1
Amiga Hardware Issues and discussion / Re: I made a modified accelerator, it works, but has a problem
« Last post by SpeedGeek on April 28, 2024, 03:18:53 PM »That's the impression that I got reading the 68000 datasheet and noting that the Amiga has no actual M6800 peripherals.
Definitely.
That's a point. Currently /VMA is unconnected on the PCB (and per the schematic), I guess there's an internal pullup...? What's the consequence of the short E pulse?
I did not find the pound sign (#) in the list of standard boolean operators. What did you mean?
I'm not familiar with the .D and .T notation. What did you mean? ETERM and EDTACK symbols are connected to what?
I found a dedicated divider part that will generate the proper frequency E pulse in under 1ns. That will definitely get it working, at least on par with the unmodified 14MHz accelerator. I like the fact that it keeps with the minimalist spirit of the original project in OP. But latching with the rising edge of the 7MHz clock is definitely something that I can do. I also found a decade counter that would let me generate it from the 7MHz reference.
Yeah, I had a hunch about that, are you able to quantify what is affected by the pulse width?
Nice project, useful toward "Amiga-like computer from scratch" projects that don't use any original Amiga components (which will be my final goal, eventually). Thanks for providing the code. I wouldn't mind learning that type of PLD syntax/notation. I'm more fluent with VHDL and Verilog.
So the unmodified 14MHz accelerator has a problem I noted a while back. I have an ide68k+mem68k sandwich, and it does not work at all with the 14MHz accelerator, I get flashing colors when I power on the machine. At 28MHz I get a black screen with this peripheral. Unfortunately I don't have them separately so I don't know which of them has the problem, I got this peripheral with one of my A1000. It does work properly with the stock 7MHz 68000. Do you have any idea what could be the issue? I might give it a closer look next week to see which signals are connected so I could eliminate some factors. I will be building two kinds of IDE interface, so need to put together BOMs for that.
I need to gather parts lists from multiple projects, then I'll be able to continue with this. I like to build large carts to save on shipping. I really like the lists feature on DigiKey as it lets me keep BOMs for projects.
Sorry, about the confusion on the logic equations: .R is the standard Boolean syntax for a registered output and + for a logical OR. (CUPL syntax uses .D and # respectively) .T was the correct standard Boolean syntax for a Tri-State output.
I have not used standard Boolean syntax in many years because CUPL is what Commodore used and it offers some more advanced features which can simplify the logic equations.
Regarding the signal names, these are new signals created and referenced to run the E clock cycle outside of the CPU. You mentioned that /VMA was not connected? That's strange because the 8520 CIAs won't respond to a cycle without /VMA asserted.
Regarding the duty cycle performance issue, that is well explained on the linked thread. It all comes down to wait states. The responding device becomes active and completes it's cycle when the E clock is high.
If you are highly motivated and you have the spare time, you can build everything you like with standard TTL logic and counters. But I'm much too lazy and I prefer to use GALs and CPLDs. Good luck.