ATF16v8 for 2.06 decoding

[olivier.jan]

I haven't managed to do a proper address decoder for TOS 2.06 using an ATF16v8b and I'm not too sure why.
The exact same code (down to the PIN numbers) works flawlessly with an ATF22v10C, but with a 16v8 the machine (an H5) doesn't boot and stays on a very colourful screen.
The only difference I've found between the two GALs is that the 16v8 has a weak pull-up to VCC on all pins while the 22v10 hasn't.
If anyone has a clue on the issue, I have a stock of 16v8 and only a few 22v10.
I simply connect the GAL directly to the CPU socket and the ROM2 signal (JP13 on the H5).

Here is the pretty simple code:

Code: Select all

Name     STROMDEC ;
PartNo   00 ;
Date     12/23/2020 ;
Revision 01 ;
Designer Olivier ;
Company  home ;
Assembly None ;
Location home ;
Device   g22v10 ;



/* *************** CONNECTIONS *******************  */
/* ATARI A23..A16 = GAL 1..8                        */ 
/* ATARI AS       = GAL 9                           */ 
/* ATARI RW       = GAL 13                          */ 
/* ATARI ROM2     = GAL 14                          */ 
/* ATARI DTACK    = GAL 17                          */ 
/* ROM CE         = GAL 15                          */ 
/* ************************************************ */ 


/* *************** INPUT PINS ******************** */

PIN   [1..8] = [A16..23] ;
PIN      9   = !AS ;
PIN      13  = RW ;
PIN      14  = !ROM2 ;

/* *************** OUTPUT PINS ******************** */

PIN      17  = !DTACK ;
PIN      18  = !CE ;

FIELD TOS = [A23..16];

/* *************** DECLARATIONS AND INTERMEDIATE VARIABLE DEFINITIONS ******************** */

CE256 =  RW & AS & TOS:'h'[E00000..E3FFFF] ;


/* *************** ROM DECODER ******************** */

DTACK = CE256 ;
DTACK.oe = CE256 ;
CE    = CE256 # ROM2 ;

[exxos]

I'm not 100% sure, but the 16v8 does not have tristate outputs ?

Edit: there are also some odd bugs that you shouldn't control OE and the variable with the same event...

You should have something like

DTACK = 'B'0;
DTACK.OE = CE256;

Or define DTACK as open collector in atmel options..

EDIT2:
Also be aware of the number of logic terms per pin.. They vary from 6-14 ( dont remember exactly offhand ) depending which output pin you are using..

A16-23 is 8 terms, plus AS & RW = 10 terms ..

[stephen_usher]

Indeed, it's not tri-state, so I had to use a 74LS transceiver in my decode circuit for the bus interface.

[exxos]

Indeed, it's not tri-state, so I had to use a 74LS transceiver in my decode circuit for the bus interface.

Could work around DTACK and just drive it with a diode ...

[stephen_usher]

Here's my schematic. (Note that I have the ROM on the same board as the decoder and it has the ability to switch between ROM images in a 512K ROM or use the whole ROM.)

_ROE is the ROM /OE line.

ROM-decode.jpg

And here's my GAL code:

Code: Select all

Name     TOS 2.06 ROM decoder (512K) ;
PartNo   00 ;
Date     24/10/2020 ;
Revision 01 ;
Designer Engineer ;
Company  Personal ;
Assembly None ;
Location  ;
Device   g16v8 ;

/* *************** INPUT PINS *********************/
PIN  1   =  A18                       ; /*                                 */ 
PIN  2   =  A19                       ; /*                                 */ 
PIN  3   =  A20                       ; /*                                 */ 
PIN  4   =  A21                       ; /*                                 */ 
PIN  5   =  A22                       ; /*                                 */ 
PIN  6   =  A23                       ; /*                                 */ 
PIN  7   =  !ROM2                       ; /*                                 */ 
PIN  8   =  !AS                       ; /*                                 */ 
PIN  9   =  RW                       ; /*                                 */ 

/* *************** OUTPUT PINS *********************/
PIN  19   =  !ROE                       ; /*                                 */ 



SELECTED=RW & AS & !A19 & !A20 & A21 & A22 & A23;

ROE=SELECTED # ROM2;

P.S. The 74LS244 is also useful for delaying /DTACK slightly so as to make sure that the output from the ROM is stable on the bus, as per the timing diagram.

[olivier.jan]

Thanks a lot, I was really thinking the 16v8 was just a smaller version of 22v10, but clearly the features are different and tri-state doesn’t behave in the same way. I’ll use a 22v10 this time and will use a 16v8 + LS244 next time.

[olivier.jan]

Clearly the 22v10 is easier, took me less than one hour to get it working properly on a perfboard.

77401D0D-9C2F-4A45-800F-7DFB28A30456.jpeg

[PhilC]

@olivier.jan nicely done.