• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Notebook enable second fan

L-C

Joined
Oct 3, 2021
Messages
32 (0.04/day)
Hi, I have an HP pavilion 15 with Ryzen 3500u and want to increase the cooling to run the CPU at an higher TDP, which is 15W at stock with a 25W boost.

I already changed the stock heatsink with a double heatpipe one of another model with good results, but now I wonder, since the motherboard have a second fan connector and it seems connected if it's possible to enable and control a second fan. The fan spins briefly at power on but then nothing.

The moderboard uses an ENE KB9028Q C as controller but no datasheet is avaible. The idea was to find the registers that control the second fan and let Notebook FanControl do the job but failed even to find the first fan registers. Also I don't even know if it's doable that way.

Another idea was to route the pwm signal of the first fan to the second, but I couldn't find the pinout of the fan and don't know if the order is standard or not. And I don't know if the controller is able to drive 2 fans.
I have thought to split all the cables but don't know if the 5V can provide enough power, each fan may need up to 0.5A.

Any suggestion is appreciated, I think the pwm splitter is the easier solution but I need to know which cable it is. The software solution is the more flexible since it will be possible to custom program the fan behaviour, but without the datasheet a was not able to do it.
 

Attachments

  • IMG_20210929_234703468~2.jpg
    IMG_20210929_234703468~2.jpg
    485 KB · Views: 272
  • IMG_20210929_220225835~3.jpg
    IMG_20210929_220225835~3.jpg
    1.5 MB · Views: 221
Joined
Aug 28, 2021
Messages
364 (0.39/day)
if it's possible to enable and control a second fan. The fan spins briefly at power on but then nothing.
It’s not impossible, but unless you have schematics it will be difficult to know where the pwm signal is coming from on the board . That second fan is most likely triggered by gpu temperature sensor.. but in your case there no gpu , so it’s not going to work automatically unless you fake that signal on the ene chip input.
to route the pwm signal of the first fan to the second
this is what I would do , it’s will look cleaner and you won’t have to add any circuitry to the board to control the second fan
but I couldn't find the pinout of the fan and don't know if the order is standard or not.
here’s how to do it :

find the ground on the fan connector, using continuity mode , you can find 5v using voltmeter mode ....

Btw on most laptops I’ve worked with pwm was on the blue cable , you might try that out first and save some time
And I don't know if the controller is able to drive 2 fans.
I have thought to split all the cables but don't know if the 5V can provide enough power, each fan may need up to 0.5A.
I’ve done this before too and it worked , but I don’t recommend it , get the power from somewhere else , your laptop may not have enough current available in that connector
 

eidairaman1

The Exiled Airman
Joined
Jul 2, 2007
Messages
40,435 (6.61/day)
Location
Republic of Texas (True Patriot)
System Name PCGOD
Processor AMD FX 8350@ 5.0GHz
Motherboard Asus TUF 990FX Sabertooth R2 2901 Bios
Cooling Scythe Ashura, 2×BitFenix 230mm Spectre Pro LED (Blue,Green), 2x BitFenix 140mm Spectre Pro LED
Memory 16 GB Gskill Ripjaws X 2133 (2400 OC, 10-10-12-20-20, 1T, 1.65V)
Video Card(s) AMD Radeon 290 Sapphire Vapor-X
Storage Samsung 840 Pro 256GB, WD Velociraptor 1TB
Display(s) NEC Multisync LCD 1700V (Display Port Adapter)
Case AeroCool Xpredator Evil Blue Edition
Audio Device(s) Creative Labs Sound Blaster ZxR
Power Supply Seasonic 1250 XM2 Series (XP3)
Mouse Roccat Kone XTD
Keyboard Roccat Ryos MK Pro
Software Windows 7 Pro 64
It probably only kicks in once a certain temperature threshold is reached, that is how everything is to reduce noise and save power.

Hi, I have an HP pavilion 15 with Ryzen 3500u and want to increase the cooling to run the CPU at an higher TDP, which is 15W at stock with a 25W boost.

I already changed the stock heatsink with a double heatpipe one of another model with good results, but now I wonder, since the motherboard have a second fan connector and it seems connected if it's possible to enable and control a second fan. The fan spins briefly at power on but then nothing.

The moderboard uses an ENE KB9028Q C as controller but no datasheet is avaible. The idea was to find the registers that control the second fan and let Notebook FanControl do the job but failed even to find the first fan registers. Also I don't even know if it's doable that way.

Another idea was to route the pwm signal of the first fan to the second, but I couldn't find the pinout of the fan and don't know if the order is standard or not. And I don't know if the controller is able to drive 2 fans.
I have thought to split all the cables but don't know if the 5V can provide enough power, each fan may need up to 0.5A.

Any suggestion is appreciated, I think the pwm splitter is the easier solution but I need to know which cable it is. The software solution is the more flexible since it will be possible to custom program the fan behaviour, but without the datasheet a was not able to do it.
Why did you censor certain parts on the mobile? This motherboard is probably used in more than 1 configuration and probably cools the chip that is near that fan connector (glue logic?)
 

L-C

Joined
Oct 3, 2021
Messages
32 (0.04/day)
Thanks for the replies guys, indeed the other notebook from which I took the spares codes had a GPU (and an Intel CPU, but the heatink mount was the same).
I also tried to overheat the CPU but the second fan didn't started spinning, maybe the fan configuration was supposed to be in the bios or needs the GPU temperature sensor to work.

I painted some part's of the picture because didn't know and didn't checked if there were some unique IDs, like mac address on the WiFi, just in case.

I will try to split the blue cable and see what happens!
 

eidairaman1

The Exiled Airman
Joined
Jul 2, 2007
Messages
40,435 (6.61/day)
Location
Republic of Texas (True Patriot)
System Name PCGOD
Processor AMD FX 8350@ 5.0GHz
Motherboard Asus TUF 990FX Sabertooth R2 2901 Bios
Cooling Scythe Ashura, 2×BitFenix 230mm Spectre Pro LED (Blue,Green), 2x BitFenix 140mm Spectre Pro LED
Memory 16 GB Gskill Ripjaws X 2133 (2400 OC, 10-10-12-20-20, 1T, 1.65V)
Video Card(s) AMD Radeon 290 Sapphire Vapor-X
Storage Samsung 840 Pro 256GB, WD Velociraptor 1TB
Display(s) NEC Multisync LCD 1700V (Display Port Adapter)
Case AeroCool Xpredator Evil Blue Edition
Audio Device(s) Creative Labs Sound Blaster ZxR
Power Supply Seasonic 1250 XM2 Series (XP3)
Mouse Roccat Kone XTD
Keyboard Roccat Ryos MK Pro
Software Windows 7 Pro 64
Thanks for the replies guys, indeed the other notebook from which I took the spares codes had a GPU (and an Intel CPU, but the heatink mount was the same).
I also tried to overheat the CPU but the second fan didn't started spinning, maybe the fan configuration was supposed to be in the bios or needs the GPU temperature sensor to work.

I painted some part's of the picture because didn't know and didn't checked if there were some unique IDs, like mac address on the WiFi, just in case.

I will try to split the blue cable and see what happens!
Mac addresses look like hexadecimal code (base 16) and are only used for comm devices

You may need a thermistor/thermostat and a firmware comparison to a laptop that uses that exact same board but has the fan attached. So on the bottom of the case is there a vent?
 

L-C

Joined
Oct 3, 2021
Messages
32 (0.04/day)
Mac addresses look like hexadecimal code (base 16) and are only used for comm devices

You may need a thermistor/thermostat and a firmware comparison to a laptop that uses that exact same board but has the fan attached. So on the bottom of the case is there a vent?
Unlikely I couldn't find another mother with the exact same board, I only found one with an Intel CPU and is slightly different.

The bottom cover have a small thin and wide area with small holes and a possibly dust filter behind it, so airflow kinda sucks, but the other laptop have the same cover.

Let us know how it goes , good luck :)
Sure, thanks!
 
Joined
Jan 5, 2006
Messages
17,689 (2.66/day)
System Name AlderLake / Laptop
Processor Intel i7 12700K P-Cores @ 5Ghz / Intel i3 7100U
Motherboard Gigabyte Z690 Aorus Master / HP 83A3 (U3E1)
Cooling Noctua NH-U12A 2 fans + Thermal Grizzly Kryonaut Extreme + 5 case fans / Fan
Memory 32GB DDR5 Corsair Dominator Platinum RGB 6000MHz CL36 / 8GB DDR4 HyperX CL13
Video Card(s) MSI RTX 2070 Super Gaming X Trio / Intel HD620
Storage Samsung 980 Pro 1TB + 970 Evo 500GB + 850 Pro 512GB + 860 Evo 1TB x2 / Samsung 256GB M.2 SSD
Display(s) 23.8" Dell S2417DG 165Hz G-Sync 1440p / 14" 1080p IPS Glossy
Case Be quiet! Silent Base 600 - Window / HP Pavilion
Audio Device(s) Panasonic SA-PMX94 / Realtek onboard + B&O speaker system / Harman Kardon Go + Play / Logitech G533
Power Supply Seasonic Focus Plus Gold 750W / Powerbrick
Mouse Logitech MX Anywhere 2 Laser wireless / Logitech M330 wireless
Keyboard RAPOO E9270P Black 5GHz wireless / HP backlit
Software Windows 11 / Windows 10
Benchmark Scores Cinebench R23 (Single Core) 1936 @ stock Cinebench R23 (Multi Core) 23006 @ stock
Seeing the photo of your laptop, I'd add another ram stick to enable dual channel memory.:)
 

L-C

Joined
Oct 3, 2021
Messages
32 (0.04/day)
Seeing the photo of your laptop, I'd add another ram stick to enable dual channel memory.:)
That was the original reason of why I opened it in the first place! :D

I am happy to report a success, now there is a second fanspin! Thanks guys! :D

I am a bit surprised to see how effective it was, maybe as much as putting the double heatpipe headink (I also used liquid metal).
The job is quite crap, under the black square felt and kapton tape the is a mess of coiled wires, but thank god I don't have to do that for a living, and I was more concerned in not bricking the laptop.

At the end I would say that it's worth it, I paid around 45$ for the parts and gave the laptop a boost in responsiveness, a small performance uplift and with lower operating temperature.
It's a bit noiser but it doesn't bother me, since I used to have and overclocked AMD FX inside a small case with a lot of noisy fans, three of which where 3000 rpm noctua industrial.

Now I am at the point where some other component is overheating (maybe the VRMs) because while trying cibebench with 35W TDP 50A currenti limit, I saw the CPU throttle to 400 MHz! So at the end I still should learn how to use nothebook fancontrol, at least I finally found the write register. Maybe blasting the fans at full speed can give me few more watts!
 

Attachments

  • IMG_20211003_230010920.jpg
    IMG_20211003_230010920.jpg
    3.1 MB · Views: 132
Joined
Aug 28, 2021
Messages
364 (0.39/day)
under the black square felt and kapton tape the is a mess of coiled wires
I only see one wire , I assume that’s the pwm signal ?

I also see that you sleeved it with that red plastic which is good , why is it going under the fan ? Did you solder it directly on the pcb under the fan motor ?

So was the pwm coming from the blue cable in your case too ?

And since it’s only one wire ( at least that’s what I can See ) how did you go about 5v and gnd ?

someone else in the future might have these questions so better have them now while you’re active :p

otherwise good job , there’s not much you can do about vrm overheating in a laptop , just take it easy on the power sliders don’t run it on high power for very long, the vrms will degrade faster :)
 

L-C

Joined
Oct 3, 2021
Messages
32 (0.04/day)
The wires come from under the fans because that is how these are made. I did just cut the PWM ones and made an "Y" connection from fan2 to fan1. The fan2 connector was fully functional so no need to find 5V and ground elsewhere.

I just added too much wire to have enough if I messed up something, but luckily the first try was the good one so and I wrapped the excess and hidden it under the kapton tape where there is the black square. Just too lazy to cut and solder again to make it shorter :D but this is why you can't see what I did with the PWM wire on the other fan.
 

Attachments

  • IMG_20211004_014828271.jpg
    IMG_20211004_014828271.jpg
    2.3 MB · Views: 82

L-C

Joined
Oct 3, 2021
Messages
32 (0.04/day)
A small update on the improved cooling of this laptop, since I now was heavily bottlenecked by, I suppose, vrm overheating, I tried to arrange something to cool them.

I recycled the old CPU heatsink and desoldered the finstack so I got the coldplate and the heatpipe. Unfortunately I set the oven on a too high temperature for that heatpipe (somewhere around 160°C) and it had swollen noticeably, i managed to flatten a bit but still was not flat.

Then I put some thermal pads over the vrm and the inductors, had to use 3mm pads on the vrm to reach the height of the inductors so the heat transfer will not be that great. I used 0.5mm pads on the inductors.

Then I positioned the heatsink so that the coldplate will be over the pads mess and the heatpipe would reach the CPU heatsink. I used 1mm pads between the finstack and the heatpipe.

Finally I put 0.5mm pads on the heatpipe over the heatsink, 1mm pad over the other tip of the heatpipe, put some copper foil on the plastic cover and closed the notebook. (In reality I did that 5 times to find tla decent setup)

I didn't expected much from this hackjob but turned out quite better than expected. Before i was running cibebench r23 with Ryzen controller set at 50A vrm current and sometimes during even the first run was throttling down to 400 MHz; now to throttle I need 65A and it throttles to 1600 MHz, 60A looks fine with 23°C room temperature.
In all tests I had a fan blowing under the laptop, but the internal fans are never spinning at 100% because now the CPU can't reach an high enough temperature, I expect slightly better result if someday finally manage to get manual fan control and blow them at full speed (or I could put somewhere a switch that interrupts the pwm signal to the fans).
 

Attachments

  • IMG_20211013_025152595.jpg
    IMG_20211013_025152595.jpg
    3.4 MB · Views: 140
  • IMG_20211013_025201902.jpg
    IMG_20211013_025201902.jpg
    3.5 MB · Views: 95
  • IMG_20211012_225857580.jpg
    IMG_20211012_225857580.jpg
    2.6 MB · Views: 87
  • IMG_20211012_225703681.jpg
    IMG_20211012_225703681.jpg
    3.5 MB · Views: 131
Joined
Aug 28, 2021
Messages
364 (0.39/day)
Nice modd there , you’re really pushing this little laptop to the limits :d , I have a few notes
I put some thermal pads over the vrm and the inductors
This , you can avoid putting pads on the inductors , they’re just a piece of copper wire that doesn’t need cooling, in a vrm (mosfet+capacitor+inductor) the mosfets are the ones that need/benefit from cooling the most , just to give you an idea about how laptop boards are designed, in most cases the motherboard itself is used as a heatsink to cool down the mosfets, ceramic capacitors and other low tdp chips
and it had swollen noticeably, i managed to flatten a bit but still was not flat.
your heat pipes can still be okay , or maybe busted and only acting as a piece of copper, here is how to check how good it is , maybe you don’t have a thermal probe so use your a piece of ice and compare how fast it melts it compared to the other heatpipes

I used 1mm pads between the finstack and the heatpipe.
If you want more and better heat transfer, you need to compress the thermal pads , they performs very poorly without pressure, if you seen a gpu thermal pad before, when you remove the heatsink they are so compressed that you can see the shape of the vram printed on it .... maybe a thermal putty can work better , and for the heatsink you can use thermal glue it’s a lot better than a loose thermal pad , something like this :

bottlenecked by, I suppose, vrm overheating,
Do you use hwinfo ? Maybe it can indicate of the vrms are overheating, if you can’t see a “vrm sensor” keep a close eye on “ambient sensor “ some laptops use that sensor as a vrm/throttling signal
or I could put somewhere a switch that interrupts the pwm signal to the fans
that would be epic lol , call it “gaming mode” and add rgb , I heard it gives 40% FPS boost :laugh:
 

L-C

Joined
Oct 3, 2021
Messages
32 (0.04/day)
Nice modd there , you’re really pushing this little laptop to the limits :d , I have a few notes

This , you can avoid putting pads on the inductors , they’re just a piece of copper wire that doesn’t need cooling, in a vrm (mosfet+capacitor+inductor) the mosfets are the ones that need/benefit from cooling the most , just to give you an idea about how laptop boards are designed, in most cases the motherboard itself is used as a heatsink to cool down the mosfets, ceramic capacitors and other low tdp chips

your heat pipes can still be okay , or maybe busted and only acting as a piece of copper, here is how to check how good it is , maybe you don’t have a thermal probe so use your a piece of ice and compare how fast it melts it compared to the other heatpipes


If you want more and better heat transfer, you need to compress the thermal pads , they performs very poorly without pressure, if you seen a gpu thermal pad before, when you remove the heatsink they are so compressed that you can see the shape of the vram printed on it .... maybe a thermal putty can work better , and for the heatsink you can use thermal glue it’s a lot better than a loose thermal pad , something like this :


Do you use hwinfo ? Maybe it can indicate of the vrms are overheating, if you can’t see a “vrm sensor” keep a close eye on “ambient sensor “ some laptops use that sensor as a vrm/throttling signal

that would be epic lol , call it “gaming mode” and add rgb , I heard it gives 40% FPS boost :laugh:
Thanks! Let me explain why ended up doing what you saw. At first I wanted to use a thinner heatpipe that would have fitted over the mosfet quite nicely, but if you see one of the pictures in the first post you can see that the components are very close each other and I was afraid of shorting something. Also that heatpipe was a long one and almost "U" shaped, it would have ended only over the plastic of one fan without reaching the copper fins.

So the plan B was to use the old heatsink to try to cool the whole area around the mosfet as best as possible. This is why I put some thermal pads over everything. There is some pressure over the pads, the under cover is even a bit bulged. Maybe the thicker 3mm pads over the mosfet are relatively less compressed but I used high quality 12 W/mK and where possible they are 0.5 mm thick, like the ones over the inductors.
I know that the inductors are fine without cooling, but extracting some heat there would help lowering the surrounding area temperature.

I didn't used thermal putty because I had this pads hanging around, and also if you gently squeeze the thicker ones they behave like putty to some extent, just don't overdo or it will break.
I didn't used thermal glue because I didn't wanted a permanent solution and the easier to find have not a great thermal conductivity.

The heatpipe was fine, I tested by melting frost in a freezer, maybe less efficient than before likely it was not a piece of copper.

Regarding throttling, unfortunately hwinfo does not set any flag on this notebook so I can't know for sure what's happening.

The RGB performance increase is tempting, but I hate unnecessary illumination! :roll:
 
Joined
Aug 28, 2021
Messages
364 (0.39/day)
I didn't used thermal glue because I didn't wanted a permanent solution and the easier to find have not a great thermal conductivity
Thermal glue is not permanent, I am talking about the one in the video, not sure about other brands ... regarding conductivity, i found it performs better than thermal pads ( not compressed) even with the very low thermal conductivity , I think because a 10 w/mk pad only gives 1~2 w/mk if it’s not properly compressed, while the glue doesn’t need compression ..... but if you’re happy with what you have ignore my suggestion, it doesn’t make sense to throw away money for very little performance

not sure if the fan shell is metal or plastic but you can add padding here
ED43AAF9-5FEF-40BC-88C9-2C75307A280E.jpeg

Regarding throttling, unfortunately hwinfo does not set any flag on this notebook so I can't know for sure what's happening.
I actually didn’t explain it properly earlier, so hwinfo, you need to keep an eye on the sensors you have and see if throttling starts when a certain temperature is reached, of course it has to be reproduced....


for example I had a dell with U series intel cpu, it didn’t have a “vrm sensor “ but I noticed that when “ambient sensor “ reached 75 degrees Celsius, cpu would slow down to 1.5ghz , i opened the case and blew air on the mobo and that sensor never reached 75 C max was 65 and cpu didn’t throttle again

what I am saying is , don’t look for hwinfo to flag a high temp , you need to correlate behavior to all the sensors and see if anything suspicious happens...

Either way , good luck and enjoy it
 
Last edited:

L-C

Joined
Oct 3, 2021
Messages
32 (0.04/day)
Thermal glue is not permanent, I am talking about the one in the video, not sure about other brands ... regarding conductivity, i found it performs better than thermal pads ( not compressed) even with the very low thermal conductivity , I think because a 10 w/mk pad only gives 1~2 w/mk if it’s not properly compressed, while the glue doesn’t need compression ..... but if you’re happy with what you have ignore my suggestion, it doesn’t make sense to throw away money for very little performance

not sure if the fan shell is metal or plastic but you can add padding here
View attachment 220704

I actually didn’t explain it properly earlier, so hwinfo, you need to keep an eye on the sensors you have and see if throttling starts when a certain temperature is reached, of course it has to be reproduced....


for example I had a dell with U series intel cpu, it didn’t have a “vrm sensor “ but I noticed that when “ambient sensor “ reached 75 degrees Celsius, cpu would slow down to 1.5ghz , i opened the case and blew air on the mobo and that sensor never reached 75 C max was 65 and cpu didn’t throttle again

what I am saying is , don’t look for hwinfo to flag a high temp , you need to correlate behavior to all the sensors and see if anything suspicious happens...

Either way , good luck and enjoy it
Sorry if I looked offensive, didn't meant that, it's my basic English fault (and also writing from a phone) and I appreciated your suggestions, and I am also sorry for misunderstanding how that thermal glue worked!

The pads and the assembly are pressed by the chassis so they are quite in compression, except for the 3mm ones over the MOSFETs which may have been pushed only half a millimeter, maybe a bit more but less than one for sure. I think there isn't much I can do unless putting an heatpipe directly over, but I think I would end up shorting something because the components are so close together.

That side of the fans is plastic, but the heatpipe that I added rests over the copper fins of the heatsink, with pads below. They are also sealing the small gap between the fan exhaust and the finstack.

I'll take a closer look on hwinfo data to try to understand what's happening as you suggested, thanks!
 

L-C

Joined
Oct 3, 2021
Messages
32 (0.04/day)
Just a quick update, yesterday night i did the "turbo fan" mod that is just interrupting with a switch the pwm signal to the fans, so they can spin to their maximum capable speed. I did not have a proper switch that could fit inside, so i removed the sdcard reader that i never used and put behind the chassis opening a normally closed non latching pushbutton switch which i could press by sticking something inside (see the shaming pictures). It's glued so i expect it to detach at some point :laugh:

Temperature wise there was quite an improvement with the obvious downside of the noise, but performance wise there was not that margin. Even if in Ryzen Controller i put stupid high values the limiting factor is now the "TDC VALUE VDD" to which i was already pretty close to the maximum 35 and that can not be altered (probably a modded bios is needed, but i like the notebook not letting me killing it).
The good thing is that now it throttles later, i could run a complete Cinebench R23 run at the stupid settings and with 23.5°C room temperature, or with safer setting i have some temperature margin for the summer. The power drawn by the cpu reported by ryzen controlle during cinebench is between 45-50W and the EDC VALUE VDD reached between 65-70A, while maxing out to 35 the locked TDC VALUE VDD.

I also tried Diablo 2 resurrected that heavily uses the struggling Vega 8 gpu and not much cpu; the gpu was already maxed without any temperature issues, but with the screaming fans the temperature dropped from 70°C to 62°C., so again very good margin for the summer.

All the tests where done with a 120mm usb fan blowing under the chassis.

The next reasonable mod i could think of could be to add a third fan, there is room in the sata bay, and drill some hole on the chassis for a cold air intake, this may give a little bit extra runtime before throttling. There is more than enough space to fit a fan and route the air around the cpu and vrm area, i almost did it with the old notebook fan, but at the end installed an old ssd (for the pagefile, the bundled nvme ssd is not a speed champion). I still might want add the third fan though, if somebody manage to find a way to increase the frequency of the Vega 8 in the Ryzen 3500u, changing the value in ryzen controller simply has no effect.

As a side note, the original 45W power supply was not up to the task anymore, it was getting very warm and could not output enough power under load, the battery was even discharging under cinebench :roll:, so i bought a new 90W original hp power supply. I was also surprised how overly engineered the board was, it was probably cheaper for HP to share as much of the design as possible across many models than use lower specced components when applicable; stock TDP was 15 W, with boosts of 25 and 30 or 35 watts. Also an heatsink meant for an intel notebook was a perfect fit on this amd one, i think that's for the same reason as before. It's awesome when company's optimizations end up favouring the customer! :D
 

Attachments

  • Pushed.jpg
    Pushed.jpg
    997.2 KB · Views: 70
  • Switch.jpg
    Switch.jpg
    940.6 KB · Views: 86
Joined
Aug 28, 2021
Messages
364 (0.39/day)
Temperature wise there was quite an improvement with the obvious downside of the noise, but performance wise there was not that margin.
You’re hitting the limit of the silicon I guess , or maybe not i am not familiar with amd APU overclockling
23.5°C room temperature
but with the screaming fans the temperature dropped from 70°C to 62°C.
62 C is very rare in laptops, I know it’s a low tdp chip but it’s just amazing
The next reasonable mod i could think of could be to add a third fan,
do you think it will make any further improvements?
I guess what you’re trying to achieve with that is replacing the 120mm usb fan , am I right ?

This will be tricky because you’ll need to work on the pressure inside the chassis and make sure the fan you add can overcome the internal pressure and be able to force push air inside .

I am curious which thermal paste you’re using here ? Maybe there is room for improvement there.
i bought a new 90W original hp power supply.
lol , this laptop has evolved into something else , I am curious how much heavier it got ? Two more heatpipes an extra fan , a 90w brick probably 500g more ?
 
  • Like
Reactions: L-C

L-C

Joined
Oct 3, 2021
Messages
32 (0.04/day)
You’re hitting the limit of the silicon I guess , or maybe not i am not familiar with amd APU overclockling
It wasn't a silicon issue because tecnically che apu was not overclocked. I don't have a deep undestanding of it but, like any other recent cpu, the behaviour of the processor can be "programmed" by setting some threshold either on cpu or motherboard side, for example the cpu TDC, cpu temperature, vrm current, vrm temperature, etc... Some of this thresholds can be changed by the sotware Ryzen Controller, so you can let the cpu end up using higher frequency (but still within design limit, which for a ryzen 3500u is 3.7 GHz) if you can cool all the involved components and they don't blow up.

Some of these thresholds can not be changed in Ryzen Controller and for the "TDC VALUE VDD" is one of them, the threshold TDC LIMIT VDD is not avaible in Ryzen Controller, i may need to use Ryzen Adjuster directly, which is the underlying program that changes the settings, or i may need a modded bios, i did not tested it yet since i am already temperature bound and it will be not useful except for some transient loads.
So, even if i would set 100W and 150A limits, eventually the allowed TDC VALUE VDD is maxed out and the cpu frequency wont increase or is lowered. I never saw more than 50W actual power drawn by the cpu.
62 C is very rare in laptops, I know it’s a low tdp chip but it’s just amazing
It is because in Diablo 2 Resurrected the cpu is lightly loaded (maybe 20-25%), it does not need much cpu power. In cinebench the cpu will eventually reach much higher temperature, but around 84°C it will throttle down to 1.6 GHz. It's strange because when i was testing the maximum capabilities of the stock cooler it was throttling no earlier than 88°C, sometimes even higher, probably the throttle point is tdp dependent.
do you think it will make any further improvements?
I guess what you’re trying to achieve with that is replacing the 120mm usb fan , am I right ?

This will be tricky because you’ll need to work on the pressure inside the chassis and make sure the fan you add can overcome the internal pressure and be able to force push air inside .
What i could further is putting another laptop fan inside the laptop where the sata drive would fit and drill holes in the chassis, it will be like a case intake fan. Maybe it will lower a bit the inside ambient temperature and add a tiny bit of airflow. I made a picture to show the hypothetical placement.
The 120mm fan is under a laptop stand and blows air to the bottom of the chassis, i did not made any invasive modifications, externally the laptop is the same as it was brand new (well, not taking into account the scratches :laugh:).
I am curious which thermal paste you’re using here ? Maybe there is room for improvement there.
I used liquid metal, so if i hopefully applied it correctly this should be the best as practical possible.
lol , this laptop has evolved into something else , I am curious how much heavier it got ? Two more heatpipes an extra fan , a 90w brick probably 500g more ?
:roll: Actually that's not as exotic as it sounds, just bought or used stock parts and put them inside, it's not fancier than a laptop designed with higher TDP. Also these 90W bricks are not much bigger and heavier than the 45W one (another 90W one i have from an older laptop is big and heavy), and the added heatsink and fan are thin and light.
 

Attachments

  • extra_fan.jpg
    extra_fan.jpg
    142.9 KB · Views: 65

L-C

Joined
Oct 3, 2021
Messages
32 (0.04/day)
Little UPDATE.

I did some tests to see the actual performance gain i could obtain, and also the drawbacks and risks involved. The default configuration was 15W continuous power draw, 25W long boost, 30W short boost, 35A vrm-current and 45W vrmmax-current. My configurations now is 45W continuous, 50W short boost and the others high enough to be never reached under the tests i did. The tests where done with a 120mm fan blowing under the chassis and the 2 internal fans at maximum speed. Ram is a not so fast 2400Mhz ddr4 in dual channel.

- Cinebench R23 (approximates, they were at least that scores):
default 3000
current 4500 (+50%)

- 3DMark Time Spy (overall / graphics / cpu):
default 755 / 679 / 2082
current 1070 / 957 / 3286 (+41,7% / +41% / +57,8%)

- Passmark performance test (the scores you see on cpubenchmark.net; CPU Mark / CPU single thread / Cross-platform mark):
default 6020 / 1406 / 9670
current 9043 / 2200 / 13576 (+50,2% / +56,5% / +40,4%)

So at the expense of power consumption and noise, this should put the 3500u at a low tier recent laptop cpu in the 15-35W range, it's not a performance champion but i feel using the notebook much "smoother". If i was able to cool it more effectively there is a slight margin, but when the cpu runs unrestricted it dumps 55-62W under cinebench r23 (basically a desktop cpu).

Gpu wise, the gain is there, performance is not good so low resolution and low detail are needed, but now you may play something that before was totally unplayable (like Diablo 2, tested at 50% full hd resolution, maxed out details and fxaa. Lowering the setting makes almost no difference. Went from 16 to 25 fps in town, on average).

On battery those settings are out of question, external overheat is signaled and the cpu goes to 400Mhz, also battery life would had been awful. I use the default settings on battery, but now it runs almost completely fanless. To handle the switch i made a .bat script that check the ac status and calls ryzen adjuster with the appropriate settings. I run the script at user login and on the ac change event.

The problem with using the laptop with the current settings is that they are very aggressive on the components, not much the cpu that should be fine if cooled properly, but the vrms and other power lines. For example, if while charging i run an high cpu load Hwinfo64 shows an increase in the charging power, so maybe some power if flowing where it's not supposed to.
Also i am on the edge of safety limits, my settings are probably fine if i either run high cpu load with low gpu load, or high gpu load with moderate cpu load. Diablo 2 or Unigine Heaven can run because are not hard on the cpu, too much of a gpu bottleneck; i could run the avx stress test on cpu-z while running the light gpu load that gpu-z does for reading the pci express link speed. But for example running Unigine Heaven and starting the cpu-z avx stress test results in an immediate shut down, so totally not safe.

Finally, the air coming out of the fan's exhaust slightly smells of overheated components, so good luck poor motherboard :roll:

P.S. I weighted the notebook and it's 2.05Kg, didn't put it on a scale before the mod but the specs says 1.85Kg so there are 200g more, around 50-60g should be the sata ssd i added, a little unknown amount one stick of ram and the remaining due to the extra cooling components. The 90W power supply weights around 95g more than the 45W one (if i remember correcly, 192g vs 97g, without the cord). Total around 15% weight increase.

The cost involved on the components needed to do the cooling upgrade are around 60$ for the heatsink and the 2 fans (previously i forgot one item had shipping costs), and around 40$ for the 90W power supply. I should add the cost of the pieces of thermal pads i used, thermalright 12.8 W/mK, sold in sheets. They are like an hard putty, not silicone. I had them laying around, but if bought now the 0.5mm thick sheet is 10 and the 3mm thick 17, i probably used 1/10 of the 3mm and maybe 50% of the 0.5mm, let's round it to another 10$. Was that worth it? Given the better user experience, and that i could actually do something that before i could not do, maybe it was and don't hide the little enjoyment in messing inside this thing :D And it all started because of a RAM upgrade... (and too much amaro lucano) o_O
 
Joined
Aug 28, 2021
Messages
364 (0.39/day)
35A vrm-current and 45W vrmmax-current.
the air coming out of the fan's exhaust slightly smells of overheated components, so good luck poor motherboard :roll:
Check online for the data sheet of your vrms , usually you just type the serial numbers engraved on the vrm , you will get more informations from the manufacturer regarding safe operating ranges , voltages/amps , there are also two distinct values , 1- a sustained load usually for 30~90 seconds and 2 - another “normal “ operating values where the current is much lower but can be sustained indefinitely .

Was that worth it?
Totally worth it , 50% increase in performance on average is great , plus the fun and satisfaction of completing a Mod successfully .
 
  • Like
Reactions: L-C

L-C

Joined
Oct 3, 2021
Messages
32 (0.04/day)
Check online for the data sheet of your vrms , usually you just type the serial numbers engraved on the vrm , you will get more informations from the manufacturer regarding safe operating ranges , voltages/amps , there are also two distinct values , 1- a sustained load usually for 30~90 seconds and 2 - another “normal “ operating values where the current is much lower but can be sustained indefinitely .


Totally worth it , 50% increase in performance on average is great , plus the fun and satisfaction of completing a Mod successfully .
I am pretty sure I am way out of specs, especially for the 45A vrmmax-current, I saw peaks over 100A and usually is constantly over 50-60A, also don't know at what temperure they are operating... whoops!

Kudos to HP for building such a strong laptop. I'll let you know when it burns to death!
If you can adequately cool the components the life may be not much reduced under reasonable limits, I used to have an 8 core and FX overclocked on a (good) 6 phases motherboard, the last coupled of years ran at 4.8 GHz on summer up to 5GHz on winter. With a (good) 10 years old 500W psu, by the way.
 
Top