Posts by Jens

Thanks!

Maybe some pictures of your hack will help others understand what type of material you've used.

Quote from Glidewell

but neither ACA Install disk nor ACA-CF card are detected/mounted.

ACA-CF *should* be mounted if you insert a card at boot time. It certainly does when we run 1.3 on ACA500plus, it's just that 1.3 is acting up if 3.1 has been booted before - that's our current headscratcher.

You might want to try a plain original 1.3 and a formatted CF card in the ACA1234 CF slot. You wrote that you already tried this, but was there a 1.3-bootable CF card inserted when you tried?

Quote from Borschtsch

Btw. I've never had any issues on my other A1200 with an indivision AGA mk2.

I don't think we're looking at any hardware issue here. It's more like "chasing a ghost", where we don't yet know what the ghost really is.

The "binary does not work any more after copying" is worth investigating. Why don't you use one of those nice new OS3.2 tools and create an MD5sum of the source and target binary? I wouldn't be surprised if they are different.

Quote from Glidewell

I wonder, how would ACA1234 mount and boot from its CF? ACA_CF is not visible under KS1.3+WB1.3.

The CF card driver and mounter have nothing to do with the virtual install disk. The install disk is mounted using a totally different device. However, the autoconfig details are geared towards working under Kick 1.3, it's just that the rest of the product (ACAtool, added software) is only good for later OS versions.

The warm boot, however, is something that reminds me of the current problem that we have with the ACA500plus. Here's where I should really hand over to Timm, as I don't remember enough details about the problem we're looking for in the A500/ACA500plus/ACA1234 configuration. Timm, does this remind you of anything, could Glidewell do any more tests to help us?

Quote from Glidewell

P.S. I think of soldering the wire link while A1200 is still open and ACA1234 is accessible from above (keyboard lifted). I hope it won't void the warranty, will it?

If you have soldered such things before and didn't destroy anything (aka if you feel fit to do the job), it won't affect your warranty. However, please understand that I can only exclude these solder spots from the warranty checks - if something gets destroyed in the process, I can't be held responsible. To any experienced hobbyist, this seems like a very simple patch, especially because there is no cutting traces, no SMD parts and only 2.54mm-pitch solder spots involved.

So you might think "piece of cake!", but it should still be executed with "this is expensive gear!" in the back of your head

"Power rail" would be GND and VCC. GND is "everything minus", including the screw of the metal back plate. VCC is the 5V rail that goes to all active parts of the card. The "teeth" are called "card edge".

Since this is a multi-layer card that has the power distribution layers outside, you may be looking for a short between the top and bottom layer, which might be caused by ripping out one of the RAM chips too fast, by over-tightening a screw or a simple solder whisker between a GND/VCC pin and one of the power layers.

Can't say off the top of my head if top or bottom is GND, but from the position of the screw in your video, I'd expect top to be GND and bottom to be Vcc. Still, the whisker may be on either side, and it may be in a totally different position (nowhere near the RAM), as a blob of solder might have dropped anywhere on the board while you worked on it.

Power supplies are a delicate topic - especially in a Checkmate case.. anyway, please read our PSU FAQ and confirm that you have chosen a DC-DC based power supply. Not that I'd expect the maker(s) of the Checkmate case to inform customers about proper PSU choices.. also, please refrain from posting "anecdotal evidence" in form of "did work so far..". If a power supply is not suitable, you should not ask why it doesn't work now, but you should ask why it worked before.

Rant aside, all products that leave here have passed a functional test - really every single one of them, otherwise they would not get the warranty ID. Protocols of the test result are stored in our database. I therefore assume that the MK3 is fine, and we can isolate the fault.

If all else fails, we can of course take a look at it here, and we also do paid repairs on out-of-warranty products if we still ahve the spare parts. For the MK2 and MK2cr versions of the flickerfixer, we have all parts for service in stock, but let's stay on the MK3 for now:

- does the computer start, do you see it booting when the flicker fixer is installed?

- do you have a possibility to make a rescue disk?

We'll refund shipping with the unit refund when it's here. Please understand that we can't send money ahead of time, and that our contracts with shipping companies do not include returns, because we're way too small for that.

Please also read the article about how to return things in order not to cause more cost than necessary:

If you need to return something to iComp for service...

I didn't find clear wording, so please allow the question: is this behaviour confirmed to be caused by the 3.2.1 ROM than? Re-phrased: Can you get rid of the error in WinUAE by changing to Kickstart 3.1?

The GUI not working on 1.3 is expected, but command line may be something that Timm can fix once we have ironed out the problems that we have with the ACA500plus.

Again, the local CF card slot of the ACA1234 should mount any inserted card and boot from it. Did you try that?

Just did a quick check on my 2630: You should get a resistance of around 100 ohms between GND and VCC, so you are really looking for a short between the power rails on your card.

The checksum routine is the same as in all Kickstart images: The sum of all longs must be -1, and that's achieved by changing the long at (end-20) of the image. For 256k images, that's offset $3F.FE8 and for 512k images, it's $7F.FE8.

You should be able to correct the checksum with the "Sumkick" tool, which is part of MKick, an old 1995 package available on Aminet.

Still, I'd expect the Autoconfig ROM of the ACA1234 to add the CF card driver and make the CF card slot of the ACA1234 available. The Autoconfig ROM uses a 4-bit loader, which is already geared towards being Kick1.3 compatible.

This sounds like a copy-failure. There is a known bug in the copy-command of WB3.0, but that was fixed in 3.1 already, and I would not expect it to be back in 3.2.1.

I'd therefore look for the fault in your mass-storage. Please describe the exact setup and also mention the MaxTransfer settings of the partition(s) in use.

Best thing I can offer is to take a look at the board here. Please ship with your PSU and chips installed, so we can measure what's going on without having to guess if the changed chips/PSU will cause a difference.

Yes, that's possible.

Could you see if deactivating the MMU library makes it work? Our rescue disk will not install the MMU lib (it's not on the floppy!), but OS3.2.1 will nag if it finds an MMU and "convince" everyone to install MMU lib, even when there's little use for it. We've had massive problems with this already.

Quote from Glidewell

blinking every 5-6 seconds.

Just to be clear: That blinking is on the Amiga power LED, not on the accelerator?

Quote from Glidewell

Flashing this doubled kickstart (instead of using acatool romfile command) would not make any difference?

It might make a difference, as much less structures will remain in memory. Autoconfig is happening really early in the boot process.

If you don't have easy access to the jumper, you might want to wait until Timm has implemented a new feature to the flash of the ACA1234, and that's a "disable autoconfig ROM on boot" when a certain condition is met, such as "hold joystick button during cold start". It is very likely that we'll need that for debugging Kickstart 1.3 on ACA500plus, and I see no problems sending that debug-version to you as well.

Wow - two valid digits after the comma? Is this a calibrated unit? I'm asking because the input stage of the C64RMK2 power should burn anything above 12.1V. If your measurement is accurate, the Z-Diode near the power input should have considerable temperature.

Well, my opinion about the Checkmate case is well-known... make sure you're using a DC-DC converter-based power supply. Anything else will cause lots of pain locating an error that seemingly wanders with new hardware installed.

Quote from Lamaman1971

Sounds to me like my only option to retain the full functionality of the MK3 is to go with Lyra - PS2 out of the mainboard and then use a vanilla PS2 keyboard - rather than converting the amiga A1200 keyboard over with the Keyrah.

Yes, that is one option. I do not know how the SUM adapter handles multiple keys at the same time, but knowing that n-key-rollover is a whole different protocol on USB, my expectation is that it won't use that mode. Our plan, however, is to implement that mode for Keyrah V3, because the Amiga keyboard is capable of making a difference of up to 15 keys (including qualifiers). I'd like the product to reflect the capabilities of the physical hardware - and that will be an improvement over Keyrah V2. Note that the n-key rollover mode on USB is an optional protocol, so the standard USB protocol will always be supported. This will work with all adapters and computers, though with the limitation of max. six keys pressed at the same time. This is very likely to be sufficient for the live config function of Indivision AGA MK3.

Conversion back to the Amiga is something we've thought about as well. PS2 has the huge benefit of very small protocol overhead and supporting n-key rollover "out of the box", so playing games with the keyboard that require multiple keys pressed at the same time is easily possible with Lyra 3 and a good keyboard. For a next-generation Lyra (which is purely hypothetical by now), a much more complex USB handling would be required in order to support n-key rollover. Since PS2 covers that all, the most sane way to solve this problem would be to add the PS2 protocol into Keyrah V3, so it will be able to interface with Lyra3. I'll talk to Tobias - he's assigned to the firmware of Keyrah V3.

Kickstart 1.3 is a 256k ROM that will have to be doubled. On the A500 and A1000, there's just a mirror of $fc.0000 to $ff.ffff that starts at $f8.0000.

You can generate a 512k file by copying the same image twice, for example with the "join" command of the Amiga (part of every WB). However, the current firmware of the ACA1234 is not compatible with Kickstart 1.3. We' still have a problem when Kickstart 1.3 is selected on an ACA500plus (it's a lot easier to flash there), and it seems like the problem is located in the ACA1234's flash ROM.

We need more testing, so if you have an image that works on an A1200 and flash it, your result will be very interesting for us.

It's on the driver disk that came with the Microbotics memory expansions. There's a DMS of that disk on BBoAH.