Adventure in AIO repair

[M.Laz]

The short verion of the backstory is that I'm trying to quiet a system that is built from an old server carcass that is running a couple Ivy Bridge based Xeons. I've seen engineers run AIOs on these CPUs at a previous job so I figured I'd do the same (completely ignorant of the air cooling options that are available out there). 
I acquire a pair refurb AIOs, learn they don't have the hardware for narrow ILM CPUs, I manage to find that I can get mounting hardware then order it. The project sits for a while and I am waaaaaayyyyyyyyy outside of any warranty date for these. One works and seems to do pretty well. The other does not work at all. 
I consider going air cooling, but the ~$150 for a pair of Noctua coolers is a bit hard to justify right now. So I figure the AIO is already broke, I'll try to fix it.

The AIOs in question are the older model Corsair Hydro H75s. 

The symptoms are a noisy pump and that one line is getting hot. I start by assuming the system has lost water while sitting around. Maybe I can add some to it.
I look for info online. I find a lot of teardowns as well as some success stories on modding AIOs. But the Aestek pumps in my AIO are not well covered in any of the vids I find. But I suspect there has to be some way for the manufacturer to at least bleed the system. 
I manage to figure out that the screw between the inlet and outlet on the pump is the bleed hole in this system. 

I try the most non-destructive thing first. The first hypothesis is that there has been evaporation in the time since I ordered the AIO and when I was actually able to get around to trying it out. 
I remove the electronics from the pump to keep them from getting wet, elevate the pump above the rad, open the bleed hole, then try to pipette some water in. 
The pump noise gets better but one hose still gets hot. So now I think there is a clog.

I consider cutting the hoses, but I want to see if I can do as little damage as possible. The connection at the pump seems a bit fragile while the connections at the rad seem more robust. So I start by simply trying to delicately pry them off. 
They are good and stuck. So I decide to try and apply some heat hoping it makes the hose more pliable. But instead, the plastic shroud seems to take up all the heat and it deforms as I try to get at the hose. So I cut off the deformed shroud. It comes off cleanly and now I have the bare hose. I try getting the hose to move while cold without success. So I heat it, and I manage to free it enough to twist the hose! Now the hose comes off the barb relatively easily. I wish I had grabbed a pic at that time because it looks like Corsair may have used a sealer or adhesive on the barb and heat was necessary to free it non-destructively. 

I manage to get the loop at least partially drained. I see some crap in the coolant but I figure it's debris from removing the hose. 
The hose I removed was from the inlet side of the pump. I did this side so I could more easily get the pump going. It's finicky but I do eventually get fluid moving through the system. But I can't get it moving reliably. 
So I remove the other hose from the rad in case there is a clog there. 

I still have problems so I remove the cold plate and flush it with distilled water. Still no luck. 
It turns out the pump is incredibly finicky. I had pushed water into the system through the inlet until it came out the outlet then tried to get the mostly filled inlet hose into the pan of water I was using as a reservoir. That only worked a small percentage of the time. But if I did that, then used the bleed hole to help prime the pump, things went a lot better that way. Note that I always removed the pump electronics when using the bleed hole so they would not get wet.

I could get a steady stream of water flowing so I used that to fill the rad and close the loop. The system now started to behave like all the info I found online. The noise from the pump would go away by elevating the radiator and shaking it, it seemed I could get the air bubbles out of the pump. 

I haven't tested it in the system at all yet, I need proper coolant for that. 
I'm going to keep this closed loop for now. My build is pretty janky and the chassis doesn't really have a place for a reservoir if I wanted to think about converting to an open loop config. 

 

Performance for the one that works is pretty good. I've been using Doom to stress the system. The one that is running the AIO would get into the high 50s, low 60s on air but stays down in the 30s on the AIO.  The caveat is that I'm not using a MB fan header to power the pump or fans so they are going full speed (still quieter than the case fans). So I'm encouraged to keep trying to fix the AIO I'm working on and see how this goes. 

I can see why the introduction here says that water cooling is a hobby in and of itself

These last pictures are just to show that I was able to get this apart 'cleanly'.
 

 

[TempestCatto]

Post updates when you can, I'm curious too! I love adventures like this.

[M.Laz]

The next step is to get replacement coolant. The copper block and aluminum rad means I need to get coolant that prevents corrosion. I was looking at some Auquatuning fluid advertised for closed loop systems. But the shops that have it for cheap have ridiculous shipping prices. When it's more reasonable, it's about the same as getting some engine coolant.
I was avoiding the engine coolant solution because of how much I'd end up buying and then needing to dispose of the excess. But it seems like that's the best option right now. 
I'll probably use hose clamps on the rad too since I don't quire trust the friction fit of the hoses anymore. I might see about messing with the loops in the future. I know I can replace the radiators, and potentially add parts like reservoirs and flow meters. But my rig is like something from a season of Scrapyard Wars and doesn't have a ton of space for these things (maybe I'll do something in the future as my tool collection grows). The weak point of the whole thing are the pumps. Maybe they can be left unpowered or have the impeller removed and function as waterblocks on their own. I think there might be potential for recycling broken AIOs into gloriously janky water cooling setups XD

 

[TempestCatto]

43 minutes ago, M.Laz said:

I was avoiding the engine coolant solution because of how much I'd end up buying and then needing to dispose of the excess. But it seems like that's the best option right now. 

Don't use antifreeze (aka engine coolant). While it does contain things like biocides, it also will corrode most of the material we use in PC cooling. Mostly tubing, acrylics, rubber, and some other stuff.

 

Plain distilled water and an add in biocide (or something similar) are your best bet. If I'm correct, most AIO's are an aluminium rad and a copper cold plate. So you may have to add something in there to prevent corrosion (not sure if that's a thing biocides do). 

[M.Laz]

3 hours ago, TempestCatto said:

Plain distilled water and an add in biocide (or something similar) are your best bet. If I'm correct, most AIO's are an aluminium rad and a copper cold plate. So you may have to add something in there to prevent corrosion (not sure if that's a thing biocides do). 

Yes. The pump is a copper plate and there is some white colored metal used as a shroud over the fins and the radiator is aluminium. I have no idea what materials are used in the hose, the pump impeller. pump housing, and other seals. 
I wanted to get this: 

https://www.aquatuning.us/water-cooling/water-additives/ready-to-use/400/aquatuning-at-protect-clear-1000ml?c=6584

It sounds ideal for refilling an AIO.

But the $50 shipping price to me is unreasonable. I've found a couple places that advertise it online at ~$20 (assuming it is in stock) which is about what the engine coolant would run and it would be something I could just get locally that also advertises corrosion resistance.

I'm open to suggestions for coolant or appropriate additives that have an available US supply for a pretty low price.

[TempestCatto]

1 hour ago, M.Laz said:

Yes. The pump is a copper plate and there is some white colored metal used as a shroud over the fins and the radiator is aluminium. I have no idea what materials are used in the hose, the pump impeller. pump housing, and other seals. 
I wanted to get this: 

https://www.aquatuning.us/water-cooling/water-additives/ready-to-use/400/aquatuning-at-protect-clear-1000ml?c=6584

It sounds ideal for refilling an AIO.

But the $50 shipping price to me is unreasonable. I've found a couple places that advertise it online at ~$20 (assuming it is in stock) which is about what the engine coolant would run and it would be something I could just get locally that also advertises corrosion resistance.

I'm open to suggestions for coolant or appropriate additives that have an available US supply for a pretty low price.

EK-CryoFuel would be a good way to go. It includes biocides and corrosion inhibitors, and is of course safe to use with all common types of PC water cooling. They have a premix solution, where you do not add any water or anything else to it. Just pour it in your loop, shake the air out, and you're good. They also have a concentrate of the same stuff. I'm not sure if the concentrate contains a higher amount of biocides and corrosion inhibitors than the premix. But here are links to both:

 

https://www.ekwb.com/shop/ek-cryofuel-clear-premix-1000ml#ow_alert_box

 

https://www.ekwb.com/shop/ek-cryofuel-clear-concentrate-100ml-rev-2#ow_alert_box

[TempestCatto]

11 hours ago, M.Laz said:

Yes. The pump is a copper plate and there is some white colored metal used as a shroud over the fins and the radiator is aluminium. I have no idea what materials are used in the hose, the pump impeller. pump housing, and other seals. 
I wanted to get this: 

https://www.aquatuning.us/water-cooling/water-additives/ready-to-use/400/aquatuning-at-protect-clear-1000ml?c=6584

It sounds ideal for refilling an AIO.

But the $50 shipping price to me is unreasonable. I've found a couple places that advertise it online at ~$20 (assuming it is in stock) which is about what the engine coolant would run and it would be something I could just get locally that also advertises corrosion resistance.

I'm open to suggestions for coolant or appropriate additives that have an available US supply for a pretty low price.

I just found that stuff on Eaby for 20 bucks, free shipping, from the US: https://www.ebay.com/itm/Aquatuning-AT-Protect-Clear-1000ml-/264472694994

 

I would just go with this, honestly. Will save you time and headache, and a lot of money by going with it. Antifreeze will eat away at the tubes and seals in the cooler and you'll have one hell of a leak, killing whatever it leaks onto.

[Blue4130]

20 hours ago, TempestCatto said:

Don't use antifreeze (aka engine coolant). While it does contain things like biocides, it also will corrode most of the material we use in PC cooling. Mostly tubing, acrylics, rubber, and some other stuff.

 

Plain distilled water and an add in biocide (or something similar) are your best bet. If I'm correct, most AIO's are an aluminium rad and a copper cold plate. So you may have to add something in there to prevent corrosion (not sure if that's a thing biocides do). 

Do you have examples of that? People at hardocp have been running engine coolant for decades and never mention that. Not to mention, cars cooling is comprised of all the materials you mentioned. (rubber, acrylics, mixed metals) 

[TempestCatto]

On 12/6/2019 at 5:17 PM, Blue4130 said:

Do you have examples of that? People at hardocp have been running engine coolant for decades and never mention that. Not to mention, cars cooling is comprised of all the materials you mentioned. (rubber, acrylics, mixed metals) 

After further research, it seems I'm mostly mistaken. Key word mostly. As long as ones loop does not contain any PETG, it should be fine. But to me it just seems wasteful in the end. If you go through with it you're best off diluting a 50/50 mix even further to 70/30 or 80/20 (water to antifreeze ratio). Consider how much waste you'll produce when doing this, and how you'll dispose of it. Around where I live (USA) I can dispose of motor oil for free at any auto parts store, but most don't take antifreeze and those that do charge a fee. It may be different for you though, so look into it.

 

 

Also regarding Jayz video I linked below, he has a follow up video with the same test bench and same antifreeze filled loop. The pump died, but because of a short in the wiring, unrelated to the antifreeze. He does however replace the antifreeze with some other kind of PC cooling solution (figured I'd save the hassle of watching the other vid).

 

https://www.overclock.net/forum/134-cooling-experiments/401891-regular-car-coolant-pc-3.html

 

 

 

I guess it's up to you. If you want to go through the hassle, by all means. But personally, I'd just stick to the premixes for PC cooling.

[Blue4130]

Not much of a Hassle, most North American garages have a bottle of antifreeze. That number goes up drastically in Canada. 

[Blue4130]

I was able to find msds sheets for swiftech liquid and a prestone antifreeze. Guess what? The hazardous ingredients are exactly the same. So unless the non-hazardous chemicals are an issue, I say all is good. Unless you want a color other than yellow. 

[Windows7ge]

4 minutes ago, TempestCatto said:

--snip--

Bringing up a topic I made over two years ago? 

 

To whomever it may concern (I haven't read OP's question) I did run that anti-freeze 50/50 mixture in that loop and it went for well over a year with only one real issue. Bubbles. The anti-freeze behaves kind of like soap. The bubbles don't pop fast they stick around and build up. Be careful what pump/ress you use as it can exasperate this problem (mainly bay ress/pump combos).

 

If we're talking about putting it in a computer I wouldn't. The stuff stinks (smells) and the toxic properties don't make it something you want around pets or small children.

[M.Laz]

The coolant from the AIO I drained smelled suspiciously like engine coolant to me. But I figured there might be some secret "11 herbs and spices" thing going on.
Color doesn't matter since there are no clear parts.
Bubbles will be an issue, I need to make sure I get the loop filled enough. I've done some practice runs with distilled water and it can be tricky. The Asetek pump is really finicky and will either not pump well, or not pump at all if there is any air in it.


I did finally locate a source for the coolant I originally wanted that was not totally overpriced. Dealing with extra engine coolant is why I didn't immediately go that route.
In the future, I might just use engine coolant instead, it's less hassle to get and generally cheaper.

I'll keep posting updates as I get things put back together and tested. While there are some nice teardowns and even some good analysis of AIOs out there, I've seen those same articles and videos skip right over the bleed hole and not even try non-destrucive teardowns. I think there might be some interesting uses for repairing or re-purposing broken AIOs.

[M.Laz]

It took a bit of work but I eventually ordered the Auquatuning coolant form OutletPC. I have it now but didn't have time to try anything until now.
While I was waiting for the coolant, it occurred to me that I might be able to spin the impeller in the pump without electronics if I use a magnet and a drill. But my brief experiment tonight didn't seem to work It might be that the small stack of flat magnets wasn't strong enough or that I didn't get the orientations right (but I tried a few). 
What I'm hoping to achieve is to submerge the pump, open the bleed hole, and spin the impeller to try and clear any air bubbles from the loop as I can. 

I may need to either get an axially magnetized doughnut shaped magnet, or try a little more complicated arrangement of the magnets I tried to use in my last experiment. 

I wonder if they use an idea similar to this in the factory to fill and bleed the loops...