By popular request, we have simplified the modifications that need to be done to the C-One in order to activate the primary IDE port for use with the Minimig core.
In three short steps:
1. Connect PCI pins B6 and B8
2. Connect PCI pins B40, B41 and B42
3. Cut off pin 28 from the IDE port.
To understand what this actually does and how it looks like in the real-world, please go to the C-One website http://www.c64upgra.de/c-one
and click on the "technical" section. I have prepared a PDF document with explanations and photos that should make the job fairly easy for someone who is familiar with a soldering iron. However, it should be mentioned that it's not a beginner's job and that if you don't feel comfortable doing hardware modifications, you should get help from an experienced friend or a professional.
On another note, Tobias is working on a way of creating soft-IRQs, thus making these modifications obsolete. These soft-IRQs are part of the software that handles hardfiles on the CF card. Since we're not sure if this will work as planned, and there's quite a few people out there who would like to use a harddisk on their Minimig, we have published V1.26 of the core with IDE support.
V1.26 also contains a bugfix in the SD-Ram controller. In all previous cores, the SD-Ram was overclocked, which caused startup problems and crashes where you would not expect them. The overclocking was extremely moderate: Timing was only off by 400 picoseconds with PAL cores, and it was off by 540 picoseconds in NTSC cores. To give you an idea of how short this time is: Light would travel a distance of about 15cm (about 6 inches) during this time. This is also the reason why the bug did not show at all temperatures and not on all boards.
Tobias has made the SD-Ram timing a lot less critical, which made it a little slower. With the reduced speed, the fixed timing slots for the three components of Minimig (Chipset and the two processors) could not be maintained, so Tobias implemented a dynamic arbitrary system for the available SD-Ram slots. The chipset always has top priority, then comes the main CPU and after that, the IO-CPU gets memory cycles.
You will notice two effects:
1. the Minimig is slightly slower than before (still faster than the original board)
2. Minimig starts reliably at any temperature
Some of you may have noticed that I did not deliver FPGA extenders and C-One boards since mid-december - I will now continue shipping, now that a solution to the startup/stability problem is found. Sorry to everyone who has been waiting.
Have fun testing!
for latest core.