Self-made VDC RAM Upgrade

[Hydrophilic]

There are few threads that relate VDC RAM Upgrade, probably the most relevantly titled is here, but it is asking about problems with existing product.  (I was encouraged to start a new thread when I tried to reply to that topic).

I've recently acquired some 64K x 4 bit DRAMs, part# 41464-10 (date code 9447).  Although they are 18-pin DIPs, I wasn't really sure this was the right part number (I always thought they were 4464's), but a photo in a recent post confirms the part# 41464-08.

However the part# in photo ends with -08 which I'm thinking means 80ns response time, while mine ends with -10 which I'm thinking means 100ns response time.  Anybody care to guess if -10s will work?

With 16MHz dot clock of VDC, I'm thinking 2MHz cycle access (assuming 8pixels/char width) which should be 1/2MHz = 500ns access time.  Of course the DRAMs use multiplexed addressing so really divide by 2 for 250ns access time.  So I think the -10 chips (100ns) I got should work...

Now obviously I could buy one of those converter boards, but as I understand the RF shield / heat sink will no longer properly fit once installed in a flat C128 (which I have).  Plus doing things yourself is more fun!  Another reason for me is to be able to switch from 16K to 64K mode, for programming compatiblity testing (if that sounds crazy, you should see my PC which offers multi-boot of DOS 6.0, Windows 3.11, '95, NT4, '98, 2K, ME, and XP... no Vista, maybe Win 7 soon)

I've heard you can simply cut out the existing 4416's and replace with 41464's.  Can anyone confirm this?  I'm concerned because the motherboard might not have all the lines connected (save money on copper etching?)

Assuming the "cut and paste" method would work, I plan to use an alternate method.  I would like to piggy-back the 41464's onto the 4416's with exception of Vcc.  Then install a switch to power either the 16K DRAMS or the 64K DRAMS.  How does that sound?

[Blacklord]

(I was encouraged to start a new thread when I tried to reply to that topic).

Hadn't been posted to in over 120 days - you can reply to it though.

Lance

[RobertB]

However the part# in photo ends with -08 which I'm thinking means 80ns response time, while mine ends with -10 which I'm thinking means 100ns response time.  Anybody care to guess if -10s will work?


(snip)

I've heard you can simply cut out the existing 4416's and replace with 41464's.  Can anyone confirm this?  I'm concerned because the motherboard might not have all the lines connected (save money on copper etching?)

Hmm, I say that these esoteric questions should be asked of a man who would definitely know... master technician Ray Carlsen.  Send him an e-mail.


Truly,
Robert Bernardo
Fresno Commodore User Group
http://videocam.net.au/fcug
The Other Group of Amigoids
http://www.calweb.com/~rabel1/
Southern California Commodore & Amiga Network
http://www.sccaners.org

[richardc64]

Leaving active signals on an IC that's not powered isn't a good idea. Worst case would be damage to both the unpowered chip(s) and the one supplying the signals. You'd be better off switching /CAS between the two pair of DRAMs, but why keep the 16Kx4s when you can just tell the VDC how much memory to use?

[Hydrophilic]

/CAS instead?  Thanks, I would hate to fry my VDC!  (I have a spare but who knows what else could fry?)

...but why keep the 16Kx4s...?

The reason is software testing / experimenting.

With 16K chips, a single byte can be read/written from multiple addresses, as you would expect.  The mapping works like you might think when the VDC is told it has 16K.  Unexpected is how the mapping works when the VDC is told it has 64K chips (but only 16K in reality).  I went into a bit of detail in this post of another thread.

Completely useless from a practical standpoint.

Thinking about it that way, maybe I shouldn't do it?  Running a long wire for a switch shouldn't pose much of problem if it is only carying +5V, but if it is carying 2MHz data it might cause / receive some RF interference.  Guess I need to think on it...