prevented the board from working (or a loosened pin).



After soldering a new socket it was time to test the board again.



This time the Amiga booted to AmigaOS without any problem. Finally.


Next I will test if the board can be upgraded to 68060.

[8bitbubsy]

Just wow... Amazing soldering/rework! When I first saw the damaged SMD pins I thought "This won't work", but after reading more and looking at the other pictures I was wrong.

[slyder]

Nice work

[Beavis]

You're a proper craftsman, chapeau!

[hese]

Thanks for the feedback.

Today I attempted to upgrade the Apollo from 68040 to 68060.

A couple of things were done to the bottom side of the board.
First the CLKEN surface mount jumper (10 Ohm resistor) was moved from GND to CLK position.
This enables the memory to run asynchronously to the CPU and allows higher CPU clock frequencies.
Also the 220 Ohm resistor under the CLKEN (82 written on the PCB) was removed.



After that two 1k Ohm resistors were soldered to the CPU socket pins.
The resistors tie the bus cycle pins LOCK (Bus Lock) and LOCKE (Bus Lock End) to EVDD.



Next the 68040 processor was replaced with 68060 (rev 1) and the 3.3V voltage regulator was added. And a jumper was put to 060 pins.
After that the Apollo was ready for a test run, so I plugged the Apollo to A4000 and powered the computer.
Did the upgrade work?



Well, it didn't. All I got was a light grey screen (not again ).
After replacing the old Boot ROM chip with one borrowed from another Apollo 060, the computer did boot.



The original Boot ROM had 040/560 writing on the label and the EPROM chip size was 32kB.
The borrowed Boot ROM had 060/560 on a 128kB chip.

Interestingly both chips have exact the same boot code, except:

• on the 128kB chip the boot code is located at 0x8000
• on the 32kB chip the boot code is located at 0x0000
• the 128kB chip has extra 72 bytes of code at the start of the ROM (doing extra hackery to get the 68060 Apollo to boot)

The upgraded Apollo 3040 works fine with 68060@66 and MACH130 chips (MACH130-15JC, MACH130-20JC, MACH130-15JC).
Good times.

[Damion]

Quite impressive, fantastic precision work.

[Cannon_Fodder_]

Nice work What is the best method desoldering the Mach sockets without damage any smd solderpad ?

[hese]

By desoldering with a hot air gun. Just keep heating the socket pins till the socket moves freely and can be lifted up.

[Cannon_Fodder_]

Thank you !