[sdiy] TC5514 polysix memory...upgrade?

[Bob Weigel]

yeah they were a little skimpy on the memory.  Seems crazy that they 
didn't just make it 8 bit with twice the memory.  Chips were spendy back 
then but people would have dug the product way more if it had 64 instead 
of 32 patches too.

Anyway THANKS for the info there.  I haven't kept up on memory chips of 
late.  I tried to build an expansion for the MKS70 some years back and 
did what looked like the obvious things in the design but for some reason

1) You can always store a patch anywhere singly.  No problem.

2) If you try to bulk load patches to a section of the CY...whatever 
that chip was.. it will follow the same pattern of corruption every 
time.  It will consistently foul the same data the same way. The "A" 
bank of patches will load fine.  THen there is gradual degredation as 
the bulk dump continues of the data.  Really bizarre.

Anyway I laid it aside.  I had the thought that maybe I could use the 
cards in the MKS30.  I just built a board with a switch to select 16 
banks so when I get time some day I'll see if it works on those things. 
What a bummer waste of time that project was.



ASSI wrote:

>On Samstag 03 Mai 2008, Bob Weigel wrote:
>  
>
>>ANyway I was thinking it'd be cool to design even a simple daughter
>>board that has leads for a switch that could be accessed somewhere
>>non-intrusive that would give the ps6 more memory by switching a
>>couple more address lines.  Does anyone know a compatible 4 bit data
>>out chip that would pull this off with 16K or something? -Bob
>>    
>>
>
>I just had to look up in the schematics (thanks, oldcrow!) why you would 
>absolutely need a 4bit SRAM.  Hilarious that they saved the second RAM 
>and really just use the lower 4bit of the data bus...  Anyway, Cypress 
>still makes 5V x4 SRAM, the 16kx4 part is the CY7C164/166.  However the 
>lower-density parts have rather high active and standby power and I'd 
>think about using a 64kx4 instead (CY7C194BN) even if you end up 
>wasting some of it's capacity.  The data sheet doesn't say anything 
>about a data retention mode, the only part where this is explicitly 
>specified is the 256x4 CY7C1006.  Even then, the max. data retention 
>current is specified at more than 1000x the current for the TC5514-APL.  
>The typical values would likely be much lower, but it's something to 
>check.  While you're at it, replace the NiCad with a lithium block and 
>the D15 diode and the R91 resistor with a Schottky diode in antiserial 
>arrangement.
>
>The Simtek nvSRAM are also a strong consideration even though they're 
>not available as x4: STK12C68 (8kx8) or STK14C88 (32kx8).  They store 
>the SRAM content into a non-volatile shadow array on power down and 
>recall the information on power up, so there is no need for a battery 
>at all.  They'd be fast enough so you could rig a read-modify-write 
>state machine with a GAL for stuffing two x4 nibbles into one x8 byte.
>
>
>Achim.
>  
>