Re: MOS8520R4 - 1988 vs. 1991

From: Segher Boessenkool <segher_at_kernel.crashing.org>
Date: Wed, 29 Jan 2020 07:41:24 -0600
Message-ID: <20200129134124.GM22482_at_gate.crashing.org>
On Wed, Jan 29, 2020 at 12:08:46PM +0000, Ed Spittles wrote:
> On Wed, 29 Jan 2020 at 11:54, Baltissen, GJPAA (Ruud) <ruud.baltissen_at_apg.nl>
> wrote:
> 
> > > Though they changed the masks which was expensive
> >
> > I can see only one reason: the result saved money in some way, enough to
> > justify the costs.
> >
> 
> Perhaps only one mask changed, which is a lower cost than changing all. As
> it changes the bond pads, it just might have been necessary because of some
> change to the bonding or packaging process.  Or, it improved yield, which
> as you say, saves money. Or, just possibly, it improved reliability, with
> the same result.

Yes, and you have to have new masks after a while *anyway*, for normal
wear and tear, so it is easy to slip in something low-risk like a pad
metallisation improvement at the same time.

> > - they didn't want to introduce new part numbers.
> >
> > If a factory had to change the type number of a car for every small change
> > during production, then we had to get customed to something like Ford Focus
> > T394.
> >
> 
> If the chip behaves the same logically, perhaps there's less of confusion
> in not bumping the revision number.  It's not uncommon to distinguish major
> and minor revisions, although I see no other revision counters on the die.

It's functionally the exact same chip, even.


Segher
Received on 2020-05-30 00:29:21

Archive generated by hypermail 2.3.0.