IIfx SIMMs & ROMs . . . hackable or . . .

Shades of the EVIL RAS Line Hack!

http://atari.nvg.org/stacyram/

Hardwarebook has all the pinouts except 64 pin . . . figures =8-/. . .. here we go again!

. . . but think spaghetti-free!

jt

. . . I'm still lookin' for some help here folks!

According to one of the TechNotes or Q&As, the IIfx memory mapping scheme addressed 128MB w/o and 1GB w/Virtual Memory enabled. I still don't know if that 128MB lower limit is due to Apple's maximum "supported" config designation and based only upon existing/tested/supportable IC storage densities of the day. As I said before, there could be ROM/Mapping, SIMM/MLB address line or cache implementation restrictions.

The shots I've added of the RAM & ROM SIMMs in my favorite IIfx hack look VERY promising.

http://homepage.mac.com/jwt1863/TrashedHacks/PhotoAlbum5.html

The 4 MB SIMM PCBs are marked "1990 NTX-4/16M," but they're populated with -80 DRAMs (IIfx compatable) and have no electronic components other than the DRAM chips. This leads me to think that the LaserWriter IINTx a/o its SIMM specifications, probably pre-dated the IIfx and that some 4MB SIMM PCBs from the IINTx may be upgradable to 4 or even 16MB IIfx compatability . . .

jt

I've got a set of 64=pin SIMMs around here somewhere. Maybe I'll have to scrounge them up and check densities and have a looksie at what-all is on them.

Pics or scans of the PCB layout/components of your 16MB SIMMs would be MUCH appreciated. Actually, images of 4MB, 8MB, 16MB and parity versions of ANY size 64-pin SIMMs would be a big help.

I've added an unretouched/unreversed version of the SIMM piccies for those who don't read mirrored .txt or have any overriding interest in lining up connections from side to side on SIMM PCB piccies.

jt

p.s. If anyone has, or can take pics or do scans of IIsi and IIci ROM SIMMs that'd be good to have on hand too. AFAIK so far, the layout for the Gamba SIMM project seems to be spot on, but I'd like to go to the photographic record too. Might as well lay out an updated/rewritable memory component ROM SIMM if I'm gonna play with 64-pin PCBs . . .

Heh!

Well, I've unearthed another bank full of NTX-4/16M SIMMs from L.A.Components, so IC removal and PCB layout is my next step. Anybody got any data, shots or scans of 16MB IIfx SIMMs or ICs to share yet?

jt

. . . well . . . there are two unused lines per IC on this PCB layout! I've gotta research the IC pinouts, but the outlook for developing a 4/16/32/64 MB IIfx SIMM PCB from this layout is lookin' better by the minute!

Direct link to first set of piccies:

http://homepage.mac.com/jwt1863/TrashedHacks/PhotoAlbum5.html

See this album for the updated component side pics:

http://homepage.mac.com/jwt1863/TrashedHacks/PhotoAlbum21.html

jt

Try the wire pull chip removal method. Its really good as the leads remain tinned and there is no chance of lifting a pad. It takes a long time though.

IM me if your interested...

Cheers

T

Won't work? Seems like it'd be scads easier than the "hacksaw method"!

Do you have the IIfx ram pinout? Just open your copy of the "Guide to the Macintosh Family Hardware" (Second Edition) to page 214. You do have that on your bookshelf, right!?!

I can't find the damn "GttMFH" online dammit!! Grrrrrr!

An excellent resource, and I thought for sure it was available for DL from Apple. But I've just spent the last hour searching without success.

If anyone has a url for DLing the "Inside Macintosh" series, especially the above I'd be most greatful for the direction. I have a alot of the InMac pages and resources, but no GttMFH in .pdf form.

jt, if you need it, just ask and I'll scan the IIfx ram pages, plus any other pages you need (within reason of course, 500+ page book!)

Dan K

. . . of course, nopro! Look at my initial post again, but a .pdf'd be real nice for massaging the pinout info into my spreadsheet with the 30 & 72 pin signal info from hardwarebook!

. . . got Designing Cards & Drivers sittin' right next to it too!

However, it's the available IC pinouts for that package (and 24 pin IC's too) that I'm most interested in, ATM. I figure I can always do a cut & jumper job on some of the mobo lines! Worst case scenario . . . some of those unused parity lines on the MLB/SIMM bus could be hac . . .

jt

edit: I just wanted to get the ICs off cleanly . . . like NOW! HEH! They're a little close together for the wire trick and I won't use a heat gun on a PCB I intend to re-use without getting some ChipQuik to use with it. The few pads and traces I lifted experimenting are reparable and the Razor Saw technique lends itself to a rotary translation for production time on the other seven SIMMs. =8-0

back again: there are a lot of individual chapters of GttMFH online (or there WERE!) and i think I snagged whatever I could. Is the Apple site mirrored on the wayback machine?

Doh!! My reading comprehension level is very low.

After reading GttMFH, I think the 128MB limit is based on the available RAM address space, no? Also, isn't the 64 pin simm just sort of an extended 30 pin simm? That is, it's on a 32 bit bus and needs the same sort of bank-of-four arrangment that 30 pin ram uses. Are there are any 30 pin simms larger than 16MBs?

Again, my reading comprehension level ain't too great . . .

Dan K

I agree, from my understanding the IIfx is sort-of an extended 30-pin. I don't remember where I read the following from, otherwise I'd provide a link, but what I do remember is the extra pins were for hardware address latching and not actual data bus paths. With that the case, 128MB's prolly the max for the IIfx'll do, just like the IIx (which is another story, damn PAL SIMMS above 1MB sticks).

I've never seen any 30-pinners larger than 16MB. I've seen auctions for them on eBay from time to time and they still hold their $ value. I've seen some sites selling 16MB IIfx 64-pinners but were going for $25 a pop, and ya need 4x just for a bank. GAK

Hope there wasn't any bubble bursting there. By all means, if you get your IIfx to go beyond 128MB that would be the win!

-Dk

dankephoto wrote:

Doh!! My reading comprehension level is very low.

You and me both!

HRMMM . . . THAT is wht I've been trying to determine, I know there are 64 pin SIMMs available in a 64 MB capacity for another platform, it's just a question of how many address lines are implemented on the Memory Controller, the MLB and the SIMM interface, AFAIK. The memory map reservation for RAM is 1 GB, it's just a question as to the reason for the specified 128 MB limit of the four banks of SIMMs.

I have a nagging suspicion that it has more to do with what Apple actually tested and was willing to support than with REAL limitations. Like the 630 series and the 605/475, the IIfx and its memory type was obsolescent/superceded by a much faster hardware generation before memory densities overtook its "supported" memory limits. Maybe it can handle larger SIMMs than existed the last time its documentation was updated, just like the "Consumer Quadras!"

Dunno, but with more than twice as many available signals on the SIMM interface, they'd have to have been retar . . . erm . . . never mind . . .

jt

. . . like I said, it could be a bust, but I'm, laying out 16 MB SIMMs and an updated ROM SIMM at the same time.

128 MB of hacked SIMMs for a few of my IIfx playthings and a handful-bushel of Programmable ROM SIMMs for trade goods should come in VERY handy! Are you saying there have been 30 pin SIMMs available on eBay that WERE larger than 16 MB?

jt

. . .

. . . wee need an emoticon for cross-eyed drooling idiot.

I just finished inputting the pinout info for the 64 pin SIMMs into my spreadsheet for comparison with the 30 and 72 pin SIMM pinouts . . . and I have less of a clue as to what's going on than EVER! =8-0

The idiots out in the orchard put THREE TIMES AS MANY UNIMPLE . . . nted pins on the 64 pin SIMM as the 72 pin SIMM! They wasted 12 freakin' lines!

It looks like all (30/64/72 pin) Mac SIMMS have the same 12 bit addressing capability. What's the largest 72 pin SIMM that was ever available?

The IIfx has a VERY strange setup with an 8 bit Data input bus AND an 8 bit Data output bus. It only has one RAS line and one CAS line like the 30 pin SIMMs, but it has separate Write enable input lines for each freakin' IC too!

* WT . . . . . . wanders off mumbling obsce . . . *

Try using a single strand of copper wire from a standard piece of cable. Its thinner than a human hair. Loop through each pin.

I need to remove a small IC. How does one go about cutting it off? Where should one cut?

Cheers

. . . the legs wrap under the IC. I've used the wire trick on TSOP and TQFP ICs . . . I didn't think using a wire would work, but I'll give it a shot next time.

http://home.arcor.de/kickstart/TKA/Tutorials/Memory/SIMM.html

From the looks of that page, this layout should only support a 4 MB SIMM config, but that doesn't seem right, given the designation. "4/16M" etched into the copper. Memory configurations are a mystery (hence: this learning exercize) to me, I must be missing something here . . .

jt

"I’ve never seen any 30-pinners larger than 16MB. I’ve seen auctions for them on eBay from time to time and they still hold their $ value...."

Actually I meant I've seen 16 meggers on ebay, not a possibly construded larger than 16MB on ebay.

It sounds like you guy are going all out with this hack. I'm really interested to see how it pans out.

Well your chop the pins off with a knife method seemed to work great on my palm lol

Barring any problems using the unimplemented parity lines on the mobo, looks like a 64->72 pin SIMM adapter hack is just what the doctor ordered . . .

Pin 59 maps straight to an empty pad on the missing parity IC, enable that and suddenly another 128 MB of memory space is accessable from the same 8 sockets for 8 bit 64 pin SIMMs.

Pin 58 is a little more complex, each bank's 1st SIMM maps to a single unimplememted parity IC pin. The same goes for the paired second third and fourth SIMMs of the two Banks, tie'em' all together, enable that and it acts the same as Pin 59, doubling the addressable space once again.

If I can synthesize one more multiplexed "bank select" line from parity lines or any other unimplemented/unutilized signal bus . . . and the fx is suddenly addressing all four banks on each of 8 128MB 72 pin SIMMs!

AFAIK . . . or understand memory mapping. All but 128 MB of it might need to be addressed as VM with a RAM disk hack . . . but it's gonna be contiguous full speed memory . . .

. . . if it works at all.

jt

Stop trying to devalue my IIfx SIMMS! LOL

Tom

LOL!

BTW: Just had another thought . . . I could probably use just those same two lines . . . and drive them high or low to synthesize all the bank select signals needed to address the IIfx's entire 1 Gigabyte mapped memory space 128 MB at a time!

Now I've gotta find that #%^*ˆ®ª CMOS Cookbook!

jt

I've added MLB detail pics of the "Parity Line Hack" with the paths and pads highlighted/color coded.

http://homepage.mac.com/jwt1863/TrashedHacks/PhotoAlbum21.html

I guess the next step will be to fire up the desoldering and soldering irons to install header pins for a prototyping board interconnect in the empty thruholes for the Parity IC.

Test procedures will be to ground the pins and then drive them high and low for each possible logic state. If the IIfx boots and runs without any apparent hardware problem under all those cases, I guess it'll be time to call in some help for the RAMDisk/VM software end of things.

Any suggestion for testing refinements or the software/logic angle would be welcome, gang.

jt

here u can buy some 4mb sticks....http://www.memoryx.net/maciifx.html

You can't just substitute 72 pin SIMMs for the 64 pin SIMMs in the IIfx, even if you hack the address busses, because the separate data-in and data-out busses on the 64 pin SIMMs are required. This allows the IIfx to buffer writes. See page 230 of the GTTMFH and trace the data-in pins. They go to the output of data buffer chips. The FMC (Fast Memory Controller) chip latches the write data and signals completion to the CPU while continuing the actual write operation. This frees the CPU in 2 cycles instead of 6.

Other uses of the data bus don't affect the write operation because the data path from the FMC to the RAM Data-In bus is it's own little cul-de-sac electrically. A following write will be delayed.

The block diagram does show 27 address lines (128 Million addresses) going to the FMC. However, I doubt that the FMC has more than 12 address lines going out to each bank of RAM. If you wanted to insert more memory, you'd probably have to replace the entire FMC.

Other 1.5 year late comments:

30 pin SIMMs are never larger than 16MB because there are only 12 address lines available and there are no pins available to implement additional address lines (unless one steals parity lines). RAM addresses are multiplexed in two parts, so 12 address wires translates to 24 actual address bit. 24 bits yields 16 Million addresses and a 30 pin SIMM has 1 byte (8 bits) per address, so there are a maximum of 16 MB on a 30 pin SIMM.

The specs I've been able to dig out so far only show 12 address lines on 72 pin SIMMs as well. This would lead to 64 MB as the maximum (24 address bits => 16 Million address X 4 bytes per address = 64 MB).

However, the 72 pin SIMMs have four RAS and four CAS lines and I suspect that these are independly controlled in such a way that 72 pin SIMMs can be built with multiple banks, all using the same address lines.

I have 128 MB 72 pin SIMMs on hand, so I know they can be that big, and I've heard of 256 MB SIMMs in the 6100.

Anyway, you won't get any such scheme working, regardless of addressing tricks, unless you provide a separate data-in and data-out bus to the RAM so that the FMC can do the write-buffering trick. Otherwise, any attempt to write data to RAM will be stomped on by other data bus activity.

Wow, nice post. Even if it's a tad late, it's better late than never. Do you have links for some of the info you've posted, or did you figure it out by studying pinouts? I'd be a bit interested in reading more about larger 72-pin SIMMs. I've got a few systems that seem limited by 1 or 2 72-pin slots. I'm sure the MC would have to be capable to read the multiple banks and densities anyway.

I'm sorry... but could somebody please point out to me what 'GTTMFH' means?

Thanks!

First post, in italics: Guide to the Macintosh® Family Hardware

You guys might want to drop by the 68kmla and search for "iifx simms" - someone just scored some rare high capacity ones, and there's been talk about reverse engineering them

... and the VMM's (Vast Memory Modules) from my 68000 Dash 30fx. I snapped some photos and put up a slideshow on my SpyMac hosted site. Have a look here.

If you'd like a better image of one of the VMM's, let me know & I can use a scanner on it next time. I had some nasty glare going on when I photographed them.

And in case you're wondering, no, I will not under any circumstances hack any of them up (I have a set of four), nor will I part with them for any meager sum!

The chips seem to be 1x4M - totalling to 16MB.
Which would make the total size available (using eight of these babies) being 128MB (that was the ceiling on the IIfx anyway wasn't it?)... so the logic on the PCB would be purely for switching between the four 4MB banks?

Nice to see something like that - I've been hunting around for larger memory modules for my fx :-)

That's quite a machine!

I finally got around to scanning the manuals that came with my Dash 30fx. The RAM installation guide is here (about 800k) and the user guide is here (about 3MB).

Maybe there's some detail in the RAM guide that might be helpful in this discussion.

I just read a bit in the memory manual. Those power reqs are steep! Making a module with more modern and lower power chips would help thelife of a machine tremendously. I still know next to nothing about designing SIMMs, but I'm pretty certain modern memory chips could be adapted into a design like the VMM.

so has there been any leads on this hack?

Someone on the MLA has recently scored some -socketed- IIfx SIMMs. I've suggested he might like to make some really nice high rez scans of them

those were 30 pin simms for his iici

Yeah, you're right. My bad

I saw some 32 MB, 64-pin SIMMs for auction on eBay. Do you think these would work in the IIfx (despite the official docs which list 16-meg as the highest)?

I was looking at those also, probably not. There are a few 64 pin memory modules out there (AST computer, GVP products of Amiga fame, etc) and they are not the same.

I was going through my RAM box, and came across six (so far) 64-pin simms. They're labelled 32-Megs each (somewhat confirmed by each chip having 24 pins).

I have no idea where I got these from. I know they're not from a IIfx.

Has anyone made progress on trying larger capacity chips in a IIfx?

64-pin simm labelling:
p/n: 10021174 0A
PCBA, SIMM, 32MB DRAM WITH WHITE STRIPE
s/n C135772

Does anyone have scans of known working 16mb simms for the IIfx?

When I've gone a bit farther recovering my cost and turning a bit of profit (to support the next exciting project, don't cha' know) I'll probably post the circuit board designs as well. However, if you have a copy of GTTMFH and the datasheet for some 16M X 1 bit memory chips, the design is almost trivially easy, though tedious to lay out.