So you say that you didn't notice the laptop temperature getting any lower, but this isn't too surprising to me, since this thing is going to be highly thermally limited.
I suspect that your setup is doing something, but the thing it's doing is allowing your laptop to sustain a certain performance level for longer. So what I would look at is some performance metric like e.g. peakflops() in a loop, or a video game FPS counter, and compare with and without the external fans.
Exactly. If you increase the capacity of a bus in a highly crowded area, the bus will still have 100% occupancy, but that doesn't mean it didn't help. The throughput increased.
Much like adding an extra lane to a highway, more people will use the highway and you'll be back to the same crawl. But technically there's more flow/capacity.
Because it’s a slightly different argument. “More people will use the highway” is not really applicable here. The bus analogy also didn’t meant “more people will take a bus”, just that if you have bus route where 20 people on each bus stop need to wait for the next bus, if you increase capacity of buses by 10, then only 10 people will be waiting on bus stops, but you will still have 100% bus utilization.
In other words, with added lane on highways, it’s sociological mechanism, which is slightly less precise analogy.
The laptop fans are also targeting a set (maximum) temperature. If you provide external cooling, the internal fans will simply work less and the temperature will remain stable. So check the RPM of the internal fans too.
Another factor is the CPU behaviour. The CPU involved in the article is intel, but IIRC for AMD Zen4 (at least for desktop) they made them boost up to a relatively high but well within spec temperature, rather than leaving performance on the table that only tweakers will see. The better your cooling capability the more you'll get but it caused a few people to freak out a little. Aspects like that are why I reserve the noisy part of my fan curve for when temperatures are going beyond normal.
The little tiny fans in the laptop will typically be able to move a small amount of air with a large pressure. The fans you have used here externally will be able to move a large amount of air with a low pressure, but since you're trying to cram that air through a small space, they will actually be able to move a minuscule amount of air with that same low pressure. That's probably mostly why you aren't seeing much improvement.
I'd say that you have two options.
The first is to try to use external fans to force more air through the laptop than the internal fans manage. You'll need high pressure (probably centrifugal rather than radial) fans to achieve this. They'll be noisy.
The alternative, which can use quieter low-pressure fans like the ones you have, is to just ensure that all the air going into the air intakes on the laptop is as cold as possible. That is, make sure the air going in hasn't just come out. And make sure there isn't any more restriction on air flow than the minimum possible, which the raised gap under the laptop will achieve. You basically just want to direct fresh cold air from far away from the laptop towards the air intakes, and pull the air emerging from the air outlets as far away from the laptop as possible. A horizontal breeze flowing across the bottom of the laptop is probably one of the best ways to achieve this.
For my home project I used foam tape, but any seal will do.
Regarding cooling I wholeheartedly recommend one Framework 16 owner's journey in making his device maintain 80°C with the CPU going full throttle (aside from other mods):
Don't you want it to do that? The air goes upwards, hits the laptop surface, and travels outwards carrying the heat?
If it's sealed, I imagine you would get a very turbulent situation, possibly far noisier, but also one where I can't intuitively picture the airflow and expected cooling flow. I should note I am very, very far from an expert here ;)
There's a lot going on here, but the gist is that even a magnesium alloy laptop chassis is a rather poor conductor of heat (for a metal) and you really have to push that air through the machine to get it out.
As is the conclusion of the article, one of the easiest things to do is simply raise the laptop more off the surface so there's less impediment to airflow.
I was recommended these by a friend and the ability to easily do that anywhere (and as a nice side-effect get a lovely angle to the keyboard) but still be able to pop the laptop in a backpack easily is amazing. They have been put on every laptop my wife and I have owned for the last 6 years.
We need way more of this content! Things that were tried and failed. I have spent so many head-scratching hours on problem solution approaches that others likely already tried and failed. In science we only celebrate success. Even then it is typically poorly documented.
Compact electronics rarely have vents "close" to the things that get hot - it's been heat pipes for a long time already; and that system is fixed and "scaled" for the included fan(s)... Not much anyone can do in practice without taking the laptop apart - at which point you'd be better off buying a (mini) PC in the first place.
Also: thermals in general are vastly more complex and tricky than our intuition tends to tell us (with the article being a great example of it!).
So I had a similar problem and what actually worked pretty well was an external vacuum fan [1]. I tested it before by just using an inflator I had to pull a vacuum and I watched the CPU temp drop so I knew this was valid.
Probably a monster chip like that has way more headroom than you expect, and can max out the extra cooling you’re giving it as well. If the goal is to avoid the 90C blast, maybe enabling thermal throttling plus using the custom cooling setup will give you (much?) less of a drop in performance than throttling with stock cooling?
I'm not surprised that this doesn't change the CPU temp.
The important heat-exchange from the CPU is usually managed by a heatsink and a radial fan, with the grille of the casing designed as fan inlet/outlet.
Blowing/sucking additional air at the inlets doesn't affect the cooling surface of the heatsink, it just blows at the radial fan.
This may make sense if there's a weakness of the radial fan (that heat is building up at the heatsink which the original fan cannot remove).
But it's much more likely that the heat is not distributed well away from the CPU, either to the heatsink (thermal paste) or by the heatsink (weak heatsink design/material).
I'd look into ways to replace the radial fan or the thermal paste
Seems P53 has plastic chassis, I bought fan stand for my laptops that have metal chassis because that is used as heat sink. With metal you don’t need magically placed fans you just want to cool the surface.
For overpowered CPU that should not be in laptop it won’t help anyway, I had core i9 and best thing to do was to sell it and just swap to core i7 because there was no home made trick to make it better even with metal chassis.
I drilled some holes in my Framework laptop's bezel.
It isn't designed to be used with the lid closed which is what I do 99% of the time. After drilling the performance with the lid closed went up 30%, with the lid open up 5%. Removing the bezel with the lid open gives extra 10% vs normal bezel.
The nice thing is these can be swapped easily (magnets) and come in different colors. I used to have 2 until I stopped caring about my laptop looking weird in public.
Well - the challenge is with the right type of fans.
I like the idea of a pump curve for a garden water pump. If you pump water without any back pressure (not even a hose connected), the pump will give a certain volume per time that is can move. If you connect a hose and pump upwards, you will reach a max pressure / max height at which no water will flow any more. Both cases are useless, but this is what the manufacturer prints on the box. Your sweet spot is in between. If you pay enough, you will get a datasheet with the pump curve in between these two points.
For simple and cheap PC fans, you are at most given the number of air volume without back pressure. I've not come across a cheap fan that gives a defined backpressure of even a curve.
For your application you would need a fan that can push air through a restriction.
I had success with high rpm server fans. They create pressure, consume 20W of power or more and also make a lot of noise. But in this case - this is not a solution
FTA: “In the end, I’m not really sure we can just put dissimilar fans in series. If you put a fan that can pump (say) 200 cubic feet per minute (CFM) in series with a fan that can only pump 100 CFM, what is the resulting output? Probably not 300 CFM.”
- "n" fans in series will increase pressure "n" times at a given flow rate, with no additional free delivery flow. Fans in series can be useful in high resistance systems.
- "n" fans in parallel will increase flow "n" times at a given pressure level, with no additional shutoff pressure generated. Fans in parallel can be useful in low resistance systems.”
Frore makes thin flat electrostatic coolers that would be perfect for laptop SSDs, but I don't know if they're available for purchase by consumers.
The main problem would be powering them. They could potentially draw power from the M.2 connector through an adapter, or soldering to an USB connector, I guess.
It's a pity laptop manufacturers haven't figured out how to dissipate heat by using the lid as radiator. I read about some experiments years ago where they managed to get rid of like 25W
Edit: yes, normal laptop with all the bits under the keyboard and heat pipe going to the lid. I can't find a better source than this patent
Interesting, were those experiments with traditional layout when motherboard is in the bottom part of the laptop? Maybe we just need to make screen heavy laptops with thin keyboards but with thicker cooling, that could also utilize convection better. Something that you put vertically and open keyboard down, I think some gaming tablets are like this now
That is so extremely cool (ha) but boy, that sounds difficult to make reliable through ~100,000 lid open/close cycles. Still, I would love this if it could be made so. I have seen so many horror stories of "I'm using my laptop in my lap on a blanket and blocking all the cooling, why is it so hot?"
Just a note: some laptop fans are not solely controlled by CPU or GPU temperature, but also by power consumption. This may explain why your laptops internal fans are still spinning up under load, even though the external fans do all the work already.
I got a laptop cpu in a desktop motherboard (a mini PC board). It was cheaper, and performs much better that the same cpu in a laptop. It was a steal. It's like shucking an hdd, an incredible value for the price.
I got one from Amazon. A 32-thread Ryzen 9 7945HX, along with a motherboard, CPU cooler, case, and PSU for $400. But I didn't trust the PSU, so I bought a better one, replaced the thermal paste with PTM 7950, and replaced the CPU fan with a Noctua.
I also had to buy SODIMM DDR5 RAM modules (just before prices skyrocketed, I got 64 GB for $180), a 2 TB Samsung NVMe SSD, and an RTX 5060. The CPU comes with an integrated GPU, but I disabled it to get better cooling.
I also wanted to replace the CPU cooler with a Noctua one, but the GPU sits too close to the CPU. Maybe someday I'll adapt a liquid-cooling radiator, but that would require 3D printing a custom adapter and making multiple trips to a 3D printing shop for trial and error. That's too much hassle.
If I had a penny for every time a heat engineer's suggestion was discarded because it's too expensive, too time consuming to implement this late into the project or too impactful on weight or size I'd be rich already. The only guys I know that are ignored more than them are the EMI filter designers ("what's EM noise anyways?")
Amazing they're called 'laptops' when you haven't been able to safely keep a laptop on the top of your lap for decades.
Even slight increases in temperature is enough to do damage to certain parts of a man, and it's really not good for your legs to have this extremely hot air blasted over it continuously.
How feasible would it be for laptop manufacturers to add a slot in a laptop's underside cover to the CPU's heat pipe, and adjust the heat pipe's shape at this location? This would allow a dock to insert its own heat pipe through it, and dissipate the heat with larger fans and/or larger heat sink.
For Macbook Air there was DIY solutions where people added thermal pads to conduct the heat to the aluminium chassis and then use external cooling on that [1]. Since Air just has passive cooling, this gave major boost for long running tasks.
or even better a PTM. i was battling high temps and stutter due to throttling on my lenovo legion with ryzen 7945hx. stock thermal compound, whatever it was, didn't cut it. regular thermal paste was slightly better. PTM finally fixed it even to a point when i can use unlimited power mode.
for half a year at least, until radiators get clogged with dust and thing requires total disassembly. i'm looking to add metal mesh as filter, see how that works
Back during COVID I had an old MacBook Pro that I was using as a desktop which would run its fans at full blast whenever I had to do a video call, which was a lot of the time. Since I no longer had any intention to use this device as a laptop, I decided the best course of action would be to try and get as much heat into the metal baseplate of the laptop as I could, and then draw that heat off by some means. I settled on using large water cooling plates intended for Bitcoin ASIC miners. It worked really well! I posted about it on MacRumors: https://forums.macrumors.com/threads/so-i-water-cooled-my-ma...
I’ve had some mild success with external piezoelectric coolers. However, the form factor isn’t ideal (weird angle, which led to a lot of returns from other people, so it’s no longer on Amazon)
So you say that you didn't notice the laptop temperature getting any lower, but this isn't too surprising to me, since this thing is going to be highly thermally limited.
I suspect that your setup is doing something, but the thing it's doing is allowing your laptop to sustain a certain performance level for longer. So what I would look at is some performance metric like e.g. peakflops() in a loop, or a video game FPS counter, and compare with and without the external fans.
Exactly. If you increase the capacity of a bus in a highly crowded area, the bus will still have 100% occupancy, but that doesn't mean it didn't help. The throughput increased.
Much like adding an extra lane to a highway, more people will use the highway and you'll be back to the same crawl. But technically there's more flow/capacity.
Curious that you are being downvoted for the car analogy when the bus analogy you're replying to is functionally identical.
Because it’s a slightly different argument. “More people will use the highway” is not really applicable here. The bus analogy also didn’t meant “more people will take a bus”, just that if you have bus route where 20 people on each bus stop need to wait for the next bus, if you increase capacity of buses by 10, then only 10 people will be waiting on bus stops, but you will still have 100% bus utilization.
In other words, with added lane on highways, it’s sociological mechanism, which is slightly less precise analogy.
The laptop fans are also targeting a set (maximum) temperature. If you provide external cooling, the internal fans will simply work less and the temperature will remain stable. So check the RPM of the internal fans too.
Another factor is the CPU behaviour. The CPU involved in the article is intel, but IIRC for AMD Zen4 (at least for desktop) they made them boost up to a relatively high but well within spec temperature, rather than leaving performance on the table that only tweakers will see. The better your cooling capability the more you'll get but it caused a few people to freak out a little. Aspects like that are why I reserve the noisy part of my fan curve for when temperatures are going beyond normal.
The little tiny fans in the laptop will typically be able to move a small amount of air with a large pressure. The fans you have used here externally will be able to move a large amount of air with a low pressure, but since you're trying to cram that air through a small space, they will actually be able to move a minuscule amount of air with that same low pressure. That's probably mostly why you aren't seeing much improvement.
I'd say that you have two options.
The first is to try to use external fans to force more air through the laptop than the internal fans manage. You'll need high pressure (probably centrifugal rather than radial) fans to achieve this. They'll be noisy.
The alternative, which can use quieter low-pressure fans like the ones you have, is to just ensure that all the air going into the air intakes on the laptop is as cold as possible. That is, make sure the air going in hasn't just come out. And make sure there isn't any more restriction on air flow than the minimum possible, which the raised gap under the laptop will achieve. You basically just want to direct fresh cold air from far away from the laptop towards the air intakes, and pull the air emerging from the air outlets as far away from the laptop as possible. A horizontal breeze flowing across the bottom of the laptop is probably one of the best ways to achieve this.
So close to greatness.
I've dabbled in this and the one thing that absolutely has to be included is a seal on the edges between the laptop and the stand like so:
https://www.ebay.com/itm/126317216434
Otherwise the air will escape to the sides.
For my home project I used foam tape, but any seal will do.
Regarding cooling I wholeheartedly recommend one Framework 16 owner's journey in making his device maintain 80°C with the CPU going full throttle (aside from other mods):
https://community.frame.work/t/all-the-modifications-ive-mad...
> Otherwise the air will escape to the sides.
Don't you want it to do that? The air goes upwards, hits the laptop surface, and travels outwards carrying the heat?
If it's sealed, I imagine you would get a very turbulent situation, possibly far noisier, but also one where I can't intuitively picture the airflow and expected cooling flow. I should note I am very, very far from an expert here ;)
There's a lot going on here, but the gist is that even a magnesium alloy laptop chassis is a rather poor conductor of heat (for a metal) and you really have to push that air through the machine to get it out.
As is the conclusion of the article, one of the easiest things to do is simply raise the laptop more off the surface so there's less impediment to airflow.
https://www.amazon.com/Computer-Keyboard-Anti-Slip-Compatibl...
I was recommended these by a friend and the ability to easily do that anywhere (and as a nice side-effect get a lovely angle to the keyboard) but still be able to pop the laptop in a backpack easily is amazing. They have been put on every laptop my wife and I have owned for the last 6 years.
We need way more of this content! Things that were tried and failed. I have spent so many head-scratching hours on problem solution approaches that others likely already tried and failed. In science we only celebrate success. Even then it is typically poorly documented.
Compact electronics rarely have vents "close" to the things that get hot - it's been heat pipes for a long time already; and that system is fixed and "scaled" for the included fan(s)... Not much anyone can do in practice without taking the laptop apart - at which point you'd be better off buying a (mini) PC in the first place.
Also: thermals in general are vastly more complex and tricky than our intuition tends to tell us (with the article being a great example of it!).
So I had a similar problem and what actually worked pretty well was an external vacuum fan [1]. I tested it before by just using an inflator I had to pull a vacuum and I watched the CPU temp drop so I knew this was valid.
[1] this is just example, there's better ones I'm sure. https://www.amazon.com/KLIM-Tempest-Laptop-Cooler-Vacuum/dp/...
Probably a monster chip like that has way more headroom than you expect, and can max out the extra cooling you’re giving it as well. If the goal is to avoid the 90C blast, maybe enabling thermal throttling plus using the custom cooling setup will give you (much?) less of a drop in performance than throttling with stock cooling?
I'm not surprised that this doesn't change the CPU temp.
The important heat-exchange from the CPU is usually managed by a heatsink and a radial fan, with the grille of the casing designed as fan inlet/outlet.
Blowing/sucking additional air at the inlets doesn't affect the cooling surface of the heatsink, it just blows at the radial fan.
This may make sense if there's a weakness of the radial fan (that heat is building up at the heatsink which the original fan cannot remove).
But it's much more likely that the heat is not distributed well away from the CPU, either to the heatsink (thermal paste) or by the heatsink (weak heatsink design/material).
I'd look into ways to replace the radial fan or the thermal paste
Seems P53 has plastic chassis, I bought fan stand for my laptops that have metal chassis because that is used as heat sink. With metal you don’t need magically placed fans you just want to cool the surface.
For overpowered CPU that should not be in laptop it won’t help anyway, I had core i9 and best thing to do was to sell it and just swap to core i7 because there was no home made trick to make it better even with metal chassis.
I drilled some holes in my Framework laptop's bezel.
It isn't designed to be used with the lid closed which is what I do 99% of the time. After drilling the performance with the lid closed went up 30%, with the lid open up 5%. Removing the bezel with the lid open gives extra 10% vs normal bezel.
The nice thing is these can be swapped easily (magnets) and come in different colors. I used to have 2 until I stopped caring about my laptop looking weird in public.
Well - the challenge is with the right type of fans.
I like the idea of a pump curve for a garden water pump. If you pump water without any back pressure (not even a hose connected), the pump will give a certain volume per time that is can move. If you connect a hose and pump upwards, you will reach a max pressure / max height at which no water will flow any more. Both cases are useless, but this is what the manufacturer prints on the box. Your sweet spot is in between. If you pay enough, you will get a datasheet with the pump curve in between these two points.
For simple and cheap PC fans, you are at most given the number of air volume without back pressure. I've not come across a cheap fan that gives a defined backpressure of even a curve.
For your application you would need a fan that can push air through a restriction.
I had success with high rpm server fans. They create pressure, consume 20W of power or more and also make a lot of noise. But in this case - this is not a solution
FTA: “In the end, I’m not really sure we can just put dissimilar fans in series. If you put a fan that can pump (say) 200 cubic feet per minute (CFM) in series with a fan that can only pump 100 CFM, what is the resulting output? Probably not 300 CFM.”
“Dissimilar” isn’t needed there.
https://www.rotron.com/Resources/Application-Notes/Series-an...:
“When applying more than one fan in a system:
- "n" fans in series will increase pressure "n" times at a given flow rate, with no additional free delivery flow. Fans in series can be useful in high resistance systems.
- "n" fans in parallel will increase flow "n" times at a given pressure level, with no additional shutoff pressure generated. Fans in parallel can be useful in low resistance systems.”
Frore makes thin flat electrostatic coolers that would be perfect for laptop SSDs, but I don't know if they're available for purchase by consumers.
The main problem would be powering them. They could potentially draw power from the M.2 connector through an adapter, or soldering to an USB connector, I guess.
It's a pity laptop manufacturers haven't figured out how to dissipate heat by using the lid as radiator. I read about some experiments years ago where they managed to get rid of like 25W
Edit: yes, normal laptop with all the bits under the keyboard and heat pipe going to the lid. I can't find a better source than this patent
https://patents.google.com/patent/US5880929A/en
But I distinctly remember seeing photos of a prototype in something like wired or the register
Interesting, were those experiments with traditional layout when motherboard is in the bottom part of the laptop? Maybe we just need to make screen heavy laptops with thin keyboards but with thicker cooling, that could also utilize convection better. Something that you put vertically and open keyboard down, I think some gaming tablets are like this now
The displays generate their own heat
That is so extremely cool (ha) but boy, that sounds difficult to make reliable through ~100,000 lid open/close cycles. Still, I would love this if it could be made so. I have seen so many horror stories of "I'm using my laptop in my lap on a blanket and blocking all the cooling, why is it so hot?"
Just a note: some laptop fans are not solely controlled by CPU or GPU temperature, but also by power consumption. This may explain why your laptops internal fans are still spinning up under load, even though the external fans do all the work already.
I got a laptop cpu in a desktop motherboard (a mini PC board). It was cheaper, and performs much better that the same cpu in a laptop. It was a steal. It's like shucking an hdd, an incredible value for the price.
A frankencpu BGA to LGA? I am aware they're sold in places like AliExpress, but in addition to it you need to flash a custom BIOS in order to post.
I got one from Amazon. A 32-thread Ryzen 9 7945HX, along with a motherboard, CPU cooler, case, and PSU for $400. But I didn't trust the PSU, so I bought a better one, replaced the thermal paste with PTM 7950, and replaced the CPU fan with a Noctua.
I also had to buy SODIMM DDR5 RAM modules (just before prices skyrocketed, I got 64 GB for $180), a 2 TB Samsung NVMe SSD, and an RTX 5060. The CPU comes with an integrated GPU, but I disabled it to get better cooling.
I also wanted to replace the CPU cooler with a Noctua one, but the GPU sits too close to the CPU. Maybe someday I'll adapt a liquid-cooling radiator, but that would require 3D printing a custom adapter and making multiple trips to a 3D printing shop for trial and error. That's too much hassle.
If I had a penny for every time a heat engineer's suggestion was discarded because it's too expensive, too time consuming to implement this late into the project or too impactful on weight or size I'd be rich already. The only guys I know that are ignored more than them are the EMI filter designers ("what's EM noise anyways?")
CPUs control frequency themselves and throttle. No need for manual throttling.
And it also might be a problem with the cpu paste and not with air.
Raising it is the most important part. After that, if you can pop the bottom off while it's running, and swapping out the thermal paste are my go-to.
The mobile hardware edge that Apple still holds on almost every front still seems substantial.
I wish there would be only one company that could seriously meet them at their functional high ground.
Amazing they're called 'laptops' when you haven't been able to safely keep a laptop on the top of your lap for decades.
Even slight increases in temperature is enough to do damage to certain parts of a man, and it's really not good for your legs to have this extremely hot air blasted over it continuously.
How feasible would it be for laptop manufacturers to add a slot in a laptop's underside cover to the CPU's heat pipe, and adjust the heat pipe's shape at this location? This would allow a dock to insert its own heat pipe through it, and dissipate the heat with larger fans and/or larger heat sink.
For Macbook Air there was DIY solutions where people added thermal pads to conduct the heat to the aluminium chassis and then use external cooling on that [1]. Since Air just has passive cooling, this gave major boost for long running tasks.
[1] https://forums.macrumors.com/threads/guide-macbook-air-heats...
It would risk the user touching the heatpipe during work, which is quite toasty under heavy load.
Not quite the same, but there are laptops with external water cooling that connect to internal piping inside the laptop when placed on the dock:
https://videocardz.com/newz/colorful-launches-water-cooled-l...
I'm using the Ikea Drönjöns laptop stand. It is made of metal so it conducts heat to wider area. That is probably more important than some extra fans.
> I picked the locations of the fans and their supports very carefully, and it’s clear that the intakes aren’t blocked.
Just double checking: Did you put the fans in the correct orientation and on the intakes instead of the exits?
Totally noob question: Couldn't the problem just be that the laptop body not conducting heat so much? I think it'd be worth to try it with a macbook.
Aren't those laptops designed to boost until it hits the max temp ? OP should test performance before and after.
keep going, it's just one negative result, try to model the airflows and heat transfers in your mind, etc.
Another thing author missed is thermal paste. I got way better cooling in my 2-year old framework by just replacing the dried out paste.
or even better a PTM. i was battling high temps and stutter due to throttling on my lenovo legion with ryzen 7945hx. stock thermal compound, whatever it was, didn't cut it. regular thermal paste was slightly better. PTM finally fixed it even to a point when i can use unlimited power mode.
for half a year at least, until radiators get clogged with dust and thing requires total disassembly. i'm looking to add metal mesh as filter, see how that works
Back during COVID I had an old MacBook Pro that I was using as a desktop which would run its fans at full blast whenever I had to do a video call, which was a lot of the time. Since I no longer had any intention to use this device as a laptop, I decided the best course of action would be to try and get as much heat into the metal baseplate of the laptop as I could, and then draw that heat off by some means. I settled on using large water cooling plates intended for Bitcoin ASIC miners. It worked really well! I posted about it on MacRumors: https://forums.macrumors.com/threads/so-i-water-cooled-my-ma...
This blog has been very popular lately. 12 submissions in the past 3 days.
wouldn't one need high static pressure fans for this to work?
I’ve had some mild success with external piezoelectric coolers. However, the form factor isn’t ideal (weird angle, which led to a lot of returns from other people, so it’s no longer on Amazon)
[dead]