Micron Xceed docs on ebay

[kreats]

Anyone see this auction on ebay?

Other than the patent docs - I haven't seen this information out there before.. Hope the person who bought them wants to share.

 

[Juror22]

I have to fix my searches - I totally missed this!  And the seller was the engineer for these cards!

I was the engineer who developed the HR upgrade and this is from my personal collection.

That is so incredibly cool!

I agree that it would be awesome if they were to share this knowledge.

 

[techknight]

Yea this needs to be exposed, Seriously. 

I wish I was the one who got that auction, I would be one of the few that would understand what to do with this stuff. Technically if the disks are good, you can create new clones of these cards. With legal issues I am sure, but for hobbyist use?? 

It appears one disk contains everything to the Mavrick Gate Array IC. Of course the design can be ported to a newer more common Xilinx or Altera chip. 

Heck besides the RAMDAC, the entire card could be fit inside an FPGA. 

 

[kreats]

I tried sending him a message on the ebay contact system but no response so far - he might have some backups. Maybe an offer of some cash might incentivize it?

Alternately, you could just ask him some questions about the card.

 

[ZaneKaminski]

Maybe we will see a post from the new owner of the materials, here or elsewhere.

Regardless, a community effort to develop a similar product would be neat. I'm proficient with PCB design and design for electromagnetic compatibility, though I don't know how to design the grayscale video circuit itself.

Relatedly, I was going to develop a faster 68030 accelerator board with a fast DDR2 cache of the entire main memory for the SE/30. Looking at the DiiMO card, an accelerator like that should actually be pretty easy, especially with a modern FPGA to implement the cache and glue logic. Hell, an entire 68030 running faster than was ever available could be done in an FPGA, but I couldn't find any free '030 soft cores, only 68000 cores. An SE accelerator using a 68000 soft core is also possible and definitely would be cheaper and smaller than with a hard 68030 and FPU. It could also be a lot faster than using a hard 68030 or even 68040. Even cooler would be the possibility of reprogramming the FPGA and adding another (digital) peripheral to the card, like some kind of a coprocessor mapped somewhere in memory. Aaaaaaaand it would be possible to max out the system RAM with such a card, even if the target Macintosh only has the minimum amount installed. (at least on the 030 systems, except for the IIfx)

So maybe a combo card could be created. It would be a shame to have grayscale but still only a 16 MHz '030! A 6- or 8- layer PCB with components mounted on each side would make it physically small, other than where large legacy stuff would have to be mounted, e.g. 68030 and 68882. Could be done in only 4 layers but it might end up to be too large to fit into a compact Mac, especially if we want another PDS on top to mount, for example, the ethernet card.

tl;dr: designing a reimplementation of the Xceed is beyond my skill set, but I can at least do the PCB for it. I'm also good with driver software but I don't know if I want to take that on. Nominally, I'm a programer, but software development is very time-consuming. Maybe the Xceed driver can be reused. A 68030 accelerator seems easy, though I would like to see at least a schematic for such a card. The cache is easy and the rest of the glue logic should be straightforward from the timing diagrams given in Motorola's '030 docs.

 

[Paralel]

I really hope the person that managed to get this does indeed share all of it. To keep it to oneself would be a tragedy.

Its too bad the guy didn't post here or another place for the 68K community and give us collectively a first whack at getting our hands on it. Its not like he did all that well on ebay.

 

[techknight]

Relatedly, I was going to develop a faster 68030 accelerator board

I was researching heavily in accelerator cards a few years ago as I wanted to make one for the macintosh portable, but the logic was slightly complicated for me. the 030 expects different types of signaling for the bus states than the 68000. Anyways... That and I have 0 knowledge in FPGAs. 

the PAK68K and a few other designs are published in the amiga world, but im sure its all the same. 

 

[Paralel]

If someone can get their hands on this stuff and can actually build new cards/grayscale adapters, I'd seriously consider getting an SE/30, since it would once again be reasonable to have it run grayscale again.

 

[ZaneKaminski]

I've got a product to finish, test, and ship, and then I'll set my sights on the accelerator board. If more details about the Xceed come up, I'll try my best to add compatible functionality to my accelerator. It should be easier with a big FPGA on the board that can handle some of both elements of the functionality. I'd eat the cost to make a few, since my aim with the accelerator board is to demonstrate my ability to implement an FPGA and DDR in a design, but the units will be somewhat costly.

Of course, my hardware design will be totally open-source. I'll make a template for an SE/30 card too. Maybe that can come soon.

 

[mrpippy]

Months ago on eBay, someone was selling schematics/documentation for a Micron IIci cache card, must have been the same guy. He also sold schematics for a NuBus image compression card at the same time as the SE/30 card.

If anyone gets a response, ask if he's got any more documents/schematics he wants to sell. I think we could throw together more than just $40

 

[Byrd]

Whomever does get the schematics (hope you're reading this, whoever you are!), it would make for a good cottage industry as it's the upgrade many of us have lusted for for years should the card and yoke harness be reproduced.  Many would be willing to pay good money for a readily available, drop-in Micron Xceed clone.

I don't know what it is that makes the SE/30 so special, nor is greyscale anything to write home about (and I assume an unaccelerated SE/30 would render 256 colours quite slowly), but it's just so cool.

JB

 

[kreats]

I think drivers and software compatibility would be the biggest hurdle for a new accelerator.  Some sort of "multi-core" switchable board would be ideal though. e.g you could put something in the CPU slot (not the PDS) of a se/30 and be able to choose whether to have a 030, 040 or supercharged variant of the two.

Adapting the vampire boards that are on the amiga would be the way to go, but they would probably say to just use macos on an amiga running fusion. They also have their own hdmi out btw.

Byrd (hey mate!): the yoke harness can easily be cloned - it's a pretty simple board. Getting all the custom wiring done is mainly the problem.  Same with gamba's twinspark.. there is only one chip on that board.

I have an xceed (+greyscale) and while it's an awesome card - it has some caveats. You get many of the same issues you get with the colour classic in terms of software compatibility (most classic stuff wants to run in 1 bit colour or wants a bigger screen). The multi card upgrades really need a strong power supply to run glitch free also.

 

[Byrd]

Hi kreats, nice to see you around, it's been a long time!  Last time we met was some crappy LAN party at La Trobe university, I think ...

I wasn't aware of the yoke harness being cloned - so that's easily made, it's the graphics card.  You'd also assume if one had an Xceed clone we'd also need a cheaper dual slot PDS card in tow ...

JB

 

[kreats]

Heh yeah - that LAN is still going I think! In Sydney now so should update that.

Last time I checked that guy wanted $200 USD for his clones - which is off the planet but I suppose there has to be profit to bother.  Sharing a design on OSHpark is the way to go I reckon.

A guide for refurbishing or replacing compact mac power supplies is the thing that is most desperately needed I reckon.

 

[ZaneKaminski]

Cloning the board would be easy. Desolder all components, stick it on a scanner, then trace over the scanned image in your PCB layout program of choice. If someone can get me such an image, I'll do it and put it up on OSH Park.

 

[kreats]

Well desoldering the components might be something I don't want to risk - but high res photos could be possible. Scan of the bottom of the board maybe too.

A straight up clone of this might be something that's achievable too?

http://jamesplee.com/images/tech/2012/SE30_2/xceed30_10.jpg

 

[ZaneKaminski]

Soldering is certainly not one of my skills... I have a friend who does my soldering, and I do his programming, haha. So I wouldn't want to suggest that you desolder the components. I will say, however, that neither the board nor the components strike me as particularly costly. The cost (excluding assembly) couldn't be more than $25. Maybe even $15. This analog stuff is not my forte. and I won't be able to understand how it really works, so it would be best to trace the board as exactly as possible in terms of the specific layout. Maybe just scanning the bottom will be sufficient.

The Daystar accelerator, on the other hand, is impossible to clone by visual examination. We would need to be able to program the CPLDs on the board--the 9 square chips with pins on all four sides. Those implement the glue logic. The rest of the chips (68030 and 68882 aside) are signal buffers and cache memory. Daystar's design from years ago doesn't really get us anywhere today. The hardest part isn't doing the schematic or laying out the circuit board, it's designing the control logic in the CPLDs. We would want to redesign the board and use modern components.

 

[kreats]

Well the powercache apparently works in the se/30 using the iicx adapter (which is a dumb adapter with no additional circuitry). I haven't inspected it but I think the same chips are present on both cards - so it would be a matter of harvesting one card to populate the other.

 

[danpoarch]

How would you implement a modern RAMDAC in this scenario? Would you look for some sort of IC that provides old-world VGA-out support for modern video interfaces and feed it the frame data from the FPGA? I've always thought sourcing the RAMDAC might be the hard part but after looking into VGA<->HDMI convertors on another quest to build a modern 15-pin video adapter, I'm thinking it might be easier than I presumed...

Edit: My start on this track was that you would at least want VGA out (thinking outside of the err... box), and at that point you might as well hop to HDMI/DVI and these kinds of multi-format ICs are present on the market today and somewhat easy to implement (at least, documented).