Issue with ACA500Plus after updating

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.

    Hey guys,


    I have several ACA500Plus's and today when doing a little Amiga maintenance and updating I realised two of them had not been updated to the latest firmware etc.


    I proceeded to flash the latest Menu Update, Install Update and Network Update on the two cards. Both seemed to go through all fine, however when finished, one card was working fine, the other has become "partially bricked".


    By partially bricked I mean the card boots to the ACA Menu just fine, but selecting any profile, or hitting F1 or F2 for the two preset profiles causes the system to crash as it is attempting to read the Kickstart ROM.


    Sure enough if I hit the TAB key for information in the ACA Menu it says "MAP ROM empty or checksum error".... Okay, somehow the ROM's have been corrupted or deleted, no biggie, I saw a file on the Icomp site for restoring them, so I downloaded the file unarc to a freshly formatted FAT CF card, plugged it into the AUX port, powered on and hit F1..... unfortunately this won't work as F1 reads the stored Kickstart 3.1 ROM..... which is now either corrupted or deleted. Oh.... Okay, what if I set Profile 1 to Mainboard ROM and switch off the MapROM option - this should just boot with the ROM on the A500+ motherboard and I can run the restore files from there.


    Well that kinda' works..... except the Mainboard ROM on this A500+ is 1.3 and so the restore image errors out. No probs. I opened the A500+ up and swapped out the KS 1.3 ROM with a KS 3.1 ROM.


    Tried again and success! It reads the restore files and flashes both the Kickstart ROMs and the WB disks, then says to remove the CF Card and reboot., which I did.


    But the problem is unresolved. It still crashes when attempting to read any ROM that should be in flash memory and ACA information reports the "MAP ROMS empty or checksum error".


    Help. :)

    Quote from abraXXious

    if I hit the TAB key for information in the ACA Menu it says "MAP ROM empty or checksum error"...

    That refers to the RAM area that contains the currently-mapped ROM, not the flash.


    Quote from abraXXious

    I saw a file on the Icomp site for restoring them,

    ...and you decided to ignore the "only use when prompted to" remark? Hmm.. OK, it *should* not make a difference, but there's an important detail, and that's "not not use this file if an A1200 accelerator is installed". You didn't mention one here, but you should really be careful if we write warnings like this in the Wiki.


    Quote from abraXXious

    But the problem is unresolved. It still crashes when attempting to read any ROM that should be in flash memory and ACA information reports the "MAP ROMS empty or checksum error".

    Again, the "report" is not about the flash, but the RAM area.


    The prime reason why an ACA500plus crashes when unpacking Kickstart ROMs is overclocking. Note that the system will be cranked up to the maximum speed that's set in the global settigns, regardless of what speed you're selecting for your launch configuration. If you are choosing a 14MHz profine, but still have enabled 42MHz in the global settings, the CPU will be run at 42MHz while unpacking the ROM to make this time as short as possible. Please set your max. speed to 14MHz and see if this resolves your problem.


    Jens

    Hello,


    ahhh, okiez, now I understand. The information is the Kickstart ROM image that has been loaded into the ACA500Plus onboard memory ready for use, not the actual firmware images? Got it. So, that's not the problem if it says it is empty or a checksum error as it is probably empty. So, when is this information populated? If, for instance I used the MAPROM option with the Mainboard Kickstart ROM option then I assume the ACA500Plus would read the mainboard KS ROM into this assigned memory on the ACA500PLus and it would then show this in the TAB information section, however I am confused as to when you would see it as to get back to the ACA500Plus config screen you would need to power off and power back on, which would then delete the memory contents from the ACA500Plus? I am probably misunderstanding how this functions, please explain.


    "This archive will restore Kickstart and OS3.1 install files, in case they got garbled. Only use on ACA500plus. Unpack to a FAT-formatted card in root directory. Insert card into AUX slot and remove any card from Boot slot. Start computer with F1. Only use this archive when instructed to do so. This is not an update!" <-- I did see the last bit, but I assumed (wrongly it would seem) that the Kickstart ROm files had become garbled (see above) and so assumed this archive was intended to fix this. It does not seem to have changed or made anything worse.... just did not affect the problem.


    I did reset the card back to its factory defaults, which I assume sets the CPU speed back to a non-overclocked state. I even manually chose to boot at 14mhz a couple of times, but this did not help. Out of interest, the card was functioning perfectly stable at 42mhz previous to updating it. Is it possible that the update uses "tighter timings" or something and so overclocking that was stable and usable with one firmware may not be stable with another?


    Just to be safe I will go and set the max speed to 14mhz and report back - thanks for your help Jens.

    Anyone would think you knew what you were doing Jens. :)


    I had everything set to 14mhz..... except the Max CPU setting in the F8 menu, which was still set to 42mhz. Changed it to 14mhz.... and it booted fine!


    Now this is the weird part. BEFORE updating to the latest firmware (1.46) the card ran at 42mhz fine. I tried setting the Max CPU speed to 28mhz and this time the ACA500Plus did not crash when copying Kickstart to RAM, but the system still crashes before getting to a Kickstart screen. So I tried 21mhz and this seems to work fine... but that is the weird part - it was running at 42mhz before. I have four ACA500Plus cards, two of which are fitted with the X-Surf 500 module, and they all run (or did) at 42mhz. Two cards already had the latest firmware on them, the other two cards I recently updated, one card still runs at 42mhz after the update, but this card now only seems to want to run at 21mhz. Not a big deal as I was intending to pair one with a spare ACA1232 I have, so I will just use this card.... but I am interested as to why the newer firmware will not allow as much of on overclocked (on this one card) - could the card have always been on the verge of not handling the overclock and the newer firmware tweaks things (timings?) just enough to push it over the edge?

    Quote from abraXXious

    So, when is this information populated?

    After the first system start; you'll see the same message on any ACA500plus when you cold-start. If you enter the menu next time (reset with LMB held), it'll show what's already loaded in the MapROM area.



    Quote from abraXXious

    Out of interest, the card was functioning perfectly stable at 42mhz previous to updating it. Is it possible that the update uses "tighter timings" or something and so overclocking that was stable and usable with one firmware may not be stable with another?

    A software update cannot change the hardware timings of the ACA500plus. It is NOT FPGA-based, but CPLD-based, and the CPLDs have fixed code, which cannot be changed with a flash update. Only our FPGA products can do that (Indivision, Chameleon and the Catweasel MK4, which was the first to use this kind of architecture).


    One thing that may have changed is your power supply's stability, and of course temperature. We've had quite a few hot days in Germany this past week, and heavily overclocking a CPU does require very stable power supply and good ventilation. 42MHz operation may have been on the edge for your card, and with higher temps, your PSU might have lowered it's output voltage enough to become a problem for your CPU.


    So the overclocking problems showing up "just now, after the update" is not a causal link, but merely correlation of things happening at (seemingly) the same time.


    Also, note that the 28MHz setting is almost as demanding as the 42MHz setting for the CPU. This is because the base clock is 85.12548MHz. To get to 28MHz with whole-number division, you need to divide by 3, but that results in the duty cycle of the clock signal to be 33/66. However, I still added the 28MHz clock rate in order to be able to mimic the speed that the Supra accelerator offered back in the days.


    This should explain why you have lost the *two* highest clock rates on this setup. However, it really seems to be on the edge, so you may be successful with a small cooler, new electrolytic caps for your power supply and a fresh voltage setting (yes, many of the original Commodore PSUs have a small potentiometer inside).


    Jens

    Okiez, I was not aware you could get back into the ACA500Plus menu from a reset, but after flashing the system performs a reset and sure enough this information is now populated. Cooliez.

    Yessum, I was aware the ACA500Plus was CPLD based, but I was wondering if there were some software programmable timings for memory access or clock skews etc that could have changed - I am not overly well versed in the internal workings of these cards, so was guessing as to anything that might have changed between updates.


    Heat wise it is around 10 degrees here, quite cool almost chilly, so can rule that out. As for the PSU, I was using a brand new after market PSU (I know your opinion on some of them is not the highest. ;) ). I measured the 5v rail and got 5.08v, so that would be fine. I believe the other variable that can cause trouble is the ripple, which SUPPOSEDLY is less than 50mv on this PSU, though I have no way of measuring it. I must say that this PSU seems to run all the Amigas I have tested on it quite stabily, and a few of them can be rather picky with voltage input. Just to rule it out I tried three other PSU's - one is another new aftermarket unit from a different manufacturer (not as good as the former in my opinion), an original A500 PSU that has been recapped, and one of the later A600 PSU's (the "light" version) which has not yet been recapped. Same results on all.


    I do have some small copper Heatsinks with self adhesive heat transfer material - I will install one to the 68k CPU this evening and see if that makes any difference, however I am starting to think that it might have always been on the very brink of functioning at 42mhz, and just some tiny variable has changed enough that it is now over the brink of working.


    Oh, I am testing it on a difference Amiga that I originally had it installed on, however it is a recently recapped A500+.... could this also be a variable?


    Sorry about the double post - I replied to the thread and didn't realise I was not logged in. I cannot delete the double post my end.

    Quote from abraXXious

    I measured the 5v rail and got 5.08v,

    I'm impresse that you have a multimeter that is calibrated to be this accurate. I guess it's 6000-count at least? Where did you measure, and under what load conditions?

    (yes, this is the intentional poking "do you actually realize what you're saying by quoting two digits after the comma")


    I highly doublt the 50mV ripple. These after market PSUs use MeanWell chassis, and they have changed their spec to "worse than 50mV" years ago. They also don't do cable drop compensation.


    Quote from abraXXious

    I am starting to think that it might have always been on the very brink of functioning at 42mhz, and just some tiny variable has changed enough that it is now over the brink of working.

    Yes, it can really be tiny things, such as a mouse adapter, a monitor solution (especially if it uses it's own power supply: These can introduce ground loops) or an internal memory expansion.


    To reduce bus load, the ACA500plus has a solder-option that lets you remove the A500 68000 CPU. There is a solder-jumper near U24 that says "close jumper if 68000 removed" - this will generate the E-Clok on the ACA500plus and feed it into the machine. This is an expert's mod only, and it is absolutely necessary to remove the 68000 CPU from the A500 main board before you can connect an ACA500plus that is modified in this way. However, if bus load and/or power is the problem, then every tiny bit helps.


    Jens

    Lol, well, you are certainly easy to impress. Can't comment as to how accurate a $60 general purpose multimeter is, but it SAYS 5.08v..... so until I catch it lying to me about borrowing my car keys without permission, I will have to give it the benefit of the doubt.


    I measured the 5v at the floppy power connector - I can measure elsewhere if you would like to suggest another point. As for load conditions, I measured with the machine running and booting to Workbench 3.1 (ACA500Plus connected and accessing the Boot CF Card). Again, happy to try a different set of variables if you like?


    "(yes, this is the intentional poking "do you actually realize what you're saying by quoting two digits after the comma")" <-- that's okay, I have pretty thick skin. :) Did I post 5.08v with a comma? Obviously should have ben a full stop possibly a typo...... or my multimeter is lying to me. ;)


    Hey, I have started using a USB Mouse adapter, I wonder if that could be the changed variable? Will try again without it.


    Totally agree - even with my relatively limited experience I don't believe ANY ripple current tolerances on mass produced power supplies unless they come with certification of individual tuning...... which I have yet to see on an Amiga PSU..... still, I live in hope that a PSU that denotes a max ripple current tolerance of 50ma MIGHT actually be closer to an acceptable tolerance (to the miggy) than a PSU that states a ripple tolerance of 200MA....... hopefully. :)


    I take it the safest bet for a reliable Amiga external PSU is an original Commodore unit freshly recapped with quality Japanese capacitors?


    I am going to do the external clock generator mod! Thanks for the heads up, will report back shortly with the results (and will try without the USB Mouse adapter).


    Thanks again for your help Jens!

    Quote from abraXXious

    Did I post 5.08v with a comma?

    Sorry, that's a cultural difference: While most English-speaking countries use "point" as the delimiter after whole-numbers, many European countries (including Germany) use a comma. So if you say "one point five" in English, we translate it to "eins komma fünf", so the "point" becomes a "comma".


    When a German engineer speaks of "digits after the comma", he most certainly refers to the precision given in a drawing. It does happen to me as wekk: I like to use two digit precision in my CAD system, as that's what I need for traces/spaces on the copper side. However, precision of the milling machines that make the outline of the boards is much less. Naturally, you have to design for tolerances, and PCB manufacturers check that.


    However, if you're leaving the PCB manufacturing bubble and go to service companies that also make non-electronics stuff (such as the plexi things I like to use), you either have to change the drawing to only use one digit precision, or you have to do some explaining first before you send out the drawing. If you don't, you either don't get an offer at all, or it's like "looks like for a medical/military product, let's add a few hundred EUR per unit" - or they get back to you and say "we can't make it according to this drawing".


    There's a lot more not-so-obvious differences in communication between Europe and other cultures - what you often see in press articles is a wrong translation of million/billion/trillion: In the US, this is 10-to-the-6, 10-to-the-9 and 10-to-the-12. However, the German word "billionen" means ten-to-the-12, because we're making a difference between "millionen" (same as in English) and "milliarden", which is 10-to-the-9 (lazy me: "to the" is short for "to the power of").


    And of course there's the whole metric vs. imperial dimensions topic.. let's not go there :-) Just let me know what I *should* have written instead of "after the comma" in order to not be misunderstood?


    Quote from abraXXious

    I take it the safest bet for a reliable Amiga external PSU is an original Commodore unit freshly recapped with quality Japanese capacitors?

    Yes, that's what I recommend until we can deliver the CA-PSU again.


    Jens

    :) We have an electrician who has worked with us for more than three decades who came over from Germany in the early 1980's, lovely fellow by the name of Jurgen Klapper, he is almost family. Anywho, when we first started using his services many moons ago he would give us the most puzzled looks when we used certain phrases and measurements pertaining to electrical supply - I guess for the same reasons you mention. He also has the most amazing stories of dodgy building electricals he came across when working in Italy..... straightened out coat hangers in the walls used to connect AC voltage to outlets which had been moved a couple of metres. :)