1
\$\begingroup\$

I have a problem understanding the following schematic, specifically the "Glue Logic(Memory Decoding)" section in the lower half middle:

circuit

This is a simple breadboard computer based on a 6502 CPU with RAM, ROM, and a D-Latch that serves as IO(seen on the lower right corner driving a single LED). The memory map is

  • $0000 - $3FFF: RAM (16K)
  • $4000 - $7FFF: I/O (16K)
  • $8000 - $FFFF: ROM (32K)

The one signal I don't understand is /RAMSEL. As shown in the schematic A15 and A14 are used to address the three chips(U2, U3, and U7). /RAMSEL feeds into /CE(chip enable) of the RAM chip(U3) which is active low, so if I what to read or write from RAM /RAMSEL has to be low(0).

Suppose I want to read the value located at $1000 in RAM(e.g., LDA $1000 in 6502 assembly). I expect the address bus to be $1000 which in binary is b0001000000000000. As can be seen the signals A15 and A14 are both low(0) making /RAMSEL high(1) through the glue logic. Therefore the signal on /CE is high, making the RAM chip not readable since it will be in high-impedance mode.

On the other hand, if I want to read from a ROM location say $F000/b1111000000000000, A15 and A14 are both high(1) and through the glue logic /ROMSEL will be low(0) (so the ROM chip can be read since its /CE pin is low) and /RAMSEL will be high(1) rendering it unreadable. Same applies for the IOSEL signal.

So as I understand this schematic, the signals for reading/writing to ROM and I/O works fine, but not for the RAM chip. Am I reading the glue logic wrong?

EDIT: This schematic is taken from this book, in case you wonder.

\$\endgroup\$
  • \$\begingroup\$ I agree. The /RAMSEL is wrong for the given memory map. U5B needs to be an OR gate \$\endgroup\$ – Icy Dec 1 '15 at 10:29
  • \$\begingroup\$ I had guessed so, but I was wondering since this comes from a book and he uses the same glue logic for another project - I wouldn't expect for such an error to happen twice. So I figured I might simply be misunderstanding the schematic but seems I'm right after all. \$\endgroup\$ – georgjz Dec 1 '15 at 10:41
  • 2
    \$\begingroup\$ It wouldn't be a Black Art if you didn't mislead the uninitiated... \$\endgroup\$ – Brian Drummond Dec 1 '15 at 10:58
1
\$\begingroup\$

You are correct, The /RAMSEL is wrong for the given memory map. U5B needs to be an OR gate.

\$\endgroup\$
  • 1
    \$\begingroup\$ Thanks, that is the right answer. Using the 74HC04 to invert A15 and A14 before feeding them int U5B will do the trick. Thanks a lot! \$\endgroup\$ – georgjz Dec 1 '15 at 11:01
3
\$\begingroup\$

Instead of address lines, the output of U5-A and output of U5-C should be fed into U5-B to get the proper ram select without using another chip or gate. (Romselect NAND ioselect = ram select)

\$\endgroup\$
  • \$\begingroup\$ Thanks, that would work too I guess, I might even try this once I come back to this project. Currently I'm working on a ROM emulator to go with it, looking good so far, but need to order the components. \$\endgroup\$ – georgjz Aug 31 '16 at 22:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.