Raijintek Rhea vs CoolerMaster TX3 EVO (misusing CPU coolers)

[Dr.EM]

Hi all,

Bit obscure but worth a shot, in short does anyone know the dimensions of the base plate and screw holes of the Raijintek Rhea?

Longer version, I have an electronics project which needs a lot of cooling and I had planned up till now to use the CoolerMaster TX3 EVO which handily provides this dimensions info at the bottom of the page (I'd still need it in hand to measure those screw locations):

http://www.coolermaster.com/cooling/cpu-air-cooler/hyper-tx3-evo/

I now notice the Rhea is very similar on paper and is cheaper. It seems it may be a bit deeper front to back once the fan is installed but the critical dimensions are those of the base plate since I plan to mount directly using screws into it. This sort of design where the mounting hardware is screwed in from beneath is perfect for this misuse! Also, does anyone know if these are usually mounting with standard M3 machine threads?

Thanks

 

[Dr.EM]

Probably an easier question, does anyone know which cooler is newer? Particularly, which cooler will still be manufactured in the near future, because I see that CoolerMaster now have the H412R model which may be a replacement to the Evo? It's too wide for my purpose but if the Evo is becoming end of line it could make the Raijintek a better choice.

Thanks

 

[doyll]

There are several coolers that have threaded holes in their base for attaching mounts to. What spacing do you need for your application?

 

[Dr.EM]

Hi,

I basically need the spacing as tight as possible, both because the PCB size is restricted and because I need the base smaller since my power transistor mounting pad is smaller than a CPU top area. This is for cooling of a linear power supply, I've put an old Arctic FP7 cooler into service on the test build and it works very well, worst case I dissipate 140W but I don't need temps as low as for a CPU, running 70-80C in this case is perfectly acceptable.

I've mostly convinced myself I need the CM Evo now since, going from the photos, the Raijintek looks like its mounting base plate is longer and that wouldn't be desirable, you can see the plate gets close to electrolytic caps which are already right at the PCB edge. But if you know of any other CPU coolers with this mounting method I'm interested to investigate them. Everything can be moved around but I won't make any huge adjustments now and I think the Raijintek is too tricky, it's also not as tall, although I've kept tall components away from underneath the cooler.

Thanks

https://imgur.com/a/1LCuRQq

 

[doyll]

Without some sort of scaling reference I have no way of knowing the size of your PCB in image. It could be 3" or 30". Are you hoping to find a cooler with base having 4 corner threaded holes you will make matching holes in PCB to use for mounting cool onto the power supply? Might need a rigid backing plate like many coolers use to avoid stress/flex of PCB.

Keep in mind, CPU cooler heatpipes are designed to function optimally in the 35-90c range. I don't have the heatpipe specs but my guess is by 100c their liquid/vapor cycle will likely loose become all vapor .. and I'm pretty sure their TDP ratings are based on temps in 80-90c range. I'm going to assume you want a cooler with high enough TDP that it can keep thing cool in 30c ambient, maybe even higher.

Air coolers have a near 1:1 ratio of air temp to component temp. So if it needs to keep power supply at 80c in 35c ambient air, cooler will keep power supply at 67c if air ambient is 22c assuming fan/s are running same speed. Will you have fan speed temperature controlled or fixed speed?

Tower coolers have significant advantage over flat downflow cooler because tower keeps heated air flowing away from cooler and not mixing with cool air into cooler. The flat / pancake coolers blow their heated exhaust in all directions over PCB and if anything turns their heated exhaust up away from PCB it can be drawn back into it's cooler fan and pre-heat the cooler's airflow.

But tower coolers exert way more forces exerted onto PCB due to leverag then pancake coolers exert. So if whatever it is you are designing is going to be moved / bounced around, maybe a pancake cooler would be better option.

While A30 may be best for your application, there are many others to choose from, and depending on number needed maybe a custom mount could be made.

 

[Dr.EM]

Hi,

Dimensions are 100x80mm. The cooler does need to have 4 holes really, some have just 2 and that would be unstable. I will be making a custom back plate from 6mm aluminium plate, just a saw and drill job.

That is an interesting point on the heatpipe vapour flow. It would be almost inconceivably rare to use this at close to that 140W dissipation but it would be good if it still performs correctly. The Arctic FP7 is a very similar cooler and was still only really warm to touch at the copper base with 90W, couldn't really get my IR thermometer in there to take a reading but estimate 50C or less. Raijintek have been good in actually specifying a thermal resistance, though we can only presume this is based on full fan speed in free air, but at 0.18C/W this yields ((140W x 0.18)+25C Ambient) = 50.2C, extremely low for a device of this type and plenty of room for maneuver if case temp is higher. CM Evo will likely be almost the same spec here.

Fans will be temp controlled, the sensing diode is integrated into this transistor package. I would allow temps to go a bit higher before ramping up the fans, the profile increases from min to max between 45-70C as it stands, it's easy to change however.

A tower suits me well, it's such a small board and has 5 mounting points that it should be robust even with the heavy load sticking out the top. The 120mm coolers are getting into dubious territory though for most motherboards imo!

This is only a personal project so for now just 2 will be made, but I plan to be able to add more, possibly with lower input Voltage to provide a smaller range of outputs more suited to digital and higher current applications, I'll ensure dissipation will stay around the same.

Thanks

 

[doyll]

Sounds like you have it well planned out, but if you've done many design project you know plans don't always work out.

The info about functional range was just info. Your cooling application is same basic range as Intel CPUs which max at 100c. AMD range is not.

If your heat source is 140w I would suggest using a cooler rated about 180w just in case it's running in warmer/hotter air environment. Keep in mind the only place you will feel much warmth is the cooler base. Reason is the phase changing is in base where liquid vaporizes and moves toward end area of heatpipes where it condenses into wick on inside of heatpipes and is wicked back to base/heat source .. so except at heat source heatpipe surface is only barely warm if warm at all.

I often use a remote senor (cheap ebay indoor/outdoor, thererium or fridge digital with wired sensor) thermometer to monitor air temps. Can be had for 2-5 quid. Might be handy for you to check were you can't get IF readings.

I test coolers as a hobby and have no idea how the calculate TDP ratings. I've tried to find some common denominators and can't, so assume they are not very accurate.

LOL. I have never seen a motherboard damaged by a cooler regardless of how big that was not the direct result of abuse, like dropping system off of desk or shipping damage, and many of us have seen how couriers handle parcels.

Sounds like you do some interesting projects.

 

[EsaT]

I basically need the spacing as tight as possible, both because the PCB size is restricted and because I need the base smaller since my power transistor mounting pad is smaller than a CPU top area.

With only small heat outputting area those outermost heatpipes might not do work that effectively.
Direct contact heatpipes are manufacturing cost cutting measure, not something giving any automatic advantage even in normal situation.
So for cooling power would be better to at least have those heatpipes spaced tightly next to each others like this:
https://img.overclockers.co.uk/media/image/thumbnail/HS06MAR_162381_800x800.jpg

I test coolers as a hobby and have no idea how the calculate TDP ratings. I've tried to find some common denominators and can't, so assume they are not very accurate.

Marketroids just pull those out from MIASS
http://www.c4vct.com/kym/humor/miass.htm
Already different speed fans make those numbers meaningless for direct comparing.

 

[Dr.EM]

Hi,

Definitely not my first project, but the surprising thing is I'm still living with my first project which was a power supply and by this point it's very inadequate. LM317/337 job with analogue pots and totally undersized heatsinks, one has turned from black to grey colour!

I'm assuming AMD still get much hotter than Intels? That used to be the case. I guess really that any cooler of these dimensions can only dissipate a certain amount of heat, increasing airflow with an extra fan or using a Delta type (noisy) fan should help but eventually the heatpipes must set a limit. I do want this to be as quiet as possible. When I checked the Arctic cooler I was touching at the base, that one actually has a small copper block which the heatpipes travel through, but it seems the direct contact is favoured, possibly because they are faster transfer but I suspect the cost of the copper is a big part of why they changed.

I could and probably should get a cheap sensor, either a full system or a thermocouple for a multimeter. IR has trouble reading off of unfinished metal surfaces anyhow, needs to be black really.

I'm not a fan of many of the mounting strategies used in PCs tbh. What seems to have happened is the cooling requirements have increased way beyond what the mounting was ever designed to handle, I'd like to see a mounting rail in the top of the case to hold the top of tower coolers, of course this would need to be standardised in some way. Far worse are graphics cards, I had a passive one (so particularly heavy) seem to destroy itself while being moved just by car, hadn't come out of the socket but it just stopped working and never worked again despite re-seating etc. To secure something of that weight from one edge only with a single screw seems insane to me, relying on the socket to bear the main mechanical load. I can say from professional experience that this is a challenge in industry where desktop PC components are being used in more mobile applications, they would usually be fine in a PC case left alone. For this project I don't foresee any issues, it's rarely moved and I could secure the tops of the coolers easily if required.

I mostly do audio projects. This project is borne out of necessity and a lack of genuinely good cheap lab supplies. Most use relay switching to somewhat increase efficiency but this scheme is by far the most common failure point, sometimes causing overvoltage from the outputs which is clearly bad news. Relays don't do well switching DC, becoming rapidly more unsuitable as Voltage increases. I tried to switch using solid state but this had reliability issues too so decided to just go full linear and provide the necessary cooling, in reality you aren't often drawing 4A at 0.5V and typically your higher current projects are the higher Voltage ones too so real world losses aren't as dire as they seem, plus this isn't something you run the project from long term, it's only for development. I can share once it is finished, but to make more progress I need to buy some coolers and measure the mounting locations otherwise I can't order any PCBs!

Thanks

 

[Dr.EM]

Esat, you replied while I had my response open. You mentioned something that I questioned, which was whether the direct contact was a cost saving ploy rather than a real benefit and it sounds as though it is.

My transistor package has around 22-23mm length to span the 3 heatpipes. They do need to be packed quite closely to get the 3 at 6mm diameter across the package, the edge ones are certain to only make partial contact, I wonder how much worse that will be, relying somewhat on the base to conduct/spread. Perhaps my Arctic FP7 cooler is better in this regard having the copper base making an effective spreader. Mounting system isn't really suitable though, are there any copper base units with universal (threaded holes) mounts?

Thanks

 

[doyll]

Marketroids just pull those out from MIASS
http://www.c4vct.com/kym/humor/miass.htm
Already different speed fans make those numbers meaningless for direct comparing.

Yeah, I agree. I've seen some coolers that cool as close to identical as I can record (less than 1c difference) when testing but have 30-40 TDP difference in spec .. or have same coolers one with 1300rpm fans and another with 2500rpm fans rated the same .. or same cooler with 800rpm fans and 1300rpm fan rated the same.

 

[EsaT]

I'm assuming AMD still get much hotter than Intels?

What seems to have happened is the cooling requirements have increased way beyond what the mounting was ever designed to handle

Actually it's Intel CPUs which are hotter running.
Intel wanted to milk consumers even more with same old dual/quad cores, so instead of soldering heatspreader on its place they put toothpaste aka standard TIM under it.
And they can't even do heatspreaders properly.
So instead of TIM just filling unevenness between surfaces of heatspreader and CPU die, whole heatspreader can be lifted up from CPU with thicker insulation between them.

Wouldn't wonder if they decided Coffee Lake name from having spilled hot coffee on their laps.
Though now with "9th gen", which is architecturally still same as 6th gen Skylake, they're apparently soldering some models.
Because with eight cores those would otherwise be constantly throttling from overheating under full load.
Already six core 8700K can reach up to 150W consumption under full load, if mobo/BIOS doesn't limit it to stay inside advertised TDP.

And that 150W class power consumption was actually reached 10+ years ago with failed architecture Pentium 4s trying to keep up with AMD's CPUs.


If direct contact heatpipes were better, then that design would certainly be used in more expensive heatsinks for trying to get that last bit of cooling power.

 

[doyll]

I used to think direct contact was not as good but in recent years Thermalright has released several direct contact cooler and they perform as well as solid copper based coolers of same basic design. One down side of direct contact is we cannot lap the their bases to match lapped CPU IHS. I haven't seen any CPU 'IHS' that is in any way a functioning heat spreader. They are all used for load distribution to get distribute cooler weight/mounting pressure from cooler base to area of CPU IHS that is supported by motherboard socket. Proof of this is simple, just look at motherboard CPU socket size and then look at CPU IHS size of CPU that mounts on that socket. The more contacts involved between chip and motherboard the bigger the socket becomes and the bigger the 'IHS' becomes. Thread Ripper is easy example of how approx. 5000 contacts need a big socket and we end up with an "IHS' that is approx. 50.5x68mm.

 

[Dr.EM]

I've looked up what's inside CPUs since I never really thought to investigate it. It seems the actual core is somewhat smaller than my transistor mounting plate, although that is more or less equivalent to the IHS of the CPU since the transistor die is really the core equivalent. If you say the IHS does little to spread heat out from the core then presumably the direct coolers work alright with a smaller package being attached. The idea of it straddling 2 of the 3 heatpipes and not making full contact is uncomfortable though, the exposed pad is in fact only 20mm long.

I'm looking to find a similar cooler with a solid copper base but am not turning up much with universal mounting holes. If I have to make my own mounting then Arctic FP7 (as I'm testing with) would be suitable but I'd prefer not to. That would basically consist of a second mounting plate sitting on top of the cooler base section to clamp it down. Be Quiet - Pure Rock Slim appears somewhat suitable, although my original plan was to save money, not spend more

 

[doyll]

Direct contact heatpipe or enclosed in base, heat still transfers through base so both work .. as long as base is flush with direct contact heatpipe.