If top blowing coolers are so good, why they are so uncommon?

[The red spirit]

I have read a lot about how top blowing coolers are good and with certain boards mandatory. They seemingly help to cool VRMs and at the same time they are usually as good as single tower 120mm coolers. And despite seemingly being great particularly for budget overclockers, they are really uncommon. So far, today only competent downdraft coolers are Scythe Choten, Noctua NH-C14 and BQ Shadow Rock TF2. If they are good and help to cool down motherboard (VRM heat is still a big problem), why they are still so uncommon? It's not like they are new or anything, Scythe made Andy Samurai Master in 2003, so they are very well known and likely one of the first big upgrade over stock Athlon XP or Pentium 4 (socket 423) heatsink. I'm pretty sure that they can be manufactured rather well and they aren't really more expensive than single tower 120mm heatsinks.

 

[Batou1986]

It has a lot to do with clearance around the socket, I had a darkrock TF for quite a while and it made it impossible to do things like swapping ram removing m.2 and unplugging the 8pin cpu connector also a GPU with a large backplate would have never fit.
That and cooling VRM's isn as much of an issue as it used to be since VRM designs have become more efficient.

 

[The red spirit]

It has a lot to do with clearance around the socket, I had a darkrock TF for quite a while and it made it impossible to do things like swapping ram removing m.2 and unplugging the 8pin cpu connector also a GPU with a large backplate would have never fit.
That and cooling VRM's isn as much of an issue as it used to be since VRM designs have become more efficient.

VRM heat is still a big problem. Just watch some HWUB or GN. Arguably it's even more common now that Intel Burning Lakes came out (Comet and Rocket Lake).

 

[jayjr1105]

I had the same thought with top blowing coolers. I had a Thermalright AXP200R for every generation of Ryzen chip I had up until the 5600X where it wasn't quite keeping up to my standards for cooling (probably more stringent than others). I bought a Scythe Mugen 5 and ID cooling SE 225 XT and I kept the ID cooling and returned the Scythe. I figure I have 4 140mm Arctic static pressure fans that are giving my case some really good airflow plus a 5600X pulling maybe 80 watts worse case scenario isn't stressing VRM's much anyway. I also buy MSI motherboards for they seem to always do well with VRM and back side VRM temperatures.
I feel like VRM's are over engineered unless you are buying D tier boards and will probably outlive the rest of your system anyway. They can handle way more heat than normal electronic components.

 

[freeagent]

5600X pulling maybe 80 watts worse case scenario

More like 140w

But that is with a manual all core OC.

Personally haven't used a top down cooler since 939. It was a Thermalright too

But for myself, I see no need for one, everything gets plenty of air in my system.

 

[jayjr1105]

More like 140w

But that is with a manual all core OC.

Personally haven't used a top down cooler since 939. It was a Thermalright too

But for myself, I see no need for one, everything gets plenty of air in my system.

Yeah, when I first enabled PBO and the chip was set to the motherboard limits, not the CPU, I almost hit around 135w as well. Was fun for a few benchmarks then I got sick of roasting sitting next to my PC so I tuned PPT EDC and PDC way down.

 

[_UV_]

Top blowing good for tight cases with "low" power CPU. If try to use it with OCed hot CPU you actually blow all the heat back to the socket and VRM.

 

[The red spirit]

I had the same thought with top blowing coolers. I had a Thermalright AXP200R for every generation of Ryzen chip I had up until the 5600X where it wasn't quite keeping up to my standards for cooling (probably more stringent than others). I bought a Scythe Mugen 5 and ID cooling SE 225 XT and I kept the ID cooling and returned the Scythe. I figure I have 4 140mm Arctic static pressure fans that are giving my case some really good airflow plus a 5600X pulling maybe 80 watts worse case scenario isn't stressing VRM's much anyway. I also buy MSI motherboards for they seem to always do well with VRM and back side VRM temperatures.
I feel like VRM's are over engineered unless you are buying D tier boards and will probably outlive the rest of your system anyway. They can handle way more heat than normal electronic components.
View attachment 214196

In my experience motherboards don't last at all. After 2-3 years they die, but yeah I buy some cheap stuff. But then gain, some clearly overclocking intended boards are nowadays really toasty and if not cooled more, won't survive long either. Some not so low end boards have overheating VRMs with stock settings. Particularly the ones based on Burning Lake platforms. There hasn't ben any magic done to VRMs since AM3+ days and with those FX chips, which consumed a tons of power.

Top blowing good for tight cases with "low" power CPU. If try to use it with OCed hot CPU you actually blow all the heat back to the socket and VRM.

I read somewhere that even hot airflow is better than no airflow.

 

[sneekypeet]

As I have been testing coolers, I have been tracking VRM temps on that system. AIOs, tower coolers, C-style, low-profile, they all end up within ~5*C of each other, and not always is the C-style cooler the better of the bunch.

 

[Vayra86]

I had the same thought with top blowing coolers. I had a Thermalright AXP200R for every generation of Ryzen chip I had up until the 5600X where it wasn't quite keeping up to my standards for cooling (probably more stringent than others). I bought a Scythe Mugen 5 and ID cooling SE 225 XT and I kept the ID cooling and returned the Scythe. I figure I have 4 140mm Arctic static pressure fans that are giving my case some really good airflow plus a 5600X pulling maybe 80 watts worse case scenario isn't stressing VRM's much anyway. I also buy MSI motherboards for they seem to always do well with VRM and back side VRM temperatures.
I feel like VRM's are over engineered unless you are buying D tier boards and will probably outlive the rest of your system anyway. They can handle way more heat than normal electronic components.
View attachment 214196

Its a misconception that with a tower you would not cool the VRM. The only thing you're not doing, is placing a fan directly on it. But at the same time, heat will only move away from its source if the surrounding area is cooler. So whether you do that with forced air (that is still created within the case) or with a larger temp delta because the air going OVER the VRM with a tower is cooler than whats underneath, is really not that impactful. However, the efficiency of airflow direction in a case with a tower and case fans, is higher, so you need lower RPM for similar results.

The places where a topflow cooler works (better) is where the case is small so the air circulation is naturally higher; the single, large fan can do a lot more work there in just getting air to move away from the hottest point. Pressure and some case fanspeed does the rest. In larger cases you just don't have to, you can create front > back / top airflow much more easily because you've got a lot more case fans working for you and in push /pull.

Size is really the biggest reason for the OPs question. In the ATX format you just have the room for a tower, most of the time, so why not use the case width to increase cooling surface area, and keep other stuff free to access.

 

[TheoneandonlyMrK]

Like any cooler, the use case verses price usually decides, after that if it works most don't care anyway and most OEM coolers are top blowers therefore the majority of users use top blowers in OEM prebuilts, few use anything else.
Since enthusiasts Are the few.

 

[mtcn77]

Top blowers work according to a different principle. The issue with air cooling is removing the stagnant air isolation by using negative pressure, not positive pressure bearing down on objects. For this principle air bearing coolers have been developed, but they didn't reach market due to higher tolerance needs in production. Yet, if you are going to use an air bearing cooler, you will not have air pocket issues since the whole unit is spinning and air cannot develop inertia around crevices.

 

[jayjr1105]

Top blowers work according to a different principle. The issue with air cooling is removing the stagnant air isolation by using negative pressure, not positive pressure bearing down on objects. For this principle air bearing coolers have been developed, but they didn't reach market due to higher tolerance needs in production. Yet, if you are going to use an air bearing cooler, you will not have air pocket issues since the whole unit is spinning and air cannot develop inertia around crevices.

So recently I've been trying 3 intake and 2 exhaust when I originally had 2 intake and 3 exhaust. Is it general rule of thumb to have more exhaust than intake for negative pressure?

 

[mtcn77]

So recently I've been trying 3 intake and 2 exhaust when I originally had 2 intake and 3 exhaust. Is it general rule of thumb to have more exhaust than intake for negative pressure?

I get what you mean, but if you do that your case seals will fail. It will collect dust from every crevice - I'm sure you heard of positive pressure rooms. I said it as an air cooler principle.

 

[trog100]

i have used a a BQ top flow cooler for the last few years.. i do run it with a case side fan though.. the side fan directly feeds the top flow cooler with room temp air.. this airflow then gets blown on the motherboard.. ram and any other bits that need cooling air..

it makes sense to me..

trog

 

[Operandi]

In my experience motherboards don't last at all. After 2-3 years they die, but yeah I buy some cheap stuff. But then gain, some clearly overclocking intended boards are nowadays really toasty and if not cooled more, won't survive long either. Some not so low end boards have overheating VRMs with stock settings. Particularly the ones based on Burning Lake platforms. There hasn't ben any magic done to VRMs since AM3+ days and with those FX chips, which consumed a tons of power.


I read somewhere that even hot airflow is better than no airflow.

2-3 years??? Maybe you have bad power or something? I've literally never had a board VRM fail, ever. My daily systems is overclocked FX 8350 that runs 24/7 and runs WCG when idle since the day I built it and several client systems that are just as old doing the same thing all using desktop boards. Now that I said that my system will probably go up in smoke tomorrow, lol.

I don't really know shit about the VRM on the Asus board for my FX but under max Prime95 load the heatsink only gets hot to the touch, so I assume its pretty decent, and my HS is a top - down 120mm Noctua (I forget the model) for what its worth. That said VRM quality is all over the place today just like it was 10 years ago but ironically most high-end overlocking boards with massive heatsinks don't even need them because their VRM is so efficient and there is 8,10, or more MOSFETs to handle the load and spread the heat. And yeah, hot air flow is better than no air flow as long as the air isn't hotter than what you trying to cool which is never going to be the case when it comes to cooling a VRM.

 

[Sithaer]

2-3 years??? Maybe you have bad power or something? I've literally never had a board VRM fail, ever. My daily systems is overclocked FX 8350 that runs 24/7 and runs WCG when idle since the day I built it and several client systems that are just as old doing the same thing all using desktop boards. Now that I said that my system will probably go up in smoke tomorrow, lol.

I don't really know shit about the VRM on the Asus board for my FX but under max Prime95 load the heatsink only gets hot to the touch, so I assume its pretty decent, and my HS is a top - down 120mm Noctua (I forget the model) for what its worth. That said VRM quality is all over the place today just like it was 10 years ago but ironically most high-end overlocking boards with massive heatsinks don't even need them because their VRM is so efficient and there is 8,10, or more MOSFETs to handle the load and spread the heat. And yeah, hot air flow is better than no air flow as long as the air isn't hotter than what you trying to cool which is never going to be the case when it comes to cooling a VRM.

Friend of mine recently upgraded from a i 5-2500k system to a 11700k and his old mobo is still alive and kicking.I mean the one he used with the 2500k and he had that for like 10 years, ASRock P67 Pro B3 according to him.

I also had no dead mobos so far tho I'm not into ocing, even used a potato sub 50$ mobo for nearly 3 years and then sold it to someone else.
Same with my bro and he did OC the crap out of his x5650 xeon for years. 'granted he had a solid mobo if I'm right but still'

In my case my B350 Strix is 3+ years old now and its all good, tbh I don't really concern myself with VRM cause I only buy CPUs in the ~200$ range at most and those don't need beefy vrms so any half decent B range mobo works.

That beint said I did consider a top down cooler not long ago but only cause I like how they look in a PC, more than the oversized tower coolers. 'this is most likely why I will buy an AIO one day cause I like their looks'

 

[tabascosauz]

They're uncommon because you can get better CPU cooling out of a conventional tower for not much more $. Especially since your average case is not height limited, around 160mm can just about fit any 120mm/140mm single or dual tower.

• Most people don't care about VRM temps. Most Asus boards don't even have VRM temp sensors because Vishay DrMOS has very little in the way of monitoring. Quite frankly, anything with a 8+2 VRM with Vishay 50A will handle any 5950X stock or PBO just fine, no need for special cooling. A 4+2 with doubled lo-side with decent discretes like 4C024/4C029N or the non-garbage Sinopower ones, will handle any 65W part OC or a 105W stock.
• Most people also don't care about RAM temps. Not all ICs are temp sensitive (really just B-die, 8Gb Rev.E, possibly 16Gb Rev.B, and DJR), and none of them except those same four can safely take more than 1.45V (except legacy E-die). You only start needing cooling if they're temp sensitive and you're pushing a lot of VDIMM, and they're dual rank / you have a >150W GPU. And most kits don't even have temp sensors.

Granted, a downdraft cooler can be very capable. But downdraft coolers are highly situational, your case and airflow setup can have a huge impact on the performance you can get out of it. If you starve it by having it face a solid panel, you're in for a bad time. If you don't have a small exhaust near it to remove the heat it dumps into the case, you're in for a bad time.

I mainly run a C14S. When configured properly, it matches or beats a U12S, which doesn't seem like much of a feat but it's like 15-20mm shorter in height. Cerberus and Cerberus X basically don't fit capable tower coolers at 149mm max height, so this is the best I can do. If I could put a U12A in there, I'd do so in a heartbeat. But I can't.

Admittedly, it's very good at doing what it's good at. Namely, cooling the RAM (I run 1.5V daily and having push-pull A14s on the cooler makes sure DIMMs never exceed 41C ever). To achieve that kind of cooling with AIO or tower coolers, you'd otherwise need to have some sort of fan sitting on the DIMMs (because that's literally what the pull fan is doing).

That, and a lot of downdraft coolers have horrible board/socket/VRM heatsink/RAM clearance. Dark Rock TF, L12S, Black Ridge, (possibly?) AXP100, etc. There's a lot of planning involved. Even with the C14S, some boards have weird socket placements that will have the C14S sticking out the top of the case or hitting the graphics card.

 

[mtcn77]

Forgot: down draft coolers spread the heat whereas horizontal coolers can push heat away from components - they are compatible with air ducts which down draft coolers are not.

 

[The red spirit]

2-3 years??? Maybe you have bad power or something? I've literally never had a board VRM fail, ever. My daily systems is overclocked FX 8350 that runs 24/7 and runs WCG when idle since the day I built it and several client systems that are just as old doing the same thing all using desktop boards. Now that I said that my system will probably go up in smoke tomorrow, lol.

Well I can list my dead boards with chips that I used them with:
Gigabyte 78lmt-s2p rev 5 (I admit I pushed this board too far, after overclocking attempt PCB around VRMs was brown and reached 149C, so it's just me being inexperienced) - FX 6300
Asrock 970 Pro 3 R2.0 (probably another overclocking victim, but this time no visual signs of that) - FX 6300
Asrock 970M Pro 3 (I tried to keep this one alive, but it died for no reason once I upgraded computer)- FX 6300
Gigabyte F2A68HM-DS2 rev 1.1 (was not overclocked, RAM slot died at first, until whole board died) - Athlon 870K and Athlon 760K

It's not clear if VRMs specifically failed, but all these boards are died quite soon. Maybe VRMs were too weak or maybe I just somehow got more defective boards.

Friend of mine recently upgraded from a i 5-2500k system to a 11700k and his old mobo is still alive and kicking.I mean the one he used with the 2500k and he had that for like 10 years, ASRock P67 Pro B3 according to him.

But at the same time i5 2500k barely sips power, meanwhile Athlon X4 chips can use 100 watts of power. I'm pretty sure that's more than i5 2500k. FX 6300 is even more demanding, it was rated at 95 watts, but in reality it can consume a lot more. Somewhere in 120-150 watt ballpark and this review shows it:

AMD FX-8350 and FX-6300 Piledriver Review

System Power Consumption. AMD is refreshing its desktop processors with Piledriver, an enhanced version of Bulldozer that focuses on improving instructions per clock and frequency -- just like it did...

www.techspot.com

Meanwhile, i5 2500k most likely was consuming less than in specs. So it is really any surprise that similar budget potato survived i5, but didn't with FX 6300?

I also had no dead mobos so far tho I'm not into ocing, even used a potato sub 50$ mobo for nearly 3 years and then sold it to someone else.
Same with my bro and he did OC the crap out of his x5650 xeon for years. 'granted he had a solid mobo if I'm right but still'

Now running X5650 OC'ed for a long time is quite decent VRM test, but I still think that stock FX 6300 might be more demanding. I still remember that some MSI boards literally had popping capacitors from FX chips, some others had warped from heat VRM PCBs. If you bought the wrong board, FX was no fun. FM2 based Athlon may not have been quite as bad, but they still had same cores, but without L3 cache. So they still heated up quite a bit. Some people overclocked them and I can't imagine that being good for VRMs. But using those FX and FX derivates with VRMs without any cooling is suicidal and sometimes even passive cooling isn't enough. That's at stock, for overclocking you need more, so basically every decent overclocking board for AM3+ socket had at least 6+2 VRMs and majority went with 8+2. Boards that supported FX 9590 were even more beefed up, but that was a truly exeptionally demanding chip, i9 10900K has nothing on it, unless you raise PLs.

 

[GreiverBlade]

ah, a thread that made me remember how my Scythe Grand Kama Cross (Rev.B) came to be my favorite "top down" (and HSF) ... aaahhhhh if only i did not sell it ...


well i love my tower ETS-T50 nonetheless ... but man ... i want to find a Grand Kama Cross again now ... to add it to my HSF collection

 

[freeagent]

That’s why I buy Asus boards, they can usually stand the test of time, at least the ROG ones..
I do have a dead mid range Asus, and I got a DOA mid range Asus so those ones aren’t that good lol

If your tower cooler uses 140mm fans a lot of the time it will cool the vrm. It’s also really hard to manage temps if you won’t let your fans go over 700 rpm

 

[Operandi]

Well I can list my dead boards with chips that I used them with:
Gigabyte 78lmt-s2p rev 5 (I admit I pushed this board too far, after overclocking attempt PCB around VRMs was brown and reached 149C, so it's just me being inexperienced) - FX 6300
Asrock 970 Pro 3 R2.0 (probably another overclocking victim, but this time no visual signs of that) - FX 6300
Asrock 970M Pro 3 (I tried to keep this one alive, but it died for no reason once I upgraded computer)- FX 6300
Gigabyte F2A68HM-DS2 rev 1.1 (was not overclocked, RAM slot died at first, until whole board died) - Athlon 870K and Athlon 760K

It's not clear if VRMs specifically failed, but all these boards are died quite soon. Maybe VRMs were too weak or maybe I just somehow got more defective boards.


But at the same time i5 2500k barely sips power, meanwhile Athlon X4 chips can use 100 watts of power. I'm pretty sure that's more than i5 2500k. FX 6300 is even more demanding, it was rated at 95 watts, but in reality it can consume a lot more. Somewhere in 120-150 watt ballpark and this review shows it:

AMD FX-8350 and FX-6300 Piledriver Review

System Power Consumption. AMD is refreshing its desktop processors with Piledriver, an enhanced version of Bulldozer that focuses on improving instructions per clock and frequency -- just like it did...

www.techspot.com

Meanwhile, i5 2500k most likely was consuming less than in specs. So it is really any surprise that similar budget potato survived i5, but didn't with FX 6300?



Now running X5650 OC'ed for a long time is quite decent VRM test, but I still think that stock FX 6300 might be more demanding. I still remember that some MSI boards literally had popping capacitors from FX chips, some others had warped from heat VRM PCBs. If you bought the wrong board, FX was no fun. FM2 based Athlon may not have been quite as bad, but they still had same cores, but without L3 cache. So they still heated up quite a bit. Some people overclocked them and I can't imagine that being good for VRMs. But using those FX and FX derivates with VRMs without any cooling is suicidal and sometimes even passive cooling isn't enough. That's at stock, for overclocking you need more, so basically every decent overclocking board for AM3+ socket had at least 6+2 VRMs and majority went with 8+2. Boards that supported FX 9590 were even more beefed up, but that was a truly exeptionally demanding chip, i9 10900K has nothing on it, unless you raise PLs.

That just seems like an exessive amount of failure to me. I've literally never had a board fail (VRM or otherwise) that wasn't related to power bad power. But if the PCB is turning brown yeah thats an overheating VRM I'm sure. My 8350 only has a 200Mhz all core OC because I always leave a 100-200Mhz buffer for stability and thats as far as it goes being stable at stock voltage, more voltage and things just get crazy power and heat wise. My UPS has display on it that tells me how much power is being consumed and when a WCG CPU workload comes up it hits 230-250 watts if I take away what the monitor pulls.

Also when it comes to VRMs you can't just count the phases and pass judgment based on that. MOSFETs vary wildly in performance and you easily have a 4 phase out perform a 8 phase, so phase count gives you something to start with but its far from the whole story.

 

[docnorth]

Any sufficient air cooler will dissipate hot air from CPU, VRM’s and RAM, if you have decent airflow. The biggest problem for VRM’s usually is liquid cooling, due to reduced hot air dissipation. You can find a simple test here on TPU, https://www.techpowerup.com/review/amd-ryzen-9-3900x-tested-on-cheap-b350-motherboard/, it’s done for a different reason, but the VRM temperatures are very interesting.

 

[mtcn77]

Any sufficient air cooler will dissipate hot air from CPU, VRM’s and RAM, if you have decent airflow.

Yeah, but not volumetrically 9 times as efficient as air bearing coolers. Cooling is about removing the stagnant layer over the fin array. You can't do that with positive pressure alone.