PLL Pinmod Overclocking Methods and Examples

[Andreche]

Hey...
I have BSEL pinmoded an Intel T2400 from 166 --> 133 (the 166 --> 200 wont boot, 945GM chipset) just to see if the multiplier looked and it didn't. So, if by any case I would be able to mod the Bios to accept the 200 MHz, will it lock the multiplier then? I mean, if the downclock didn't lock it, will the overclock lock it?

[timohour]

Hey...
I have BSEL pinmoded an Intel T2400 from 166 --> 133 (the 166 --> 200 wont boot, 945GM chipset) just to see if the multiplier looked and it didn't.

It is probably not a bios issue. and yes. It is not on the first page but I have done a research on the matter and for 945 and 965 (I am going to check soon and 45) downclock pinmod does not affect the multiplier. Only overclock does..

So, if by any case I would be able to mod the Bios to accept the 200 MHz, will it lock the multiplier then? I mean, if the downclock didn't lock it, will the overclock lock it?

try to give some more voltage to your t2400. also what's your pll and on which laptop. (you can add your laptop to your signature so it will be easier to help)

[naton]

That's true. I just pll pinmod oc my T4400 from 200 to 266 @ 2.93Ghz.
It is way cooler (tops @ 69 C) than my t9300 @ 266 and very energy efficient. (No extra voltage needed)
I think it is the R0 stepping that does this. I am gonna keep this and use the t9300 only for benches

Can you explain how did you do the PLL mod. Did you remove the resistor completely? or just unsoldered one of its ends? What tools did you use for the mod?

By the way thanks for the info regarding downcloking via pin mod. What CPU did you use for your downclock experiments?

[timohour]

Can you explain how did you do the PLL mod. Did you remove the resistor completely? or just unsoldered one of its ends? What tools did you use for the mod?

I removed every resistor leading to fslb and pulled fslb to Gnd with a small IDE cable. A screw driver and a soldering iron.

By the way thanks for the info regarding downcloking via pin mod. What CPU did you use for your downclock experiments?

Tested on PM945 with t7600
Tested on GM965 with t5450 from 166 to 133.
______________ with t8100 from 200 to 166.
______________ with QZOL from 200 to 166.
Test on progress on GM45 with t4400 from 200 to 166.

[naton]

I removed every resistor leading to fslb and pulled fslb to Gnd with a small IDE cable. A screw driver and a soldering iron.

Once the resistors removed, does the CPU still boot at its nominal FSB? for instance will a T4400 boot with FSB 200 or 266?

In your case can the mod be removed?

being able to pin mod to downclock is very interesting since one can have his own home made UVL CPU

[timohour]

Once the resistors removed, does the CPU still boot at its nominal FSB? for instance will a T4400 boot with FSB 200 or 266?

In your case can the mod be removed?

The result depends on the laptop. For example, when I did the mod on my Fsc U9200 nothing happened. This time (lenovo Y550), when I removed the 2 resistors CPU locked to its lower multiplier @ 200mhz BCLK! If I want to undo the mod I have to reconnect the resistors I removed. I still don't understand why this happens but I don't really care. However when I tried to remove resistors leading to fsla it didn't boot! This is probably caused because it is used also for a USB signal needed on boot. However there must be a work around. I can't probably go further because thaiphoon burner doesn't flash my rams so I didn't try it.

being able to pin mod to downclock is very interesting since one can have his own home made UVL CPU

I disagree with you on this point. In my opinion the higher the BCLK the better the power consumption. You can very easily make an under volt pinmod and use Throttlestop to avoid using higher multipliers. However it can be used by users who want to oc pci-e bus. They maybe can underclock the CPU and then do the pll pinmod to get back to previous frequency. I didn't test this feature yet but I am soon gonna try since I need the extra pci-e bandwidth for my eGPU.

[ziddey]

I'm looking to bring up the fsb from 133 to 200 on my compaq c762nr (c700 series for search indexing purposes), bringing my t2370 from 1.73 to 2.6.

I have 2 different sticks of ram, both rated for 333 5-5-5 operation. Hopefully, it will be able to run at 400 6-6-6 or better.

It uses the slg8sp512t:





Uploaded with ImageShack.us

The frequency select table is the same as posted on the first post:

CBA
001 133fsb
010 200fsb
011 166fsb

I figure I'll get to 166fsb, so I can change the spd of my ram (runs 266 4-4-4, so hoping it can run 333 4-4-4..)

I see fs_b is shared with test_mode. Did I read right that 3.3V would put the pll into test mode, and I need ~1V?

I'm hoping to do this without lifting any legs on the pll.

Sorry for the lousy picture, but if you can follow it, it looks like I'm removing at least the first resistor (1kohm). The second one seems to have no resistance. I suppose I could cut the trace on the pcb at the double silkscreen lines, and always be able to reconnect it on the surface if need be. That'd allow me to simply use the first test point to obtain 1V somehow.

The resistor connected to fs_a seems to be 40ohm, and the one directly next to it seems to be the only one connected to fs_a. I may be wrong though; I am using a $2 multimeter.

Anyone have any suggestions on the best way to approach this without lifting legs? Will the fact that fs_a is shared with usb be a problem?

I can try to take a better picture if it'd help.

Thanks

[naton]

I disagree with you on this point. In my opinion the higher the BCLK the better the power consumption. You can very easily make an under volt pinmod and use Throttlestop to avoid using higher multipliers. However it can be used by users who want to oc pci-e bus. They maybe can underclock the CPU and then do the pll pinmod to get back to previous frequency. I didn't test this feature yet but I am soon gonna try since I need the extra pci-e bandwidth for my eGPU.

I did undervolt via pin mod a T5300 and a T2060 a couple of years ago. Because of it's low FSB (i.e. 133), the T5300 was stable with the following voltage:
x06 ---> 0.7500v
x07 ---> 0.7500v
x08 ---> 0.7625v
x09 ---> 0.8000v
x10 ---> 0.8375v
x11 ---> 0.8750v
x12 ---> 0.9000v
x13 ---> 1.1250v
I had to go high with x13 because due to the mod I lost access to all the VIDs between 0.9000v and 1.1250v.
I had a Pentium M740 (FSB133) running on a laptop with an Intel i855 chipset. Since this chipset supports FSB100 only, the M740 was automatically downclocked from 1.73GHz to 1.3Ghz. As I result I was able to set all the VIDs in RMClock to below 0.9v. The CPU was so cool that the CPU fan never kicked in. In a sens I created a ULV CPU.

Based on the above I assumed that:
1- dropping the FSB via pin mod will allow enable the CPU to run stable at lower voltages.
2- a CPU that uses lower voltages is a ULV CPU.
3- Lower VIDs means a lower DTP.
4- Lower DTP means the fan will spin less or not at all.
5- Fan that does not spin often or not at all means power saving.
6- Lower VIDs means a CPU with a lower power consumption.
7- 5 and 6 means a better battery live.

Please feel free to correct me on the above if I'm wrong

Back to the PLL mod.
I'm slightly confused. I see that some (like yourself) removed the resistors and made a connection with a wire from the resistor pad to the ground, when others (such as nando4) connected the pad to another resistor before connecting it to the ground.
Why two techniques? and what's the difference between them?

[User Retired 2]

Back to the PLL mod.
I'm slightly confused. I see that some (like yourself) removed the resistors and made a connection with a wire from where the resistor pad to the ground, when others (such as nando4) connected the pad to another resistor before connecting it to the ground.
Why two techniques? and what's the difference between them?

I always recommend a resistor since if the PIN is a high (1) AND doesn't have any internal protective resistors, running it directly to GND would create a shortcircuit.

[timohour]

@ ziddey
You better start the 133-->166 by removing the leading resistor and pull to a VDD point (3.3V) with a small (1k) resistor. After that you should remove the fsa leading resistor and check if it boots. pls make sure that you provide enough voltage.
Good lack!!

I did undervolt via pin mod a T5300 and a T2060 a couple of years ago. Because of it's low FSB (i.e. 133), the T5300 was stable with the following voltage:
x06 ---> 0.7500v
x07 ---> 0.7500v
x08 ---> 0.7625v
x09 ---> 0.8000v
x10 ---> 0.8375v
x11 ---> 0.8750v
x12 ---> 0.9000v
x13 ---> 1.1250v
I had to go high with x13 because due to the mod I lost access to all the VIDs between 0.9000v and 1.1250v.
I had a Pentium M740 (FSB133) running on a laptop with an Intel i855 chipset. Since this chipset supports FSB100 only, the M740 was automatically downclocked from 1.73GHz to 1.3Ghz. As I result I was able to set all the VIDs in RMClock to below 0.9v. The CPU was so cool that the CPU fan never kicked in. In a sens I created a ULV CPU.

I used to have a T5450 10x166 @ 1.6GHz. I was able to do a OC pinmod on a gm965 chipset causing multiplier to lock @ 6x. This way I had a 6x200 @ 1.2GHz, running at lower voltage than the undervolted 6x166 @ 1.0ghz. I never however tested undervolting t5450 10x133 @ 1.33GHz so I am not able to make a compare between the three.
You are probably right.

Back to the PLL mod.
I'm slightly confused. I see that some (like yourself) removed the resistors and made a connection with a wire from the resistor pad to the ground, when others (such as nando4) connected the pad to another resistor before connecting it to the ground.
Why two techniques? and what's the difference between them?

They are not two techniques. The result is the same since you have a low signal to your fsl_b. However the orthodox technique is to use a resistor. I couldn't use a resistor because I was careless removing the on board resistors and there wasn't enough room to connect a resistor. Anyway the physical impedance an small resistor and a small IDE cable is almost the same.