To hell and back. Via SCSI.

My friend David knows I have a bit of a SCSI problem. Apparently the appropriate treatment is more SCSI controllers.

“I do have an interesting story,” he wrote. He had picked up some Amiga goodies, including two cards from a machine that had been destroyed in a fire. One was a Phase 5 Fastlane Z3, the other an Amiga 2000 accelerator with RAM and SCSI. Maybe these could be candidates for a reproduction, like our ReA4091?

Well, there was a small problem with that idea. The programmable logic on the Fastlane was never open sourced, and when Phase 5 went under, the contents of those GALs and CPLDs did not become available as JEDEC files either. Without those, a reproduction was unfortunately out of reach.

But maybe we could get the original working again?

The accelerator had gotten the worst of it. The PCB substrate itself was burned through, and half the gold fingers were gone. That one was beyond what I was willing to attempt.

The Fastlane, however, looked surprisingly intact. Dirty, dark, and missing a whole row of components, but intact. The machine had gotten so hot that the chips along the bottom of the card had desoldered themselves and slid off.

Well, that’s one way to do hot air rework. I prefer a little more temperature control.

David called the Fastlane “one of my grail items,” adding, “you know me and SCSI!” Yes, David. I do. And I am glad we share the same passion.

And David is a friend who spent countless hours fixing my 3D printer, and who recently supplied me with parts to bring another Amiga 3000 back to life, plus parts for one of my keyboards. So getting his grail item back into service would be a nice way to return a favor or two.

First, however, there was a lot of cleaning to do.

The board was covered in wax, melted plastic, and whatever else had found its way onto it during the fire. I suspect the wax originally protected some other components in the machine. Now it was protecting my Fastlane from being repaired. Particularly the RAM sockets took a lot of convincing.

An evening of cleaning and an ultrasonic bath later, the card still looked noticeably darker than the Fastlane in my own collection. But underneath all that mess, the traces were intact.

Two pads on the logic chips were lifting. One I managed to save, the other came off. Oh no. A quick check revealed that the missing pad belonged to an N.C. pin.

Not connected. Two beautiful words.

Getting all the replacement parts together took about a month, with a two week trip through Shenzhen, Hanoi, and Nanchang in the middle. The shopping list was seven 74F245 bus transceivers, two DS1000 delay lines, a 74F04, a 74F32, and a 74HCT191. That last one turned out to be the hardest to find. A couple of tantalum capacitors from my garage happened to fit perfectly, too.

See? Keeping all those parts around does serve a purpose.

Along the way, one of the half-desoldered 74F245s had a surprise for me. There were six little resistors around its legs, and at first I thought they had fallen off somewhere else on the board and gotten stuck there in the heat.

Only after taking the chip off with hot air and cleaning it could I see what was going on. Six legs had been carefully bent upwards, with 10 ohm resistors attached at a right angle between the legs and the PCB. A very elegant little rework. Somebody must have spent quite some time doing these by hand.

My guess is that the series resistors were there to reduce ringing on some signal lines. But I could not find any documentation, and my own Fastlane does not have this modification. I have never seen it on another one either. If you know the story behind this, please leave a comment!

I did not recreate the modification on this card. So far it has worked without it, although that does not tell us why it was needed in the first place.

Another evening with the soldering iron, and all the components were back where they belonged.

Time for the part where we find out whether I have repaired a SCSI controller or built a small electric heater.

My Amiga 3000 volunteered for testing. It does this a lot, being the machine I use to test all my A4092 controllers, although this particular candidate made me more than a little nervous. It had already been in one fire. I did not need a sequel.

So out came the thermal camera that attaches to my phone, for what I like to call debugging with a thermal camera. Turn the machine on and watch what gets warm. Reasonably even temperatures are encouraging. Something staying unexpectedly cold deserves investigation. Something getting very hot very quickly means turn it off NOW.

For the first test I left the SCSI drive disconnected. All I wanted was for the card to show up in the early boot menu.

Power on. Watch the camera. Hold breath.

And… there it was!

Alright. That was already more than I had dared to expect. Let’s give it something to boot from.

I connected a ZuluSCSI with an emulated hard drive on its SD card, and turned the Amiga on again.

Bingooo! It boots!

No further repairs, no mysterious crashes to chase down first. It just booted. Several more boots and a bunch of file copies later, the Fastlane was still happily doing its job.

The card has since made it back to David. “Got it today and it looks fantastic!!” he wrote.

Still a little darker than the one in my collection. But after everything this card has been through, I think it has earned that.

To hell and back. And straight into the early boot menu.