Incase anyone is interested in how Collin’s magic works in detail and compared to a stock G3... here’s my vid on his masterpiece
https://www.youtube.com/watch?v=lBO55EiyFts
https://www.youtube.com/watch?v=lBO55EiyFts
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
replacing the pearl in the clamshell. DIY iBook G3 Clamshell G4 CPU upgrade
- Thread starter LightBulbFun
- Start date
- Sort by reaction score
very cool video
just a couple things.
im curious where did you get 533Mhz from? with the machine running at 500Mhz (66.6*7.5) the only thing being overclocked is the L2 cache (from 150Mhz to 250Mhz) otherwise everything else is running in spec and nothing is being overclocked and nothings running at 533Mhz
and if you decide to overclock the FSB you must lower the multiplier as 100*7.5=750Mhz and I doubt your 7410 can run at that not without a few too many volts and some LN2
(for a 100Mhz Bus speed you could do 5x or 5.5x multiplier which is 500Mhz or 550Mhz)also its worth mentioning, that Core image always says "Software" on a non core image capable GPU regardless of if the Drivers installed.
installing the Rage Pro drivers in Leopard helps a small bit with pushing the UI around but not much else. as the Rage Pro GPU does not support OpenGL under Leopard or any version of OS X sadly. (Apple actually got sued over this by G3 beige users https://arstechnica.com/civis/viewtopic.php?f=2&t=1976)
in regards to the Mac OS 8.6 thing. I dont think its caused by the G4 mod. I know @AphoticD ran into the same issue when he was rebuilding his iBooks battery. so im thinking its Power thing. My Pismo also did the same thing and I noticed 8.6 on my Pismo does not detect the battery properly.
have you tried booting 8.6 without the battery installed or trying another battery in the G4 clamshell? or manually imaging 8.6 over bypassing the install CD. thats how I got 8.6 onto my Pismo since the install CD would always switch off just before the desktop for me.
Not exactly Clamshell related, but I've just gotten through doing this same upgrade on a few PowerBook G3 Pismo CPU cards people have sent to me. I made a video of the whole process of the first one, which you can view here, and I also have some pictures of the process as well. It went very well, all machines ran perfectly with the G4 CPU installed.
Last edited:
Amazing work! Just watched the video through and now I can see how much goes into this. I could feel the tension at the moment you lifted the tape on the stencil and the whole thing lifted up and then... *Calmly* "Unfortunately, I had to redo that whole process again".. Kudos! The term "painstaking process" is probably a better fit than just a "lengthy process".
Well, I might be tempted to invest in some rework gear and give this a go myself. I'm feeling like I have now acquired (and upgraded) all of the PowerPC systems in the way I wanted, so my next step will be to play around with hardware mods like this!
Also tempted. I reflowed my HD5870 card yesterday. It took 4 attempts before it worked. I think I sort of have a feel for when the BGA solder 'takes' having witnessed it at close quarters.
The first three attempts were done by guesswork. I looked for signs of the Amtech flux melting and as soon as I saw the flux smoking, I stopped. For the final attempt, I shone my iPhone's flashlight at the 0.5mm gap between the chip and the PCB, crouched down and looked out for the signs that the solder had actually melted. What I noticed is that when the solder actually melts the chip drops down suddenly so that there is no longer any visible gap between it and the PCB. I let the heat carry on for another 10-15 seconds to ensure that all the solder balls had an equal opportunity to melt, not just those on the periphery. A couple of seconds after removing the air gun I saw the chip shoot back up and the solder balls reform into place. Quite a remarkable sight. I couldn't see any bridging so screwed everything together again and repasted. This time it worked.
What I learnt was that despite CPUs' going into shutdown mode much above 100C they can take a lot more external heat without frying.
The first three attempts were done by guesswork. I looked for signs of the Amtech flux melting and as soon as I saw the flux smoking, I stopped. For the final attempt, I shone my iPhone's flashlight at the 0.5mm gap between the chip and the PCB, crouched down and looked out for the signs that the solder had actually melted. What I noticed is that when the solder actually melts the chip drops down suddenly so that there is no longer any visible gap between it and the PCB. I let the heat carry on for another 10-15 seconds to ensure that all the solder balls had an equal opportunity to melt, not just those on the periphery. A couple of seconds after removing the air gun I saw the chip shoot back up and the solder balls reform into place. Quite a remarkable sight. I couldn't see any bridging so screwed everything together again and repasted. This time it worked.
What I learnt was that despite CPUs' going into shutdown mode much above 100C they can take a lot more external heat without frying.
I hate to tell you this, but "reflowing" a GPU is not a proper fix... If a GPU fails, it's the chip itself that has failed, not the solder underneath it. "Reflowing" can temporarily get the chip to work again, but it will almost certainly fail again in probably less than a month.
I accept what you're saying here, as you obviously have a great deal of experience. I'd like however to become a little more knowledgeable into why some GPU reflows are successful for just a few weeks, whilst some have reported still going strong after 15 months or so?
I'm asking this because only a very short while ago I wasn't even familiar with the word reflow, but recently as a gratis favour to a near total stranger with a sick Mac (and especially as an interesting challenge for myself), I dissected his 27" iMac and performed a reflow on the ATI Radeon HD4850. I was under no illusions that I could sucessfully revive it, but had nothing to loose. Link HERE. Ok, it was just over a month ago, but I'm keeping close contact with the owner to see exactly how long it survives.
I understand that it is a temporary fix, although with my usage case I am hoping for the longer end of temporary but I don't understand how reflowing can revive a failed chip in the first place.
There's an excellent video by Louis Rossmann from the Rossmann Repair Group on this subject here. Not gonna embed it here since it does contain language that would be censored on this forum
He talks about reballing, but reflowing is pretty much the same deal.
@dosdude1 is at it again, this time upgrading a Trayloader iMac G3. He's figured out how to change the core voltage on the card, which is necessary when the G3 chip on the card is a ARX Motorola 750, which run at higher voltages that will damage the Motorola 7400 G4 CPU (more information on these G3 chips and Vcore can be found earlier in this thread). @LightBulbFun and I have been trying to figure out what sets the Vcore on these cards, so I am very glad @dosdude1 and him figured out that a set of specific resistors does this.
indeed it was very fun working out what set the Core voltage the trayloaders with @dosdude1
I think @dosdude1 will be doing a full write up here about the work me and him did to figure all this out
so im not going to go into too much detail as this is his work so to speak, but I will say im very chuffed that he was able to slap a 7400 onto a 233Mhz MPC750A Bondi Blue CPU card and im glad my theory about the 7400 working on a MPC750 board actually holds true
(anyone got a spare Kanga logic-board? )(if we can figure out what sets the L2 cache voltage then its might be possible to slap a 7410 on, but not all early G3 macs have L2 cache voltage control from what I have read a few of them just run the L2 right off the main 3.3V rail)
on a slightly separate note I have ordered a TS-100 soldering iron, im eager to try it out as it has had really good reviews and its compatible with Hakko "T12/T15" tips which work way better then the 936 clones do, ill let you guys know how the iron works out when it arrives
Last edited:
I'm curious as to how these programmable irons stack up. The TS-100 looks about as slim as my Antex and that just doesn't have the thermal mass to be genuinely useful.
For those in the UK, Maplin stocks the Hakko FX-888D and current prices are now at eBay levels. If they cut them any more and your local branch still has any, they might be worth a punt.
For those in the UK, Maplin stocks the Hakko FX-888D and current prices are now at eBay levels. If they cut them any more and your local branch still has any, they might be worth a punt.
the thing about the TS-100 is it uses Tips which use the same tech the Hakko FX951 does (and is also compatible with 951 tips) in that the heating element is built right into the tip your using giving WAY more thermal mass and ability to quickly transfer heat into whatever your working on
a traditional iron like a FX888 or the many 936 clones will always have a gap between the heating element it self and the tip which severely limits thermal mass/the ability to transfer heat quickly and efficiently.
hence why im looking forward to my TS100 arriving it should be a massive upgrade from my 936 clone and it "only" cost me £50 for the iron and 2 tips of my choosing.
im hoping that the iron works out, so I can then actually move out of the experimental/prototyping phase and actual do G4 CPU upgrades for other people eventually I hope to make enough Cash to get a Quick 861D hot air station and a bottom heater so I can do big chips like intel CPUs on intel Mac logic-boards and such
(my 858D hot air station can Just about manage a 360 Pin CBGA package but thats it, anything bigger and it doesn't work well or at all, for example it took me forever to remove a 483 pin CBGA XPC7455A from a 1Ghz MDD single CPU Card using the 858D, and the card it self lost a few pads)And it's been getting glowingly positive reviews from people in the business, like Louis Rossman, who I know really doesn't mince his words when he doesn't like something, so that's saying something!the thing about the TS-100 is it uses Tips which use the same tech the Hakko FX951 does (and is also compatible with 951 tips) in that the heating element is built right into the tip your using giving WAY more thermal mass and ability to quickly transfer heat into whatever your working on
a traditional iron like a FX888 or the many 936 clones will always have a gap between the heating element it self and the tip which severely limits thermal mass/the ability to transfer heat quickly and efficiently.
Well, I have just looked at Rossman's review and glowing is overegging it. He certainly thought the iron was very good and it had some cool features while being comfortable to use despite appearences but there are certain downsides: You need to source a stable 24V power supply, which he believes will cost the same as the iron and of the TS tips on offer he only rated a couple. While Hakko tips work, they are overlong and somewhat unwieldy.
For those on a budget without a power supply to hand, he recommended a knock-off Hakko FX-951, otherwise the total cost of the TS-100 rises to some $50 more with fewer tips to choose from.
A very good product but not for everyone.
did ya see his follow up video to his review? he addresses and corrects a few things he said there (like the cost of a PSU), plus I (and most people doing this probably) have a 19V PC laptop charger which should work well for powering the iron
(as for the fewer tips available, the TS-100 is compatible with Hakko T12/T15 (aka 951) tips and you can even 3D print an adapter that lets you use the Hakko grip thingy but most importantly for me the TS-100 works with that Broom head style tip for cleaning up lots of BGA pads in one go )
Just going to be honest here... Stunning, simply amazing! I would love to do this with my iBook G3 Clamshell, but lack the proper equipment (soldering stuff ), maybe in the future though. What you did is simply amazing, actually breathtaking to see an iBook Clamshell natively running OS X 10.5 Leopard! iBook Clamshells are truly a priceless piece of machinery.
), maybe in the future though. What you did is simply amazing, actually breathtaking to see an iBook Clamshell natively running OS X 10.5 Leopard! iBook Clamshells are truly a priceless piece of machinery.
I'm actually working on that project... I have the 7448 soldered on, but the system will no longer POST. It needs a firmware update in order to properly work with the CPU. I'll be attempting updating the firmware very soon.
@LightBulbFun
Regarding your comment, "im hoping that the iron works out, so I can then actually move out of the experimental/prototyping phase and actual do G4 CPU upgrades for other people".
If/when that happens, expect me to be one of the first to come knocking at your door.
Regarding your comment, "im hoping that the iron works out, so I can then actually move out of the experimental/prototyping phase and actual do G4 CPU upgrades for other people".
If/when that happens, expect me to be one of the first to come knocking at your door.
I already do this for people located in the US... If you are located in the US, feel free to let me know. I'd be happy to upgrade a card/machine for you.@LightBulbFun
Regarding your comment, "im hoping that the iron works out, so I can then actually move out of the experimental/prototyping phase and actual do G4 CPU upgrades for other people".
If/when that happens, expect me to be one of the first to come knocking at your door.
Ok thanks. As I'm in Europe (France), location-wise it will probably be easier to await LightBulbFun once he's up and running in production mode.
Ooohhh!
Get the kinks worked out, and take my money. I'd LOVE to have a 400mhz G4 Bondi Blue iMac
The kinks have already been worked out! I'll upgrade your CPU card if you're located in the US. I have plenty of G4 CPUs available.Ooohhh!
Get the kinks worked out, and take my money. I'd LOVE to have a 400mhz G4 Bondi Blue iMac
I decided to set the PLL configuration of my G4-upgraded Tray-loading iMac CPU card to run the G4 at its stock 400MHz clock speed... And it runs great! Seems to get a little warmer under load, but still is well within a reasonable operating temperature.
