IBM Thinkpad 600E bios mod for processor update

[Sharedoc]

Beatoem,

Great news confirming the mods. Yes, you are right, the nearby resistor to 8% resistor gives about 5% overclock as well.

How I did 133MHz? I used so called turbo-PLL circuit, which is a Japanese originated overclocking trick. You chance the 14.31818MHz standard XTAL of the clock circuit with faster XTAL. But then you have to feed correct reference clocks to the main board from a "piggyback" clock generator . It is a trick that serious overclockers use. There is not so much room for the piggyback in a laptop, so I used the PCMCIA cage to host the extra clock generator and fed the clocks by wire for the mainboard. I didn't get 24/48MHz clock working reliably by wires so USB does not work correctly if I use the turbo-PLL. I still have turbo-PLL installed in my TP600E but I run it now everyday with 108MHz FSB.

With this 700MHZ PIII I have now, the fastest speed it run stabile was 833MHz. But now it won't run at higher multiplier anymore (due to unlucky visit at 600X motherbord)

I am looking to find a cheap 850MHz PIII-MMC2-processor to try to hit 1 Gig.

[beatoem]

Beatoem,

Great news confirming the mods. Yes, you are right, the nearby resistor to 8% resistor gives about 5% overclock as well.

How I did 133MHz? I used so called turbo-PLL circuit, which is a Japanese originated overclocking trick. You chance the 14.31818MHz standard XTAL of the clock circuit with faster XTAL. But then you have to feed correct reference clocks to the main board from a "piggyback" clock generator . It is a trick that serious overclockers use. There is not so much room for the piggyback in a laptop, so I used the PCMCIA cage to host the extra clock generator and fed the clocks by wire for the mainboard. I didn't get 24/48MHz clock working reliably by wires so USB does not work correctly if I use the turbo-PLL. I still have turbo-PLL installed in my TP600E but I run it now everyday with 108MHz FSB.

With this 700MHZ PIII I have now, the fastest speed it run stabile was 833MHz. But now it won't run at higher multiplier anymore (due to unlucky visit at 600X motherbord)

I am looking to find a cheap 850MHz PIII-MMC2-processor to try to hit 1 Gig.

obviously your at the forefront of tp600 developments,priviledged to communicate with you,that pll trick certainly requires absolute surgical skills,i have 2 tp600e,p111@702+p111 celeron @486mhz,both running @108fsb +really im impressed with the improvement the celeron gave over the p11 400 really,at his stage i'm trying to work out why pleap control panel fails automaticaly load the cache at startup ,it comes with a failed to load etc at start up in xp ro,so what i done is uncheck the enable at startup on the menu +now it boots with no startup errors +then i just click on the icon to trigger the cache,did you have this experience with the software? that 700 cpu you got, that wont revert to the full speed since you installed on the 600x have you considered tempoarily removing the cpu+revert it back to speedstep active + replay the procedure again?
at this stage i will look for a p111 750 to upgrade my rig from the current 650cpu ,am pondering if i add another1kohm resistor to r157 , that may constitute in a 8+5% increase in the fsb,have you tried c21 to r153 at all? are r 153,r157+r164 the only options the pll will support?conversely do you think its possible to modify the mmc2 p11(66fsb)cpu sereis
to default to 100 fsb?why i ask is because i have some mmc2 366,400,433(66fsb)cpus to mod some hp 4150 (non b)laptop,problem with those models are they wont boot with the 433 /450 celeron+p111 500 mmc2 either +only allow the 366/400 p11 to boot,despite the 100 fsb sdram i fitted.then i figured if the cpu can default to 100 fsb the hp bios will detect the id string atleast,(seens that i heard one chap atleast got the p111mmc2 450 to work on the hp,they indeed work on 100fsb)i am looking into a way to get the bios to allow the faster mmc2 cpus to boot on that 4150,i beleive on fcpga skt 370 coppermines you can modifie the bsel contacts to allow the processor to function @66,100,133 fsb,i cant see why the same cant be done with mmc2 66/100 fsb cpus,100-133 of couse may require the fster plls 133fsb
shuch as in your rig,but the 66fsb cpus can potentially reap a 41% gain
+maybe no bios mods needed.because the id strings are unchanged.
this would b a good thing for many notebook overclockers,thus get some use out of a perfectly good cpu (before the upgrades).
there are 750 mmc2 around $70-80 on ebay,i havent seen any 800-850 for while,but the 750 can give you around 1ghz on your rig.
do you have a dock 111 setup?i got one running raid +that really perks up the performance.i highly recommend for any tp 600/765/770 owner,am planning to try out a voodoo 11 on the dock soon,i'm rapt these 600s,they are very strong+versatile machines.sorry to read about the bad cpu experience you got with that dud cpu,i only hope the trader refunded you or compromised for the hassle.

regards ja

[Sharedoc]

TP600E clock generator can only generate 66, 69, 71, 100, 105, and 108MHz fsb using the frequency selection inputs. So if you want a higher than 108 you have to change the XTAL and make a turbo-PLL.

The 66/100MHz selection signal to motherboard clock generator comes from the MMC2 processor module. I have traced it down in the MMC2 connector and you can change it from 66 to 100 MHZ setting by cutting folio and hardwiring. But the resulting 50% overclocking to PII/Celeron processor is too large. The processor won't boot.

It might work if you can change the PII/Celeron also to use 100MHz bus, but
I don't know if it is possible. It may be multiplier locked. Search the web, I believe Intel used Dixon core also in desktops, so maybe some overclockers tried to change the multiplier.

HP notebooks: if PIII don't work, it may be a problem with the Bios not recognising the CPUID string that the processor returns when executing the CPUID instruction in the Bios code. Changing the FSB frequency does not help if this is the case.

[beatoem]

The 66/100MHz selection signal to motherboard clock generator comes from the MMC2 processor module. I have traced it down in the MMC2 connector and you can change it from 66 to 100 MHZ setting by cutting folio and hardwiring.

HP notebooks: if PIII don't work, it may be a problem with the Bios not recognising the CPUID string that the processor returns when executing the CPUID instruction in the Bios code. Changing the FSB frequency does not help if this is the case.[/quote]

are you saying by cutting folio +hardwiring the mmc2 cpu it can send the 100 fsb signal to the mainboard?assuming thats the case which parts do you change?it is possible to get a33% overclock on the 66fsb cpu,with the mods,it may allow it to work at 100 fsb,conversely if the hp bios knows the id string +assuming the cpu (modded to 100fsb works ok)then its possible to boost he hp 4150 speed.i have just ordered a p111450 to try on the hp conversely i got a spare 300 mmc 2 am willing to see if it can give 450 mghz with the fsb mod,if any 66fsb cpu can reach a 100 fsb it would it would be the 300 type.

[cotopaxi]

i have a piece of information that seems relevant to ray of bandang's method. it is from the IBM hardware maintanance manual, TP 600/600E, under the RAM cover there are two round copper connections on the PCB, shown on pg. 106 as "password pads". there is an illustration of these being bridged (apparently) using a small tweezer with 45 degree tips. i haven't tried ray's procedure yet,( i have 600e) but i suspect these contacts may correspond with the EEPROM contacts. write me for this manual, in PDF format

[Sharedoc]

It is the 600E mainboard clock generator that makes the FSB clocks. The frequency selection signal 66/100MHz originates at MMC2-processor module and it is routedvia MMC2 connector. The Intel MMC2 spec document lists the connector signals and you can find it from there and cut folio and connect to either VIO (+3.3V) or GND via 1k resistor.

I remember trying PII 366MHz overclocking by forcing the FSB from standard 66MHz to 100MHz using the above method: it did not boot. So I cannot give high hopes for success.

BTW 66-> 100MHz is 50% overclock. Haven't seen 50% overclock done since Celeron 300A times

[beatoem]

It is the 600E mainboard clock generator that makes the FSB clocks. The frequency selection signal 66/100MHz originates at MMMC2-processor module and it is routedvia MMMC2 connector. The Intel MMMC2 spec document lists the connector signals and you can find it from there and cut folio and connect to either VIO (+3.3V) or GND via 1k resistor.

I remember trying PII 366MHz overclocking by forcing the FSB from standard 66MHz to 100MHz using the above method: it did not boot. So I cannot give high hopes for success.

BTW 66-> 100MHz is 50% overclock. Haven't seen 50% overclock done since Celeron 300A times

did you use the 100 ram at that time,i got a 300p11+a 433 celeron
they are both different to th 366 maybe the celeron with less cache
can work@100,in which turn i would like to give it a go
i can t quite comprehend which pins to address other than vio +gnd
do you know which rows of the cpu i need to modifie+which bit needs cutting?i have the mmc 2 diagram handy any help will be appreciated.
regards ja

[Sharedoc]

MMC2 connector pin 34B. Signal name is FQS. 0=66MHz, 1= 100MHz.

According to Intel spec document it should be hardwired with 10kohm resistor to GND at PII MMC2 module.

If you are lucky, you don't have to cut any folio, just use 1kohm to connect the pin to Vio (3.3V). Use resistor, don't connect directly to Vio.

[beatoem]

MMC2 connector pin 34B. Signal name is FQS. 0=66MHz, 1= 100MHz.

According to Intel spec document it should be hardwired with 10kohm resistor to GND at PII MMC2 module.

If you are lucky, you don't have to cut any folio, just use 1kohm to connect the pin to Vio (3.3V). Use resistor, don't connect directly to Vio.

ould have you sussed out r73+r71?i am pondering this first on a 300 cpu+if no go i'll try your version,the other seems easier to me ,gee the p3
memory scores are impressive?how time consuming was the mod yu performed on the turbo pll?i gots me a 750 cpu that should give me 800+
on 108,but i guess any faster +very hot cpu,conversely i think the 700 mgz is actually running cooler than the p11 400 cpu?dids you find this the case with 108+p111 combo?

[Sharedoc]

Building a Turbo-Pll to a laptop takes easily 2 - 3 days. It requires removing the motherboard from the case, because clock generator is on the lower side of the motherboard. Also requires cutting folios and soldering wires to surface mounted resistors. It would be good to have an extra motherboard in case something goes wrong. Also you need an extra clock generator to feed the reference frequencies. This can be obtained by removing a piece of motherboard with a clock generator from an old motherboard. You may also need digital voltage/resistance meter or oscilloscope to check connections. There is a big risk of ruining the motherboard if soldering goes bad. I have ruined several motherboards when building these.

PIII's need to run cooler than PII's. To overclock MMC2 PIII you actually get better results if you lower Vcore from the standard 1.60 volts down to 1.45 ... 1.50 to make it run cooler. This can be done by modifying the resistors that set the Geyserville code to the voltage regulator on the MMC2 module.

[beatoem]

Building a Turbo-Pll to a laptop takes easily 2 - 3 days. It requires removing the motherboard from the case, because clock generator is on the lower side of the motherboard. Also requires cutting folios and soldering wires to surface mounted resistors. It would be good to have an extra motherboard in case something goes wrong. Also you need an extra clock generator to feed the reference frequencies. This can be obtained by removing a piece of motherboard with a clock generator from an old motherboard. You may also need digital voltage/resistance meter or oscilloscope to check connections. There is a big risk of ruining the motherboard if soldering goes bad. I have ruined several motherboards when building these.

PIII's need to run cooler than PII's. To overclock MMC2 PIII you actually get better results if you lower Vcore from the standard 1.60 volts down to 1.45 ... 1.50 to make it run cooler. This can be done by modifying the resistors that set the Geyserville code to the voltage regulator on the MMC2 module.

definetaly the lower the voltage is the way to go on notebooks,however is there a way i can lower voltage on my p111 speedstep module without it reverting back to the slower speed?if that can be done that would surely benefit the battery side of things,thus make the package nicer overall.I got 3 spare tp600 mainboards(all password protected)but atleast i can see if the mainboard boots to error codes atleast,surely thats better as you said than destroying a few mainboards,obviously easier said than done.
heres the latest on p11 mmc fsb mod,apparently i removed r73 on the 300(256) cpu,i+placed the resistor to r71,and 450 mghz it resulted,i ran a high stress burn in test for 40 mins but then it rebooted,the cpu wasn't very hot so i suspect it needs a tad more voltage,is there a way to bump up the vcore on the mmc 2 cpu?i'd say going by the burn in test it needs probably .05 volt increase,then again we know that xp pro is a pig of an os +win2k,me,98 are alot easier on mmc2 cpu loading,i tried the same cpu on the hp +it runs ok (now i'm pondering if heres any way one can mod the id string of the cpu to fool the hp bios into accepting any mmc2 cpu,i modded another 300 cpu(same spec) +it seems to run stroner than than the first cpu i modded,time will tell how they cope with cd burning+playing movies,etc but seems to be ok as far as general running of the basic tasks on hand.also tried the cpu on 108 fsb +it booted to the bios s 500mghz but froze,i undone the tp600e fsb back to original settings so i can evaluate the 366,400,433 mmc2 cpus with 100 fsb,reality wise the 300 02 sereis cpus can be modded to 100 fsb+work,proving them 100%stable is another matter(not that it doesn't seem to be stable in xp,but rather in high stress testing it did reboot on its own accord after a while,i beleive the 266 mmc 2 should go to 100fsb too,albeit only to 400mgz, fact remains a p2 @100fsb is a better than a p2 @66fsb(assuming 100fsb mod is as stable.theres also the 333,but have no such cpu on hand,irrespective i'm rapt in your findings +feedback without legends like you id be running a 433 celeron on my tp at the most.

[Sharedoc]

In PII-MMC2-modules

http://www.cpu-collection.de/?tn=1&l0=c ... n-diecache

http://www.cpu-collection.de/?l0=i&i=1680&s=big&tb=1&n=

Vcore can be adjusted by resistors R17 and R18, whose ratio defines Vcore. Standard Vcore is 1.65 volts. Example: if add 33kohm resistor in parallel to R18 you get Vcore=1.58volts. If you want to increase Vcore, add a 33k resistor in parallel to R17.

In newer PIII MMC2 modules there is binary coded Vcore setting (Geyserville coding). Vcore is regulated by ADP3420 regulator. You can find the voltage coding from the data sheet

http://www.analog.com/UploadedFiles/Dat ... 3421_a.pdf

Follow the signal lines from the circuit, and you can find resistors/vacant resistor slots on the top side of the MMC2 module. You can remove/add resistors and change the Vcore. Use a digital voltmeter to check the Vcore.

[beatoem]

In PII-MMC2

Vcore can be adjusted by resistors R17 and R18, whose ratio defines Vcore. Standard Vcore is 1.65 volts. Example: if add 33kohm resistor in parallel to R18 you get Vcore=1.58volts. If you want to increase Vcore, add a 33k resistor in parallel to R17.

In newer PIII MMC2 modules there is binary coded Vcore setting (Geyserville coding). Vcore is regulated by ADP3420 regulator. You can find the voltage coding from the data sheet

http://www.analog.com/UploadedFiles/Dat ... 3421_a.pdf

Follow the signal lines from the circuit, and you can find resistors/vacant resistor slots on the top side of the MMC2 module. You can remove/add resistors and change the Vcore. Use a digital voltmeter to check the Vcore.

basically this little project is for the p11 punters whom would like to upgrade mmc2 cpu on the tp 600e laptops but without bios mods+xtra software required to enable the cache,also it helps for those notebooks that can't identify the faster cpu id strings. in that case modifying a cpu thats id string is supported in the notbooks bios will fool the notebook to run anyway (assuming the cpu is capable in the first place)i had a celeron running at 466mghz (66fb)no bios mod+128kb cache fully stable on the tp600,conversely i would like see 100fsb+ can be acheived with the p2 (66fsb)cpus,like pentium 3/celerom coppermine sereis mmc2 cpus(100fsb+.we already know the tp600e/770z have unlimited boundrys as far as really perking up performance thanks to sharedocs ingeluarty.whilst i'm at the preliminary stages i beleive stability can be enhanced if i can find optimum vcore voltage to run the p2 300/450 cpu @100fsb,thanks for your input,very much appreciate sharedoc
Yes, i can see the 2 slots in r17 are vacant,conversely r18 has a 700ohm resistor affixed that being the case ,i affix a 33 kohm resistor
to the vacant r17 holes +that should yeild me approx 1.7-1.725 volts?
if it is a 33k ohm resistor i have to go to tandy +gets me some 33kohm resistors first.also where can i measure the vcore of the cpu from?is it cr1+ and earth?you speak good english for a fin+have sound knowledge in electronic circuits surely you must be an electronics engineer

[Sharedoc]

http://www.cpu-collection.de/?l0=i&i=1680&s=big&tb=1&n=

Vcore can be measured over C33

[sarkeizen]

Has anyone given thought to attempting to put a uPGA socket on the MMC?