Posts by Jens

RTG works nicely on A2000 systems with 68030@25MHz and 3.5MB/second performance to the framebuffer, so CPU requirements are rather low. Further, if there's some 2D acceleration and hardware-sprite for the mouse pointer, P96 can really shine - one of the reasons that we've acquired ownership of P96.

DanyPPC, your observation of slowing down standard screens with 256 colours is correct. However, that's not due to the larger amount of colours, but due to the extremely reduced bandwidth available to the CPU. The effect practically accelerates itself, because by increasing colour depth, you not only increase the amount of data that the CPU must handle, but you also reduce bandwidth at the same time. This does not happen on a properly-designed RTG solution.

Integrating RTG with the Indivision series is not a plan, but an RTG card would have to be integrated with the Indivision output, so a single monitor can be used for both. I do have ideas, but no actual design path, as there's other things that need to fit the A1200 as well. Space inside an A1200 desktop is scarce!

I keep telling people to look for PSUs from vendors with a product liability insurance, but it appears like that's not available anywhere.

I already talked to my agent at the product liability insurance for coverage of power supplies, and we came to the conclusion that our own design is out of the question, as it would require tests and paperwork of well over 40.000,- EUR. So the plan is to make a DC-DC converter that takes 19V or 24V from a laptop PSU and converts it to the voltages that the Amiga requires.

I can't say when this will be available - might be summer 2019.

Thanks for the hint - that's not realistic any more. Both the A1200 and the A4000D versions need a re-design, and current projects have taken up more time than anticipated.

Quote from mbruines197703

Can the Indivision ECS V2 be piggy bagged to use 2 monitors like the earlier model? And if so is it possible to mix the indivision V1 with the indivision V2?

No, the 2-monitor mode is not supported any more.

Quote from alex

"Can only be used with an Apollo accelerator card if you have an A604 or A604n."

Is this correct?

It's in the product description, so sure, this is correct. This applies especially to the A603, where the Apollo accelerator would conflict with Indivision ECS V2.

Quote from jab76

Hmmm, I read the wiki and it says that the 1M Chip Ram option is only available if you have a ram expansion of exactly 512k. This must be a new requirement, because with the older firmware it worked just fine.

This requirement is not new - it was there ever since the feature was introduced in 2013 with the first ACA500. If some firmware is offering the ChipMap feature although the trapdoor expansion is not exactly 512k, it must have been a bug (although we have not gotten any report about this - probably because people didn't activate it, or found that the system is not stable with the combination).

What you can try is to remove the Gary-adapter that makes the expansion 2M. If the expansion is built with some common sense, it should automatically downgrade to 512k, and you won't have more expenses.

Your picture is showing a mechanical damage of the connector, either caused by your A1200 not being chamfered, or by the card being tilted when pushing it in. This is not covered by warranty.

What's the warratny ID of the card? Just asking to see the age and sales path.

There are still a few units in safety stock that will be released in the next two weeks. A new production run will be started next year.

That's a known weakness of the shop site if you are using a mobile device: The register page is not linked to from the mobile-compatible view. Just follow the link that Tobias has posted.

Your picture shows that the fix that makes the unit more immune against ripple is already applied (and even a few extra caps that were later removed from that fix-list). This clearly shows that both your PSUs react the same way to a high load: They add ripple pretty badly.

Do you have a chance to re-cap these PSUs? Are they switch mode (light type) or transformer type (heavy) PSUs?

Thomas RIchter actually took part in creating OS 3.1.4, and I'd expect his MMU library to be part of the OS, which attempts to create a clean MMU setup. During development, I did communicate the way that the "common ACA register interface" works, so Hyperion should have taken this into account. I even got a request from a betatester of the OS who owns a card with this register interface. So what remains now is to remind them of that fact and have Hyperion fix the bug.

Could you please try the ACAtune output using the old Kickstart, ideally without startup-sequence? Just to make sure that an MMU setting of the new OS is not interferring with the ACA registers.

Quote from frame

This confuses me. I got my card end of October but the firmware update was released on Nov 8th?

That's a different type of firmware: The CPLD code is something you can't update yourself. The update that was published earlier this month is a software update that's stored in flash memory of the card. So yes, these terms are kind of confusing; the output you got from the tools is referring to CPLD code, which has been newly developed for the ACA1221 when the ACA1221ec was under development.

Quote from frame

I got one additional question: Because I use a physical 3.1.4 ROM, I don't need to use the MapRom feature. The Wiki states that MapRom speeds up the system dramatically by copying the ROM to RAM. What would be the benefit to set the jumper to parked position?

If you have a physical ROM, then you can safely set the jumper to the "MapROM" position. The recommendation of "parking position" is mainly for safety reasons, as we've had quite a few requests from users who were trying to load the kick file. If you have an older physical ROM and have the jumper set to "MapROM", loading a file will fail.

This does sound like a power supply issue, so at this point, I have to assume that your twp PSUs both have the same weakness. After all, the Amiga with two accelerators and the flicker fixer does draw considerable power.

a close-up photo of the components under Denise will show me more. There has been a change in assembly of these flickerfixers to make them more immune against power supply ripple. So please remove the Denise chip and post a new photo.

The tools properly see that it's an ACA1221ec, yet they still output text that ONLY applies to the ACA1221, which is not what you have. This is either a bug of the software, or it's a mix of software that you have installed on your computer. In any case, the CPU mask 0E30G that is listed is for the "full" 68020 CPU, but the ACA1221ec uses the 68ec020 processor. In fact, the firmware upgrade for the ACA1221 was actually made during development of the ACA1221ec, so your ACA1221ec already has the latest code in the CPLDs.

In other words: You can safely ignore this message, as it does not apply to your card. You might want to check if the latest ACA1233n software fixes this; it was developed on the same basis, and it's attempting to unify all accelerator software in a single archive.

If you are having stability issues, please try the obvious first: Check if your power supply is good, maybe leave the trapdoor open for additional cooling.

Jens

The 68030 does not need active cooling - not even the 55MHz cards that were available until recently. If you're worried about heat, you can just leave the trapdoor open, but since space is very tight under the keyboard, there's hardly any room for a heat sink. The holes are actually meant to carry such a heat sink, but since the 68030 CPU has a very high junction temperature spec (after all, they've been used in aviation), I never took the time to design a heat sink.

That $FEFEFE colour trick may not fix all problems that an A1200 may have when using S-Hires/Double-CAS modes. bitter, that's definitely not Lisa "not liking the workout", but a plain assembly error that was made by whoever produced the A1200 back in the days. There is a number of memos from Commodore on the topic of timing fixes, but none of them is really complete.

I did some research myself to find&fix the errors. If you have SMD soldering equipment, you can make your A1200 compatible with S-Hires modes:

1) The negative bias voltage of the Alice-chip is already supported with a ceramic capacitor, but that's too small. I'm talking about C2X on the bottom side of the A1200 board, about "under Alice". You just add a 10µF ceramic cap on top (no need to remove the old one).

2) The CCK clock needs to be delayed a bit, which can be done with a 100pF capaitor to Vcc. Indivision AGA MK2 is prepared for such a modification: You can just solder the 100pF cap onto the unpopulated place R24. However, that's probably the only part of your setup that's still in the warranty period, so you might want to choose any other place on the A1200 board where CCK and Vcc are close to each other.

It may well be that some network drivers have been removed from the package when we licensed it, as there may be a license clash with E3B, who did finance certain drivers back in the days.

It's a bad idea mixing different versions of drivers and the stack. Please only use the drivers that are included in the Poseidon archive that we provide in the Wiki. Trying weird combinations is not part of the product support.

Power supply design is more than just slapping a few parts together, and if you look close at what ripple is caused by, you may get the impression that the capacitor type has been changed with the change of the ripple spec.

Since you can only add a capacitor "after a cable", you need to take the cable into account for your calculation, and that cable is a resistor and a coil. So you might even make things worse, because the combination of a coil and a capacitor can start to oscillate. Although this is fairly unlikely, it's still not impossible without knowing the exact details such as conversion frequency and regulation rail of the unit - and you won't get any recommendation from my side for modifying a power supply, as I ahve already given the best recommendation: Go for a DC-DC type PSU, as that completely eliminates cross-regulation headaches.

I know of a few people who use that part, but I cannot recommend them any more, as they recently changed their ripple spec.

It is possibe, but the external USB connectors may not be used in that case. Remember it's just a 2-port host controller, and you're adding USB ports to the same signal lines, which may only be used by one device at a time.