T60 hardware mod to undervolt CPU below 0.95V [updated +photos]

[ziemeck]

I have searched for any technical details regarding possibilities of making hardware mod on T60 to undervolt CPU below 0.95V, but failed. Apparently the blog or forum that hosted description of that mod is long gone. I think this forum is visited by quite a lot of hardware-competent people. I only know for sure that you can go down to 0.85V with a hardware mod, but I am not even sure if a bios mod is also required. If anybody could help me with any detail I would be very obliged.
-------
EDITED:

The hardware mod as far as I could research involves connecting 13th pin (PGOOD2) of MAXIM 1540E as displayed on the pictures here: https://www.dropbox.com/s/9rmiv9ux8xi4x ... d.zip?dl=1

The mod reduces minimum voltage to 0.85V and maximum voltage to 1.1375V (while standard range is 0.95 to 1.2V). It seems that after the mod you can get T7400 to oparate 100% stable at 2.166GHz/x13/0.9750V (vs standard stable minimum 1.0375V).

Backstory: I have purchased a T60 for parts in a really pitiful state (broken bottom chassis including frame and left hinge). I did not expect the motherboard survived, judging the state of frame, but it did. I have noticed a bridge was made, but thought it was some repair. After I mounted the motherboard into different chassis and run NHC i was stunned - completly non-standard voltages were available. The previous owner purchased the machine as used and had no idea the motherboard was repaired/modded - and has no longer any contact with the person he purchased this machine from.

Questions:
1. I have thoroughly inspected the motherboard and found no other bridges/ alterations on it. Is the bridge visible on the photos all that is needed to cause observed voltage range alteration?
2. Is there any risk that the presented bridge ALONE might damage motherboard - or it is safe to try it on a spare motherboard?
___________________
EDITED:
Here are photos of this T60 while running NHC, CPU-Z and Throttlestop 6.0 at 0.975V/x13/2.16 GHZ (full CPU load) and at 0.85V/x6/998Mhz:
https://www.dropbox.com/s/07a3hvw0syjr7 ... 6.zip?dl=1

[Screamer]

I don't see how would a BIOS modification extend the processor's voltage range, at least on laptops that is.

By the way, do you have a Merom or a Yonah? If you have a Merom in your T60, take a look at the voltage identification definition range from this datasheet. If you have a Yonah in your T60, use this datasheet instead.

Before I get into the details of performing this modification (a very thin strand of magnet wire is required), could you tell me what is your stock high frequency mode voltage for your processor? That would be good to know, since I do not want to guess the wrong voltage.

EDIT: Nevermind, I guess that is not needed.

Alright, I am going to assume that the stock high frequency mode voltage of your processor is 1.2V. This means a VID6-VID5-VID4-VID3-VID2-VID1-VID0 combination of 0-0-1-1-0-0-0, in respective order.

I would suggest to keep the processor's stock voltage intact, since you would need it to boot up in the first place. I don't think that would be stable or advisable to hold an undervolt in the boot-up stage.

First of all, you need a very short 32/36/38 gauge magnet wire, just to be safe. I would also recommend to get a pair of tweezers, to reduce the chances of inhaling or dropping the wire (which is somewhat difficult to pick up). I had inhaled a wire by accident before, so I would strongly advise you to get a pair of tweezers.

If you want a voltage range of 1.2V-0.75V, connect pin VID4 to the closest Vcc pin that you can find on the socket. Be careful though, you must not let the magnet wire touch any of the surrounding pins.

Code: Select all

===========VID4 is always high=========
VID6-VID5-VID4-VID3-VID2-VID1-VID0
[0]--[0]--[1]--[1]--[0]--[0]--[0] -> 1.2000V
[0]--[0]--[1]--[1]--[0]--[0]--[1] -> 1.1875V
[0]--[0]--[1]--[1]--[0]--[1]--[0] -> 1.1750V
[0]--[0]--[1]--[1]--[0]--[1]--[1] -> 1.1625V
[0]--[0]--[1]--[1]--[1]--[0]--[0] -> 1.1500V
[0]--[0]--[1]--[1]--[1]--[0]--[1] -> 1.1375V
[0]--[0]--[1]--[1]--[1]--[1]--[0] -> 1.1250V
[0]--[0]--[1]--[1]--[1]--[1]--[1] -> 1.1125V
[0]--[1]--[1]--[0]--[0]--[0]--[0] -> 0.9000V
[0]--[1]--[1]--[0]--[0]--[0]--[1] -> 0.8875V
[0]--[1]--[1]--[0]--[0]--[1]--[0] -> 0.8750V
[0]--[1]--[1]--[0]--[0]--[1]--[1] -> 0.8625V
[0]--[1]--[1]--[0]--[1]--[0]--[0] -> 0.8500V
[0]--[1]--[1]--[0]--[1]--[0]--[1] -> 0.8375V
[0]--[1]--[1]--[0]--[1]--[1]--[0] -> 0.8250V
[0]--[1]--[1]--[0]--[1]--[1]--[1] -> 0.8125V
[0]--[1]--[1]--[1]--[0]--[0]--[0] -> 0.8000V
[0]--[1]--[1]--[1]--[0]--[0]--[1] -> 0.7875V
[0]--[1]--[1]--[1]--[0]--[1]--[0] -> 0.7750V
[0]--[1]--[1]--[1]--[0]--[1]--[1] -> 0.7625V
[0]--[1]--[1]--[1]--[1]--[0]--[0] -> 0.7500V

This simple modification forces pin VID4 to be always high, which would affect the other VID combinations that sets pin VID4 low. In reality, this means that VID combinations from 1.1V to 0.9125V would become unusable, due to pin VID4 being permanently set to high. Also, please take into account that this modification will not be detected by any program that reads your processor's voltages. They are only limited to reading Intel's specified VID range, which means that you would have to remember which VID combination has been modified.

While there are programs that can read the actual voltage input that is being fed to the processor, the T60's SuperIO chip unfortunately does not seem to have it enabled (or it was non-existent to begin with).

[ziemeck]

Thank you for reply, Screamer. I can see you're pretty knowledgeable regarding pinmods. I have extended opening post to contain all the information I could supply. This required me to completly dissassemble my T60 to make photos of the reverse side of the motherboard. I have also made photos of screen with opened NHC voltage ranges for CPU.
As you can see the hardware mod seems to rely solely on bridging MAXIM 1540E 13th pin to some resistor

I don't see how would a BIOS modification extend the processor's voltage range, at least on laptops that is.

Neither do I.

Also, please take into account that this modification will not be detected by any program that reads your processor's voltages. They are only limited to reading Intel's specified VID range, which means that you would have to remember which VID combination has been modified.

While there are programs that can read the actual voltage input that is being fed to the processor, the T60's SuperIO chip unfortunately does not seem to have it enabled (or it was non-existent to begin with).

This makes me wonder - is the mod I presented above doing anything with voltages or just some voltage monitoring circuit is getting fooled? Any ideas?

[Screamer]

No problem, and that modification was actually a quick fix that Lenovo had came up with. That was a long time ago though, so that T60 of yours may be an early production model.

I don't think that it is messing with the voltage input, at least it should not be doing something such as that.

Could you run Throttlestop to confirm that the voltage selection has changed? I have a feeling that NHC might be inaccurate.

[ziemeck]

Thank you for explaining the "green wire mystery".

Could you run Throttlestop to confirm that the voltage selection has changed? I have a feeling that NHC might be inaccurate.

I could not run Throttlestop above 6.0 (8.3 gives "unsupported CPU" error, 8.7 "not valid win32 application" error). 6.0 gives (in the VID field) exactly the same readouts NHC does.

I have run also CPU-Z (1.87.0.x32) and minimum voltage reported is 0.85V as in NHC, also CPU 2.16/x13 operating voltage is same as reported by NHC: 0.975V. Additionaly CPU-Z reports Max VID 1.138 V (max voltage reported by NHC was 1.1375 V, so my guess it is just a round-up). I can upload photos as proof if you wish.

The voltage range shift might be a side-effect of the "green wire patch", but to confim that another patched motherboard is necessary it seems. Whether such a patch applied on later revisions of these motherboards (my is X1400 41v9920 USI-SZN, but I have a spare motherboard with identical FRU, but no patch applied - this one reports "standard" voltage range) could produce similar effect is an open question.

[Screamer]

Strange, Throttlestop 8.50 did happen to work on my Core Duo T2700. I don't see why would a Core 2 Duo T7400 be considered an 'unsupported CPU', especially if a Core Duo T2700 is still in the list of 'supported' processors for Throttlestop.

I see, and thanks, but there is no need for proof. CPU-Z reporting that voltage alone, is good enough for proof.

Indeed, there is a possibility that it could be. Although, finding such a motherboard with the same 'fix' applied is not going to be easy. I don't think that there were a lot of motherboards, that had the same 'fix' applied to them.

[Lister]

This is very interesting, I would say there is another possibility to change the hardware voltage of the processor. It looks similar to adjusting to increase the power of the processor described in this thread https://thinkpad-forum.de/threads/13201 ... et-und-T61 on the German forum. You haven't found an interrupted route on the board like at 1066fsb mod http://abload.de/img/266-fsbe2cow.jpg? Thanks.

[ziemeck]

Strange, Throttlestop 8.50 did happen to work on my Core Duo T2700. I don't see why would a Core 2 Duo T7400 be considered an 'unsupported CPU', especially if a Core Duo T2700 is still in the list of 'supported' processors for Throttlestop.

I see, and thanks, but there is no need for proof. CPU-Z reporting that voltage alone, is good enough for proof.

Indeed, there is a possibility that it could be. Although, finding such a motherboard with the same 'fix' applied is not going to be easy. I don't think that there were a lot of motherboards, that had the same 'fix' applied to them.

I am using 32bit WindowsXP , maybe Throttlestop dropped support at some point after 6.0. At the bottom I link photos of CPU-Z, ThrottleStop6.0 and NHC altogether.

It looks similar to adjusting to increase the power of the processor described in this thread https://thinkpad-forum.de/threads/13201 ... et-und-T61 on the German forum. You haven't found an interrupted route on the board like at 1066fsb mod http://abload.de/img/266-fsbe2cow.jpg? Thanks.

I knew this mod you refer to, but as I said, I was looking for any modifications, inspected soldered capacitors and resistors and aside from green wire there were no modifications.

Screamer, Lister, please note: the motherboard underwent some serious shock, the frame and chassis were broken so one could assume it is possible some solders could have broken and do not contact. This is impossible to find though. I have tested this motherboard extensively and it is 100% stable at reported voltages no matter how long stressed (to stress I used looping 2x fullHD movies in 2 instances of VLC, at 2.16Ghz it run for 12 hours till I gave up). If I will muster some courage I will try to replicate the green wire "fix" on the other motherboard I have.

In the meantime- here are photos of this T60 while running NHC, CPU-Z and Throttlestop 6.0 at 0.975V/x13/2.16 GHZ (full CPU load) and at 0.85V/x6/998Mhz:
https://www.dropbox.com/s/07a3hvw0syjr7 ... 6.zip?dl=1

[Lister]

You tried a processor in another board, if the works with a lower voltage (0.85-0.95)? Can't it be a better processor that handles these lower voltages? I use IBM_ECW to undervolt a processor, i think this is better than Throttlestop or NHC.

[Screamer]

Very interesting results, but I recall that Throttlestop 8.50 was still supported under Windows XP with Service Pack 3 and 2000 with Service Pack 4. In fact, I am currently using a laptop with Windows XP Professional Service Pack 3 that is capable of running Throttlestop 8.50. I believe there may be something off with your installation, if this is the case.

Serious shock? It can't be too severe for a component to drop off, right?

100% stable? Could I suggest running Prime95 v29.4b7 on 'Small FFTs' till it catches an error?

[ziemeck]

Very interesting results, but I recall that Throttlestop 8.50 was still supported under Windows XP with Service Pack 3 and 2000 with Service Pack 4. In fact, I am currently using a laptop with Windows XP Professional Service Pack 3 that is capable of running Throttlestop 8.50. I believe there may be something off with your installation, if this is the case.

Nope, the whole trick is you used it on Penryn machine. It seems Penryn is the oldest supported CPU for versions above 6.0. I have verified that on a Merom T61 running Vista Business.

Serious shock? It can't be too severe for a component to drop off, right?

But some solder(s) might have cracked ...

S100% stable? Could I suggest running Prime95 v29.4b7 on 'Small FFTs' till it catches an error?

Ok, will do it someday.

Should anybody be interested - on the revision of the motherboard that had the patch done there's no connection to 13th pin (PGOOD2) of MAXIM 1540E except for the "green wire". I have reapplied the "green wire patch" on the other x1400 motherboard I had, but it was a later revision, with 13th pin connected to motherboards circiutry. There was no change of the reported (NHC) voltages, so it seems one must sever the circuit exiting from PGOOD2 and then MAYBE the "green wire patch" MIGHT work. Luckily there's a place where both cutting the path and resoldering it should be easy, so at least testing does not involve irreversible damage to the MB.

[Screamer]

That does not make any sense, because any derivative of Yonah worked with Throttlestop 8.50. I tested it ages ago, when I still had a Socket M laptop. Furthermore, there is also this image that shows Throttlestop 8.70 working with a Merom.

I suppose that you may be right, considering that some heavy impacts do shatter/damage solder joints at times.

[ziemeck]

That does not make any sense, because any derivative of Yonah worked with Throttlestop 8.50. I tested it ages ago, when I still had a Socket M laptop. Furthermore, there is also this image that shows Throttlestop 8.70 working with a Merom.

No matter what "makes sense" according to you it is impossible to run Throttlestop above 6.0 on any of my Merom machines, no matter which x32 OS I am using, WinXP or Vista Business. None of screenshots you supplied comes from XP. This debate on Throttlestop is immaterial anyways since what really mattered - voltage ranges - were confirmed. If you are that persistent on Throttlestop, I can supply you with video documenting that on my T60 with x32 WinXP updated up to "end of support" message no Throttlestop above version 6.0 is able to run. Unfortunalely you are not able to provide me with any video proving you can run Throttlestop above 6.0 on any T60 machine, so let's consider it QED.

[Screamer]

Seeing as this would veer off-topic in any second if you or I were to continue this 'debate', then I am inclined to agree with you that it is QED.

I was only merely trying to tell you that something was strangely off with your current installation, but as said before, you had already gotten what you needed. My apologies for focusing on something that is clearly unimportant in this topic.