2716 EPROM replacement with diode matrix?

[neazoi]

Hello,
I am onto a project that uses a 2716 EPROM to store micro-instruction logic. Only 64 addresses are used from this EPROM to store the code, so I was thinking to replace the EPROM with a diode matrix.
Could you suggest some examples, ideas or schematics of how to do this?

 

[betwixt]

That's a lot of diodes, you possibly need 8 x 64 = 512 although you may be able to miss some out, depending on the code it has to hold.

2716 is old, I still have some here but thats because i'm getting old too :lol:

You can use larger capacity devices which are cheaper and easier to program, just tie the unused address lines to ground and treat them as a smaller memory.

Brian.

 

[neazoi]

That's a lot of diodes, you possibly need 8 x 64 = 512 although you may be able to miss some out, depending on the code it has to hold.

2716 is old, I still have some here but thats because i'm getting old too :lol:

You can use larger capacity devices which are cheaper and easier to program, just tie the unused address lines to ground and treat them as a smaller memory.

Brian.

- - - Updated - - -


The micro instruction code has mostly 1 in the bits. there are only a few zeros. If I could make a reverse diode matrix I could limit the use of diodes to only 31.
Any ideas?

Also what do you propose as an address decoder?
In the original schematic 6 pins are used for the address, so I need a 6 to 64 decoder.
I would like to use these ttl 74xx series 3-to-8 decoders (I do not remember the exact number now) but I do not know how to connect them.
A way I think, is to connect all the inputs of the 3-to-8 decoder chips (8 of them) in parallel. Then use a 9th chip and connect it's outputs to the chip enable pins of each of the decoder chips.

As I said it has to be compatible with the 27c16 eproms..

Any help would be appreciated.

 

[betwixt]

You can do it that way, the idea is that you parallel the lower address bits to each 74LS138 and use another to decode the high bits. Each of the lower address decoders has an enable input which you connect to the output of the high address decoder. There can be problems when using a diode matrix like this though, particularly with the speed they operate at. The combined decoding process and diode selection may be much slower than you could achieve with a real EPROM.

Is there a reason why a real 27C16 or larger more modern device can't be used?

Brian.

 

[Alan69]

What you're talking should work, but sounds like a hot mess. What's the speed required, or is speed even an issue, just that it decodes..

Take a small PIC or similar with enough I/O for the ins and outs, and just a simple look up table and put the data out continuously. It will likely be faster than the 2716 if speed is an issue, and won't matter if not. For sure it will be something cheap, simple, and available vs all those parts.

 

[IanP]

- - - Updated - - -



The micro instruction code has mostly 1 in the bits. there are only a few zeros. If I could make a reverse diode matrix I could limit the use of diodes to only 31.
Any ideas?

Also what do you propose as an address decoder?
In the original schematic 6 pins are used for the address, so I need a 6 to 64 decoder.
I would like to use these ttl 74xx series 3-to-8 decoders (I do not remember the exact number now) but I do not know how to connect them.
A way I think, is to connect all the inputs of the 3-to-8 decoder chips (8 of them) in parallel. Then use a 9th chip and connect it's outputs to the chip enable pins of each of the decoder chips.

As I said it has to be compatible with the 27c16 eproms..

Any help would be appreciated.

It’s a terribly complicated solution, but this is how it may go:

You need n-to-2{to-the-power-of} decoder, so for 64 unique addresses the decoder will be 6-to-64 (or n-to 2!n), as it is done in a typical ROM, where after the address decoder each output line of decoder has 8 connections through a fusible diode... A diode n-end connects to a data bit output...

The 6-to-64 address decoder can be done with, say, 1x 74139 and 4x 74154.
74138s are not really a good choice!

Once you have working decoder, you can create a matrix with diodes to produce the desired 8-bit D0-D7 output.

:wink:
IanP

 

[neazoi]

Thank you all for the replies,
what short of speed would you expect if using 74F series in the decoder and 1n4148 diodes for the diode matrix?

 

[betwixt]

Well, the 2716 isn't a particularly fast device anyway, it is after all, about 35 years old!

The speed of a diode matrix is difficult to guess, the problem isn't the address decoding although an 'F' device will help a little, it's the fact that the output of the matrix will be passive rather than active. In the real IC, the cells produce a charge which is sensed and used to drive a totem-pole output but in the diode matrix you have to rely on current through the diodes and pull-up resistors to provide the logic level. It is analogous to DTL logic, the technology before TTL took over. the way speeds were increased in DTL days was by lowering the pull-up resistor values but that makes them very power inefficient.

I'm not sure if small PROMS are still avaialable, they may be a viable alternative as you only use a few addresses, you could certainly use PALs but then you would need a programming unit for them.

Brian.

- - - Updated - - -