ACA1234 CF slot booting woes

Hi,

I just installed ACA1234 to my A1200. O ordered it few weeks ago, so it should be pretty much the latest version.

ACA seems to be working just fine, it is just that I haven't been able to successfully boot from the ACA1234 CF card slot after installing WB3.1. I've tried several CF cards, Transcend x133 4,16 and 32GB versions, Sandisk Extreme 32GB, Kingston 4GB CF and even Transcend CF300 Industrial 256MB card. Kingston came from my old A600 installation, so it definitely has been working on Amiga before. Every card behaves pretty much the same, no boot.

If I use the Amiga IDE slot with original Conner HDD which has WB 3.0 installed, everything works just fine and computer boots from it as usual and I can successfully run for example WHDLoad demos from it. ACA1234install mounts OK, ACATool works and even the CF Card in ACA1234 slot and its two partitions are mounted correctly and can be accessed normally. It just doesn't boot from there and pretty much always stops at Software Failure window C:AddDataTypes Program Failed (error #80000004). IdeSpeedup is at default.

My system is with 1D4 MB revision and I just recently recapped it and removed the two recommended surface mount capacitors from the underside of the board. It still has the original 3.0 KS ROMs onboard. No other mods or accessories are installed, except RTC on the ACA clockport and currently I have the Conner HDD connected to the internal IDE.

I'm suspecting that for some reason the KS3.1 maprom function is not working and that is causing the issue when trying to boot from the newly installed WB3.1. The thing is, I don't understand what I'm doing wrong here, so help is appreciated.

Thanks!

Quote from Jens

If this happens with all cards, the file may just be faulty in the flash of the ACA1234. You could try replacing the file with the version from your older drive - if that gets you further, I can send you a script/archive that will re-flash the relevant blocks of the flash to correct whatever is wrong in there.

Dear lord, thank you! I was pulling my hair off for the best part off this afternoon by swapping all those cards, using the jumper etc. This was something that didn't even cross my mind. I did copy the aforementioned file from the old WB3.0 installation to new WB3.1 CF and it now it boots just fine.

There is something else going on still.

This morning I turned the computer on to continue installing software to system and possibly upgrading the OS. Computer didn't boot up, but showed a red screen which to me applies to a bad ROM (I had the MAPROM function with 3.1 turned on from yesterday). At this point I didn't have anything connected to internal IDE. I removed the ACA1234 and machine booted normally to insert floppy screen. I used the jumper to reset the ACA EEPROM and I got the card to boot again and practically did the installation again from the scratch like yesterday. It again could work for few reboots and restarts, but then soon I managed to get it into a red screen or no boot black screen situation. The green "HDD led" on the card does turn on and blinks, so it definitely tries to do something, even if it doesn't boot correctly.

I then returned the computer to its "original" state when I got the system, that is with an old school Fast RAM / FPU card (with a 4MB stick, no FPU) and the aforementioned HDD. System works like a champ and I see no issues with the operation.

I'm pretty clueless what could be causing all this. That corrupted OS file is definitely a thing, but I'm starting to think there is something else going on with the ACA EEPROM or flash in general.

Quote from Jens

This "red screen" is something we have addressed in a fairly early update. Did you check all updates in our Wiki?

Nope. I didn't expect an update being necessary as the card was just recently shipped from you. But I'll reinstall the card and check the firmware today.

Some more testing. I did the latest update ACA1234upd220228.lha and it didn't give me any errors while doing so.

I haven't seen a single red screen after that, but I still struggle with booting from ACA. I have made these observations:

- Having Internal IDE disabled on enabled doesn't change the problem behaviour (described later) if IDE slot is empty

- If I have system powered off for a longer period of time (few minutes or more), it doesn't boot correctly anymore. Short power off/on cycle, and it boots normally if system has booted before. This is the reason I yesterday thought it worked fine after copying that one file: I had system powered a long time and just few short power cycles in the mean time. Corrupted file is definitely a thing though, so copying that file from WB3.0 solved that particular error I mentioned.

- When it doesn't boot normally, after a half a minute or so ACA green led lights up and shuts down soon afterwards. It then takes a while and this repeats until I make power on/off cycles few times and it finally boots at some point

- If I have short power off/on cycle and IDE is disabled from ACATool, It boots quickly to CF card as it should

So, system works like it should, when I manage to boot it and if I have a short power off/on cycle. But it is unstable if I have system shut down for a longer period of time. When it doesn't seem to boot at all, the green led on the ACA lights up and blinks once in a while like the accelerator is trying to boot, but for some reason it doesn't. After 2-3 or4 power off/on cycles, I usually get it to boot at some point.

I have tested the system with three different PSUs, Amiga 500, A1200 and Electroware. All have exactly the same results and symptoms, so no differences there. Old PSUs have not been recapped yet, but I've tested them on other Amigas and they seem to work just fine. I have also used Electroware A LOT with my pimped up A600 with Furia, ChipRAM/RTC, CF HDD and SUM600 adapter and it works there perfectly. It is stable as rock and don't remember a single crash with it.

Edit: I just had one no-boot issue, where ACA green led stayed on and it didn't boot. So in the case of no boot, light may turn off and light up later again, or like in this case, it just stayed on.

Quote from Jens

I won't argue about know-bad PSUs, so please leave Elektroware out of the equation. The Elektroware PSU might actually be the root cause of the flash memory being corrupted.

This does seem like a PSU-related problem, so if you've planned to re-cap (one of) the Commodore PSUs, now would be the time.

What else is in the system, what's your monitor solution?

Amiga is connected to CM8833mk2 with RGB SCART cable. CM8833mk2 has SCART adapter on it. There are nothing unusual in the system otherwise. Only accessory it has currently besides the ACA1234 and your RTC module is MouSTer USB mouse adapter with wireless dongle. I've also tested it unplugged and so far hasn't noticed that it would make any difference. I can't see how that would affect to the operation anyways, it is such a low power unit after all.

I run more tests with original PSUs this evening: let the system sit for awhile and see if it boots any differently in the long run. The thing is, although ACA probably consumes somewhat more juice, but that old Conner HDD with that memory expansion surely adds some wattage to the equation too and system just works with those: No issues whatsoever and this with all three PSUs. Also, when I get the system to boot, to my knowledge it runs fine and I have tested some WHDLoad demos for example and haven't experienced any crashes. One had some mild graphics anomalies in one section, but that may be just due to the fact that it doesn't like the accelerators to begin with.

Yeah, I have planned recapping those. I don't have those beefy PSU caps in my parts bin, but I guess it is time to crack the suckers open and put an order to those bad boys.

Edit: I also remembered that I have also A600 PSU, so I'll test with that too.

I just want to thank you already for your active support on this issue, awesome and thumbs up!

Interesting topic regarding monitor playing a role! One thing I wouldn't have even considered myself. I can test the system just fine without the monitor to rule that out, because I have "learned" even without looking at the monitor when it boots up. The main being the FDD starts its usual clicking and I know that it is in the WB. I have powered the system from the same multi socket, but I also tried to put the PSU to a separate wall socket before, so the Amiga was powered from a different source to the monitor.

HDD is 2,5" old laptop type of drive. I'm not actually sure of the capacity, I didn't bother to check it in the WB because I don't have any plans to use it in long term, and the sticker doesn't say it, even the type number is somewhat obscure. It may be a 250 megs.

I just recapped the system in monday with Panasonic caps. It didn't have any disastrous leaks like my A600 did, but it was on the verge of cap leaks. One SMD capacitor had little bit visible "stuff" on one leg and three caps or so started to smell when I started to heat them indicating that they had pushed some stuff out too. My A600 few years ago was a mess and smelled like a tuna fish canning factory every time it was powered on and it warmed a little bit. Of course that was before I did the recap and cleaned the board.

I will report further Demo findings, if there are any, after I get this thing sorted out. Demo was Real by Complex:

https://www.youtube.com/watch?v=MD-k8Oxkalc

If I remember correctly I had sort of vertical blue bar appear in a scene starting at 2:55. It was only visible in a blue textured part little bit left from the center, so it wasn't for the whole screen, more like an anomaly on the colors of blue texture on the inside of the wall. The other was in a scene starting at 3:15, where at some angles some of the yellow triangles disappeared for a while. Both to me looked like some sort of timing issues with the demo. I also tested couple other demos, Full Moon and BreathTaker (both also AGA demos) and both worked just fine with several runs.

I have now tested few times yesterday using pretty much all the PSUs, but mostly my A600 PSU as it is the beefiest Commodore PSU I have (3,5A@5V).

Earlier in my testing I had the case open because of fiddling with the HDD and it was also easier to remove/install the ACA if I needed to use the Jumper. I assembled the system yesterday fully as it was evident that I can get it to boot. So now I had also possibility to see power led in action.

Same symptoms continue and this is how it goes:

- When I power on the system after a longer period of time, it fails to boot and power led is blinking 7 times. It goes through this cycle few times and then boots first displaying the error message in the attachment

- I clocked few times the time it takes to finish the cycle before error message is displayed: with IDE off it was 1 min 20 seconds. With IDE on, I clocked 1 min 50 seconds. There is variance and with IDE on, I got at least once the system boot on a first try.

- After error message, system proceeds to the WB normally after waiting perhaps 10 seconds or so, or pressing the mouse button

- On WB, behavior seems normal

- If computer boots to WB without any issues, no error message is displayed: it is only after the failed boot attempt (that is one or more cycles of blinking power LED)

So, I couldn't make anything conclusive about the booting issue, except that it is relatively consistent. Again, in the original configuration I didn't experience a single issue with booting or operation and when the system finally boots with ACA, operation seems to be normal, although I don't have long term experience at this point.

Quote from Jens

Please do try the keyboard reset if you see a non-start after longer power-off.

That didn't help and I have a feeling that during a failing boot, it even doesn't read the keyboard, or at least constantly.

But, I did the following discovery:

I set the processor speed to 25MHz from ACATool (processor in my card is a 33MHz variant). I have now tried several "cold" starts and boom, it goes to WB as it should and there are no issues.

This got me thinking, that what if it isn't about power per se, but there is some issue with a processor and the failure of the boot is actually about thermal interactions and change of resistance or leak current somewhere in the die? In other words, short power cycle works because processor die is already warm after the machine has been on for a while and it has reached some operating equilibrium. Lowering the clock speed introduces improved stability and it works now correctly as it should.

I tested overclocking the processor and you are right: With 40Mhz the system boots like a champ like with 25MHz.

And guess what? When I set it back to 33MHz the problem becomes immediately back. So to me, it seems to work without a hitch at 25MHz and 40MHz, but not at the nominal speed of 33MHz (haven't tried the higher settings).

I don't understand this.

I just sent mail to you at shop@icomp.de.