PDA

View Full Version : Quicksilver Oveclock with Dipswitches!



bonsai
4th April 2005, 11:49 PM
Hey All!

Just finished modding my quicksilver 733mhz cpu. The beauty of these cpus is that there is no l3 cache to worry about (i know its a downside, but meh.).

The problem I had was i spent wayy too much time fiddling on the back of the pcb moving those tiny resistors! So I invested into making a dip switch so that I could change the multipliers without having to arguously go back and forth unplugging, unscrewing, soldering etc etc.

Anyway the results speak for themselves. The resistors that were contorling the multipliers before were 900ohm. So instead, I put a 1.2kv ohm resistor over one side of the solder pads (yes they share the same bridge. Then connted it to a dip switch.

Funnily enough, I actually found stability issues using a lower ohm resistor. Strange but it would only boot a couple of times, change the resistor over and BOOM works perfectly!

Now in the first image, you'll see another resistor ive soldered on, this is to boost the core voltage from 1.9v to the g4 rated max of 2.1v. Once again, i found stability issues just using a peice of wire and strapping it between the two solderpoints. So in went the 460ohm resistor.

The result was a nice overclock from 733mhz to 933mhz! although it would not boot at 1ghz :( Might have to use larger resistors.... its strange but it didn't want to boot at 800mhz when using smaller resistors.

Anyway without further adue here's the pictures! (sorry, quite proud of my soldering work!)

Also, if the admins want me to write this up as an article with the multipliers, voltage settings and resistors to use and whatnot, i'll be happy to! I know I had alot of trouble trying to find this information out on my own having to ussually refer up to 4 or 5 sites to find out which information was correct and what was very very incorrect!

http://web.1earth.net/~alex/mac/quicksilver1.jpg

http://web.1earth.net/~alex/mac/quicksilver2.jpg

josh64
5th April 2005, 12:48 AM
Nicely done! I wish I had your soldering skills!

Kildare
5th April 2005, 12:53 AM
Nice work :)
Starting to look into the Mac side of overclocking as I've pretty much overclocked every PC I've ever owned...

Atomic
5th April 2005, 01:41 AM
Yes a job well done but a word of warning regarding voltage boosts and extreme CPU speeds. Older revisions of the 7410 or 7450 (i assume the CPU is one of these) just weren't built to handle extended periods of operation at such high voltages/frequencies. Just be mindful that this will greatly reduce the life of your CPU's.

Also, some advice about adding a resistor to your dip switch setup. Personally I would remove the resistor altogether and connect your common wire directly to the board. You will more than likely find it to be more stable at various speed settings. Plus that resistor (and i cant see the colour bands well) probably does not have tight tolerances thus causing its reistance to vary during operation. Apples SMD resistors more than likely operate at extremely tight tolerances.

To anyone interested in overclocking - a rule of thumb is to be prepared to fork out money to buy a CPU board at any time. Due to the nature of the mac overclock, you are constantly touching, soldering and fiddling with tiny internal components that are extremely prone to damage. Im not saying you shouldn't do it - far from it, I love fiddling too but i have years of soldering experience and feel very confident in this area, although I was never naive to think something wouldn't go wrong.

kim jong il
5th April 2005, 06:13 AM
Originally posted by bonsai@Apr 4 2005, 11:49 PM
Hey All!

Just finished modding my quicksilver 733mhz cpu. The beauty of these cpus is that there is no l3 cache to worry about (i know its a downside, but meh.).

The problem I had was i spent wayy too much time fiddling on the back of the pcb moving those tiny resistors! So I invested into making a dip switch so that I could change the multipliers without having to arguously go back and forth unplugging, unscrewing, soldering etc etc.

Anyway the results speak for themselves. The resistors that were contorling the multipliers before were 900ohm. So instead, I put a 1.2kv ohm resistor over one side of the solder pads (yes they share the same bridge. Then connted it to a dip switch.

Funnily enough, I actually found stability issues using a lower ohm resistor. Strange but it would only boot a couple of times, change the resistor over and BOOM works perfectly!

Now in the first image, you'll see another resistor ive soldered on, this is to boost the core voltage from 1.9v to the g4 rated max of 2.1v. Once again, i found stability issues just using a peice of wire and strapping it between the two solderpoints. So in went the 460ohm resistor.

The result was a nice overclock from 733mhz to 933mhz! although it would not boot at 1ghz :( Might have to use larger resistors.... its strange but it didn't want to boot at 800mhz when using smaller resistors.


A couple of points to note on a subject close to my heart:

The maximum CPU core voltage is processor dependent (as you should know from your G4 400/500 overclock where the PPC 7400 has a maximum voltage limit of 3.1V; coming out of the factory set at 2.1V and I, for example, boosted mine to 2.6V) Check for details of your particular processor always.

The reason you will have stability and boot problems with low value resistors is that the function the resistors fulfill is a pull high, pull low setup. With the value too low the machine will have trouble establishing whether it is a pull low or pull high value situation/setting and will be varying your PLL ratios each time you boot, (not to mention probably on the fly too if this is possible) creating boot and stability issues. Explained a bit differently your machine will have trouble determining whether a particular resistor is in place resulting in the wrong PLL multiplier being set. I have found no literature that discusses the threshold resistance at which point it changes from low to high.

Please instruct people to anchor the ribbon cable to the daughterboard really securely (superglue is good) :) Stickytape can only be used for testing really. Failure to address this one simple issue can see you(one) peel the circuit traces of the PCB so easily it is frightening. Again some daughterboards are worse than others.

Finally, nice overclock :)

cheers, kim

bonsai
5th April 2005, 08:10 AM
Originally posted by Atomic@Apr 5 2005, 01:41 AM
Yes a job well done but a word of warning regarding voltage boosts and extreme CPU speeds. Older revisions of the 7410 or 7450 (i assume the CPU is one of these) just weren't built to handle extended periods of operation at such high voltages/frequencies. Just be mindful that this will greatly reduce the life of your CPU's.

Also, some advice about adding a resistor to your dip switch setup. Personally I would remove the resistor altogether and connect your common wire directly to the board. You will more than likely find it to be more stable at various speed settings. Plus that resistor (and i cant see the colour bands well) probably does not have tight tolerances thus causing its reistance to vary during operation. Apples SMD resistors more than likely operate at extremely tight tolerances.

To anyone interested in overclocking - a rule of thumb is to be prepared to fork out money to buy a CPU board at any time. Due to the nature of the mac overclock, you are constantly touching, soldering and fiddling with tiny internal components that are extremely prone to damage. Im not saying you shouldn't do it - far from it, I love fiddling too but i have years of soldering experience and feel very confident in this area, although I was never naive to think something wouldn't go wrong.
Very true what you say, You will degredate the life of a cpu, however i believe (don't quote me on this) but the life of the average cpu is around 15 years. If you overclock and remove 2/3 of the life out of the cpu, there's still 5 years left and thats at 100% load 24/7. In a 2 - 3 years this g4 cpu will definitly be sherrif long-gone.

And to be prepared to loose your cpu + mainboard is very crucial, however do not let it deter a curious mind.Remember that Galileo (spelling?) went against the orders of the church who would have killed him, but he still made some of the original discoveries of our solar system.

Actually the computer would not boot at all when connecting it directly to the common wire. Unfortunatley, my old man pinched my 1/2watt resistors so I was only left with my 1/4 watt box to use when I did the mod, however a quick trip to Jaycar will solve this and annoy me that the fuel used to get there will cost more than the resistors!

bonsai
5th April 2005, 08:16 AM
Originally posted by kim jong il@Apr 5 2005, 06:13 AM
A couple of points to note on a subject close to my heart:

The maximum CPU core voltage is processor dependent (as you should know from your G4 400/500 overclock where the PPC 7400 has a maximum voltage limit of 3.1V; coming out of the factory set at 2.1V and I, for example, boosted mine to 2.6V) Check for details of your particular processor always.

The reason you will have stability and boot problems with low value resistors is that the function the resistors fulfill is a pull high, pull low setup. With the value too low the machine will have trouble establishing whether it is a pull low or pull high value situation/setting and will be varying your PLL ratios each time you boot, (not to mention probably on the fly too if this is possible) creating boot and stability issues. Explained a bit differently your machine will have trouble determining whether a particular resistor is in place resulting in the wrong PLL multiplier being set. I have found no literature that discusses the threshold resistance at which point it changes from low to high.

Please instruct people to anchor the ribbon cable to the daughterboard really securely (superglue is good) :) Stickytape can only be used for testing really. Failure to address this one simple issue can see you(one) peel the circuit traces of the PCB so easily it is frightening. Again some daughterboards are worse than others.

Finally, nice overclock :)

cheers, kim
A couple of very good points!

Thanks for clearing up the resistor information! Armed with this and a trip to Jaycar I know what i'll be doing to stablise this overclock which im shocked didn't occur to me before! (however you'll have to wait till its done ;))

Oh yes! I did forget to mention that, the sticky tape is only for temporary purposes. It is around the resistor to stop creating any shorts that may lead to the processors death only because i've also run out of heat-shrink tube. It is crucial to securely adhere the wires to the board. I've learnt this all too much with soldering resistors directly onto a board!

Thanks for the feedback once again :)

Rayd
5th April 2005, 11:32 PM
love it! i cant even do normal soldering :)