Can a Fan on the Glass Stop Your Phone Throttling?

This page contains affiliate links. If you buy through one we may earn a commission, at no cost to you.

They work, but the question has three answers rather than one, because the three things people mean by working behave completely differently. On the surface temperature you can feel, an active cooler does a great deal. On sustained performance in a long session, it does something real and smaller than the surface would suggest. On a thermal spike in the first minute of hard use, it does almost nothing at all, and that is arithmetic rather than opinion.

The reason is that throttling is triggered by the temperature of the silicon, and between the silicon and the air there is a chain of thermal resistances. An accessory on the glass can only change the last link.

Two temperatures, and only one of them matters to the firmware

A phone has at least two thermal limits and they are nothing like each other. The junction temperature is the temperature of the processor die itself, and the firmware starts reducing clock speeds when it approaches a threshold that is typically somewhere in the region of 85 to 95 degrees Celsius. The skin temperature is what your hand reads, and it has a separate and much lower limit, in the low to mid forties, set by touch comfort and safety rather than by silicon.

So a cooler that takes the back of your phone from uncomfortably hot to pleasantly cool has moved a number that was never the trigger. That is not a trick, because the two are linked, but the link is through a chain of resistances and the size of each link decides how much of the improvement arrives where it counts.

Think of it as a ladder from the die to the room. Heat leaves the transistors, crosses the die, crosses the package, spreads sideways through whatever heat spreading the phone has, usually a graphite sheet or a vapour chamber, reaches the inside of the back glass or aluminium, crosses that, crosses your protective cover if you have one fitted, and finally leaves into the air by convection and radiation. Each step has a thermal resistance in kelvins per watt, and because they are in series they add.

Working the numbers, so you can check them

Start with the outside link, which is the only one you can actually calculate from public physics. A phone presents something in the region of 0.015 square metres of surface once you count both faces and the edges. Natural convection from a surface that size at a realistic temperature rise gives a heat transfer coefficient of about 5 watts per square metre per kelvin, and radiation from a surface of emissivity around 0.9 at these temperatures adds roughly another 6, so call the total 12. One divided by 12 times 0.015 is 5.6 kelvins per watt. That is the resistance from the phone's skin to the room, and it is the number every passive phone lives with.

Sanity check it against experience. A phone running a demanding game dissipates somewhere around 5 watts. Five watts times 5.6 kelvins per watt is a 28 kelvin rise, so in a 22 degree room the skin settles near 50. That is about what a hot phone measures, which is a reasonable sign the model is not nonsense.

Now the cover. One and a half millimetres of polymer with a thermal conductivity around 0.25 watts per metre per kelvin, over the roughly 0.009 square metres of the back, is a thickness divided by conductivity times area: 0.0015 over 0.25 times 0.009, which is 0.67 kelvins per watt. Not nothing. A three millimetre case is double that.

Then the internal link, from the die out to the skin, and this is the one nobody publishes. For a phone with a vapour chamber, 2 kelvins per watt is a plausible figure, and it is an illustration rather than a specification. Hold that thought, because the fact that it is unpublished is the honest limit of this whole post.

Finally the accessory. An active cooler clamped over maybe 25 square centimetres of the back replaces natural convection over that patch with something far more effective, and the illustrative figures in the drawing take the whole outside link from 5.6 kelvins per watt down to about 4.0, which is a 29 percent improvement in that step.

The dilution, which is the actual mechanism

Here is the part that is usually got wrong, including in the way this problem is normally described.

At a fixed power, reducing the outside resistance by some number of kelvins per watt lowers the skin temperature and the junction temperature by exactly the same number of degrees, because they are in series and the same watts are flowing through both. The accessory does not somehow move the surface more than the chip in absolute terms. That part of the folk explanation is simply wrong.

What is true, and is the right way to think about it, is that once a phone is throttling it is not running at a fixed power at all. It is in a control loop that holds the junction at its limit by reducing the clock speed, and therefore the power. So the question is not how much cooler the chip gets. It is how many watts the phone can shed while sitting at its limit, because that is what the clock speed is proportional to.

That number is the temperature budget divided by the total resistance. With a 90 degree limit and a 22 degree room the budget is 68 kelvins, and dividing it by each of the four totals in the drawing gives 8.2, 9.0, 10.2 and 11.3 watts. Which means: taking the cover off is worth about 9 percent more sustained power and costs you nothing, the accessory is worth about 24 percent, and the two together are worth about 38 percent.

And the dilution falls out of the same division. The accessory improved the outside link by 29 percent, but the outside link is only part of the chain: 1.6 kelvins per watt out of a 7.6 kelvin per watt total is 21 percent, so a 29 percent improvement in the part it can reach becomes a 21 percent improvement overall. The 2.0 kelvins per watt of internal resistance is a quarter of the baseline chain and it is untouchable. That is the ceiling on every accessory ever made for the outside of a phone, and it is why a phone with a better vapour chamber beats any phone with a fan bolted to it.

Which also corrects the worklist assumption I started with, and the common advice: taking the cover off is not the bigger lever here. For a thin cover the accessory is worth roughly two and a half times as much. What is true is that the cover is free and it stacks, so it should always be the first thing you do and never the only one.

Why none of this helps in the first minute

This is the arithmetic that settles the spike question, and it is the cleanest number in the post.

A phone weighs about 190 grams of mixed glass, metal, polymer and battery, with an average specific heat somewhere around 800 joules per kilogram per kelvin, so its heat capacity is about 152 joules per kelvin. Multiply that by the 5.6 kelvins per watt of thermal resistance and you get a time constant of about 850 seconds, which is roughly 14 minutes.

Now watch what happens in the first minute of hard use. The phone generates 5 watts for 60 seconds, which is 300 joules. At that point the temperature has only risen a couple of degrees, so the heat actually leaving is the rise divided by the resistance, which is about 0.36 watts, or 21 joules over the minute. Seven percent has left and 93 percent has gone into warming the phone up. Switch the accessory on and the heat leaving becomes about 0.5 watts, or 30 joules. The difference between having the cooler and not having it, over that entire first minute, is 9 joules out of 300, which is six hundredths of a degree.

So a cooler cannot stop a thermal spike. Not because it is weak, but because during a spike almost nothing is leaving the phone by any route: the thermal mass is absorbing the energy and the cooling path is barely engaged. The cooler only begins to matter as the phone approaches equilibrium, which on a 14 minute time constant means about ten minutes in and fully after half an hour. Our own guide to why phones overheat reaches the right conclusion from a different direction, saying free fixes cover most everyday cases and an active cooler is for hours of gaming or streaming rather than an everyday essential. The 14 minutes is why that advice is correct.

The product, and the sentence of ours that needs fixing

Our COOBILE magnetic phone cooling fan is a semiconductor cooler, which means a thermoelectric or Peltier module rather than only a fan, and the listing publishes the number that makes the physics legible: a 5 volt, 3 amp power input. That is up to 15 watts.

Hold that against the roughly 5 watts the phone is producing. A thermoelectric module does not create cold; it pumps heat from one face to the other using electricity, and its hot face then has to reject everything it pumped plus every watt of electricity it consumed doing so. So this accessory is adding as much as 15 watts of its own heat to the assembly in order to move some fraction of the phone's 5, and the small fan and heatsink on its back are rejecting up to 20 watts in total. That is not a criticism. It is how the device achieves what a fan alone cannot, which is driving its contact face below the temperature of the surrounding air. A fan can only ever approach ambient. A thermoelectric plate can go below it.

Which is also, exactly, where the second half of the original question lives. Our own record lists as a drawback that it can cause condensation in humid conditions, and the dew point says when. In a 25 degree room at 60 percent relative humidity, the air holds a vapour pressure of about 1.9 kilopascals, and the temperature at which that becomes saturated is about 16.6 degrees. Drive the contact face below 16.6 and water condenses on it, and on whatever it is touching. A fan cannot do that at any humidity because it cannot go below ambient; a thermoelectric cooler can, and that is the single mechanism behind every story about moisture under a phone cooler. Run it in a dry room, or at a lower setting, and the physics simply does not arise.

Now the sentence of ours that needs correcting. Our record for that product lists among its advantages that it "prevents thermal throttling in long sessions", and our overheating guide says the thermoelectric plate "turns genuinely cold against the phone and pulls heat out fast, holding off throttling". Both overstate it in the same way. Nothing fitted to the outside of a phone prevents throttling, because a quarter of the thermal chain is inside the phone and unreachable, and the accessory's own effect on the reachable part works out at about 24 percent more sustained power on the figures above. Twenty four percent more power at the limit is genuinely worth having and it is a different claim from prevention. The honest version is that it raises the ceiling the phone throttles down to, and delays reaching it, and does neither in the first minute. Record in the COOBILE phone cooling fan review.

The same fan, in a machine where it has leverage

Put a fan where the geometry suits it and the whole calculation changes, which is the clearest way to see what is wrong with the glass sandwich.

Our VZXAB 120mm RGB case fan publishes 1650 rpm, about 21.6 dBA, a 120 by 120 by 25 mm frame and a 4 pin connector drawing 0.30 amps. At 12 volts that is 3.6 watts of fan to serve a processor that might be dissipating thirty times that.

The reason it works is not that it is a better fan. It is where it sits in the chain. A desktop processor has a heatsink clamped directly to it with thermal compound in between, so the resistance from die to heatsink fin is a fraction of a kelvin per watt, perhaps a tenth of what a phone has from die to skin. And a case fan does not reduce a resistance at all: it lowers the temperature of the air that the heatsink is breathing. Lowering the ambient by five degrees lowers the junction by five degrees, undiluted, because it is a shift in the baseline rather than a change in one rung of a ladder. A phone accessory has to work through the ladder; a case fan works on the thing the ladder is measured from.

Same component, opposite leverage, and that is the whole difference between a machine designed to be cooled and a sealed slab designed to be thin. Our record notes the RGB is self cycling rather than software synced and that it is an airflow fan rather than a high static pressure one, which matters if you were putting it on a radiator. The 21.6 dBA figure is also worth reading carefully, and we took that apart in why a 33 decibel fan is louder than a 30 decibel one. Record in the VZXAB 120mm RGB case fan review, and the airflow argument is in PC cooling and case fans.

The honest limit

The 2.0 kelvins per watt internal figure that sets the ceiling on all of this is not published by any phone maker, and no cooling accessory publishes a thermal resistance, a heat pumping capacity, or a coefficient of performance either. The COOBILE publishes its electrical input and nothing about what it does with it. So the numbers in this post are a worked model from standard heat transfer, calibrated against the fact that a hot phone really does sit near 50 degrees, and not a measurement of any product. The shape of the conclusion is robust because it depends only on the internal link being a significant fraction of the chain, which it must be in a sealed device. The exact percentages are not.

Which leaves the free moves, and for this problem they are unusually strong, because the loop that throttles your phone is a power loop and power is the one thing you control directly.

Take the cover off. Nine percent more sustained power for nothing, on the arithmetic above, and it stacks with everything else.

Then reduce the watts instead of fighting them. Dropping a game's frame rate cap from 60 to 30 roughly halves the load the GPU is carrying, and halving the power halves the temperature rise through the entire chain, every rung of it, including the 2.0 kelvins per watt no accessory can reach. No external cooler can come close to that, because the accessory is working on one link and the frame rate is working on the numerator. Dropping the resolution or the shadow quality does the same thing. Turning the screen brightness down helps more than people expect, because at full brightness the display is a couple of watts on its own, and it is a couple of watts in the worst possible place, spread across the surface that is supposed to be shedding heat.

And do not charge while you play. A phone charging fast is dissipating several watts in the battery and the power path, which lands in the thermal budget alongside everything else and comes off your performance pound for pound. Our overheating guide already says this and it is the single highest value thing on the list.

The limit worth keeping in view: an active cooler is a real tool for a real situation, which is hours of sustained load at a performance level you refuse to compromise, on mains power, in a dry room, held still. If that is you, the 24 percent is worth paying for and the condensation caveat is worth respecting. If you are reaching for one because your phone gets hot for two minutes during a video call or in the sun, the arithmetic says it will do essentially nothing, and the reason is that nothing had time to leave the phone anyway.