30 Apr
2009
30 Apr
'09
7:29 p.m.
I was thinking about that very subject just this week
- I have a
TRS-80 Model III that I think has a flaky 4116. I have a chip tester,
but it only tests 64Kbit, 256Kbit and 1Mbit DIP DRAMs (i.e. -
single-supply parts). Is this hack as simple as eliminating the -5V
and +12V to the socket, or is there more to it? I have tubes and
tubes of NOS 4164s, and could easily build a 1-2-socket stacker with
missing pins to plug a 4164 into a 4116 socket if that's all it takes.
The differences are :
Pin 4116 4164
1 -5V N/C
8 +12V +5V
9 +5V A7
IIRC (but do check this). (Note that pin 16 is ground -- the main power
and ground pins are still the diagonally-opposite ones you'd expect, but
they're the other way round to the normal configuration!)
So what you need to do is isolate pin 1, disconnect pin 8 from the PCB
and connect it to pin 9 (or another source of 5V). This will use 1/4 of
the chip (A7 will be tied high, rememebr it's a multiplexed address, so
effectively 2 pits of the full address are forced to '1's).
One problem could be the refresh. Obviously with A7 tied high, the chip
only sees a 7 bit refresh (on pins A0-A6). Some 4164s only needed a 7 bit
refresh anyway (so no problem), of the others, many will refresh the
cells in use if A7 is tied high (it may not refresh some of the cells
that are never used, in the remaining 3/4 of the chip, but that shouldn't
matter). FWIW, I've done this a couple of times using randomly selected
4164s, and had no problems
I normally do it by a little kludge. Namely bend out pins 1 and 8, and
solder a jumper wire on top of the package between the bent-out pin 8 and
and pin 9. Stick it in the 4116 socket, making sure that pins 1 and 8 do
not make contact with the socket.
-tony