[Simh] DEC VT emulators on MAME

[Timothe Litt]

> Since they're windowless, they are not EPROM (remember what the E
> stands for), but plain ROMs.
Nope.  I meant exactly what I wrote. 

I wrote "they're windowless, so OTP" - one-time programmable.   
Internally they ARE UVEPROMs.  They are exactly the same dice (silicon)
as the windowed 27C256.

But with no window, they can't be erased.

This was mostly a marketing thing - allowed a lower volume price for
production; the windows were expensive to produce.  Once volume is
attained, the package/packaging can cost more than the die.  Initially,
windows could only be produced in ceramic packages, which are way more
expensive than plastic.  (Plus, command a premium because of their
thermal characteristics - e.g. for mil spec applications.)  I think
eventually someone came up with a plastic package with a window.  (It's
hard to include a window because plastic packages are injection
molded.)  But it's still more expensive. 

By using the same dice, there was absolutely no chance that switching to
the OTP parts would introduce a bug.  And no chance that the adhesive on
a label over the window would dry out, allowing unintended erasure from
fluorescent lighting/daylight.  So reduced cost for everyone was a winner.

You could debug with many-times reusable parts, which saved time and
money.  For volume, you used electrically the same part at a much lower
cost.  Both to the semiconductor manufacturer and to the product that
used it.  And the manufacturer could program the parts at final test
very inexpensively since no handling was required.  (Of course,
distributors tried to horn in by offering faster turnaround.)

In any case, the bottom line is that they ARE (OTP) (UV)EPROMs.  The
dice name the part.  The fact that the E isn't accessible in that
package does not change the fact or the designation.  I know, it's
weird.  But marketing is like that.

As for which signal you use for what - it doesn't matter.  OE puts the
chip into a low power state just as effectively as CS - assuming that
the part isn't in programming or ID mode.  Since the part is never
written (in the terminal), this effectively gives you 2 CS pins
(effectively ANDed), and thus decoding requires at most an inverter.

The 27C256 is a 32K x 8 part; it has no A15 (but the cartridge socket does.)

Keven pointed out that the odd chip is probably the character generator
ROM - thus the separate address and data bus - and it doesn't need a CS
or OE.  It's always reading something.

As I've written before, rather than guessing, a few minutes with an
ohmmeter can sort all this out.

I'm leaving that - and further exploration - as an exercise to the reader.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.trailing-edge.com/pipermail/simh/attachments/20170418/2975962d/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4577 bytes
Desc: S/MIME Cryptographic Signature
URL: <http://mailman.trailing-edge.com/pipermail/simh/attachments/20170418/2975962d/attachment-0001.bin>