Board top
Board back
PCB Print
Another Gal
Hi Res version, Board top
- 2592 x 1936, 2,249K
Hi Res version, Board back
- 2592 x 1936, 2,030K
Hi Res version, Jumpers
- 2592 x 1936, 1,864K
Hi Res version, Gal?
- 2592 x 1936, 1,623K
Hi Res version, PCB Print
- 2592 x 1936, 1,854K
Hi Res version, Stickers
- 2592 x 1936, 1,451K
Hi Res version, Another Gal
- 2592 x 1936, 1,811K
Hi Res version, 68030
- 2592 x 1936, 1,893K
Hi Res version, 68020
- 2592 x 1936, 1,843K
Hi Res version, Stickers
- 2592 x 1936, 1,495K

This is the CPU card that was designed and built for the Amiga A3200/A3400 prototype computer that was never
released*. It is a dual CPU board containing a 68020 and 68030 CPU. While being very similar to a standard A3630, it
is certainly a different card. The best way to describe the difference, is that this card is more complete than the A3630.

* Info added 12/27/2013:This was not intended on going in an Amiga 4000 or 3000. I believe It went into a computer that had a very limited release the A2200. There WERE two versions.  One with an 020 and one with an 030.  Both promised the ability to upgrade to 040/060's like the 3000/4000. They apparently, both also had an A1200 CPU Slot which would have made it exceptionally unique that it could use two different types of CPU upgrades. I think this CPU Board was one (both) that were made for both versions of the 2200. A very rare group of people actually got to own some of these.  Commodore made these to be "AGA versions" of the 3000 (if you consider the 1200 an AGA version of the 500 and a 4000 to be aga version of a 2000)...  so it was a very weird idea.  I do believe those boards here, were the intented stock cpus for the 2200. New info submitted by Anthony fox

Differences noted are:

- Has the 68020 CPU onboard that isn't on a normal A3630
- More links are present that aren't on a normal A3630
- Extra GAL present that isn't on a normal 3630
- Stickers on back read differentely to a normal A3630

The 68030 on this board has been tested to work fine. The 68020 hasn't. This is because two important chips - U860
and U152 - are missing from Amiga 4000 motherboards, which are needed to interface with this card. These two
chips are present on the A3200/3400 prototype motherboard.

It may be possible to upgrade a normal A3630 to be functionally equivalent to this card.

The GAL is for the MAPROM function and the card itself is identical to an A3630, but these are fully populated.

To get the 020 working, you have to change a LOT of jumpers around. The trick is probably disabling the MAPROM in 020 mode. You can look at the A3630 schematics and figure it out, it's the same board. Some of the boards around, came from boxes in the NewTek warehouse, so they could be prototypes, but they looked like final shipping products.QA stickers, different barcodes and slightly different sticker labels.

How many of these cards exist is not known, but is estimated at a very low number.

Additional information known as of 2020/2021 (written by Robert Miranda):

- The board was intended as a prototype 68020/68030 CPU option for A3000/A4000-like system configrations then under design.  It was cheaper to insert this card in the CPU slot than solder expensive surface-mount CPU/FPU parts directly on the test motherboards.
- The 68020 configuration reqired 2 parts on the motherboard as noted above (component pads are present on early A4000 production revisions).  This 68020 logic was present on prototype motherboards which included the Motorola DSP and other variations then under consideration.
- Only 1 CPU and 1 FPU socket can be used at a time.
- The board uses INT clocking from the motherboard, implying 25MHz CPU operation (in most common cases)
- A 68881 or 68882 FPU is compatible with either CPU.
- The oscillator position selected the FPU clocking if not the same speed as the CPU.
- The MapROM feature (provided by the onboard GAL) operated the same as it did on the A3640, and helped with development of newer Kickstart/component feature testing without requiring the repeated erase/reburn of Kickstart ROMs.  It provided soft-load options for development Kickstart images similar to the early A3000 SuperKickstart solution.
- Production revisions (A3630) eventually showed up on the A4000D/T 030 systems, and omitted the the 68020 CPU and related parts altogether, used either a full 68030 or the 68EC030, and may have left the FPU socket unpopulated (to be sourced by the dealer/user at a later time).

John Hertel reverse-engineered this PCB as a personal project in the later 201x's.  His final reduced-component version was a CPU-only PGA 68030 card with a prototype signal header for other developers to make use of in their own personal projects.

Page contributors: Anthony Fox, John Perkins, PR, Robert Miranda (GVP Tech Support)
Updated: 7/24/2021 . Added: 3/2/2013