Individual Computers - New accelerator cards for A600 & A1200

[fitzsteve]

Originally Posted by Riempie

I've bought a 300W Power Supply from AmigaKit. That should do it right? Still I have the issues.
http://amigakit.leamancomputing.com/...roducts_id=866

I use the same PSU to drive an A1200 with BPPC/BVision, Subway USB & Delfina Sound Card! It should be plenty for an ACA1231

[Bamiga2002]

Originally Posted by fitzsteve

I use the same PSU to drive an A1200 with BPPC/BVision, Subway USB & Delfina Sound Card! It should be plenty for an ACA1231

Does it have enough juice for both systems at the same time if you are using the same PSU?

Nice looking mod

[Schoenfeld]

It's not about the "amount of juice" - it has never been with an Amiga. A fully equipped A4000 takes way less than 80W.

It's ONLY about the quality of the 5V rail: Ripple must be low, because the Amiga has no point-of-load regulation. You can only measure ripple with an oscilloscope.

Jens

[som99]

Originally Posted by Schoenfeld

It's not about the "amount of juice" - it has never been with an Amiga. A fully equipped A4000 takes way less than 80W.

It's ONLY about the quality of the 5V rail: Ripple must be low, because the Amiga has no point-of-load regulation. You can only measure ripple with an oscilloscope.

Jens

Ofc the wattage of the PSU is not the important factor since the Wattage is counted on all rails, but the Ampere is important, for example my A1200 needs 10-15+ amp on the 5v rail and the best quality ATX PSU I could find fitting inside the Amiga power brick had 25A on the 5v rail so im safe forever, Ive tried with a ATX PSU with 10A on the 5v rail and it was not enough for my expanded A1200.

[Schoenfeld]

Originally Posted by som99

for example my A1200 needs 10-15+ amp on the 5v rail

This is utter nonsense. Do you have a towered A1200 with a 25-Zorro-slot expansion, fully equipped with power-hungry cards?

An A1200 board takes 1.7A on 5V while booting, and 1.5A when idle.
ACA1231 takes 0.56A under full load, 0.45A when idle
Indivision AGA MK2 takes 0.45A on highest output pixelclock and DVI enabled

What do you do with the spare 12 amps?

Originally Posted by som99

and the best quality ATX PSU I could find fitting inside the Amiga power brick had 25A on the 5v rail so im safe forever, Ive tried with a ATX PSU with 10A on the 5v rail and it was not enough for my expanded A1200.

ATX PSUs are not the proper choice for an Amiga, because they rely on point-of-load regulation inside the supplied computer. It was way before the ATX standard that the 5V rail was used inside the computer without further regulation (back when the Amiga was new and PCs used the power supplies with P8/P9 connectors). Nowadays, the 5V rail is not critical any more, so regulation is not sufficient any more.

The old P8/P9 PSUs are too old to still supply low-ripple power in the year 2012. I must admit that the only way is currently to refurbish the original A1200 PSU with new electrolytic caps, but this should only be done by a professinoal - the primary side of a PSU contains voltages that are way too dangerous to be handled by a hobbyist (no matter how experienced).

Jens

[som99]

Originally Posted by Schoenfeld

This is utter nonsense. Do you have a towered A1200 with a 25-Zorro-slot expansion, fully equipped with power-hungry cards?

An A1200 board takes 1.7A on 5V while booting, and 1.5A when idle.
ACA1231 takes 0.56A under full load, 0.45A when idle
Indivision AGA MK2 takes 0.45A on highest output pixelclock and DVI enabled

What do you do with the spare 12 amps?


ATX PSUs are not the proper choice for an Amiga, because they rely on point-of-load regulation inside the supplied computer. It was way before the ATX standard that the 5V rail was used inside the computer without further regulation (back when the Amiga was new and PCs used the power supplies with P8/P9 connectors). Nowadays, the 5V rail is not critical any more, so regulation is not sufficient any more.

The old P8/P9 PSUs are too old to still supply low-ripple power in the year 2012. I must admit that the only way is currently to refurbish the original A1200 PSU with new electrolytic caps, but this should only be done by a professinoal - the primary side of a PSU contains voltages that are way too dangerous to be handled by a hobbyist (no matter how experienced).

Jens

Strange, then I do not know why my Amiga wont boot of the 10A PSU, worked until I put the indivision in then I could not get it to boot, it froze and behaved strange just before the AmigaOS was loaded fully, replaced the PSU with a stronger one and bam worked.

[Schoenfeld]

Originally Posted by som99

Strange, then I do not know why my Amiga wont boot of the 10A PSU,

The answer is right in the posting that you've quoted: ATX PSUs don't treat the 5V rail as a critical line. Regulation on that rail is average, but not good enough to maintain stable operation of a computer that is 5V-based.

Jens

[Riempie]

I don't understand anything you guys are talking about, but I assume AmigaKit knows what they're doing when making their own PSU's?

[Bamiga2002]

I have a good quality ATX PSU in my Mediator setup that's given no problems as of yet. Maybe the mediator helps here dunno...

[som99]

Originally Posted by Schoenfeld

The answer is right in the posting that you've quoted: ATX PSUs don't treat the 5V rail as a critical line. Regulation on that rail is average, but not good enough to maintain stable operation of a computer that is 5V-based.

Jens

How kinky is an Amiga then, since I have no problem with high quality ATX PSUs manufactured by FSP. If I check decent FSP PSUs I see that the ripple on low loads (around 25 percent load) on the 5v rail is 9mV is that low ammount really a problem, seems decent enough.

[Toni Wilen]

There are two major "types" of PSU. (Or more like 3, third being Crap, blows up when it is loaded with half of rated capacity, bad regulation, too much ripple)

Non-crap cheaper ones are group regulated (Only load level on 12v line is used for regulation), more expensive ones are independently regulated (each rail has its own regulator).

EDIT: Also different PSUs have different min load requirements.

All this makes choosing the best "Amiga compatible" PSU very very difficult.

[som99]

Originally Posted by Toni Wilen

There are two major "types" of PSU. (Or more like 3, third being Crap, blows up when it is loaded with half of rated capacity, bad regulation, too much ripple)

Non-crap cheaper ones are group regulated (Only load level on 12v line is used for regulation), more expensive ones are independently regulated (each rail has its own regulator).

EDIT: Also different PSUs have different min load requirements.

All this makes choosing the best "Amiga compatible" PSU very very difficult.

Thanks, I do not know what I was thinking, If I just had taken time thinking I would know that the Amiga ofc can't drain that many Amperes, my bad not thinking straight.
Also had completely forgotten about thinking on the minimum load factor so thanks for remaining me.

Then it seems that the best way to go when using ATX PSUs must be a quality small form factor ones, since the small mATX ones are rated at a lot lower wattage they are made for low consuming systems (HTPC:s) and would fit the Amiga good as long as it has a decent 5v rail (good reviews show the ripple at all individual rails under x% amount of load), then I think my move to the mATX FSP one was a smart choice

Edit: Also sorry Jens, was not thinking straight, been working my ass of at work and replying from work stressed and I have not read what you wrote properly.

[Retrofan]

Then if it isn't necessary that amount of amps as Jens says, Zetr0 was wrong http://www.amibay.com/showthread.php...ing+guide+post

Amigakit's psus have also 20A in the 5v line, and the one I tried with 12A there also wasn't booting. The 3 ones I've got working have 21A there.

I don't say he's wrong, I don't know what their ripples are.

[Schoenfeld]

New version of ACAtune here.

Jens

[Retrofan]

Tested. Jens you are great. Thanks for still supporting the ACA1230/56.

You know that with my configuration (Amikit) I coudn't make Acatune to recognize the ACA; "card not supported".

You've made it now to recognize it ... I've got to try different combinations, but my question is why it says it's a 1230/28 while it's an ACA1230/56. Although I suposse it won't affect it.





Edit: With that command I was getting 18.70mips with SysSpeed, the same that without Acatune. Now with:

C:ACATune -maprom * -z2cache on -fastchip on -fastz2 on -cache on -burst on >NIL:
I've got 18.78.

Edit: Finally I've left:
C:ACATune -maprom * -fastmem on -chipcache on -z2cache on -fastchip on -fastz2 on -cache on -burst on >NIL:
Although I get Data burst as off.

[Methanoid]

Jens

Are any of your new cards going to be "small" like ACA1231? I know you said the pictures were of prototype cards so the final cards could be different....

[Schoenfeld]

Retrofan:
I strongly recommend to switch OFF chipcache. Your card should show up as 56MHz if that's what it actually is. Try cleaning the connector, and re-check that it really provides the computation power that you expect from the 56MHz card. If in doubt, post a Sysinfo screenshot.

Methanoid:
Size of the ACA620 and ACA1220/ACA1232 is the same as the prototypes. Only the colour is different, and some things have been added. The ACA620 is the first one to come out of production sometime next week. I have added a freezer framework to the memory controller and also found a way to make almost all of the 16MByte physical memory available to special software: After running ACAtune with the "maxmem" option, you have a total of 11.3MByte Fastmem, a fast kickstart "maprom" of up to 1.5MByte, 1060KBytes of freezer memory and about 2.5MByte for special software (hoping to find someone who writes a RAD: -like driver for that).

On the mechanical side, the mass-produced ACA620 got a ventilation hole for the 68EC020 processor.

During development of the ACA620, I found a way to optimize cache for the 68020 processor, and changed something in the circuit of the ACA1220 that lets me put that optimization into ACA1220 as well. I have also added a level-7 Interrupt connector, but can't tell (yet) if the freezer-functionality will fit the logic. That's the kind of changes that you can expect - no change in board size ;-)

Jens

[Retrofan]

Thanks Jens. In fact with SysInfo I noticed that I had the Fast Ata disabled, now I've got it with Pio 3, as with 4 I was getting messages of DH0 read attempt outside partition, that I must see, anyway this new Cf is giving problems from the beginning, but I bought it after the problem with ACATune and card not supported.
Well, with SysInfo I've got 12,24Mips. This is an screenshot if you find something rare. There's no need to clean any connector 'cause if I just change the Cf Hd with another Workbench I get the Acatune showing it's an ACA1230/56. Now I don't have chipcache, but ACATune keeps telling it's another ACA (and it's the same with -chipcache off).

... Anyway it seems the mips are right, aren't they?

[mfilos]

Cheers for the update Jens
Will test it with my ACA630@30 and report back.
I updated the summary of all ACATune versions along with Changelog readme file as well

http://eab.abime.net/showpost.php?p=...postcount=2316

[mfilos]

OK I tested the new version and seems to work just fine on my ACA630@30MHz.
I found out a bad side-effect though :S
Under WinUAE all versions of ACATune (until this one) reported that Card is not present and booting continued succesfully (since it was >NIL).
In last version and for reasons that only Jens or the Developers know ACATune reports that it found an ACA accelerator reporting ACA630@25 if the chipset is ECS and A600 and ACA1230@28 if the chipset is AGA and A1200!!!

You can see an exampled screenshot that I took just now from WinUAE and a test environment of my A600 system.



...and the same environment just changing the chipset to AGA and A1200:




That bug or feature (dunno what it is tbh) prevents WinUAE from loading (making an endless black booting loop) until you disable ACATune from startup-sequence!
It would be nice if this bug or feature would be removed, cause it's rather irritating having to disabled ACATune if you're using your backup under WinUAE and then re-enabling it if you transfer it back into the Real Amiga!