My hardware project :: GCart 2011

TLovskog · Post by **TLovskog** » Tue Dec 13, 2011 3:35 pm

pcbcart is what I usually use. This time it was dorkbotpdx.org and their PCB sharing ...

TLovskog · Post by **TLovskog** » Mon Feb 06, 2012 3:15 pm

It took an absurd amount of time, but I have understood that I was not alone. Alot of problems and vacations came in the way. I think I have waited 3 months in total for these three boards. A personal record for me.

Well anyway here they are and they where very cheap and seem to be of high quality, so I guess that is fine all in all. Mext stop soldering.

I will do this in a couple of steps ...

1) The Xilinx circuit, buffers and SRAM. Once that works ...
2) PIC24, SD CARD and RTC with battery. This will test the booting and code injection.
3) MP3, OLED with buttons and WiFi.

Post by **joshuadenmark** » Mon Feb 06, 2012 3:28 pm

Hi Thomas

Very fine looking boards, like the colors.

Happy soldering

moijk · Post by **moijk** » Sun Feb 12, 2012 4:02 pm

I'm all in for one when it is ready.

The first experience I ever had with a computer was the vic 20. I was no more than 3-4 at that time. I recall I was too young to excel in any games, so I watched my older sister. I recall her favorite game was miner 2049er, and she was so upset when my father wanted to upgrade the vic 20 to the c64 that was just out. I don't recall if my sister ever got around to play that much with the c64 because I grew up enough to get the required reflexes and she lost her place beside the joystick.

The vic 20 was sold, and for years I never thought about it. but now I want to play more with the computer that lit the computer interest in the family. I have a c64, plus4, c128d, amiga 500, 600, 4000, cpc and a bunch of consoles so it's not that unusual project for me.

(well, that was an intro as well)

SkydivinGirl · Post by **SkydivinGirl** » Sun Feb 12, 2012 5:30 pm

Wow! Can't believe I missed this project. I'm very excited to see how it turns out.

Heather

Mike · Post by **Mike** » Tue Feb 14, 2012 3:26 pm

Hi, Thomas,

has the PCB of the 'GCart 8k Internal' also found its way to you?

Greetings,

Michael

TLovskog · Post by **TLovskog** » Tue Feb 14, 2012 3:50 pm

Unfortunately that was canceled.

However, that was just as well. Although I knew Xilinx was about to terminate some of the 5V versions of the CPLD used (XC9500), I still used it. However, they have now issued an end-of-life for the whole 5V series so I took the time to design in another part.

It is starting to be a bit tricky finding modern parts that handles 5V. 5V tolerant inputs is still ok, but driving the good old 5V TTL is harder. In the Cartridge I added 5V-3V3 buffers/levelshifters, since it would probably be a good idea to buffer the address/data bus anyway.

However for the internal part it was wiser to try to find a 5V CPLD, since the rest of the parts where 5V only.

I also had a chance to modify the placement so it should fit more (all?) models of the VIC 20 PCB.

In any case. I am right now finishing the new boards and will hopefully send them away this weekend.

TLovskog · Post by **TLovskog** » Tue Feb 14, 2012 3:53 pm

@joshuadenmark, @moijk, @SkydivinGirl ...

Thanks for the cheerful comments. This is not the fastest moving project of mine, but we are getting there slowly.

Mike · Post by **Mike** » Tue Feb 14, 2012 5:26 pm

TLovskog wrote:Unfortunately that was canceled. [...] In any case. I am right now finishing the new boards and will hopefully send them away this weekend.

Oh, that means I'll have to wait another few weeks for the first prospective re-implementation of my VFLI mod. Sigh.

At least the redesign means you don't have to rely on end-of-life parts ... Good luck.

Leeeeee · Post by **Leeeeee** » Tue Feb 14, 2012 8:13 pm

but driving the good old 5V TTL is harder

You can drive 5V TTL perfectly well with 3.3V CMOS.

Lee.

TLovskog · Post by **TLovskog** » Wed Feb 15, 2012 3:22 am

Leeeeee:
Oops. That didn't come out right. My thought was for this case (where I have an unknown environment to interface to externally) it can be hard or uncertain. It would be easier if I had done the total system design.

I know that 3V3 should be perfectly ok (equal to or above the 2.4V needed for a HIGH with TTL). I do have a component that has a minimum Vhi output at 2.4V with a fairly low output drive capacity. It would most probably be ok to drive the TTLs inside the VIC, but nevertheless.

I wanted to be on the safe side and make this look as much as an "normal" cartridge as I could. I have little control over what extra expansions there might be.

I only intend to do this prototype and then some minor fixes for the "production run".

But still the generic statement is wrong. Thank you for pointing that error out.

TLovskog · Post by **TLovskog** » Wed Mar 07, 2012 1:42 pm

The first prototype of the Internal 8k expansion is scheduled to go to manufacturing from the pool service at 12th of March. The lead time for the 4 layer boards is a bit long, but much less expensive then my usual PCBCart.

BASIC features are ...
- Full 8k BLK0 that is usable by the VIC chip for graphics.
- 32 1k pages for color RAM to be able to use Mikes (or a variant of it, because the different control over the extended colors) excellent work for high-res, high color pictures. 32 is actually overkill, but the SRAM come in 32k packages.
- 4 x 4k banks of RAM that can be mapped to $8000 as character generators, or use the original ROM.
- Some experimental features including auto starting programs (setup, picture viewer, ...) from the internal modification if you use a FRAM and even more usable RAM for BASIC.

Optional feature ...
Programmable RGB 5mm LED driver to replace the Red LED in the VIC Case. What about a nice pink heartbeat to signal that the VIC is ready to rock ...

The goal is that it should be easy to add to a VIC, especially if you have the character ROM in a socket already today.
7 small wires from the board is needed for the basic features. 2 more if you want to be able to completely turn the card of with software, and maybe one more for the experimental features.

... if everything works out as expected ...

Mike · Post by **Mike** » Wed Mar 07, 2012 2:06 pm

TLovskog wrote:- 32 1k pages for color RAM to be able to use Mikes (or a variant of it, because the different control over the extended colors) excellent work for high-res, high color pictures. 32 is actually overkill, but the SRAM come in 32k packages.

This is mainly a matter of replacing the bank-select access in the original display driver of my VFLI graphics mode (which uses port B of VIA #1) with the registers used by your internal 8K expansion. VIC access to bitmap data should work straightaway without any issues.

Ideally, the 32 banks should be selected with the lower 5 bits of a register which is both readable and writable and no other functions in the upper 3 bits (so a single INC can do the job). The second set of 16 banks then simply receives a copy of the data put into the first set of 16 banks. Do you still have split registers which specify the VIC and CPU colour RAM bank?

Greetings,

Michael

TLovskog · Post by **TLovskog** » Wed Mar 07, 2012 2:17 pm

This is mainly a matter of replacing the bank-select access in the original display driver of my VFLI graphics mode (which uses port B of VIA #1) with the registers used by your internal 8K expansion. VIC access to bitmap data should work straightaway without any issues.

Yes I know, very small change. Should be trivial.

Ideally, the 32 banks should be selected with the lower 5 bits of a register which is both readable and writable and no other functions in the upper 3 bits (so a single INC can do the job). The second set of 16 banks then simply receives a copy of the data put into the first set of 16 banks.

I actually have a register with b0-b4 for the colour, then a guard bit as b5 that is ignored when writing and always reads 0, then b6-b7 is used. This still makes a INC work, which of course is preferred.

Greetings
Lövskog

Mike · Post by **Mike** » Wed Mar 07, 2012 2:56 pm

Great news!

TLovskog wrote:a guard bit as b5 that is ignored when writing and always reads 0, [...] This still makes a INC work, which of course is preferred.

Ah, disturbingly elegant!

Looking forward to the results ...