Uzebox Forums

Good morning/afternoon/evening, is there a board (preferably the gerber files) with most expansion slots? like for the 128k SPI RAM, Uzenet, uzebox mouse/keyboard etc. (preferably without the AD725, can be scart instead) Also is the Atmega1284p supported? Have a great day/night

Hi,

I'm not aware of a SCART version of the Uzebox with the SPI RAM and ESP8266. Your best bet is the F5 revision on the downloads page. There's a header port for direct access to RGB and sync signals so it's should be easy to add a SCART connector and avoid the AD725. The mouse and keyboard uses the regular SNES controller ports. I made many tests in the past on the atmega1284 and it can't be overclocked in a reliable manner.

Cheers!

uze6666 wrote: ↑Wed Oct 20, 2021 5:04 pmI made many tests in the past on the atmega1284 and it can't be overclocked in a reliable manner.

I'm wondering if it only has to do with the 64KB+64KB flash banking/addressing circuitry because I don't see the SRAM itself having an issue and everything else seem identical to a 644.
( I need to check all the IOs and see if they really are identical, just did a quick glance )

Did you try taking a 644 build Hex/Bin and duplicating the low 64KB of flash ROM into the high 64KB including the bootloader?
Or a 1284 build with a custom linker map file that limits everything to the top 64KB and then duplicate the ROM into the lower 64KB

If the extra single high bit of address is the culprit that could fix the issue. Even if that high bit is marginally too slow the flash would read the same data.

I don't have a 1284 on hand to test tho.

Hrm, thinking about it a bit more, if that works the bootloader would need some extra custom code to jump to the lower 64KB to flash the lower bank to prevent that high bit from flipping during the flashing of a new binary to make sure both 64KBs have the same content.

It'd be a lot of effort just to get a 1284 working at that clock but ...

My tests used just the lower 64k of flash and 4k of ram compiled both with the 644 and 1284 as the mcu. Even with that, it was unreliable so I didn't bother to tests the upper banks. Out of a batch of 10 chips, none worked overclocked at 5V. Maybe 1-2 worked at 5.5 and perhaps 2-3 at 6V. Some would not execute anything (blank screen) and some that ran had severe video glitches.