What type of clock crystal is fitted to the aca 1233n?

Caution: Non registered users only see threads and messages in the currently selected language, which is determined by their browser Please create an account and log in to see all content by default. This is a limitation of the forum software.
Don't Panic. Please wash hands.
  • It's an oscillator, 5V type. If you're thinking of overclocking, you should know that the memory is already operated near it's specified limit. Going beyond 80MHz means to reduce margin for setup/hold timing, and going beyond 83MHz means to violate RAM timing parameters. I strongly recommend keeping it at 80MHz.

  • I’d seen someone using 100mhz parts and an 50mhz 030. I have a 50mhz 030 from a dead blizzard card. Thought I’d try it, can always put it back if it doesn’t work.


    Have you got a part number for the 80mhz oscillator? That way I can cross reference to a 100mhz part. I’m well out of warranty period.

  • another question.


    someone was saying it’s possible to run the fpu from the same timing as the cpu. Essentially I think they were using links in place of the crystal. Is this possible?

  • Have you got a part number for the 80mhz oscillator?

    Not at hand, but any XO75-sized oscillator should work.


    I’d seen someone using 100mhz parts and an 50mhz 030.

    Sure you can use that, but you'll either find that it doesn't start at all, or that it starts being unreliable when temperature is above freezing temps. I give this zero chance of working at room temperature, let alone the elevated temps in the trapdoor of the A1200. The RAM chips are just not fast enough to open a row and then 10ns later (given your 100MHz idea) initiate an access by opening a column in that row. 12.5ns is already "reading between the lines" of the data sheet, and if you go along that path, you'll find that 83.33MHz is the maximum you can squeeze out of the RAMs. The CPU may run at 50MHz, but with the memory clocked in sync, you'll violate the specs by going any higher. You'd either have to change the SD-Ram controller to insert another NOP command, but that will result in another wait state for every memory access, so your effective performance is reduced.


    In other words: The command sequence that the SD-Ram controller of the 40MHz card is sending to the RAM chips is so fast that it is equivalent to the memory running at 160MHz. If you increase the base clock to 100MHz, these memories would have to be 200MHz-compatible with the shortened access sequence that I'm using. Just trust me: It does not work reliably.

    someone was saying it’s possible to run the fpu from the same timing as the cpu. Essentially I think they were using links in place of the crystal. Is this possible?

    Yes, the solder-jumper near the FPU will put the CPU clock on the clock input of the FPU, so you can avoid using an extra oscillator. If you close that solder-jumper, you MUST NOT insert an oscillator, as that would short out the two clock signals.

  • How’s he managed it?

    He didn't. He relies on the RAMs to not fail, although they are operated far out of spec. The card *will* fail, and the buyer may even think that there's a defect or a CPU problem, but it'll turn out to be random memory errors, and it'll get worse over time (remember that RAM does age, and it gets slower over time, reducing timing margins).


    I can understand that you want to get the best out of this accelerator, but I want you to be satisfied with it, and I don't want you to have an unstable Amiga because you wanted a few more percent of performance. It's much more fun if it "just works".

  • Guess so. It’s just a shame my blizzard mk3 packed up. That ran at 50mhz with a 50mhz 68882 fpu. 16mb sim. Those were happy days.

  • how did they achieve 50mhz in the old days?

    Memory was not sync - remember this is SD-Ram.


    And what memory does the 55mhz card you sell use?

    It is using the same memory, but a different memory controller in the CPLD: Memory runs at 55MHz and one more wait state is inserted to have proper timing margins as required by datasheets of all used components. As a result, all parts of the 55MHz card are within specification, and only the CPU is moderately overclocked.


    If you just put a 100MHz oscilaltor on the 40MHz card, you will operate three components outside their spec:


    - the RAM (as explained yesterday)

    - the CPU (that's 40MHz-rated!)

    - the CPLD (timing simulation of the RAM controller requires 90,9MHz or lower)


    I hope that you see that there is a difference between "works by design" and "works by accident". The seller on Amibay obviously got lucky and had enough time to run a benchmark before the card crashed for the first time. However, it will crash when warming up. Timing margins are required in order to cover a wide temperature range. It is calculations like these (that fill large libre office calc sheets in my case) that make or break a stable design. If you don't do that and clearly see "green" for every single timing parameter, the product is not good for the field.

  • Ok so it is potentially possible but only if the CPLD firmware is changed, and a 50mhz cpu is used (which I have).


    or are the CPLD’s totally different.

  • We can re-program the CPLD here, but you'll of course have to pay for that, as you haven't bought the license for that SD-Ram controller with your slower card.

  • I’m guessing there is a header for programming this. I have several different programmers including a tl866ii plus and usb to avr.


    To save it getting lost in the post and also I’m self isolating from covid due to health issues. If I paid for the reprogramming either I could upload the file if you emailed it or you could do it remotely with quick assist.

  • I'm sorry, but the files don't leave this workshop. Re-programming can only be done here, or if we meet at an Amiga show and I have the programming equipment with me.

  • I’m in UK.


    just a thought when the covid nightmare is over. If I send it back would you need the 50mhz cpu from my blizzard, would it effectively become a 1233n 55, would it get the ide speeder ability, and how much would it cost with return to UK shipping?

  • We don't do that kind of service - I'd re-program the CPLD, but I won't exchange a CPU and as a result, become responsible for the whole thing while I haven't tested the CPU for compatibility with 55MHz clock.


    If you want the IDE speeder for your 40MHz card, we can add that for the difference between the old and the new card (that's a few cents shy of 60,- EUR). If you want both, the IDE speeder and 55MHz, I recommend to sell your 40MHz card and buy a new 55MHz one.

  • Ok scrapping the 50mhz idea and looking at the fpu clock.


    if I use a separate crystal can it be a square type? As I have a spare one. It looks like both sizes are installable.

  • Yes, both sizes will work. It should be a 5V-canned type; please double-check the pin#1 mark. For the 8-pin variant, please keep pin#1 of the 14-pin case free.

    • :)
    • :(
    • ;)
    • :P
    • ^^
    • :D
    • ;(
    • X(
    • :*
    • :|
    • 8o
    • =O
    • <X
    • ||
    • :/
    • :S
    • X/
    • 8)
    • ?(
    • :huh:
    • :rolleyes:
    • :love:
    • 8|
    • :cursing:
    • :thumbdown:
    • :thumbup:
    • :sleeping:
    • :whistling:
    • :evil:
    • :saint:
    • <3
    Marks thread as resolved after post creation.