New to liquid cooling- questions about pump/fan control and additional water blocks on the mobo

[sorican]

Okay, so this is my first build for a liquid cooled PC. I'm not entirely unfamiliar with liquid cooling- I use it a fair amount in high-end drones and R/C aircraft that I sell, so I'm fairly familiar with the physical mechanics and basic components of a loop. I'm not deluding myself into thinking the similarities end any further than that.

I appologize if I seem like I'm just shotgunning questions. I've read the guides and theres a few specific details that I'm concerned about, going forward. I have read the stickies, and will probably re-read them several times more yet. I greatly appreciate people taking the time to look at this and lending their advice.

I've recently had reason to begin a series of upgrades on the rig, which on the high-load end, I use to run CFD, tapping the GPU's for running the math problems. The chip is running hotter than I would like, and the only real change I could make is going to a Noctua NH-D15, but it's too large for my case to be practical, and I was going to make the switch to a custom loop rig when I did the next series of upgrades any how, which is to go for SLI 1080's, and I'm going to be getting parts that can handle that expansion.
My current air-cooled rig is as follows:

Spoiler

Phanteks Entho Pro case (split window, if it matters)
MSI z370 gaming M5 motherboard
Intel i7 8600k currently throttled to the base 3.7ghz, which under modest load maintains a temperature at 40-50 deg. c.
64 gig corsair vengeance ram, 2400 mhz.
MSI geforce 1060 6gb aero, not overclocked currently.

I have two intake fans- a 200m front fan, and a 140 bottom fan mounted forward with a duct redirecting the flow towards the back of the case- about a third the air flow is under the PSU cover, while the rest is above..

I've four exhaust fans- a 140mm rear fan mounted high, and 3 top mounted 120's. the CPU cooler fans are programed to just go constantly on full.

I generally run Debian as the OS, though occasionally boot into win 10 (eww) and ArchLinux.

ambient temperature in the room is about 20-22 d. c, while case temps are about 24-ish, but this case does go to do exhibits with me, were temperatures may be higher. (but I'm almost never running high loads here. certainly not if it gets past 30c.

Okay. onto the questions.
First, the methods of controlling pumps and fans. my expectation is to set the pump to an appropriate flow rate that does the job, and leave it there, which is why I'm looking at the pump with the on board controls, for convenience.

This leaves the fans open to question. the 'easy' option I'm looking at is to plug them into a system-fan header, forked by a fan-hub providing additional PSU power to it, running them at an acceptable static speed. Is this basically how it's generally set up, or am I missing something?

Second important question is about adding auxilarly water blocks to the motherboard. The northbridge is now handled by the CPU. That leaves the southbridge and the mosfets.

The z370 has some fairly extensive cladding over them- which looks to be at least sheathed in plastic, though I assume it has some heat sinks and such like on it. the southbridge has glowy LED's, too. Provided enough airflow, should I even consider replacing these stock coolers with waterblocks? right now they're staying cool enough, but I'm concerned about loosing the air flow from the CPU fans and plugging the top three with a radiator.

I'm not even certain what the standard 'acceptable' temps are on the mosfets and southbridge, or memory- though I'd assume the sticks are okay.

Third, I've not settled on resevoir type. On one hand, I care more about functionality... and for a res that means ease of filling. Are drive bay res's- specifically the ones visible on the front- any more convenient in that regard? is there a certain minimal volume the resevoirs should hold, and do the larger resevoirs hold any substantial advantages?

In my R/C set ups, having a larger res means they can 'hold' more heat, but I'm pretty sure in pc's that's not the problem.

Fourth, does the hard tubing provide any useful benifits past looking neater? this will be a displayed case, so neatness counts, but, not that much.

and finally, because bling is important, do they make LED uv's that put light into the tubes? (I.e. making the tubes themselves glow more) or is that effect strictly from UV lighting in the case?

Edited to add another question: is there a functional difference between acrylic and coper-topped waterblocks?

I greatly appreciate your advice- and that this was possibly a giant wall of text.

 

[n0ns3ns3]

1. plunging the fans and pump to the MB header is totally fine. you can configure their speed according to CPU temperature. The better option would be to include temperature sensor for the liquid and use the liquid temps for the fan speed control. Unfortunately, this MB does not have any built in thermal probe connectors, so you will have to rely on external devices that connect via USB (like aquaero or vision touch from aquacomputer). since you are using Linux, there could be a problem reading the data from them.
Personally I don't like the PWM pumps, hence D5 Vario with manual control.
2. Generally, additional cooling to the MB or RAM is not required and in most cases is done for aesthetics or fun. VRMs might benefit from liquid cooling in very hot environment (over 30C ambient/room) without airflow over the heatsinks. And even then it will require massive CPU overclock and some unrealistic load that will pull ~150w through the VRM for at least 30 minutes continuously. The MOSFETs are designed to work up to 105-125C (internal package temp) so there is really nothing to worry about. With this specific MB, the VRM is 10 phase which supposed to be totally fine without additional cooling even under stress.
3. best reservoirs are tube ones preferably over 10CM long. if you are looking for the most versatile, I'd recommend to look at the Singularity Computers reservoirs and pump tops - it's probably the most flexible solution out there.
Bay reservoirs tend to recycle air into the loop which makes filling and bleeding the loop of air much longer process.
And you are totally correct that the volume of the liquid is not important and will only affect the time required for the loop to reach the max temp or cool down. It will not affect the max temp in any way.
4. There are some benefits for the hard tubing:
a. tubes will be much shorter - it's actually creates cleaner look and do not block the airflow.
b. no kinking
c. 100% plactisizer free - some soft tubing may/will release some gunk to the liquid that eventually will block the microfins in CPU block.
d. harder the tubes tend to be more chemical agnostic. the most common threat is polypropylene glycol found in most coolants. high % might dissolve soft tubing and some hard tubes like PETG.
d. harder tubes are more durable - scratch resistance etc.
that's why I ended up using glass tubes. no scratches, no staining, no worries about chemicals etc.
5. there are fittings and reservoirs that have lighting built in.
6. Theoretically, there is a difference even in heat transfer, but it's not measurable in modern blocks. the only practical advantages off all metal would be durability for CPU blocks and the option for adjustable geometry like in those: https://shop.aquacomputer.de/product_info.php?products_id=3561
For GPU, they are simply too heavy to be practical.

 

[sorican]

Thank you for your excellent (read: awesome!) reply.

It pretty much answered all my questions. The only other questions are...ridiculous to the extreme.

like how many rad's would it take to run with no radiator fans... or what's the worst things you've seen people try for coolant. (caught a youtube video of some guy trying milk, poweraid and orange juice. it was in reply to a 'hypothetical' question of using mountain dew. got me wondering. because, you know.)


.... ah crap. I spoke too soon. how much liquid does a typical loop contain? some how, I feel like the typical bottle of coolant with a a liter of fluid is more of a top off- a .6x10cm resevoir is conveniently sized to hold a liter on it's own. never mind the internal capacity of a 3-fan radiator.

Distilled water is looking increasingly cost effective. especially considering that I'd pity the fool who introduced my system to sub-freezing temperatures (or 100 deg. c.), which is the only real benefit of glycol coolants.

...on second thought... who needs LN2? it's winter...

 

[n0ns3ns3]

Any substance liquid enough to be pushed through the loop fast enough by a given pump can be used. so your imagination is the limit to what can be put into the loop. but most of them can't be used for a long term for various reasons.

making passive loop is possible but would require special rads. with common ones it will not work under load. there must be an airflow through the rad to make it work. on the positive side, some time ago, a concept loop without pump was shown.

My loop contains slightly more than 1L. radiators and blocks contain relatively small amount of liquid. most of it in res and tubes.

glycol main purpose in the coolant is to be "anti corrosion" and "biocide". though there is nothing wrong to run the loop on distilled water. it is indeed very cost effective - but who cares about the cost of liquid when you spend somewhere between 350 and a 1000$ on the loop itself.

 

[sorican]

but who cares about the cost of liquid when you spend somewhere between 350 and a 1000$ on the loop itself.

people are funny like that?

Actually, I was just thinking in terms of performance. By the time I add biocide and all the good stuff, I'll imagine that it'll come out to be even enough.

Again thank you for your information.