Re: external ram expansion modules for 16k 264 series machines

On 2012-07-30 18:14, Gerrit Heitsch wrote:
> On the C116/C16 they could have used the spare 74LS125 gate together
> with the spare 74LS02 gate and 2 pullup resistors to the same effect.
> Very strange that they didn't since it wouldn't have meant any costly
> extra parts.

Yep.

>> I couldn't believe that the known external ram modules simply suppress
>> CAS' by some ugly hacky way, but as it seems this is indeed the case.
> 
> Might explain some of the dead TEDs...

Arranged another quick test; connected a trimmer between Vcc and CAS'
(using solder pads close to the expansion port), watched CAS' on the
o'scope, and tuned the trimmer until obtaining similar-ish waveforms to
what I had seen previously. Then measured the trimmer's resistance. It
took around 64 ohms to suppress CAS', and a little bit less to obtain
very similar waveforms. So, 1.) the cartridge might just pull the CAS'
line up by a simple resistor, the strange waveform is just "normal". 2.)
Supposed that we use a pullup of 56 ohms, and we make at least 2.4 V on
CAS' (standard TTL V_IH minimum), that makes it 5-2.4=2.6V on the
resistor, therefore, current is ~0.046A. That is, the TED has to sink
around 46 milliamps of current on its CAS' pin, whenever it goes low.
Not bad, obviously way above any absolute maximum ratings. Personally, I
have no doubts that this could kill some unlucky TEDs.

> These expansions I can only see working by letting the output drivers of
> the RAM chips fight it out during a read cycle and hope that the
> internal RAMs lose by using 41464 with stronger drivers than the TMS4416
> posess. At least this expansion should only be able to damage the
> internal RAMs or its own RAMs but not TED.

The data outputs of industry standard 4*16k and 4*64k chips are both
rated to drive 2 standard (74xx) TTL loads, no difference in fan-out
specs, that is, factory data doesn't seem to indicate that 64k chips
could/should overdrive 16k chips (safely, or at all) in this scenario,
at least in a predictably large percent of cases. There must be some
other trick here IMHO.


Levente


       Message was sent through the cbm-hackers mailing list