tom@microsoft.UUCP (Tom MCCONNELL) (05/13/91)
Here's a question for you hardware guru's(Dave H?) out there.
I have an A1000 with a rejuvenator, flicker fixer, and an Expansion
Technologies Zorro II expansion box, which has a Trumpcard Pro and
Quantom HD in it.
I have an opportunity to get a 2630 Commodore 68030 acclerator for
_cheap_! The only problem is I don't have a CPU slot!
But I DO have a fully compliant Zorro II slot.
Somewhere I heard about someone hacking a 2630 onto their expansion
port, which is essentially a different form of the Zorro slot, right?
Is it possible to do the same thing in a Zorro slot?
What would be involved in doing so?
Hardware Guru's, here's your chance to show your stuff!
Endlessly curious...
-Tom
--
"Hey COW!" _____________((
/ ** o \ ______ _______________________
/| **** \ _/moo...\ | Tom McConnell |
/ | ** /\__/ \______/ | uunet!microsoft!tom |daveh@cbmvax.commodore.com (Dave Haynie) (05/16/91)
In article <72302@microsoft.UUCP> tom@microsoft.UUCP (Tom MCCONNELL) writes: >I have an opportunity to get a 2630 Commodore 68030 acclerator for >_cheap_! The only problem is I don't have a CPU slot! >But I DO have a fully compliant Zorro II slot. >Somewhere I heard about someone hacking a 2630 onto their expansion >port, which is essentially a different form of the Zorro slot, right? >Is it possible to do the same thing in a Zorro slot? Ok, here's the deal. The 86 pin edge on the A500 or A2000 is actually more similar to an A2000 coprocessor slot than a Zorro II slot. In fact, the A2000's coprocessor slot has nearly the same pinout as the "edge", only with a few more signals than available on the 86 pin edge. More importantly, it sits on the local bus (physically connected to the 68000) and, since no internal DMA device exist on 500s or 1000s, the 68000 can be overridden by a properly connected A2000 coprocessor card, even though the 500 or 1000 lack the full "coprocessor interface" of the 2000. The Zorro II bus doesn't provide the required signals to really manage the same kind of thing. Plus, there will be a set of buffers between any Zorro II slot and the local bus, which get in the way. The largest problem is that the Amiga systems aren't designed to be completely controlled by a Zorro II bus master. Zorro II doesn't get the full Paula managed interrupt lines, nor does it have a way to become a "primary" master, which is what the A2000's Coprocessor interface provides. Essentially you have two kinds of bus masters. A secondary master, like a DMA driven hard disk controller, requests the bus, gets a grant, runs as many bus cycles as it needs, and then gets off. It gets to keep the bus as long as it likes, but as long as it's on the bus, no other bus master gets to use the bus. A primary master, like the 68000 or an A2000 coprocessor device, receives all requests from secondary masters. When the primary master receives a request, it grants the bus and gets off. When no secondary master is requesting the bus, the primary master becomes the "default" bus master, essentially taking all the bus time not used by secondary masters. -- Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: hazy BIX: hazy "That's me in the corner, that's me in the spotlight" -R.E.M.
dltaylor@cns.SanDiego.NCR.COM (Dan Taylor) (05/16/91)
In <72302@microsoft.UUCP> tom@microsoft.UUCP (Tom MCCONNELL) writes: >But I DO have a fully compliant Zorro II slot. >Somewhere I heard about someone hacking a 2630 onto their expansion >port, which is essentially a different form of the Zorro slot, right? A ZorroII slot IS NOT THE SAME as the processor slot. In fact, the processor slot is more like a gender reversed version of the connector on the side of your A1000. To the best of my knowledge, what you need to do is create a small box that fits BETWEEN your A1000 and the Expansion Technologies. This is so any DMA devices there will still work, after the A2630 takes over from the A1000. You will need a female 86-pin connector (0.125" centers) for the A1000, another for the A2630, and a MALE connector to extend to the ET box. There is enough power in the A1000 to run EITHER an A2630, OR the ET box, but NOT both. Some ET boxes had supplies, some didn't; I think it was an add-on. Make sure the ET box is set up for it's own power supply. The A1000 pinouts are almost exactly 1 to 1 with the A2630 (I don't have the schematics here, but I'm going to build my own box like this), and I can't think of any conflicts, either power or signal. If you really plan to do this, get the docs from C=. DISCLAIMER: This is not something to try, unless you are sure of all of the consequences. I (and NCR, since they know nothing of this) will not be responsible for any damage ensuing from attempting this project. Good Luck! Dan Taylor /* My opinions, not NCR's. */
vollath@Informatik.TU-Muenchen.DE (Ulrich Vollath) (05/16/91)
In article <21607@cbmvax.commodore.com>, daveh@cbmvax.commodore.com (Dave Haynie) writes: |> Ok, here's the deal. The 86 pin edge on the A500 or A2000 is actually more ^^^^^ You probably mean A1000? So there is a way to stuff these neat Processor cards in my good old A1000? :-) If someone has already successfully done this, I would really like to get some Information about this. -- Ulrich Vollath email: vollath@lan.informatik.tu-muenchen.de Munich Institute of Technology voice: 004989/2105-8102 (result: -8099.62992874)
daveh@cbmvax.commodore.com (Dave Haynie) (05/16/91)
In article <1991May16.063945.7032@Informatik.TU-Muenchen.DE> vollath@Informatik.TU-Muenchen.DE (Ulrich Vollath) writes: >In article <21607@cbmvax.commodore.com>, daveh@cbmvax.commodore.com (Dave Haynie) writes: >|> Ok, here's the deal. The 86 pin edge on the A500 or A2000 is actually more >You probably mean A1000? So there is a way to stuff these neat Processor >cards in my good old A1000? :-) Yeah, that's the one. If you don't have any real DMA devices hooked up, you can at least in theory hook an A26x0 or probably other accelerator card to an A1000 or A500. While these machines don't have the coprocessor interface, that isn't an issue unless there's other DMA around. So the CBR/CBG lines can go directly to the BR/BG lines of the 86 pin edge (they naturally correspond, anyway). There is no /BOSS line on the 86 pin edge, but there is /BGACK, which again, in the absense of any other 68000 style DMA, can be used. This requires that pin 62 of the 86 pin edge be run to pin 20 of the accelerator board. And it's a good idea to connect pin 62 of the accelerator board to +5V, now that I think of it. The one final missing signal is the 7MHz clock. This could be tapped off the internal 68000, or created from the CDAC clock. CDAC is a 7MHz clock that trails the 7M clock by 90 degrees, or 35ns. If you can delay it by another 270 degrees, or 105ns, you get 7M back. The best way to achieve this is by running CDAC though a fast inverter, like a 74F04, and then a 30ns delay line. You might be able to get by with a 100ns delay line, instead, though I don't know for sure (that's right on the edge of CDAC to 7M tolerance that the A26x0 requires). As for how to put this all together mechanically so it works good, I really don't know, that's something you'll have to work out on your own. The key idea from the electronic point of view is to keep wires as short as possible. >Ulrich Vollath email: vollath@lan.informatik.tu-muenchen.de >Munich Institute of Technology voice: 004989/2105-8102 (result: -8099.62992874) -- Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: hazy BIX: hazy "That's me in the corner, that's me in the spotlight" -R.E.M.
monty@sagpd1 (05/21/91)
In article <21643@cbmvax.commodore.com> daveh@cbmvax.commodore.com (Dave Haynie) writes: >A1000 or A500. While these machines don't have the coprocessor interface, that >isn't an issue unless there's other DMA around. So the CBR/CBG lines can go >directly to the BR/BG lines of the 86 pin edge (they naturally correspond, >anyway). There is no /BOSS line on the 86 pin edge, but there is /BGACK, which >again, in the absense of any other 68000 style DMA, can be used. This requires Dave, Does this mean a DMA device not on the accelertor, or no DMA at all? I guess what I am asking is how the new GVP all in one '030/SCSI/RAM board would adapt to this scheme. Is it possible with out major hacking? So far all I've seen in this hack is some connectors and a minor bit of wiring. This type of board would eliminate needing the expansion chassis in most cases. It might be an interesting exercise to hack the GVP board to an A500/A1000. I understand that the official C= policy may forbid any further comment on this type of hack but I hope they haven't tied your hands that much. Thanks for any additional comments, Monty Saine
monty@sagpd1 (05/22/91)
In article <1991May21.161508.7078@sagpd1> monty@sagpd1.UUCP (Monty Saine) writes: >In article <21643@cbmvax.commodore.com> daveh@cbmvax.commodore.com (Dave Haynie) writes: >>A1000 or A500. While these machines don't have the coprocessor interface, that >>isn't an issue unless there's other DMA around. So the CBR/CBG lines can go >>directly to the BR/BG lines of the 86 pin edge (they naturally correspond, >>anyway). There is no /BOSS line on the 86 pin edge, but there is /BGACK, which >>again, in the absense of any other 68000 style DMA, can be used. This requires Dave, Does this mean a DMA device not on the accelertor, or no DMA at all? I guess what I am asking is how the new GVP all in one '030/SCSI/RAM board would adapt to this scheme. Is it possible with out major hacking? So far all I've seen in this hack is some connectors and a minor bit of wiring. This type of board would eliminate needing the expansion chassis in most cases. It might be an interesting exercise to hack the GVP board to an A500/A1000. I understand that the official C= policy may forbid any further comment on this type of hack but I hope they haven't tied your hands that much. Thanks for any additional comments, Monty Saine To the net, I appoligize for following up to my own article, but our news reader would not let me cancel my previous article. Now to the additional point to the above. Dave, is the missing clock you were talking about the 28 Mhz on pin 9 of the 86 pin?? I could find no reference to an 7 Mhz clock. The only pins that are referred to as being different between the A500/A1000/B2000 are 9, 11, 20, 60, and 64. Again sorry for the followup to my own post .. blame hp-ux and rn. Monty Saine
daveh@cbmvax.commodore.com (Dave Haynie) (05/23/91)
In article <1991May21.161508.7078@sagpd1> monty@sagpd1.UUCP (Monty Saine) writes: >In article <21643@cbmvax.commodore.com> daveh@cbmvax.commodore.com (Dave Haynie) writes: >>A1000 or A500. While these machines don't have the coprocessor interface, that >>isn't an issue unless there's other DMA around. So the CBR/CBG lines can go >>directly to the BR/BG lines of the 86 pin edge (they naturally correspond, >>anyway). There is no /BOSS line on the 86 pin edge, but there is /BGACK, which >>again, in the absense of any other 68000 style DMA, can be used. This requires > Dave, > Does this mean a DMA device not on the accelertor, or no DMA at all? I guess > what I am asking is how the new GVP all in one '030/SCSI/RAM board would > adapt to this scheme. Is it possible with out major hacking? I don't know that much about how that did the GVP board, but I think that it pretty much has to resolve any bus mastering between 68030 and DMA controller on-board. That would imply that it should work just dandy hooked up in this way. That is, of course, only a guess, but I can't really see how they would have done it differently, unless they're madmen or something. > So far all I've seen in this hack is some connectors and a minor bit of > wiring. Careful building of connectors and wiring, I might add. We're talking about running most of the 68000 bus somewhere, and that bus doesn't want to go very far, otherwise it picks up noise, capacitance, inductance, and all kinds of dirty, mean, nasty stuff that makes it not work any more. > This type of board would eliminate needing the expansion chassis in most > cases. It might be an interesting exercise to hack the GVP board to an > A500/A1000. Yeah, it might just work. You know, technology does march on. What used to take a whole box could very well fit on a single card. In fact, that's kind of the idea of the A3000 in the first place. We reached the point at which it made sense to integrate the whole 32-bit/68030 system, rather than provide it as an add-on. So the A3000 can handle more on the motherboard than most people hooked up to their A2000 or A500. The new GVP card probably isn't much behind in levels of integration, I guess they do even offer memory over the basic 8Meg expansion of the A2000/Zorro II setup. Since you could build such a connector without harming your A500 or GVP, it would be an interesting experiment. I imagine you'd need a power supply improvement for an A500, but probably not for an A1000 (though that guy at GVP still holds the record for power consumption on 2Meg Zorro II RAM cards, and probably always will...). > Monty Saine -- Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: hazy BIX: hazy "That's me in the corner, that's me in the spotlight" -R.E.M.