SuperDuoHack Brainstorming Session =8-D

[Trash80toHP_Mini]

Keep this in mind when considering this project's unrealistic (at best) goal, & all these seemingly unrelated lines of research:

The first G-3 IC Packages were designed to be PIN-FOR-PIN compatible/replacements for the PROC in the 2300c.

[Osgeld]

what are you talking about?

[LCGuy]

He's talking about potential hardware hacks to increase the performance of a Duo 2300. (such as upgrading the rather pedestrian 100 Mhz 603e to an early G3)

It'd make sense, really - if I recall the G3 was based off the 603e, not to mention that I'm fairly sure I remember reading that Apple was planning on releasing a Gossamer based system with a 603e, keeping the chips pin compatible would allow them to use the same daughtercard design for both the 603e and the G3, saving them money at the time they needed it the most.

[MacJunky]

You mean the PPC 740? I am not sure what exact package the CPU in the duo is but there was some talk a while back about putting a 740 into a PM6400/6500. Unfortunately it never went anywhere. I think one of the issues was availability of the 740 but I cannot remember exactly.]

I just want to say that it would be really cool to see someone actually get the CPU and do it.

[Byrd]

While you're at it ...

Could you investigate the possibility of increasing the multiplier of the 2300c, ie. from 100Mhz to 117 or 133? And even better ... a total keyboard replacement for the slush-boxes they call "keyboards" on Duo portables!

Good luck.

JB

[Trash80toHP_Mini]

Could you investigate the possibility of increasing the multiplier of the 2300c, ie. from 100Mhz to 117 or 133?

That IS part of the project, but to increase it to a much faster G-3 multiplier. The DuoMauler I mentioned in the Unimplemented Pads Thread as as having successfully transplanted the Video IC, could never find the "bus clock multiplier resistors," which I suspect were intentionally hidden at far spread corners of the MoBo to avoid just what you are suggesting.

If I were to diddle around with a small change like that, i'd probably just desolder the requisite legs of the CPU and hardwire the correct resistors to the freed CPU legs. I'm NOT above doing something as inelegant as that or the DeclROM PowerPinHack, because it was "down-n-dirty," worked just fine for testing and it seemed like too much trouble to find documentation on the "chip select line" which WOULD have been a much more elegant solution . . . and will be discussed in an upcoming thread.

... a total keyboard replacement for the slush-boxes they call "keyboards" on Duo portables!

You've just gotta find the last revision KBD . . . . . . and then learn to live with it.

[Trash80toHP_Mini]

You mean the PPC 740? I am not sure what exact package the CPU in the duo is but there was some talk a while back about putting a 740 into a PM6400/6500. Unfortunately it never went anywhere. I think one of the issues was availability of the 740 but I cannot remember exactly.

If you're talking about the early G-3 that was specifically designed to be "pad" compatible with the 603e in the 2300c, there IS a source for the necessary CPU's, but it's a crapshoot in every case as to whether the "donor" has the 603e Package CPU in the Duo* or the later QFP (?) version of the G-3 CPU.

I'm going to have to research the identity of DuoMauler so I can get back in touch with him to give him proper credit . . .

. . . and to see if he's made any more progress!

. . . and post links to the DementedDuoHackin'Topics over on 'fritter.
. . . and to my MLA parallel threads remaining in our own archives.

I just want to say that it would be really cool to see someone actually get the CPU and do it.

It'll have to be a TEAM effort, I probably can't manage to do all the research on my own, so I'll be posting threads to define all the necessary lines of research and ask for volunteers to be part of this particular 68kMLA "Mission" and I'll be posting some very Interesting sideline research topics, like the DuoDock II: UNIMPLEMENTED PADS PROJECT!*

I'll edit in proper links to the various threads and I've finally got "right-sized" piccies loaded on HP_Mini! I'll be heading to the Library to do some serious work on whipping these threads into shape and posting the requisite pics!

[Trash80toHP_Mini]

Now that I've got my piccies downsized from 10 MegaPixels to , , ,
. . . something less likely to freeze up a comrade's computer!

THIS is what I'd call an inelegant approach to hacking, but as I said, it was down-n-dirty AND it worked for testing purposes!



The same kind of thing could be done to the tiny legs on the 603e, except each of the bus multiplier pins would be be wired to the proper resistor circuitry to achieve your wildest overclocking fantasies.

The 603e is well documented, IIRC, the resistance tables are available for the clock multiplier settings too . . .
. . . but the ole' noggin' not QUITE fully online as yet.

[Trash80toHP_Mini]

Erm . . . there's been just a slight change of gears: SuperDuo is no longer an upgrade project of existing Apple hardware!

It's now an Open Source Project aimed at developing an entirely new PowerBook Duo form factor MoBo connected to a much more usable, much higher definition LCD!

I think that I may have found my I/O subsystem:



Technical specs at: http://www.embeddedarm.com/products/board-detail.php?product=TS-7500

My lil' PowerDuo baby's plastics are gonna' house a NetBook's NIGHTMARE!

Unless a more modern LCD - and WiFi antennage will fit into the thinner Grayscale Duo's Lid!

Oopsie! I forgot to mention the modern battery technology . . . . . . my bad!

[Bunsen]

increasing the multiplier

The DuoMauler I mentioned / could never find the "bus clock multiplier resistors,"

Would that be Tom's House of Hardware Hacking Horrors?

i'd probably just desolder the requisite legs of the CPU and hardwire the correct resistors to the freed CPU legs

DIP switch, if there's room - makes experimenting with different settings a heck of a lot easier.

total keyboard replacement

I'm with Byrd on this one. My vague thoughts along those lines include finding a decent mini keyboard at the swap meets, remapping the matrix (cut and join, cut and join), and connecting it to the Duo internal ADB ribbons.

it's a crapshoot in every case as to whether the "donor" has the 603e Package CPU in the Duo* or the later QFP (?) version of the G-3

Sparkfun Electronics and Australian ebay seller stuffdownunder both seem to have supplies of QFP-to-x adapter boards/sockets.

an entirely new PowerBook Duo form factor MoBo connected to a much more usable, much higher definition LCD!

That's a big shift in focus, jt, and a huge leap in (in)feasibility - from a theoretically possible CPU transplant to designing and making an entire new SFF logic board from scratch. I hope you won't be abandoning your original line of enquiry completely.

Why a new logic, rather than just stuffing a Mini- or nano-ITX board in there? What defines "Duo form factor" to you? Actual Duo docking port compatibility (IMHO madness)? Or a work-alike but all-new docking system based on newer components, like say Mini-PCI or Cardbus/ExpressCard to PCI/PCIx/PCIe? Or no form of docking, but a self-contained system inside a Duo shell?

I think that I may have found my I/O subsystem:/ http://www.embeddedarm.com/products/board-detail.php?product=TS-7500

Nice board - but why just for an "I/O subsystem"? Why not pop an extra $30 for a complete ARM board with better specs? I guess I'm not following your new direction 100% just yet.

a more modern LCD - and WiFi antennage / modern battery technology

MiniITX (or similar) board with LVDS out. Compatible LVDS LCD. 12V worth of AA Lithium rechargables; lithium charger board/s. Bake on high for 30 minutes and cool before serving.

Unless I'm missing something here ...

[Trash80toHP_Mini]

increasing the multiplier

The DuoMauler I mentioned / could never find the "bus clock multiplier resistors,"

Would that be Tom's House of Hardware Hacking Horrors?

Probably, the name rings a few bells.

total keyboard replacement

I'm with Byrd on this one. My vague thoughts along those lines include finding a decent mini keyboard at the swap meets, remapping the matrix (cut and join, cut and join), and connecting it to the Duo internal ADB ribbons.

That's one reason I'm seriously considering the PB100 as my baseline unit, I've already hacked a 1400 KBD onto BabyPB . . . and she's got the ultimate in spherically correct pointing devices when it comes to size/usability. BTW, there are MUCH more efficient means of prototyping KBDs, one of which is to fab a fiberglas copper-clad etched PCB replacement for the bottom membrane, thus stiffening said spongeboard!

. . . and why would the interface need to be ADB?

it's a crapshoot in every case as to whether the "donor" has the 603e Package CPU in the Duo* or the later QFP (?) version of the G-3

Sparkfun Electronics and Australian ebay seller stuffdownunder both seem to have supplies of QFP-to-x adapter boards/sockets.

At what cost? If not astronomical, will the adapted CPU interface efficiently with the Duo's heatsink? . . . methinks NOT!

an entirely new PowerBook Duo form factor MoBo connected to a much more usable, much higher definition LCD!

That's a big shift in focus, jt, and a huge leap in (in)feasibility - from a theoretically possible CPU transplant to designing and making an entire new SFF logic board from scratch.

Not really, it makes the hack MUCH easier!

Why a new logic, rather than just stuffing a Mini- or nano-ITX board in there? What defines "Duo form factor" to you?

That'd be a "carrier" board in the Duo form factor that "bolts right up" to the Duo's magnesium frame and has the same (or improved replacements) I/O and battery connections (or an expansion bay and INTERNAL battery combo) and places a VERY CAPABLE CPU under the magnesium heat sink or near enough to it that a solid gold (or better heat transfer medium, if available) adapter could take up the slack.

I think that I may have found my I/O subsystem:/ http://www.embeddedarm.com/products/board-detail.php?product=TS-7500

Nice board - but why just for an "I/O subsystem"? Why not pop an extra $30 for a complete ARM board with better specs? I guess I'm not following your new direction 100% just yet.

Will said ARM outperform the Atom ChipSet in HP_Mini? All I'm looking for at this point is an open Source board that is capable of fulfilling all the functions of the PBX Bridge IC and ALL the slow bus subsystems of the "ubuntuDu . . . erm . . . SuperDuo!" The CPU Card/Memory, fast i/O bus would be another matter entirely.

a more modern LCD - and WiFi antennage / modern battery technology

MiniITX (or similar) board with LVDS out. Compatible LVDS LCD. 12V worth of AA Lithium rechargables; lithium charger board/s. Bake on high for 30 minutes and cool before serving.

I'm not yet familiar with those form factors or their heating/cooling/power/space/location requirements, but you're getting the general idea, it's just that all the parts need to be tied together by a Duo Form Factor carrier (think prototyping or even breadboarding level PCB) "MoBo" with as little circuitry on it as possible, which won't be much considering modern interfaces.

BTW: FPGA technology will (should) allow implementation of LVDS out among other things.

Unless I'm missing something here ...

I'm sure we're BOTH missing a LOT here at this point, hence the title "brainstorming session!"

As well as my request for help in narrowing down the choices and defining areas for further inquiry!

[Franklinstein]

AFAIK, the PowerPC 740/750 were available only in a BGA package, never in a QFP; I don't even think that the 603ev could be found in a QFP.

As far as a BGA-to-QFP adapter, it's possible that the whole shebang could be thin enough to fit with the original heatsink, as the 7X0 in a BGA is slightly thinner than a QFP, and the adapter shouldn't add too much height. I'd be even more willing to go with an adapter if it could use a 603ev with an adapter-mounted bus-level cache (the 740 might not work with a bus-level cache and cannot use a backside cache, and the 750 wouldn't be pin compatible, though with that adapter, it could potentially be made to be, and could also have a fast backside cache).

Personally, if I could figure out how to change the multiplier, I'd just put a 200MHz 603e QFP (found in PM6400/200s) in there and call it good. That way, most everything stays original, and you get a decent speed bump.

If you're looking for donor Macs, these have QFP 603s:
52/62XX (75-100MHz 603)
53/63XX (100-160MHz 603e)
54/64XX (180-200MHz 603e)

Some clones have them, as well, like the PowerComputing PowerBase 180, but those were often variable.
Most others have BGAs, like the PB1400 and later, 55/6500, 4400 (or whatever your local version may be), and the TAM.

[Trash80toHP_Mini]

AFAIK, the PowerPC 740/750 were available only in a BGA package, never in a QFP; I don't even think that the 603ev could be found in a QFP.

According to what I've read about the G-3, probably on Motorola's site, it was intentionally designed to be pad compatible with the QFP packaged 603e.

What DuoMauler has communicated to me, is that he has found a piece of donor hardware that contains either a QFP or a BGA G-3, depending on bog knows what! Date of manufacture would be my guess, but he says you can't tell what you've bid on until you recieve it and crack the case.

The resistance tables for the clock multiplier for every PPC proc is an open, published spec/pinout. Since these are NOT high speed signal lines, a more elegant version (that's not a very high bar, check the photo above) of my power pin hack should do the job very nicely. pulling the appropriate legs high or low to establish a dependable clock multiplier can likely be done w/o generating the kind of RFI produced when "hotwiring" memory addressing lines on the fast I/O bus.

Got any more ideas? Simply switching to a faster 603e never even occurred to me!

[Franklinstein]

I'd be very interested in seeing examples of the device of your donor parts candidates, because PPC 740/750 spec/order sheets from both IBM and Freescale list the chip as being available only as a FCBGA of some type. Of course, there could exist some limited QFPs, but I doubt there would be many, as both the 603 and 604 were using BGAs long before the 740/750 was introduced. In fact, there are pictures somewhere of a prototype PB5300 (I think it was code-named Mustang or something) with a BGA 603.
But yes, the 740 is fully pad-compatible with the 603, so good luck finding a QFP version.

As for clocking the processors by directly modifying leads, that's not too bad of an idea, especially in lieu of a proper resistor config. Maybe I'll try it on a spare 5300 board when I'm home.

However, seeing as how the early PB1400s wouldn't use 166MHz processors (they'd always clock to a max of 133MHz), and with the issue being attributed to the ROMs, do you imagine that there could be a similar problem in the 2300 and 5300?

[Osgeld]

make a bga to qfp breakout board and pop it in the toaster oven

[Trash80toHP_Mini]

I'd be very interested in seeing examples of the device of your donor parts candidates <snip>

I'm pretty sure the posts & piccies re the donor are over in the 'fritter archives, I'll check my various HDDs from that era for the pics, as I'm POSITIVE that I'd have glommed 'em!

Of course, there could exist some limited QFPs, but I doubt there would be many, as both the 603 and 604 were using BGAs long before the 740/750 was introduced. In fact, there are pictures somewhere of a prototype PB5300 (I think it was code-named Mustang or something) with a BGA 603.

I'm only going by what DuoMauler has said over on 'fritter and in emails/pms to me, but I'm CERTAIN he knew exactly what he was talking about!

But yes, the 740 is fully pad-compatible with the 603, so good luck finding a QFP version.

I think I may be able to find one, but reworking a QFP on a PCB is very likely beyond my present skill set. Designing a "carrier board" and installing modern components would be right up my alley though! I first conceived of this approach for upgrading the "Luggable" which has LOTS of PCB acreage to play around in . . . but I can't even afford a DOA unit and I'm NOT hackin' my pristine NASA property plated Portable!

'sides, I've got Duos up the ole' WAZOO!

As for clocking the processors by directly modifying leads, that's not too bad of an idea, especially in lieu of a proper resistor config. Maybe I'll try it on a spare 5300 board when I'm home.

THX, I thought it was pretty nifty too, just wish I'd have thought of it back in the day! DuoMauler woulda' JUMPED on that one! He had QFP G-3s AND the skills to rework the MoBo! I've just gotta get into contact with him again!

However, seeing as how the early PB1400s wouldn't use 166MHz processors (they'd always clock to a max of 133MHz), and with the issue being attributed to the ROMs, do you imagine that there could be a similar problem in the 2300 and 5300?

I doubt that even Apple would have bothered with such nonsense on PBs w/o upgradeable ProcCards on board. Besides, Apple has ALWAYS had enough accidental issues w/ROMs. Remember the '030 "dirty ROMS" fiasco and the (likely multi-millionaires) who developed and patented Mode32?

[Bunsen]

BTW, there are MUCH more efficient means of prototyping KBDs, one of which is to fab a fiberglas copper-clad etched PCB replacement for the bottom membrane, thus stiffening said spongeboard!

Yep, that works too

. . . and why would the interface need to be ADB?

Just for clarity, this thread is about two different hacks:

1. G3 on original Duo logic board
2. New logic in Duo shell

Correct? ADB would only be required for the first hack.

will the adapted CPU interface efficiently with the Duo's heatsink?

I would assume replacing the original barely adequate heatsink was part of the equation when contemplating a CPU transplant.

a "carrier" board in the Duo form factor that "bolts right up" to the Duo's magnesium frame and has the same (or improved replacements) I/O and battery connections (or an expansion bay and INTERNAL battery combo) and places a VERY CAPABLE CPU under the magnesium heat sink

Right, now I follow you. You want to make up a carrier board for an available logic board that breaks its I/O out to the Duo's physical ports, and to the LCD. Sounds eminently doable. Similar to the "stealth" casemods that have been performed on desktop machines, and eerily similar to a plan I've been slowly brewing for the much more spacious shell of a PB520/540.

AFAIK, that heatsink is just a sheet of stamped aluminium, not magnesium. I have in the past had visions of a drop-in replacement in stamped copper, a low-profile finned copper heatsink, and/or a kludged in heatsink/heatpipe/fan combo out of another laptop.

http://www.embeddedarm.com/products/board-detail.php?product=TS-7500

Nice board - but why just for an "I/O subsystem"? Why not pop an extra $30 for a complete ARM board with better specs?

Will said ARM outperform the Atom ChipSet in HP_Mini?

Probably not, no. But even the one you post above would outrun a Duo, and give a low-end G3 a decent run for its money.

I'm not all that familiar with the ARM series, but from what I gather they're decently efficient little *nix platforms, especially in watts/flops terms. Raw clockspeed is not that useful a comparison.

an open Source board

Again, for clarity:

1. A board whose hardware design is itself open-sourced
2. A board capable of running an open-source OS

Which is it? The first is a very, very small class of boards; the second, very very large.

that is capable of fulfilling all the functions of the PBX Bridge IC and ALL the slow bus subsystems of the "ubuntuDu . . . erm . . . SuperDuo!" The CPU Card/Memory, fast i/O bus would be another matter entirely

Nope, you've lost me again. The embedded boards you seem to be discussing are all highly integrated - I'm not aware of any that place the CPU and/or any I/O on daughterboards.

/edit/ Okay, IO daughterboards for Mini-ITX are here and useful.

MiniITX / Compatible LVDS LCD.

I'm not yet familiar with those form factors or their heating/cooling/power/space/location requirements

Mini-ITX is just a board form factor specification, like ATX. As such, there are boards available from a number of manufacturers, with everything imaginable on them, from low-power, low-heat 500MHz Pentium III derived CPUs, to ARMs, to embedded PPCs , integrated media processors (combined GPUs, codecs and audio DSPs), to dual sockets for a pair of quad core Xeons. Power, heat, cost, ease of development and availability vary accordingly.

http://en.wikipedia.org/wiki/Mini-ITX
"17 x 17 cm (or 6.7 x 6.7 inches)"

There are also nano- and pico-ITX boards - rarer, smaller range, higher price/performance

Judicious searching and reading here should prove enlightening (especially the FAQ and shop) and inspiring (especially the right sidebar under Projects ):
http://www.mini-itx.com/

BTW: FPGA technology will (should) allow implementation of LVDS out among other things.

Quite a large number of MiniITX boards have LVDS out for LCD video built-in - so you can connect directly to a bare LCD panel without the added, and substantial, expense of a DVI or VGA converter/controller board.

It does seem likely that the easiest and cheapest solution might be to rip apart a netbook and stuff that into the Duo.

[Trash80toHP_Mini]

BTW, there are MUCH more efficient means of prototyping KBDs, one of which is to fab a fiberglas copper-clad etched PCB replacement for the bottom membrane, thus stiffening said spongeboard!

Yep, that works too

. . . and why would the interface need to be ADB?

Just for clarity, this thread is about two different hacks:

1. G3 on original Duo logic board
2. New logic in Duo shell

Correct? ADB would only be required for the first hack."]

I'm just along for the ride on the G-3 Duo now that I've got the "Open Source" Bug, but yes, ADB would only be required for the AppleCentricHack.

The ubuntuCentridcHack is another story entirely!

will the adapted CPU interface efficiently with the Duo's heatsink?

I would assume replacing the original barely adequate heatsink was part of the equation when contemplating a CPU transplant.

I wasn't so sure the Duo's heatsink arrangement was all that bad.

]a "carrier" board in the Duo form factor that "bolts right up" to the Duo's magnesium frame and has the same (or improved replacements) I/O and battery connections (or an expansion bay and INTERNAL battery combo) and places a VERY CAPABLE CPU under the magnesium heat sink

Right, now I follow you. You want to make up a carrier board for an available logic board that breaks its I/O out to the Duo's physical ports, and to the LCD. Sounds eminently doable. Similar to the "stealth" casemods that have been performed on desktop machines, and eerily similar to a plan I've been slowly brewing for the much more spacious shell of a PB520/540.

Sick minds think alike!

AFAIK, that heatsink is just a sheet of stamped aluminium, not magnesium. I have in the past had visions of a drop-in replacement in stamped copper, a low-profile finned copper heatsink, and/or a kludged in heatsink/heatpipe/fan combo out of another laptop.

IIRC, the entire magnesium frame of the Duo is the heatsink, the little projection is a heatpipe to the chassis.

It does seem likely that the easiest and cheapest solution might be to rip apart a netbook and stuff that into the Duo.

Thought of it, rejected it, there's room for a bigger screen and a MUCH better proc than the current crop of NetBooks offer!

Not worried about easy OR cheap, as I have plenty of time and no money ATM. When the design is right, I should have the money!

[Bunsen]

Duo's heatsink

I realise now it's the 1400's heatsink that has been floating before my eyes. I've seen that much more often than the Duo one, and I had them mixed up. And it is to the 1400 that the following was addressed:

a drop-in replacement in stamped copper, a low-profile finned copper heatsink, and/or a kludged in heatsink/heatpipe/fan

I must take a gander inside a Duo again to refresh my memory.

the entire magnesium frame of the Duo is the heatsink, the little projection is a heatpipe to the chassis.

Do you actually mean a heatpipe as defined here, or just a chunk of metal?

[Trash80toHP_Mini]

the entire magnesium frame of the Duo is the heatsink, the little projection is a heatpipe to the chassis.

Do you actually mean a heatpipe as defined here, or just a chunk of metal?

My Bad, I just meant that it's an integral part of the magnesium chassis and transfers heat to the rest of the chassis, pretty slick way to do it actually, just not conducive to moving the CPU.

Haven't had one open for a while either, so I'm going on memory too, therefore I might also be mistaken . . . =8-\

[Bunsen]

Have a read of that article. Heatpipes are kinda awesome. They allow you to have your CPU in one spot, and your main heatsink in another.

[Trash80toHP_Mini]

I'll check it out. Meanwhile, I've got a Professional Linux Programmer looking into other boards for the DuoNetBookNemisisHack.™ He was impressed with the Duo's hack potential and its incredible factor. He said it ought to be a piece of cake and has already identified several very promising ProcCards!

I doubt he'll have much time to devote to the project, but we've now got one wild code jockey, one certifiable hardware hacker/KBD Butcher/Rudimentary PCB Designer-Prototyper in on the project so far!

I ran into my brother's friend, a professional Multilayer PCB Designer, who may troubleshoot that end of things . . .
. . . HOPEFULLY . . . . . . and my M. E. middle lil' bro, who does high end Blade Server Packaging for a living, knows rapid prototyping inside-n-out. If I twist his arm just a lil' he might be of great assistance to the project . . . or not!

. . . but he'll usually give me an informed opinion of my plans, even though my hacking bores him to tears.

If I can get a few Open Source Maniacs involved on the Software end of things, this might actually become a credible project!

Does ubuntuDuoHack™ have a better ring to it?

[Bunsen]

DuoBuntu?

[Trash80toHP_Mini]

DuoBuntu?

[Franklinstein]

You guys are really going to town on this project. It seems that you've rapidly progressed from relatively simple upgrades to a completely different machine encased within a Duo shell. Either way, the outcome is relevant to my interests.

... but reworking a QFP on a PCB is very likely beyond my present skill set.

Reworking SMT devices is remarkably easier than doing so with BGAs, though the principles are similar: the devices are installed and removed with heated air.

If you've got access to proper equipment (like a reflow station), it's almost trivial to replace an SMT device such as the QFP processor on those boards. Plus, if there's a mistake, it's super-easy to touch up the affected areas, unlike a BGA, which would require a complete do-over to correct even the smallest of errors.

If you're strictly an at-home hobbyist, then the same general end can be achieved with a good heat gun. Simply apply the heat gun to the desired area, slowly lowering it until the solder melts, then pull off the device. Then, slowly remove the heat gun before installing the new device and repeating the process, making sure that the device doesn't slip from its pads or anything, and you're pretty much done.
*Ideally, you'd clean the pads before placing the new device, but then you'd have to go and solder every pad by hand, since you likely won't have new solder paste or the original application tools. This way, you just need to go in and touch up any imperfections when you're done. Or, clean the pads and hand-solder. Whichever.

Obviously, there's more to proper heat gun hackery than what I've included here, but these are the basics. For more in-depth instruction on heat-gunning stuff, there are plenty of resources on the intarwebz, particularly pages dealing with iBook GPU reflowing.