Why new A1200 accelerator cards suffer timing issues

[manic23]

Apologies if this has been discussed before, but I am very keen to understand why a lot of new A1200 accelerator cards seem to suffer from this bothersome timing problem. My understanding is that most of the original cards that were available back in the day like the Blizzard cards for example did not have these issues, so why do new ones? am I missing something?

I know it's not actually the cards that has the issues, I understand that it is do do with timing bugs on some A1200 motherboards, but if manufactures back in the day managed to make boards that played nicely with these boards, then why do modern cards not?

[orange]

it would probably make production more expensive and design more complicated/time consuming.
iirc, I had problems with old Apollo 030 turbo card before applying timing fix.

[Loedown]

I don't own an A1200 so I am just going on speculation.

I often program Microchip parts and Microchip sometimes need to release new versions of their chips ( new versions of silicon ) as the old one may have had a design flaw. So a part may be called PIC16F84 as the generic part number but it may have a letter after the base nomenclature of the part and it may be called PIC16F84B and there will be a note in Errata about what has been fixed. I recall an Intel problem with some processors which were unable to use a certain command.

Bottom line is that when code is executed by the processor in the A1200 it may be interpreted slightly differently by some of the custom chips which may have had a workaround applied in early versions ( think A1200 Gayle issues with early motherboards ). As the circuitry may need to be modified this can have an effect on chip timings which are usually pretty critical and if an accelerator relies on a certain signal at a certain time and it's been slightly delayed to fix another issue suddenly your accelerator won't work because the window for the signal to be received is missed.

Hope that all makes sense.

[NovaCoder]

My old ACA1230 56Mhz was a very fast card which could even beat an 040 on some tests.

I think to achieve this kind of speed, these cards are less tolerant of timing issues that seem to effect lots of 1200 boards.

[klx300r]

Originally Posted by NovaCoder

My old ACA1230 56Mhz was a very fast card which could even beat an 040 on some tests.

I think to achieve this kind of speed, these cards are less tolerant of timing issues that seem to effect lots of 1200 boards.

I've got an ACA1230@56 in my 1200 with 1d4 MB & have no timing issues so far BUT the question is how does one know if you have a timing issue? do you get random crashes, heating problems etc ??

[Nipedley]

Isn't the jist of it that the new cards push the technology to the limit much moreso than earlier cards, so small peculiarities in different amiga motherboards or underlying issues that you wouldn't normally notice are exacerbated.

I've not had any problems with my ACA1230/56 1d4 either but understand issues to be graphical corruption or crashing

[desiv]

Yeah, never had any issues (knocking on wood) with my ACA1230/28.
I think Nipedley is basically right, although the timing fix was around before the "new" cards.
So this was a known issue before. Just not with some of the more popular accelerators, like the Blizzards.
I think a lot of it had to do with Commodore's poor quality control in the 1200 line.
Some boards require the timing fix and some don't, even if they are the same revision.

desiv

[amigakit.com]

Older cards used an asyncronous design which was more forgiving on timing tolerances. The modern accelerators use a 3-1-1 synchronous design which will expose the timing problems on some A1200s.

[SilentBob]

Thanks for the clarification on this, Amigakit. I've actually been wondering the same, but never got around to asking.