HP Elitebook 8570W - is it still good ? Hardware modifications

[GuitarG]

18 minutes ago, Daedalus said:

 not to be difficult, UK users cant see IMGUR pictures, so all i see is a blue block.  Awesome progress though. Shame the 8570W journey ends. 

Did anyone uncover the HP LVDS cable pinout diagram? i seem to recall its in an older post (i'll go hunt for it.  Happy to try recreate the cable, just need to get a bundle of LDVS to sacrifice 😄

 

[Daedalus]

On 8/17/2026 at 6:15 PM, GuitarG said:

 

am i assuming from the diagram above no interposer board is needed? just looking at sources for the cables at present. LVDS elitebook ones seem viable on ebay. is there any IPS displays preferred that fit the 8570W lid best (just to focus searches)
Also has anyone got the interposer dreamcolor board cable assembly part number too?

@GuitarG i salute your devotion to the cause and effort so far, a shame the 8570W is retired from your flock.  Enjoy the next challenge 😉

[FrancoTecno]

On 11/17/2025 at 3:30 PM, Vladka76 said:

Interesting. The FBGA code "D9RVX" on your chips means the model name MT41K512M8RG-107:N, officially also 1866 (from the datasheet: the "-107" marking = option "1.07ns@CL=13 (DDR3-1866)"). This chip's frequency is the same as one I've encountered, but the FBGA code on those I've seen was slightly different from this one.
What's preventing good memory chips from running at full speed? Perhaps poor-quality resistor arrays on the data buses (all my original modules have ceramic, white resistor arrays)? Or the firmware on the SPD flash IC? It's sad. This is often the case with the Chinese – they copy almost everything well, but somewhere there's always a flaw that ruins everything...

Another my photo. 🙂 Compare the quality of the sticker on the other side of the two modules (original on top, counterfeit on the bottom). The original has an aluminum base and a black ink printing, while the counterfeit has a paper base and white ink with a low-resolution print. And a resistor arrays are different in color and material...

i leave the official PDF of those DRAM family chips, the -107 revision (D9RVX) are rated for 1866MHz at 13-13-13, a jedec profile that matches with the DRAM oficial specificacion isnt specified in the SPD of these fake DIMMs, so probably you need to adjust primary timings and secondary ones manually to make them working at 1866, right now i managed to make them working at 2133MHz, but i need more research into the official PDF to make a fitable timing table at 2133 according with the oficial specifications of the DRAM modules to avoid stability issues

micron4Gb_DDR3L.pdf

[FrancoTecno]

the SPD data on these chinese modules is pretty uncertain, because the reported latency in miliseconds contained in the SPD is lower than the official specificacion by micron in that PDF... even the jedec profiles are very uncertain because they dont have a 1866MHz jedec profile when the DRAM chips are officially of 1866MHz

[FrancoTecno]

my best now.. i want to increase the dram voltage a little bit to go up to 2400 but i dont find any type of RAM voltage control in my unlocked bios, anyway 2200MHz is very well for DDR3

[MODDER1]

Hi, 

 

the system i worked on so far, "working fine" i would say, its a pure copper plate ajusted to the components of the card, i did dig some cavities for "R22" components in it to allow  a  direct and good contact between the plate and the GPU die of the 980M. At idle i hit 32-33° for the GPU and 46-47° for hot spot with PTM7950, the plate and heatpipe being cool i would say, but then under load the temp continue to spike and hit the 92° limit and the card start to throttle. I think may be its the heatpipes, that i don't cool effectively underload due to a bad alignement with the fins, or not enough heatpipes? And what's happening is that the heat is not pushed outside quick and effectively enough maybe? What you guys think the improvement is there ? 

 

Also, can someone confirm that the hotspot on the GPU is also at the level of the Die? Thanks

[Daedalus]

Nice work, Not bad.  Thats kinda how i was envisaging my dual fan setup to look like. My version was keeping the fan to GPU pipe but also having the inverse like you have for the GPU to second fan. 
assume all LMP solder joints there. why no primary fan to GPU plate heatpipe? GPU maybe overloading 2nd fans ability to exhaust heat 

one question, why not have the 2nd fan nearer the SSD bay? (assume due to heatpipes not working in flipped over layout. 

Also where is your second fan run from, i see the PWM feed from main fain header, but are you using +5v and gnd from DVD? (as 980M install requires a brutal haircut of the DVD port)
i planned on taking feed from USB sub-board (that black plastic covered board bottom left)

on the underside how is fan intake working? i have a round hole in my DVD bay floor and a grille in my access panel now  🙂 

Regards video output is that via internal display on 980M? which Vbios.
How did you bend your heatpipes?

 

Apologies for the critique. love the progress youve made. 

[MODDER1]

6 hours ago, Daedalus said:

one question, why not have the 2nd fan nearer the SSD bay? (assume due to heatpipes not working in flipped over layout. 

There are also some cables running there, the upper zone is the best i think for fan intake.

 

7 hours ago, Daedalus said:

Also where is your second fan run from, i see the PWM feed from main fain header, but are you using +5v and gnd from DVD? (as 980M install requires a brutal haircut of the DVD port)

So PWM from fan 1, 5V et GND from DVD port pins, and yes it requires a brutal haircut ahah, but pretty simple using hot air station. 

 

7 hours ago, Daedalus said:

on the underside how is fan intake working? i have a round hole in my DVD bay floor and a grille in my access panel now  🙂 

This is the way, i haven't done it yet from my side, but running the laptop without the keyboard for now. 

 

7 hours ago, Daedalus said:

Regards video output is that via internal display on 980M? which Vbios.

Yes, it's an alienware vBIOS i think, not sure though, but running smoothly till now. 

 

7 hours ago, Daedalus said:

How did you bend your heatpipes?

By hands, pretty simple, but you have to be gentle tho. 

 

7 hours ago, Daedalus said:

Apologies for the critique. love the progress youve made. 

Why, don't, that why i'm there for😉, Thanks!

[MODDER1]

7 hours ago, Daedalus said:

why no primary fan to GPU plate heatpipe? GPU maybe overloading 2nd fans ability to exhaust heat 

And yes, i was thinkin like adapting a system for each of the cards, i also saw a pub from GuitarG earlier, who wanted a solution just like that.

Then i also had an observation that the exhaust air was hot, but not hot enough (like it's not really pulling away much heat), so i might investigate on this also.

 

[janepa]

Hi

The very best thing you can do in this case is to close the loop by soldering on shorter heat pipes.

 

As a next step, you could ask someone with a 3D printer to make a simple tube bender, or even buy one and use 5mm heat pipes to bend them as needed and flatten them to the required thickness at the contact (soldering)  points.

 

https://www.ebay.co.uk/itm/283251571555

 

There’s also the option to adjust the fan blades themselves their number and size.

 

The basic idea however is very good. Great work.

[janepa]

 

Hotspot temp. sensors on a GPU chip.

[Daedalus]

11 hours ago, janepa said:

Hi

The very best thing you can do in this case is to close the loop by soldering on shorter heat pipes.

 

As a next step, you could ask someone with a 3D printer to make a simple tube bender, or even buy one and use 5mm heat pipes to bend them as needed and flatten them to the required thickness at the contact (soldering)  points.

 

https://www.ebay.co.uk/itm/283251571555

 

There’s also the option to adjust the fan blades themselves their number and size.

 

The basic idea however is very good. Great work.

if memory recalls i dont think linking the pipes back to GPU copper plate would be efficient, i get the logic of hot running to cold point (fan) though. I always understood heatpipes work best with a temp gradient from one end to other. looping would increase copper mass, you might benefit from two pipes ending up at exhaust grille. just my 2c.

 

Is that a 3d printed fan? i ask as i had sooooo many Ali fans working out which would fit in DVD bay then the headache of supplying different brand fans with HP happy power. 

[janepa]

Yes, years ago I built a similar model with a loop and it worked well, but in the end I decided to go a different way because of easier installation and removal needed.

 

Your idea is also possible of course. There are several options available.

 

No, this is the original fan that was damaged it was missing a few blades it was basically a scrap. Here you have the option to remove the all original blades keep the fan core and attach a newly designed ring with blades shaped as needed.

 

I also had a problem controlling the fans, but I eventually solved it by using my own small control electronics.