My hardware project :: GCart 2011

[TLovskog]

Set up the environment for developing the hardware description for the small programmable logic.

At least nothing is blocking the bus for the VIC-20 and it starts, so I guess the smoke test is passed. The reset switch also works, although that would have been pretty embarrassing if you failed on.

The front PCB in the 3-way expansion is a test board for easy measuring of the complete cartridge bus with additional pins for various signals.

The PCB in the back is the simpler cheap memory expansion. I use that as a vehicle for developing the programmable logic.

On top of the VIC 20 is the emulator for the Dialog GreenPAK programmable logic.

On the screen you see the VIC 20 in PIP on my monitor.

[TLovskog]

Quick test with only BLK1 connected to the RAM through the logic and it seams also to work.

[TLovskog]

All banks including RAM1-3 and IO RAM tested and works.

Next will be to test the config and bank latch, run some extended memory testing, and start working on the real logic.

BoM including PCB is still under 5 USD for this simple expansion.

That is probably it for the weekend.

[TLovskog]

Another weekend with some time for junior. Soon time to start on the real cartridge.

Anyway. The logic fabric is finished, but there is one minor thing to fix. All the bank switching, long/short reset, write protect, etc works very well.

The firmware for setting up the configurations also comes along just fine. Next stop is the handling of flashing the flash memory.

[TLovskog]

A new version of the hardware has been sent to production. Only a few smaller changes ...
- Replaced the octal latch with a edge triggered variant to save one delay line in the GreenPAK.
- Added an external single gate inverter for RD generation to the flash memory.
- Added the possibility for a larger flash memory, up to 64 banks of 8k and 512k in total. This will increase the cost with 1.5 USD, but it is an option. Today the BoM with PCB is 4.98 USD.

Also the configuration tool is going forward.

[crusti]

Take me money now, I need/want one.

[TLovskog]

Thanks!

[Vic20-Ian]

I'll take 10 @ $5!

Seriously I am interested in several depending on the final price.

[TLovskog]

Thanks!

I will do my utmost to keep the cost down. A challenge I enjoy, and which are part of my daily job.

[MCes]

Next stop is the handling of flashing the flash memory.

Is the flash memory writable by VIC 20 (and how)? because I read from the datasheet that it need particular HW procedures....

[TLovskog]

And to answer the question ...

Why the funky PCB shape?

Well, with this shape I can get most out of the PCB production panel. The area is reduced to 68% per pair. This should yield 23 customer panels (with 2 product PCB each) on each production panel. This also speeds up (= lower placement cost) the PCB-A manufacturing. Albeit, in this case, perhaps negligible.

[TLovskog]

Is the flash memory writable by VIC 20 (and how)? because I read from the datasheet that it need particular HW procedures....

Yes it is.

The FLASH needs a "Magic Sequence" to unlock the different commands. Chip erase, sector erase (4k), ID readout, and programing.

You write (for these FLASH memories) $AA to $5555 then $55 to $2AAA, then the command to $5555. After that you can program a byte or read the ID (Chip manufacturer and partnumber code - a byte each). For chip and sector erase you need one more sequence of 3 writes. This is probably the "Are you really sure?" kind of protection.

It is of course a bit more messy since in this case the FLASH is banked into 16 or 64 banks of 8k located at the VIC-20 address of $A000-$BFFF. So you will have to select bank 2 ($4000 - $5FFF) and then write $AA to $B555, then select bank 1 ($2000 - $3FFF) and write $AA to $AAAA, and so on.

I have not yet written the FW for flashing, but I have successfully read out the ID of both 128k, 256k and 512k FLASH memory.

[brain]

I'd fix this by using an 29F040b, which only requires 12 bit unlock codes. It's still 512kB, and I believe it still available. That way, the unlock can be done from any bank.

And, I think they are less expensive that your current option.

Jim

[TLovskog]

Thanks for your comment Jim and yes, that is true.

However to the best of my knowledge the 29F040b is an obsolete part, although 29F040C is active.

This is originally aimed for 128k FLASH and the 39SF010 is the cheapest one I could find in TSOP-32. At least from the ordinary distributors. I try to avoid ebay and aliexpress as well as new old stock. This part is then pin compatible with a 512k FLASH, which is available as an option.

That said. The real benefit for going with the newer 39SF010 and 39SF040 is that the sectors are only 4k bytes compared to the 29F010/29F040 that have 64k bytes sectors/blocks. That way I can easily erase one “cartridge” slot of 8k and reprogram only that.

Switching banks in this case is perhaps annoying and messy, but trivial.

[brain]

Does the cart allow FLASH in any bank besides 5?

Jim