Protecting the cartridge
Entry posted by mos6507
752 views
Thought this would be a good time to post a progress report with Chimera...
There were some complications delaying the completion of the PCB layout for Chimera version 10 (the LPC2368 version).
First off, we're using a 2-layer board and that is asking a little too much to accomodate the larger ARM chip and the extra signals (known as the Super Wide Bus between it and the CPLD. We spent time debating what to do with the center screw-hole which was cramping the connections and whether we could move things around or expand the board into every nook and cranny of the case.
Secondly, Delicon tells me about a risk-factor that had been in the back of my mind. In the original design, the pin assignments were fixed. You had two dongles, one RS232 a piece, and the extra signals (I2C and SPI respectively) would show up as solder-points inside the DB9's shroud. But all that went out the window when we discovered we could reconfigure the pins on the fly to make the expansions operate as Atari 2600 controller ports or other things. So Delicon is like "what happens if you were playing game X that used the ports like this and then you switch to game Y that uses them like that but you don't first swap the dongles?" It looked like under some conditions, the answer to the question would be "smoke".
So when he told me that, I suggested that we use a single 20-pin header and implement an intelligent breakout-box that would automatically route the right signals to the right labelled port. If the ARM didn't detect the breakout box, it would freeze the configuration to the standard dongle. Delicon didn't go for this. It would be more convenient, but arguably more expensive. He suggested we put a CPLD or ARM on every dongle. I didn't like that it increased cost and may have required power. Then I think I came up with a good solution...
The ARM has 6 A/D converters. 4 are accounted for (for paddle support) but we could use the other two to measure the resistance of a resistor in each dongle. The resistance would identify the dongle attached. Whenever a game initializes and attempts to reassign the pins, the ARM checks the dongles to make sure they are set up right. If there is a mismatch, the VCS will be notified and it can throw up a prompt.
Delicon thinks this may be a valid solution but he has to come up with a way to use the A/D converters without disrupting the "Super Wide Bus" because the 2 extra A/Ds are being used. The bus can be moved back one pin but that's not enough to scan both dongles independently.
Another complication is that in order to do this, you need 1 extra pin per dongle. Nobody makes 12-pin headers. The next step up is 14-pins. That makes it so the headers can't fit two to a side. So we started coming up with strategies to cram them in there but it looks like the only solution is to move to a 4-layer board, after which Delicon says we can have the dongles symmetrical on the left and right side of the cart. Unfortunately, 4-layer boards are twice as expensive to make, but the main cost of these carts is the CPLD and ARM chips, so it may still work out okay. So I have approved the move to a 4-layer board, at least for this revision.
Lastly, he told me that because of startup costs, it's cheaper for him to design all the dongles we'll use now rather than later. So here is the list:
- Dual RS232
- Dual PS/2
- Controller Port
- Genesis adapter (maybe)
- PC Gameport w/MIDI (uses both expansion headers)
The gameport adapter may be cool. Games like River Raid, Oystron, Star Raiders, Phaser Patrol, Solaris, Radar Lock, etc... might benefit from a conversion to an analog stick. (Then someone could try porting Battlesphere.) Paul Slocum could use the VCS as a MIDI device or games could drive outboard synths.
Like us on Facebook
Follow us on Twitter
AtariAge on Instagram
Watch us on YouTube
Chat on Discord
2 Comments
Recommended Comments