ACA500plus & ACA1234 & Gotek HxC or FlashFloppy Crash

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.

    This is not a 'bug' report just an observation I've had with my:
    - Ramixx500 motherboard CPU removed, ROM removed
    - Indivision ECS v2
    - A2048 Ranger port RAM board
    - Rob Cranley Zorro-II / PSU control board v2
    - ACA500plus (external clock mod)
    - ACA1234
    - X-Surf-500
    - Gbapii++ RTG graphics card
    - SwissBit CF card's
    - Gotek's with HxC and FF firmware
    - Homemade PSU

    'IF' I boot with the ACA1234 'off', every clock speed of the ACA500plus, I can read\write disk files on my Gotek drive without issue.
    'IF' I boot with the ACA1234 'on', every clock speed of the ACA1234 and every attempted Gotek drive access, locks up the computer with the floppy light stuck on.
    I'll try again with an actually floppy drive at some point...

    It's not a huge issue...
    As I just fall back to the ACA500plus to read\write Gotek disk files...
    And I have access to the NAS over the network...
    But ... it's kind of a weird issue, so I thought I'd share to see if others have seen this ?
    I'm kind of wondering if the ACA500plus somehow blocks the ACA1234 from floppy functions ?
    I'll just leave this observation here :)

    Since there's quite a bit of self-made stuff involved, I'd like to first make sure that grounding and shielding is properly taken care of. Can you post a diagram of your grounding and shielding concept?


    Jens

    The address bus of the 100-pin slots (I refuse to call them Zorro on the Checkmate case) is not buffered, and I really have no intention of supporting that thing. I tried to educate the vendor about the importance of replicating the A2000 Zorro bus 100%, and making the power supply with point-of-load regulators, but he didn't hear any of my points, always came back with "but it works!", even in this forum.


    Your setup is a good example why it's not a good idea to deviate from best practises.


    I'm happy to design a proper PSU and Zorro board for the Checkmate board, but this would have to be paid for. If you drum up enough customers, you may have enough money to get something with iComp quality and support.


    Jens

    My A500 setup is an amazingly fun Amiga...

    If it wasn't for the crazy expense, I wish everyone could do this :)


    I already have a rock solid PSU,

    Independent regulation on each supply rail

    First order LC (film) filtering on +5V@10A rail

    On the latest revision, I separated the audio power from system power.

    I'm now getting the audio +/- 12V from a separate pair of +/-15V SMPS and using Analog Devices highest quality precision linear regulators to step that down to +\- 12V through a second order LC (film) filter, exclusively for Paula's audio out amplifier circuit.

    Those two linear regulator IC's alone add over $30 CDN to the PSU BOM

    I don't really hear the difference of course... but It's 'impressive' (maybe that's not the word) to burn this much effort and money in the name of audiophilia ... for 1980's digital sound... ;)


    The mains power switch is an NKK on-off-mom rocker switch, so when I use this same PSU design in my A2000, I now have soft-power-on-off without the need to bodge an ATX style power button to the case.

    All supply rails are monitored and the monitor IC controls the fan, will shutdown the PSU in over-voltage, over-current, over-temp conditions using the primary SSR and logs performance and reports out to the Amiga via I2C

    Such a power supply would need too many quality and feature compromises to make it cost effective for commercial production.

    And I think I know your mind well enough...

    You'd not want to go to production with something if you couldn't have the best offering... :D


    Making a better version of the 100pin riser board with a Buster socket or Bluster CPLD, buffer ICs and Zorro II RC termination would be great.

    I could see paying ~$200 CDN for such a board.

    I imagine a fair number of the Checkmate case purchasers would opt to upgrade to that, especially as we could then have two Zorro slots (case supports this) and not just one 100pin slot.

    I don't think I'm the guy to round them all up though and get them to pool their money.

    Also... not entirely sure that this alone would resolve my OP.


    I'll just leave this post up, in case anyone else spots an issue with their Gotek in the ACA500plus & ACA1234 accelerator in a Checkmate case... And If I find a fix on my journey, I'll post that too.

    We did have a few cases where Gotek drives caused funny behaviour with ACA500plus. To date, I don't have any good explanation - other than a power problem. I never owned a Gotek, but from pictures I can see that a linear regulator is used, but minimal to non-existent decoupling is on the board. You might want to try removing the Gotek unit and use a real floppy drive to see if there is any change in behaviour.


    From your description, it seems like you'd know how to build an LC filter to add some decoupling to the Gotek power setup. You won't be able to decouple the GND side, as there's quite a bit of GND lines on the 34-pin floppy connector - any effort of decoupling the GND-wire of the power connector will be voided by that. However, it may be worth a try to add grounding to the Gotek drive, again, just to see if there's a difference. After all, it's a high-volume product that may be designed to be extremely cost-effective - possibly down to a level where a design review for EMI wasn't done.


    Jens

    From what I can see, these Gotek boards have all been hobby projects, especially the good looking ones, that have OpenSource Hardware badges.

    I do not expect there has ever been EMI (conducted or radiated) emissions testing... or anything that resembles formal testing..

    "Just works for me, should work for thee"


    In my case... they 'function' ... but who's to say how marginally...

    except when they don't function, which is when I have the ACA1234 running.


    I think I'm picking up the floppy power for the Checkmate from the motherboard right now.

    I can move it over to the PSU's dedicated drive power header.. might make a difference, slightly stronger 5V rail.

    While power (good, bad, or silly) is a common theme on the iComp support site, my initial thoughts are signalling issues or timing.

    I've yet to have bad power as the root of my own setup issues.


    I did not see this issue with the OEM A500 Motherboard.. I think it appeared after the Raemixx A500++ board showed up.

    So, It could be related to the Raemixx A500++ motherboard and its drive switcher as well..




    I've got lots of places to probe and lots of things I can swap and try...

    You should just leave this drive switcher in standard setting, as the functionality is in the ACA500plus already. That said, I can't think of that side of the signalling having any influence on the CPU timing. The "works marginally" hypothesis is probably the closest, where any tiny influence on address/data bus load, grounding or power noise will make a difference and push it over the edge.


    The original motherboards went through a very visible evolution, with Rev.3 having almost no ground areas, and the newer it gets, the more GND pours and via stitching you see.


    My personal experience is that the Rev.5 NTSC board is the most tricky one to get working right, but that might have been the special sample I have with a 8370 NTSC OCS Agnus. It was my first "hard nut to crack" during ACA500plus development, and it was the final nail in the coffin of "make it work with a synchronous clock". While I got this to work 100% reliable, I never saw a Raemixx A500++ in the flesh, so I can't say how it'll perform.


    Jens