Address decoding upgrade with 74HC139

[AndersNielsen]

Michael wrote:

I wonder if a decoder method using a 'magnitude comparator' IC might be of use? Here's an untested method that comes close to your memory map and I/O requirements. While the method is meant to use a 64K or 128K RAM IC to provide much more than 32K of RAM, you could use a 32K RAM IC for your memory map along with one of those spare inverter gates for the active low chip select.

Stay safe. Have fun. Cheerful regards, Mike, K8LH

Interesting! Simply comparing the address high byte with the intended page is certainly a good one chip solution to do what I wanted But yeah, the single quantity 5€ price tag is pretty aggressive for a budget build like mine.

[floobydust]

Well, I would be inclined to look at a PLD (which are also pretty retro), something like a 22V10. If you have a way to program one of these, you can likely replace a good number of chips and have a more flexible solution. Even new ATF22V10 chips are relatively inexpensive, especially if you compare against multiple chip/socket costs.

[Michael]

AndersNielsen wrote:

Michael wrote:

I wonder if a decoder method using a 'magnitude comparator' IC might be of use? Here's an untested method that comes close to your memory map and I/O requirements. While the method is meant to use a 64K or 128K RAM IC to provide much more than 32K of RAM, you could use a 32K RAM IC for your memory map along with one of those spare inverter gates for the active low chip select.

Stay safe. Have fun. Cheerful regards, Mike, K8LH

Interesting! Simply comparing the address high byte with the intended page is certainly a good one chip solution to do what I wanted But yeah, the single quantity 5€ price tag is pretty aggressive for a budget build like mine.

I'm also on a limited budget and the 74HC682 price shocked me, too, but I bought a handful of chips from a vendor on AliExpress for about $1.19 each (including shipping) back in September 2021. I also have a cache of skinny 64K RAM chips and wide 128K RAM chips from AliExpress vendors which cost less than $1 each (including shipping)...

The 74HC688 is quite capable and more reasonably priced and might be worth a look.

Cheerful regards...

[AndersNielsen]

floobydust wrote:

Well, I would be inclined to look at a PLD (which are also pretty retro), something like a 22V10. If you have a way to program one of these, you can likely replace a good number of chips and have a more flexible solution. Even new ATF22V10 chips are relatively inexpensive, especially if you compare against multiple chip/socket costs.

I certainly wouldn't be opposed to doing that for a one off build but since this project is specifically aimed at helping "newbies" - and people with lots of old stuff - make a cheap and easy build, I'm afraid the PLD's complicate things too much.
Still have to play with them though - just haven't gotten around to it yet.

Michael wrote:

I'm also on a limited budget and the 74HC682 price shocked me, too, but I bought a handful of chips from a vendor on AliExpress for about $1.19 each (including shipping) back in September 2021. I also have a cache of skinny 64K RAM chips and wide 128K RAM chips from AliExpress vendors which cost less than $1 each (including shipping)...

The 74HC688 is quite capable and more reasonably priced and might be worth a look.

Certainly getting a few the next time I stock up on recycled parts from aliexpress - looks like a part with plenty of uses

[BigDumbDinosaur]

floobydust wrote:

Well, I would be inclined to look at a PLD (which are also pretty retro), something like a 22V10. If you have a way to program one of these, you can likely replace a good number of chips and have a more flexible solution. Even new ATF22V10 chips are relatively inexpensive, especially if you compare against multiple chip/socket costs.

Also, if you correctly design your logic, use of a PLD tends to produce lower overall propagation times. The ATF22V10 is available in speeds down to 7.5ns pin-to-pin. Using purely combinatorial logic and no pin nodes, a GAL will readily outperform discrete gates.

[BillO]

floobydust wrote:

Well, I would be inclined to look at a PLD (which are also pretty retro), something like a 22V10. If you have a way to program one of these, you can likely replace a good number of chips and have a more flexible solution. Even new ATF22V10 chips are relatively inexpensive, especially if you compare against multiple chip/socket costs.

I'd have to agree with this. Once I started using GALS I never went back. It's not a cheat. You still have to figure out the logic, but you can replace a half dozen chips with one and drastically reduce propagation delay as well as enjoy big advantages in PCB layout and being able to make changes without having to redo the copper.

[BillO]

AndersNielsen wrote:

I'm afraid the PLD's complicate things too much.
Still have to play with them though - just haven't gotten around to it yet.

Once you begin to play with them you will find that the reality is quite the opposite. They simplify things to a great extent. There is a reason modern digital electronics is dominated by FPGAs, CPLDs and ASICs.

[AndersNielsen]

BillO wrote:

AndersNielsen wrote:

I'm afraid the PLD's complicate things too much.
Still have to play with them though - just haven't gotten around to it yet.

Once you begin to play with them you will find that the reality is quite the opposite. They simplify things to a great extent. There is a reason modern digital electronics is dominated by FPGAs, CPLDs and ASICs.

Programmable logic is amazing - however, besides already programming the ROM the barrier of entry would be too high with a PLD in the mix. I also previously had a tiny MCU in the mix to take some load off the video but that also had to be sacrificed to make it easier to assemble, program and solder
If I was programming them myself or selling them that wouldn’t be a problem, but since I’m expecting people to source the parts and program it themselves it’s a different matter.

[floobydust]

Well, it is your project of course, so you can choose how to proceed going forward.

Also, if you're expecting folks to be able to build anything (old parts or new), having some minimum set of known working tools is a must. Also, there's no guarantee that the old parts folks scavenge up will be 100% working, and diagnosing bad parts when building something for the first time is not something a novice/beginner will have much luck at.

Granted, I applaud your efforts and the enormous amount of time you're putting into this. Neat project, but I still think a simple PLD would simplify the design, and a newbie wouldn't need to care much about the PLD programming, much like they don't need to care about how the CPU itself works internally. A lower parts count is more attractive for a beginner as well, especially for a entry level project.

Just my $0.02 of course.

[AndersNielsen]

floobydust wrote:

Just my $0.02 of course.

I appreciate it a lot

It's probably not a huge audience.

I expect the people to jump on board will have an interest in soldering the board themselves, I expect them to have a TL866 II - which afaik can probably program some PLD's as well - or similar to program the ROMs (or to get help from someone who can - not completely ruling out selling a kit but I doubt I have the time). I expect them to install CC65 and clone the github repo to assemble the source - ROM's ready in literally three commands on Linux/MacOS.
I expect them to have an interest in learning 6502 assembly and/or BASIC - or just using it as an MCU with a keyboard, VGA, lots of I/O, wireless, etc, for benchtop tasks like playing with an I2C/SPI device, and that kind of thing..

I don't expect them to completely understand the hardware when putting it together. I don't expect them to setup another toolchain for another MCU, a PLD, FPGA etc just to get it running.

Maybe a PLD wouldn't add much, software wise, but I guess I'm going for more of a 1977 vibe, keeping it to 7400 series logic and simple ROMs. ... conceptually anyway. And then cheating a bit with the 90's ROMs and SRAMs.

[plasmo]

You can also build a simple 6502 SBC without any glue logic. The hardware is simple, but explaining the bootstrap process is more complicated. My feeling, however, is that people can grasp complexity in software better than elaborated hardware. viewtopic.php?f=4&t=6455&hilit=Zg6502
Bill

[Michael]

Another untested 2-chip decoder method inserts a single page of I/O and four strobes into the 32K RAM address space (within pages $02..$7F) using the less expensive 74HC688 'magnitude comparator' IC but it does require a 64K or 128K SRAM chip. You could also use your 74HC30 + Invertor gate in place of the 74HC688 for a fixed page of I/O at $7F00..$7FFF.

Cheerful regards, Mike

[Michael]

One last idea, if I may? If you're simply trying to reduce the chip count and you don't mind dedicating an 8K block for I/O, how about a single 74xx139 decoder/glue IC and borrow Daryl Rictor's trick from the "6502 Badge" and use the ROM write strobe to write to the '273? Drawing below (yes, it still requires a 64K or 128K SRAM IC)...

Regards...

[BigEd]

(It's quite possible that one person would offer to program up a CPLD for another person. That sort of cooperation really helps to lower barriers. Same with EPROM of course.)

[plasmo]

If the CPLD is EPM7064SLC44 or EPM7032SLC44 (I even have few EPM7128SLC84 spare), I'm happy to supply the part and program it to user's specification. However, shipping (even within USA) will be the big part of the overall cost.
Bill