ACA1234

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.


Also users that are not logged in can not create new threads. This is a, unfortunately needed, counter measure against spam. Please create an account and log in to start new threads.

Don't Panic. Please wash hands.
  • Hi there,
    First post here.


    I just got an A1200. I'm quite interested on the ACA1234. I saw the previous posts and it seems it's about to be released.

    Can I join a waitlist or pre-order it?

    Another question if I may, is it possible to install the card without having to open the computer?
    The 'warranty seal' is intact in this unit, I'd like to keep it that way for as long as possible.

    Thank you!

  • I just got an A1200. I'm quite interested on the ACA1234. I saw the previous posts and it seems it's about to be released.

    Can I join a waitlist or pre-order it?

    We've had a strict "pre-order" policy for over 20 years now: We do not take any money unless a delivery date is known for sure. Since "waiting list" and "pre-order" is usually tied to "accepting money", I am often reminded of cases where customers have lost money because they fell for bloomy promises, spent money and never got what they were promised - this started in the late 1990s and still happens today, with crowd funding platforms making collecting money really easy.


    I'd like to make people aware that mass-producing "anything" is really hard, and you should be extremely careful about who you give your money to, and under what conditions. I like to keep things simple: You pay, we deliver on the date we've promised (or earlier, as the current pre-order session for the C64 cases shows). While mass-production is running and software still needs to be written, I don't want to accept any money and put pressure on the software authors, as working under pressure usually does not result in quality software.


    Another question if I may, is it possible to install the card without having to open the computer?

    Yes, the ACA1234 only requires you to open the trapdoor on the bottom of the computer. This can be done without removing a single screw.


    is it possible to use custom kickstart roms instead of using whatever is builtin physically in the computer?

    Yes. The ACA1234 comes with a license to Kickstart 3.1 and WB3.1, and it lets you install a CF card with OS3.1 even if you don't have the installation disks, or your internal drive is defective. If you have a license to newer versions of the OS (such as 3.1.4 or 3.2), you can take these Kickstart files and write them to the flash of the ACA1234, which is even better than a physical ROM, as it's a lot faster when mapped to the fast RAM of the accelerator.


    And before you ask: The internal IDE of the A1200 can be accelerated, but it can also be switched off, so you don't have the power-up-wait for possible drives that are connected there. So whatever there is connected to your internal port - be it an old drive or nothing - won't interfere if you have your OS on the local CF of the ACA1234.

  • Thanks for taking the time to reply to all my questions Jens.

    I do really appreciate all the efforts the iComp team puts in to release such amazing products. I've been following you since I've started restoring and playing around with my C64.
    I just wanted to be sure I can get an ACA1234 (50Mhz :)) as soon as it's released. I'm subscribed to product updates, so I'll keep an eye on it.


    I think for the moment I'll buy an ACA1221lc with an RTC module, it will give a quick way of expanding my A1200 until the ACA1234 is ready.
    Also, I've purchased AmigaOS 3.2, I've read about MapROM on the wiki so I'll be using this feature to run the 3.2 kickstart.


    Thanks again!

  • Hello, Jens,


    do the ACA1234 accelerator cards all have a PGA socket? There were also CPUs soldered directly. I am particularly interested in the 25MHz variant, since I still have two 50MHz CPu, I would like to use mine and give the others the chance to get the faster ones. Many Thanks:-)

  • All CPUs are soldered to the board. The cost of adding a socket is too high.

    Thank you for your quick response. But the design would still be PGA for all types as you can see in the picture? Okay, did I understand correctly when reading the product description that you can overclock the CPU, buy a license key if you are successful and then exchange the CPU for one with the corresponding clock frequency? Thank you very much again for your answer.

  • But the design would still be PGA for all types as you can see in the picture?

    Yes. I have an alternative board in the CAD system for QFP CPUs, but those have not been produced yet (and I currently don't see a need).


    Okay, did I understand correctly when reading the product description that you can overclock the CPU, buy a license key if you are successful and then exchange the CPU for one with the corresponding clock frequency?

    Correct. However, it may not even be required to exchange the CPU. If a 25MHz CPU has the 1F91C mask, it's very likely that it just works fine with 40MHz, maybe even 50MHz. However, I don't want to add all the extra cost of testing CPUs, as that would also have to be added to the ones that don't allow overclocking.

  • Hi Jens/iComp team!


    I don't want to rush anything, I understand that planning the release of any new hardware takes a lot of effort, but I was wondering if you have an update on when the ACA1234 will be available.


    I still have a lot of things to do on my A1200, including fitting the Indivision Mk3 that will arrive soon.

    Right now my only working upgrade is an ACA1221LC, and even though is an entry level one, I'm loving it, it's pretty fast for a 68020!
    I really like the solid way that you do things, so looking forward to the next step, happy to support you!


    Thanks!:thumbup:

  • As mentioned in a different thread of this forum, we're investigating why some people had trouble with the IDE-speeder. It seems like not many people have reported these problems, as the IDE-speeder feature was always advertised as "depends heavily on the IDE device you are using".


    This only seems to be "half the truth" - these days, most people are using CF cards, and the vast majority is using an unbuffered adapter to connect them to the A1200. We have identified this as the main cause for IDE speeder problems: While CF cards are pretty much always fast enough to deal with the speeds that our IDE speeder will run them, they don't have strong-enough signal drivers to handle the high capacity of the unbuffered address/data bus in the A1200 computer. So yes, it depends on the "compatibility" of the CF card, but that can be improved with a simple thing that I thought everyone has by now: A fully buffered interface.


    This turned out to be a major misinformation on my side. While I was the first who has designed a buffered interface for the A600/A1200 in 1996, it has been copied many thousand times in the past 25 years. However, with CD drives out of fashion and a single large CF card containing all the Amiga software that is out there, most people have "slimmed down" their system with a simple CF card adapter. And this simplicity is what's causing trouble.


    Now you might wonder why we're putting so much time into investigating this, especially because the ACA1234 has a local CF card slot that is much faster than even the accelerated A1200-IDE can offer. The simple answer is: Not everyone will have an ACA1234, but stay with their existing accelerator. We therefore want to offer a proper CF card adapter the A600 and A1200. One that solves any and all problems you might have with CF cards. This development - and also ACA1234 testing - was held up when a bug in Kickstart 3.2 caused all kinds of problems that we had to locate first (it even seemed to affect IDE compatibility, but it turned out to be a side effect only).


    This is all on the edge of being solved:


    - Kickstart V47.97 (beta version, not public) resolves startup problems with real ROMs and MapROM-from-file

    - latest CF card adapter prototype (#2) makes CF cards work that even kept the computer from starting

    - that same adapter prototype makes IDE speeder work for all accelerators in the field that have the IDE speeder, with all CF cards in our test scenarios; and that's a A LOT of cards, including non-industrial, photo-type and no-name ones.


    Needless to say that there's a deeper reason, and that's the limited availability of 68030 CPUs. I really don't want to replace all 68030 (and lower) accelerators with the ACA1234, as there is not enough supply. Instead, I want to reduce the number of reasons why people might want to buy the ACA1234 - and one of those reasons may have been that the IDE speeder of the existing accel is not working as expected.

  • If it helps... right now I'm using a CF2IDE+ adapter, a Verbatim 4GB 233X CF card and my ACA1221LC running at 40MHz.

    According to SysInfo, with the IDE speeder enabled I get around 6MB/sec; if I disable it, I got around 1.2MB/sec.

    No issues so far.

    Edited once, last by Jens: No direct shop links in this forum - I don't want to deal with the legal implications! ().

  • That adapter is not buffered. YOu have 22 critical signals that need to be buffered and isolated from other things like the clock port and PCMCIA. This is an easy task with a few discrete TTL chips.


    Thanks for the hint at Verbatim CF cards. I don't have any in my stack, but the performance sounds awesome.

  • A technician's hour is 85,- EUR. If I calculate 45 minutes for removing the CPU and cleaning it up, 20 minutes for soldering the socket, 10 minutes for inserting and testing, plus a 2% chance of losing the card completely in the process, the cost would be at least 110.47 EUR (25MHz card taken as basis) - plus VAT if you're in Europe. Plus the cost of the socket, but that's the smallest share of the cost :-)


    Especially if you're buying the 50MHz card anyway, you will get a gold/ceramic CPU with 50MHz rating on it, so there is nothing "above" that. Please pardon the question, but what's the point of a socket? It will only lift the CPU high enough for the trapdoor to not close properly, and it'll keep you from adding a small cooler because of the added height.

  • Please pardon the question, but what's the point of a socket? It will only lift the CPU high enough for the trapdoor to not close properly, and it'll keep you from adding a small cooler because of the added height.

    Nothing wrong with asking questions, nothing to be pardoned.
    Like many of these boards, this one will not going in an A1200, I have no immediate physical space restrictions.
    The ACA500+ will be the 'host' system in my case, as I'm sure will many others.

    Socketing the CPU from early on makes the card more serviceable here in North America where the iComp warrantee is of greatly reduced value given the cost of return shipping. If I had to return an iComp product under warrantee, I'd likely just use my credit cards extended purchase protection and later its two year warrantee extension in order to bypass all the shipping expense and simply get a refund for my bad luck.


    Sadly, based on your response, all the cards must already have been populated, because there is no longer the more cost effective option of sending a socket in place of an MCU to the contract manufacturer. Meaning that a socket results in a re-and-re and a potentially damaged board.
    So no reasonable way for iComp to make a little little more margin, and no reasonable way for me to pay for a little more serviceable card.

    ;(sad, but life goes on.

  • Socketing the CPU from early on makes the card more serviceable

    I've been making accelerators for well over a decade now, and we've only ever has 68ec020 CPUs fail - the QFP ones. I have yet to see a 68030 in PGA package fail that is NOT a Chinese fake (I have a few fakes here, mostly just re-labelled with a higher frequency).


    That said, in terms of service, the ACA1234 will be fully documented with spare parts, function, how to troubleshoot, and I'll even offer pre-programmed CPLDs. This will probably be the most serviceable accelerator I've ever made.

    Sadly, based on your response, all the cards must already have been populated,

    Correct - I have finalized the QC procedure today, and staff has started programming the CPLDs. MCUs will be programmed after that, and then we're all set for the day when the 8M flash can be filled with all the software I want to include. Pre-orders will be opened this month.

  • The last reply was more than 365 days ago, this thread is most likely obsolete. It is recommended to create a new thread instead.