• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

What's inside a Cayman Gatorbox CS?

cheesestraws

Well-known member
The Cayman Gatorbox series are AppleTalk routers with lots of bells and whistles, like being able to proxy NFS shares to AFP and being able to proxy PAP printers to lpr clients on Ethernet. I got mine out to play with AirTalk interop testing, and thought I'd take it to bits to check what state it's in inside (mine was actually a barn find, I think: it was covered in hay when I got it and some of the finer hay had got inside—either that or it was beard hair, which I prefer not to contemplate too closely). While I had it open I thought I'd take a photo for you folks.

Everything is on one board, and here it is:

RenderedImage.jpg

Some very brief notes:
  • It's a 68000.
  • The serial circuitry which drives LocalTalk and the serial port is very, very close to a Mac's, presumably for maximum compatibility? It uses an 8530 SCC and what I think are the same RS485 drivers. Even at the time this was far from inevitable; there were other serial controllers that could do SDLC-type framing, and plenty of options for RS485.
  • The two bits that physically use the most of the board are the Ethernet interface (at the bottom right) and the flash and ROMs (bottom left, above the SIMM slots all the way up to the CPU). There's an interestingly homogenous set of ROM-like things: the two labelled EPROMs are, I assume, the boot ROMs, because they are only ultra-violet erasable. Below them, and above and to the right are some Intel flash chips, which I believe contain the application code. The device is flash-upgradeable over LocalTalk, so this would make sense. In fact, the first revision of that didn't store its firmware on board at all, and had to load it from a TFTP server, which sounds inconvenient. The EEPROM (top left, under the CPU), by a process of elimination, presumably contains the bitstreams for...
  • ... the two AT&T FPGAs. I'm not sure why I was surprised to find these here, perhaps because this is fundamentally a router with only two physical interfaces, one of which is restricted to 230kbit/sec. I assume these are used for doing fast-path stuff on the Ethernet side: they're right next to the Ethernet gubbins, and let's face it, it's not hard for even a fairly slow CPU to keep up with LocalTalk.
 

chelseayr

Well-known member
I don't have much to technically say but either way your remark about at&t for ethernet makes sense as the ethernet itself at times is generally a little hard on resources (I don't meant this in an apple-orange manner but: it does make for amusing contrasts where you have a 6mhz 16bit general processor for the entire computer/os but a 25mhz 32bit dedicated processor simply to handle running the ethernet card itself otherwise)
 

cheesestraws

Well-known member
Oh, absolutely—and @Bolle also pointed out that the 8390 Ethernet chip has some specific bus requirements which at least one of the FPGAs is more likely for than for fastpath switching
 

NJRoadfan

Well-known member
The guts are pretty much the same on the Shiva/Kinetics Fastpath. A 68k, a SCC, an Ethernet chip (I think its a DP8390 too), a ROM, and some RAM.

The GatorBox might have a bit more hardware grunt since it did NFS-AFP and LPD-PAP protocol translation as well.
 

lisa2

Well-known member
Note that while flash memory is very common today, in 1989 this was one of the very 1st devices to use a large amount of flash memory like this ( it holds the OS image that in the eariler version was loaded into RAM from a server ), these chips were very expensive at the time. The RAM is upgradable. The FPA's are just for Glue logic. There is a jumper on the board to reset the SN. The power is supplied using a standard 4 pin molex connector like a disk drive, the voltages are the same, but the pin-out is different. The 68000 runs at 10 Mhz, faster than many of the macs that connected to it. If this unit does not work, I see why in the photo. Look at the corrosion under the 40Mhz oscillator, the same thing happened to one of my gatorboxes, I cleaned this up and replaced the oscillator and she works fine again.

More information is here:

 

cheesestraws

Well-known member
If this unit does not work, I see why in the photo

This unit works absolutely fine: if it stops, I'll keep a close eye on that. I did clean up that area a bit after this photo was taken, but it hasn't got any worse since I got the box. What is it up there that causes the corrosion? I couldn't see anything odd going on, I assumed it was moisture ingress at some point in its life
 

lisa2

Well-known member
One of my Gatorboxes worked fine when I got it, then one day while I was using it just quit. Looking at the board I could not see anything obviously wrong. Lucky I had a 2nd working unit. After many-hours of comparison, I found the voltage on the bypass cap next to the 40Mhz oscillator was much lower than on the working unit. I removed the oscillator and found corrosion much like I see on your board. Cleaned it up and replaced the oscillator and she works again! This part is right under one of the vent slots in the case, I assumed that some liquid had spilled onto the board ( looked like coffee to me ) and got under that component. Now I don't know, maybe something from manufacturing or a bad batch of parts causes this corrosion.
 

lisa2

Well-known member
Tonight I will check my 2nd CS unit and see if it has any corrosion in that area and report back. I also have a GatorStar GXR it runs faster and uses a different oscillator, I will try to check that also.
 

cheesestraws

Well-known member
Thanks! I'll be very interested to hear back. I remembered you know a lot about these, always good to hear about them from you
 

NJRoadfan

Well-known member
The FastPath 4 went with a basic AppleTalk Phase 1 router in ROM and required an upload of K-STAR into battery backed RAM for anything more advanced. The Fastpath 5 went with a really basic boot loader and required everything to be uploaded into BRAM. Overall, the GatorBox is a MUCH more complex device inside. The Fastpaths use mostly 74 series logic and a few GALs for glue logic.
 

olePigeon

Well-known member
I'd like to get ahold of a GatorBox. I found the manuals and disks at a recycler, then spent many hours sifting through their old hardware, but couldn't find it.
 

lisa2

Well-known member
I checked my Gatorboxes and I have no new corrosion.

I also found the photos of the original issue I had. While I had thought this was due to something being spilled on the board, now I wonder if this something that leaked out of the oscillator package. The PCB itself is conformally coated.
 

Attachments

  • BadCrystal.JPG
    BadCrystal.JPG
    1.1 MB · Views: 32
  • Problems.JPG
    Problems.JPG
    134.5 KB · Views: 30
  • PCB.JPG
    PCB.JPG
    281.2 KB · Views: 35

cheesestraws

Well-known member
Yes, that's interesting: that's definitely the same area that's looking dodgy in mine. Perhaps I should replace that oscillator and cap in mine as a precautionary measure—what do you think?
 

olePigeon

Well-known member
As the old saying goes, "If it looks like crap, time to recap!"

Or it would be, if it were an old saying. :D
 

CC_333

Well-known member
Or it would be, if it were an old saying.
It's getting there!

Considering that this has been an ongoing concern for many 80s and early-mid 90s Macs for close to two decades now (or three, if you count that the 128/512/Plus analog boards were having issues with caps as early as 1986 or so!), it might as well be an old saying!

c
 
Corrosion may be due to dissimilar metals touching. There is a plastic insulator typically used under crystal packages like this. The old metal aluminum caps and transistor packages also used plastic insulators.
 
Top