New 512K flash cartridge design, rewritable by 4A

[+Ksarul]

Looks like I need to go on a bulk chip hunt again. . .   

[+FarmerPotato]

For

2 hours ago, Ksarul said:

Looks like I need to go on a bulk chip hunt again. . .   

For ATF22LV10?

 

These are a good replacement for PAL and GAL lines. Still new from Mouser at $1.45.  But they are a dead end. For one thing, just 10 bits of memory tied to the outputs. The ATF750 and ATF1500 offer more (hidden flip flops), but they aren't programmable in the TL866.

I think my next step is ice40lp384-SG32 .. it's a tiny FPGA for $1.31. Gives you 21 IOs. Only downside is two degrees harder to solder on. Needs an external EEPROM.  But Lattice has been very good to me in a lot of ways.  Somehow I overlooked this little baby member of the family. 

I'm obsessed with minimizing costs as a design goal.

 

[+Ksarul]

I tend to look for chips that I need at the 500-1,000 chip level, as then I have enough to last me a long time. . .and it keeps my cost per chip pretty low, as long as I'm willing to deal with the upfront transaction costs.

[RickyDean]

On 7/20/2019 at 5:39 PM, Ksarul said:

I tend to look for chips that I need at the 500-1,000 chip level, as then I have enough to last me a long time. . .and it keeps my cost per chip pretty low, as long as I'm willing to deal with the upfront transaction costs.

Yeh... that's my problem, I'm so broke, that when I need to buy chips, I have to get them in the 10 pieces level, just getting the first price break. Hopefully things will get better soon.

[+Ksarul]

I just bought 1,200 of the ATF22V10s for about 75 cents each. These are the DIP chips though, not the PLCCs. I'm still waiting to see if the PLCCs come through. . .but now I will definitely have enough of them for this and one other project I was working on that needed 22V10s.

[+Ksarul]

500 of the PLCC chips cost me a few dollars more than 1,200 of the DIP chips did, but I now have both of them on the way. Now to work an upgrade into the UberGROM board to see if I can update the ROM side in system. . .without affecting the 1284.

[+FarmerPotato]

1 hour ago, Ksarul said:

500 of the PLCC chips cost me a few dollars more than 1,200 of the DIP chips did, but I now have both of them on the way. Now to work an upgrade into the UberGROM board to see if I can update the ROM side in system. . .without affecting the 1284.

holy crap that's a lot of chips.

 

but.. 22LV10C-JU10 is $1.28 at Mouser, so 500x would be $640. This has 5V input tolerance according to the data sheet.

 

I hope you are not going all-in on 22LV10 on just my recommendation. I'm still evaluating them in-system. I've only fully tested at 3.3V. Waiting for cartridge boards.

 

 

[+Ksarul]

One note, I am using the V10 variant, not the LV10, as I try to stay with five volt components where possible. In the case of the DIP chips, I have two other boards that use them, so it was just an opportunity to get a bunch of them. Note also that I have to add about $40 in taxes to a $640 purchase, so once that and shipping (if any) are calculated in, the cost is a little higher. These just give me a good reason to begin learning the necessary coding rules for PLDs.

[+Ksarul]

Your PLD has me on a fun adventure, redesigning the ROM side of the UberGROM to make it programmable in-system. I had most of what needed to be there in place already--and the PLCC will fit perfectly in the slot now used by the 74LS378. Now I just have to move all of the wires around to make everything connect efficiently.  This is my idea of real hobby fun. Many thanks for the useful prodding! 

 

I also went to the Post Office this morning to pick up the 500 PLCC chips I ordered, so that item is in place too.    I'm still waiting on the arrival of the DIP variants, but they are for something else I've been working on off and on.

[+Ksarul]

I made it fit! I also have jumpers in place to disable programming either or both sides of the board. The 1284 already had a functional jumper to do that, but the two jumpers that were needed to set the ROM side were only usable in the default (disabled) positions on earlier iterations of the UberGROM.

 

I did notice that the two Mode pins have no connection in your design. Will that remain that way, or should I try and pull them out to some header pins for future use?  LEDs, maybe?

 

I guess I should order a few test boards now. . . 

[+InsaneMultitasker]

On 7/22/2019 at 3:02 PM, FarmerPotato said:

holy crap that's a lot of chips.

 

Ksarul is the TI-99/4A  "incarnation" of Mouser

 

[+FarmerPotato]

33 minutes ago, Ksarul said:

I made it fit! I also have jumpers in place to disable programming either or both sides of the board. The 1284 already had a functional jumper to do that, but the two jumpers that were needed to set the ROM side were only usable in the default (disabled) positions on earlier iterations of the UberGROM.

 

I did notice that the two Mode pins have no connection in your design. Will that remain that way, or should I try and pull them out to some header pins for future use?  LEDs, maybe?

 

I guess I should order a few test boards now. . . 

That's great!

 

Correct, M0 and M1 do not connect to anything. In my design, they feed back internally and are not used externally. The Output Logic block contains the flip-flop. If you use the flip-flop for storage, that is also one output pin. That's right, you get at most 10 bits of storage.

 

Later chips like the ATF750 and ATF1500 have 20 or 40 internal flip-flops, "buried" memory bits. 

 

You could connect an LED; I didn't want to try to steal 20mA from the logic high though. You could drive a transistor. Tell me if there is an easy way to do it. I should just test it on my tester breadboard, duh.

 

I hope you will verify any WinCUPL code before building your hardware.

 

[+FarmerPotato]

My next design will likely use the ICE40LP384, which still has at most 21 I/O pins for same money as the 22V10, but is an FPGA. With that, I could see multi-plexing 16 address lines in (or just 13) in two bytes, plus outputting to a serial latch using just 2 or 3 lines, so the whole banking logic takes 12 I/Os. External logic would be 2 LV245A, and a LV165, driving the flash, and 8-pin EEPROM. That leaves 9 I/Os for other purposes. I've already built this in Verilog and verified timing. 

 

Anyhow, FORTI-2 has a 22V10 working quite well now.  It minimizes use of the really valuable I/O pins on the main ICE40HX4K, which is the FPGA that creates the digital sounds. 

 

ICE40HX4K has 7680 cells (it's really an 8K), the ICE40LP384 has... 384. Priced accordingly, for different jobs.

 

[+FarmerPotato]

21 minutes ago, InsaneMultitasker said:

Ksarul is the TI-99/4A  "incarnation" of Mouser

 

Ksarul set me up with chips! Like you said.

[+acadiel]

Ksarul set me up with chips! Like you said.

Definitely. His basement must be packed!

[+Ksarul]

Understatement. . .I got the 1,200 DIP chips I had on order today. My next goal is to try and build a SAMS card with as many of the Russian clone chips from the USSR days as I can find to see whether they work like they should. . .that should be fun. I have some of the chips I need already, and some more on the way, but not all of the chips I need had clones--just most of them. 

Edited July 30, 2019 by Ksarul

[GDMike]

And you make one heck of a great SAMs card.. mine has worked absolutely beautiful and that means all areas, as Forth does very well with this card, as does Rich's basic from what I hear, but I need to send you this ramdisk let you make it a Zombie Godzilla Hercules FrankenDrive!! Ok, 384K would be ok. Lol