2 posts / 0 new
Last post
PLA Manufacturing & History

Hi Bil, quick question for you.

I recall from your previous comments that Commodore learned to make their own PLA's by copying the Signetics parts that were used in the earlier 64's. My question is were the later ones, such as the one used in the plus4, C116's & Commodore 232's still programmed post production or were the gates fixed on the mask?

cheers, Rob 

The NMOS versions were never

The NMOS versions were never programmable themselves that I ever knew  as the original Signetics parts were fuse based technology. So from the beginning our PLA's were "programmed" or had their font determined at chip production time.
 
The way we turned in the settings/font to the chip people was weused the same form that we used to specify the 82S100 "1's", "0's" and "X's" used to configure columns of AND'ed terms with rows OR'ed together.
 
So instead of fuses we used “pre-ohmic contacts” and “diffusion slugs”.  Using 6 standard layers in the NMOS process the diffusion slug was layer 2  and is deep in the chip compared to the pre-ohmic contact which was layer 5.  The layer 5 contact connected the metal layer 6 to the polysilicon layer 4, so basically it programmed the interconnect between transistors.
 
Micheal Angelina, the head of the Calma department, which was the name of the IC layout department, would take the sheet and run scripts or macros or somehow create a PG Tape for MOS for the two layers we wanted to program.
 
We would leverage the fact that one set of the programmable variables was up near the top of the layers; we would run a split lot where we would do half the lot in all layers 1-6 layers but hold half the lot after layer 3.  I think we also created some spare terms (rows) that had some diffusion in useful places but otherwise didn’t do anything unless we changed the contacts.
 
On the C128 we did this where we missed a “1” under IO bleed through mode of Ultramax mode and so International  Soccer wouldn’t play, well actually it played you just couldn’t see it.
 
We had had a split lot held back and so with some creative term additions (I think we could add “1”s or something like that) we were able to fix the second half of the lot, basically took less than a week to get the second batch.
 
We used to joke that they would forget the layer 7 for the PLA’s, where 7 was actually the silicon Nitride used for passivation to seal the chip.  The early chips especially had holes in the passivation layer and would absorb water and then act very strange.
 
Bil
 
 

Log in or register to post comments