Different AGNUSes 8372 vs 8375

[Matt_H]

If you have a 2MB Agnus in your machine that works, why replace it?

My understanding is that the only differences between Agnus (apart from 512K/1MB/2MB) are for compatibility and that there are no performance differences.

Best information is probably to go right to the source: see if you can find a PDF of a Commodore parts catalog and look up the different part numbers.

[Matt_H]

Every 3000/3000T motherboard picture in the usual hardware databases has an 8372-series Agnus, so probably best to leave it.

As to why there were so many revisions? I can only guess, but probably some combination of FCC compliance, cost-reduction, and compatibility with new board revisions.

[Matt_H]

I'd agree with Matt, many of the variants appear to accommodate specific Amiga models.  For example, later 8375 versions added a 14MHz clock output that was needed for Gayle that was used in the A600.  It also appears that they might have been trying to work around limitations of not quite having enough pins on the PLCC84 package to do everything.

Some variants of 8375 are the same die internally, but bonded differently to change functions of specific pins.  Have a look at this specification document: http://amiga.serveftp.net/Datasheets/390544-01_8375_Agnus_specification.pdf

Needless to say, many of the variants have different pinouts, so they're not always directly exchangeable.

I'm not aware of anything major that the 8375 'fixes' from the 8372.  Though I am aware of the many headaches I've had with repairing boards from customers who have helpfully decided to play "Agnus Roulette" by swapping out some random variant of the part, can't get the board working, and send it for repair, neglecting to mention that wee gem of information.  I then spend hours trying to figure out what's going on with it, eventually discovering the aforementioned game of roulette.  In addition to whatever the original fault was.

My advice is that if the current part is working, then don't mess with it.

I wonder if this explains some of the design choices of the 500+. Even a standard A500 can be made to use an 8372B 2MB Agnus with sufficient rework (Amazing Computing, February 1993). If Commodore had built a bunch of 8375s for the 600, maybe they designed the rev 8A to use the chips they had on hand rather than do another run of "obsolete" 1MB Agnus chips or do a run of pre-modified 6A boards and tap into the 8372-series stock intended for the 3000.

[Matt_H]

This looks interesting.

https://www.buffee.ca/a-little-side-project/

A new Agnus

Weed

Nice! We need more projects like this to relieve supply constraints of the custom chips.