Another X61T/s overclocking thread: BSEL + Full VRM

[wileE]

Some of you may have seen the threads on 51nb about overclocking with simply using a Dell WLAN card DW1506 or 1705.

On the german forum user sephi23 found what causes the OC with the WLAN cards.

The Dell cards bridge pin 24 (3.3Vaux) and pin 52 (3.3V). Lenovo WLAN cards do not bridge these pins.

So you can easily do the OC with a wire bridge on the mainboard or on a WLAN card. Such a card is also nice for testing to find a board that works well with the OC.

670KB pic warning: http://imgur.com/Rp3cZry

[xiphmont]

Some of you my have seen the threads on 51nb about overclocking with simply using a Dell WLAN card DW1506 or 1705.

On the german forum user sephi23 found what causes the OC with the WLAN cards.

The Dell cards bridge pin 24 (3.3Vaux) and pin 52 (3.3V). Lenovo WLAN cards do not bridge these pins.

So you can easily do the OC with a wire bridge on the mainboard or on a WLAN card. Such a card is also nice for testing to find a board that works well with the OC.

670KB pic warning: http://imgur.com/Rp3cZry

What an interesting and wonderfully alarming approach!

Let's see... Looking at the schematics, what's happening on the motherboard is that the main 3.3V rail is being bridged to the 3AMT rail, which normally is brought up partway through the power-on process after the clock select lines are already being asserted. When it's bridged, 3AMT is powered immediately and the clockgen chip boots early before the clock select lines are driven. Since the clockgen only samples the clock request at power-on, it reads 'nothing' at the three BSEL pins which typically becomes 0,0,0, corresponding to a 266MHz FSB which the clockgen maintains even after the BSEL lines change later. I think I have that all correct.

I wonder if this technique is prone to the same 'undriven is not always the same as zero' problem as simply cutting the BSEL1 trace. I've never seen it mentioned anywhere, but a few of the motherboards I had read the cut trace as a '1' some of the time, not '0'. I had to actually pull the floating line down for the mod to be reliable.

...I suppose I could test it as I still have those motherboards

[wileE]

Nice too see that somebody seems to understand what going on on the board.

The OC with the pin 24 - 52 bridge does not work on boards with the cut trace. Tried that on two of mine that have the switch to turn off the OC.

[xiphmont]

The OC with the pin 24 - 52 bridge does not work on boards with the cut trace. Tried that on two of mine that have the switch to turn off the OC.

Oh, that's good actually. It suggests the lines are actively pulled low at that point, increasing the liklihood the bridge hack is reliable.

[DaKKS]

Ordered a 1705 for my x61s. Needed a card anyway. Lets see if it works. If it doesnt or its not stable i can just swap with my other laptop, no biggie.