WorkLog: Custom Water Cooling Aorus Z370

[crashtech]

@lane42 , that would be pretty hilarious!

Given this new info, I am going to get a 10ml vial of benzalkonium chloride solution, a few drops per liter will supposedly enable the use of 100% distilled water, which should slightly increase performance. Since I'll have a clear reservoir, it will be pretty easy to keep track of any gunk or biological activity, if the disinfectant doesn't work, I'll wash out the loop and refill with >25% glycol instead.

 

[aigomorla]

AFAIR, silver coils and nickel plated components should not be used together.

meh... its actually a mix bag... depends on what the nickle coating is actually made of.
Back when vendors didnt cheap on nickle, and swiftech was heavy on the chrome, it was safe.
However now they cut every last corner possible, and yeah, nickle is kinda dangerous with silver, as you don't know the other alloy.
But the reaction between nickle and silver is very low, and at worst what it does is just strips it, but you wont see corrosion on the level of alu + copper.

I still use nickle and silver because i really dont care what my blocks look inside derlin, i only care that the outside doesnt tarnish like copper does.

I seem to remember some guy's use to just use Distilled water and a teaspoon of Bleach.

Bleach??? we dont use that... bleach can play havok on your Tubing. If you dont want to use silver, you can always use cupramine.
https://www.amazon.com/Seachem-6710...s=copper+safe&qid=1551429872&s=gateway&sr=8-4

like 3 drops should be good.... or you can use PT nuke....
https://www.amazon.com/Petras-Tech-...words=pt+nuke&qid=1551429987&s=gateway&sr=8-2

Personally id use Cupramine over PT Nuke...

That sounds like a really bad idea on the surface, bleach works by liberating chlorine gas,

Nah the Chlorine would eventually evap out though the tubing unless your running 100% rigid.
Chlorine also has no impact to rubber or silicon seals.
What it can do tho is make your clear plexi yellow, strip the dye off tubing, or making your tubing turn yellow quickly.

Edit 2, searching the entire subforum for "bleach" also brings up mostly rather old posts

Probably before the days of using copper 2 sulfate or cupramine.
Copper 2 Sulfate comes mostly from swimming pools, while cupramine comes from fish tanks.
Both are antimicrobial, and will work, just use it very sparingly...

 

[crashtech]

meh... its actually a mix bag... depends on what the nickle coating is actually made of.
Back when vendors didnt cheap on nickle, and swiftech was heavy on the chrome, it was safe.
However now they cut every last corner possible, and yeah, nickle is kinda dangerous with silver, as you don't know the other alloy.
But the reaction between nickle and silver is very low, and at worst what it does is just strips it, but you wont see corrosion on the level of alu + copper.

I still use nickle and silver because i really dont care what my blocks look inside derlin, i only care that the outside doesnt tarnish like copper does.

Is it possible to determine if the nickel alloy in the original style EK-Supremacy EVO (plexi top) is okay to use with the coil?

I never thought that the coolant mix/corrosion protection method would be the sticking point. At the time I am ready to fill the loop, I'll have four materials available to potentially use:

1. Silver coil
2. Benzalkonium Chloride solution
3. Glycol based coolant
4. Bleach

Bleach seems to be ruled out at this point.

 

[crashtech]

The X4 Photon 170 pump just arrived. Is it safe to run it dry just for a few seconds? I want to see if crafting a vibration isolation system will be immediately necessary.

 

[lane42]

I never said to use it

 

[StefanR5R]

Is it possible to determine if the nickel alloy in the original style EK-Supremacy EVO (plexi top) is okay to use with the coil?

I don't know how their nickel plating is composed chemically. But at least mechanically or with regard to their manufacturing process, it doesn't seem like top quality. I had to open an EKWB GPU water block once for cleaning after ~1 year use, and the copper shone through the plating at parts of the inner surface.

(The block needed to be cleaned because a sealant which EK had applied in my first modular AIO washed out of the seals and into the water blocks.)

The X4 Photon 170 pump just arrived. Is it safe to run it dry just for a few seconds? I want to see if crafting a vibration isolation system will be immediately necessary.

I wouldn't risk it. Damage to the pump can develop very quickly when running it dry, I'm afraid.

Besides, such a test probably wouldn't tell you much (being short and not matching operating conditions), and IMO you can always add vibration isolation later if it turns out desirable.

Edit, you could build a minimal loop (reservoir--pump combo together with just one tube between outlet and inlet) and test-run it with (distilled) water outside the computer, or inside at the planned mounting point. Hydraulic resistance wouldn't be the same as in the completed loop, but it may still give you a fair impression.

 

[aigomorla]

Glycol will leave jelly deposits in your injector plate on your supremacy over time.
We used to call this the swiftech mystery jelly.

And no i do not trust eK unless we go way back with nickle plating.
Like back when eddy was probably still the owner.

Benzalkonium Chloride is doable, again, use it sparingly... no more then 5-6 drops depending on vol. of water.

Also if your not mixing metals, you really dont need to worry about corrosion.
Copper is safe with brass, nickle, and chrome. Actually copper might react a tiny bit with nickle, again... depends on what the alloy is.
This is force net for vendors to make you use there junk premix if you want to preserve your nickle warranty.

And yes never run your pump dry.... its asking for trouble, as you can scratch the heck out of your impeller housing and damage the ceramic bearing ontop.

 

[StefanR5R]

Glycol will leave jelly deposits in your injector plate on your supremacy over time.
We used to call this the swiftech mystery jelly.

How so? Did esterification happen in these loops? 8-/
My suspicion after my one-off experience with an EKWB Predator: That jelly may have been other stuff which doesn't belong into a PC water cooling loop, namely plumber's sealant.

 

[aigomorla]

The propylene glycol gets strung out literally and clumps out though injector plates with changes to temps.

Didnt see this problem when waterblocks were simple without injectors, but they are complex as heck now.
This jelly was first reported on the G4 Storm blocks which introduced impingement jets. They were notorious for having to be cleaned out monthly because of this.

Injectors are usually the plates you see that sits ontop of a cold plate.
This silver plate you see...

So after time of being constantly stringed, it just clumps up and solidifies... and turns into mystery jelly.
It also has to due with PG concentration too.
The higher PG concentration you have, the higher clump out.

Another guess was it could be plasticizer from the tubing.
However people still reported this problem when running plasticizer free tubing.

 

[crashtech]

I never said to use it

No worries, I am very grateful that you took time to participate! Any and all ideas should be welcome to be aired out, even if they contain a whiff of chlorine!

I don't know how their nickel plating is composed chemically. But at least mechanically or with regard to their manufacturing process, it doesn't seem like top quality. I had to open an EKWB GPU water block once for cleaning after ~1 year use, and the copper shone through the plating at parts of the inner surface.

(The block needed to be cleaned because a sealant which EK had applied in my first modular AIO washed out of the seals and into the water blocks.)


I wouldn't risk it. Damage to the pump can develop very quickly when running it dry, I'm afraid.

Besides, such a test probably wouldn't tell you much (being short and not matching operating conditions), and IMO you can always add vibration isolation later if it turns out desirable.

Edit, you could build a minimal loop (reservoir--pump combo together with just one tube between outlet and inlet) and test-run it with (distilled) water outside the computer, or inside at the planned mounting point. Hydraulic resistance wouldn't be the same as in the completed loop, but it may still give you a fair impression.

Given your post and aigamorla's which follows, I will refrain from using the silver coil, although it is a cute little thing and has the aura of an elegantly simple solution. Also I won't run the pump dry. It's typical of all liquid pumps in my experience not to like running dry, I don't know why I thought a PC cooling pump would be any different. I may run it in a bypass loop temporarily just to get a feel for the noise and vibration which will of course change under load.

Glycol will leave jelly deposits in your injector plate on your supremacy over time.
We used to call this the swiftech mystery jelly.

And no i do not trust eK unless we go way back with nickle plating.
Like back when eddy was probably still the owner.

Benzalkonium Chloride is doable, again, use it sparingly... no more then 5-6 drops depending on vol. of water.

Also if your not mixing metals, you really dont need to worry about corrosion.
Copper is safe with brass, nickle, and chrome. Actually copper might react a tiny bit with nickle, again... depends on what the alloy is.
This is force net for vendors to make you use there junk premix if you want to preserve your nickle warranty.

And yes never run your pump dry.... its asking for trouble, as you can scratch the heck out of your impeller housing and damage the ceramic bearing ontop.

Thanks for the valuable insights! It seems that sticking to distilled water with a microbicidal agent may be the best initial plan. I won't run the pump dry, though I may run a quick bypass loop just to see if it's obnoxious. Most fluid pumps in my area of expertise have some kind of vibration isolation built into the design, and I have installed pumps both with and without rubber isolation with dramatic differences in noise level, though in fairness this has been in enclosed metal cabins which tend to amplify vibration sources attached to them.

 

[StefanR5R]

The propylene glycol gets strung out literally and clumps out though injector plates with changes to temps.

Didnt see this problem when waterblocks were simple without injectors, but they are complex as heck now.

Glysantin is not based on propylene glycol, but on monoethylene glycol. Is that safer? (Against precipitations, I mean. It's certainly not safer than propylene glycol if ingested...)

BTW, another risk of precipitations is when one switches from one additive to another without rinsing the loop carefully to remove all remnants of the former additive. Also, it's recommended to rinse radiators with (distilled) water before first use, in order to get rid of any potential residues from manufacturing which really shouldn't be in there. On a related note, since @crashtech and I decided on neoprene tubing rather than PVC tubing, we are safe against plasticizer being released into the loop.

 

[crashtech]

The water block arrived today. I forgot to get a before picture, so here is the pic from the ebay listing:

What I thought was copper corrosion is actually color from the previous coolant. Also there was a bunch of hairy looking stuff blocking the microchannels. It makes me wonder if that's why it was sold, because it would not be working very well in the condition it was in. I don't know the established protocol for cleaning these blocks, but based on experience I made a mix of methanol and water with a few drops of dishwashing liquid. It seems to be removing the blue stains as well as can be expected. The cleaning and brushing is being followed by plain water rinse, then a distilled water rinse, then blowing dry with compressed air. In process photos:

It's not perfect. There seems to still be a slight tinge of blue dye staining the plexi, maybe it's my imagination. Also there is a touch of light corrosion inside the cold plate where it would be difficult to remove. I could have it blasted with some extra fine glass bead but decided to leave it alone for now. For the price ($30), it meets my expectations.

Also my stuff from EKWB arrived. I rinsed the tubing with distilled water, blew it out with air, then taped up the ends. It didn't seem necessary but now there's nothing in there for sure.

The fittings take a 9mm hex key to tighten, which is annoying. I don't have one and it doesn't seem like anyone locally does either.

 

[StefanR5R]

Thumbs up for the cleanup. It's a bit of a shame how marketing makes people put unsuitable substances into cooling loops.

As for the fittings, you may be able to sufficiently tighten them manually, without the Allen key. That said, I have been using a key myself most of the time for this type of fittings and found it helpful. If you get such a key after all, take care not to overtighten the fittings. Obviously, the threads in the plexi (or in acetal parts if you have any) would be damaged if too much torque is applied to the fitting.

 

[aigomorla]

mah that's not bad...
i made the mistake of getting a completely used waterblock and here is how it came..

with enough work, and using vinegar with a brush.

 

[crashtech]

I ordered a 9mm hex key and it should be in this week, but fortunately my neighbor happened to have a very complete set and he has graciously allowed me to borrow the 9mm, which he said he never has had occasion to use and probably will never. It's just not a size you see, at least in this part of the world.

From the looks of it, the o-rings may not even want anything beyond hand tight. A more robust design has a recessed o-ring receiver groove so that the metal parts can be torqued without destroying the o-ring. Hand tight is just asking to come loose with vibration and thermal cycling

I'm glad my block does not look that that one, aigo! I am much happier now that you showed us that photo, actually!

 

[TennesseeTony]

I ordered a 9mm hex key ......

5/16 was too small? 11/32nds (probably an odd size)? 3/8ths is definitely too big.

 

[crashtech]

As it turned out, I didn't really need the hex key because anything more than hand tight would just damage the o-rings. My son came over and helped me out with it last night and we got it going but did not take photos like we were supposed to. I'll post a before and after when the new fans arrive though. Also I have yet to mount the block to the bare die, the lid and stock retention mechanism are still in place to get some baseline figures.

One issue that has cropped up, unrelated to the build, is that one of the cores on my 8700K has been and still is running consistently at least 18° C over the next hottest core, sometimes more . That's excessive in my view, what do you guys think?

 

[aigomorla]

do not over tighten...
the oring damage is not something you need to worry, its stripping the threads on the copper plate.
Copper is soft, it strips very easily, and once you strip it, the plate is almost as good as done, as you would need to rethread the tap, and use larger screws.

 

[crashtech]

I don't think we got anywhere close to stripping threads, basically the tool was a socket for 3/8" drive and we used it with no ratchet, just gripping the tool and tightening about as hand tight as they would go without excess effort. I think the only place we worried at all was in the plexi waterblock, but so far so good.

 

[TennesseeTony]

...... My son came over and helped me out with it ......

Tell Indydude we miss him!

 

[crashtech]

Tell Indydude we miss him!

He's still mostly a gamer, but I think once he settles down in his new digs a DC rig might be in the offing. If things get better on my end I might donate a 3930K rig if he'll promise to help theTeAm.

 

[crashtech]

Regarding the overheating core, this is what I am talking about:

Core 4 is consistently hotter than all the others. I'd blame it on the mount or TIM, but I don't know if that core is far off enough to one end of the die, it's really odd to me that the problem is so isolated.

 

[StefanR5R]

I wondered if Intel's "Turbo Boost Max Technology 3.0" driver was in use and core 4 happened to be the favored core... but a look into ARK tells me that Coffee Lake supports only Turbo Boost 2.0.

 

[crashtech]

Turns out that the block was mounted improperly. One of us made a mistake in determining how to mount the studs to the backplate, washers ended up being required to take up some slack due to some slight mismatches caused by the backing plate not EKWB brand. Here is a shot of it running with a 4.7GHZ AVX (llrGCW) load, which is about a worst case scenario for what I'm doing:

 

[StefanR5R]

Here is a shot of it running with a 4.7GHZ AVX (llrGCW) load, which is about a worst case scenario for what I'm doing:

I browsed your llrGCW results list; there were 2 invalid tasks on March 6-7, and 1 still inconclusive on March 9. Unfortunately, with LLR it is hard to tell whether instabilities are to be attributed to RAM or to the processor.