Status update

Chimera Cartridge

Entry posted by mos6507 June 11, 2008

1,186 views

OK, we're at enough of a milestone here to justify an official blog post.

Here is what's been going on.

First off, we didn't get the accuracy in the paddle reads that we wanted from the new isolated ADCs. Delicon tried to implement a more 2600-like reading approach using the ARM's "Capture" facilities but it also is a little jittery. In both cases, close but no cigar. So instead of holding up the project just for paddles, we decided to adjust the expansion port pinout to align the capture pins to be symmetrical with the assumption that those would be the preferred way to read pots. There may be some things that can be done on the dongle itself to make the reads better, or we could just put a microcontroller directly on the dongles which was Delicon's original idea. However, I think we've reached closure as far as optimizing the board layout to make the most accurate ADC reads. So the ADC pads will still be there but will likely not be used directly by any dongles.

The serial dongles have two extra headers, one is a straight pass-through, and one is a daisy-chain. The pass-through is meant to provide access to any ancillary pins that are not used (like the heretoforementioned ADCs). The daisy chain header is meant for stacking a second serial dongle on top in order to gain access to the 2nd serial channel. That's because each expansion port supports two serial channels. We decided to go this way so that you could just own two serial dongles and just change around how they are attached. If a game wants two serials on one side, you stack them. Or you can split them apart and go one per side.

There are two types of serial dongles, a regular RS232 DB9 and one with an integrated serial to USB adapter that you can just plug n play into a PC.

There is a jumper inside the dongle to force the ARM into bootloader mode and upload new firmware.

The dongles will use the familiar AtariVox case design.

One of the big differences with this board is a new larger CPLD with more pins on it. This was necessary in order to implement the active buskeeper circuit. This is used to enable magic writes and to clean up noise issues we experienced with the last proto. Delicon is on a system with RF but he claims the noise is gone. When I get my cart I will verify.

For the last few days he's been reimplementing the banking schemes and he has been able to run the 512K Fu Kung demo, so we know the wiring to the 512K SRAM chip is valid.

He still has to reimplement the auto-CPLD reprogramming piece into the firmware, the ANSI menu, and the queue support.

Besides making sure the video noise is gone, the main thing we need to test together is the queues. Last time we were experiencing some data corruption issues and some timing constraints. So this time around it's going to be make or break on the queues. The general concensus is that the hardware can not be further improved so whatever winds up being feasible with queues is as good as it's going to get. If we switch to the 2378 chip we will have twice as much internal ARM RAM to work with so even if the SRAM queues are kind of limited, it should not be that much of a constraint since so much more will fit inside the ARM. Time will tell.

Originally we were hoping to send the spare boards of this rev out to beta testers but we decided to wait until the next board rev with the revised expansion pinout. It would just be too much of a hassle otherwise with the dongles having to change. But between now and then we should be able to still get some good testing in and maybe be able to post some demo videos again of the hardware actually working.

In summary, we have yet to find any actual "bugs" in the board design. Given the density of the connections between the various components, that's a major accomplishment.

So we will put it through its paces and see if we can bring it all the way to the point of being a functional multicart with the onscreen menu. If we can do that, we'll have a releasable product. Then we can start the beta program and I can start to work on the actual production details, the packaging, the cases, the parts and assembly plans, etc...