Project Viking

Although I am not up to going full water yet, I am impressed with this build....and also with your build log Ohjay.

Nice to see someone admit to being human....and owning up to errors along the way. :)

I think there are many of us who learn far more about the pitfalls one can fall into, and how to avoid them, when you hear all about others hiccups. :)
Yeah better to be honest about your mistakes so someone else might not have to do the same ones. :)
And even then you might end up doing it anyway(like me ;) )

Erghhhh. Have the same res and know your pain! Mine was slightly worse as it did not leak but rather gave out in one go when the res was filled to about 60%, wow I nearly cried as it went all over my PSU :/

I double and triple check now on this res to make sure I can't pull it off, even with a little bit of force applied.

All part of the learning curve I guess :)
Yeah seeing all that water going out of the res was not a fun experience. Thankfully I had nothing under power and most of it just ran through the fans and the rad and out of the case. But it still wasn't fun.




So. I went ahead with my plan for draining the reservoir, i.e. removing the tube from the pump. Got a bit of spill but thankfully only a fraction of the water in the reservoir.
Now I need to figure out how to redo the pipes, but I think I'll have to get a 90 degree bracket for the pump. The piping will have to be very, very creative otherwise.

Gotta love the setbacks :rolleyes: :D
 
Thanks all!

Been away from the forums for a wee while and now I'm back I have to say what an awesome job you are doing.
Anything with a Primo as a starter is win win in my book. While I wasn't totally sold on the colour scheme at first I'd like to congratulate you for pulling it off.

Don't fancy your chances of plugging anything in to the top of the motherboard with that rad though ! I guess you'll be removing it for that?
Thank you, that really means a lot to me when it comes from the person who, in my book, has one of the best Primo builds out there. :D

Thanks for the heads up regarding mobo versus rad but I think I'll manage. The only things that needs plugging in there are the 12V ATX connectors and I think I'll be able to do it. I've been cursed with tiny girlish hands which sort of helps in these situations. :rolleyes: :D


I promised you an image TechMinerUK. I'm sorry that it took a day extra to get done. Also it's not the best but hopefully you'll get an idea of the slight overhang of the revolvers on these 16/10 Monsoon fittings:
qgM7cU2.jpg
It's not much but it's there.


I spent the rest of the day(and night) yesterday thinking about which direction I wanted to go with the pump.
In the end I decided to go for the tricky and potentially harder solution of just redoing the tubes without changing the way the pump was mounted.
So today I went ahead and did some bends:
VZITp3S.jpg

They're currently drying after some thorough cleaning. The closest one might not look tricky but the fact that several bends are so close together makes it a real challenge to get the bends just right.
Tomorrow they'll be installed and time permitting I'll try to do a leak test that hopefully won't end with disaster... again ;)
 
Thank you for getting the photo, the Revolvers fit in quite well with the angled fittings. Also for a set of difficult tubing you've done a great job as the tubing appears to be intact with no collapsing or straining from cold bending. :D
Thanks!

I'll see if I can get an even better image later on, but I don't think there's a better option in terms of looks.
You just have to be careful with color matching revolvers with the Monsoon fittings, from what I've seen the revolvers have more bold colors while the Monsoons are a bit washed out.

There actually is a tiny bit of straining on a couple of tubes but I don't think it's visible to the naked eye but you can feel it with your fingers.




So today I have some good news and some bad news.

The good news is that the tubes I made yesterday fit like a dream and have quite interesting aesthetics:
DxdvjMo.jpg

koRa9rK.jpg

I'll try to give a better representation of how tight a fit this tube really is in my final photos:
8WNGRJX.jpg
Precision, precision, precision. :D

After that I went ahead and started filling the loop again and with the pump pushing water the right way it all seems good. I only ran it for 10 minutes or so(more on that in second) but I had no visible leaks and everything seemed to be working fine.

But then the bad.
I ran out of coolant :rolleyes:
With the pump running the res is only filled up to about a quarter. No shop in Sweden has the same coolant in stock and I suppose it might be bad to mix different brands of coolant so I have to order from abroad.
With a bit of luck I'll get it by the end of next week...

I could use some advice from you, my dear readers...
With the way I've set up my loop I have a small issue when I turn off the pump and that is that water surges back into the reservoir "backwards".
As I mentioned, with the pump running the res is only filled by 1/4 but with the pump off it's completely full.
From my understanding it's good to leave some room for expansion in the reservoir but will that work with my loop?
And how should I go about filling the system? Should I just keep the pump running and fill the res until it's full, try to get as much air out as possible and then seal the system(still with the pump running)?
 
Thanks all!

colinmcr, I'll definitely keep you updated on how it goes in terms of venting, I fully expect it to be a nightmare.

OJ46, did you have your inlet pipe at the top like I do or is it on the side if the reservoir?
Edit: I suddenly remembered your build log and found the answer there ;)


A small update then...
The last bits n bobs for the cables arrived two days ago so they are almost finished.
The coolant arrives tomorrow so if there are no other setbacks I think I'll be able to finish the build this weekend :D
 
Last edited:
Thanks!

The coolant arrived as promised plus I got something extra that I've been putting off, more on that later today(together with an image or two).

I might have time to fill and vent the loop today as well, we'll see. Either way there will be some sort of update tonight ;)
 
You are doing a great job.
You could even do a window on the rear side with the pump :)

Liking the colourtheme.
Neat idea! :)
It won't happen at this time but maybe when I revisit this build later to fix a thing or two.



I promised you an update and here it is.
First of all, the goodies that arrived today. Dogmeat is once again happy since there was a little baggie of goodies in the box:
pheWo6i.jpg

I've been looking for a good solution to monitor water temps through software and was just about to start building my own using an Arduino or Raspberry Pi when I was recommended to get an Aquaero XT instead. Said and done, that's what I did.
I might end up making my own fan controller / temp monitor down the line, but this works for me at the moment.



Remember the SLI bridge and the red plexiglass?
The plexi is now removed:
vuZQqQt.jpg

68rKXpa.jpg

There's a few things to take note of here. The sheet of plexi is not very big and it's just wedged underneath the top and bottom edges around the logo. Also they've seemed to use some sort of glue or adhesive tape on it as well.
Second of all, the outer layer of metal(the brushed aluminum) is really, really, really thin.

vMHfWcq.jpg

Now, if anyone reads this and want to replicate this... What I did was that I brought out my trusty heatgun and heated the plexi so I could cut through it(see previous image). Totally unnecessary!
What you should do is simply heat up the plexi until it becomes very pliable and then just push it out of the top and bottom edges. I used a tiny screwdriver and pushed through the logo(on the frontside) just to get the plexi to lift a little, then I used the hole I had drilled previously to use as an anchor point for when I pulled the plexi out of there.
I do not recommend pushing through the logo like I did because the risk of bending or even snapping the metal right off is quite great. Better to have an anchor point so you can pull the plexi out instead.
Oh, and as I mentioned. DO NOT cut the plexi, it serves no purpose and if you're careless you might even cut through the brushed aluminum piece(see previous images and the marks I made into it by cutting).

With that done it was time to change that color. My idea all along had been to use some lighting filters that are used for spotlights(professionally) or camera flashes. My brother who's working in that field hooked me up with some samples. This is just half of them I think:
NbSfhfk.jpg

Over 150 different colors, hues and transparancies. You can easily buy something similar on eBay for pennies.

Sorted through them and selected a few purples that most closely resembled the tubing in color. Not knowing how the LED light would change the color I thought it would be good to have a few different ones at the ready:
6eDs83G.jpg

All cut(with a razor) and installed, easy peasy:
3qmyXR7.jpg

One of the good things about this Asus ROG SLI Bridge is that it comes with a power cable in case your GPU's can't light the led by themselves. So that made it easy to test the color and see if it was right without installing the SLI Bridge and powering up the computer.
Excitement!
1c7QBhn.jpg

Fail.
Turns out the LED is also red. Not only the plexi but the actual LED too. Not at all what I expected. But since I had all those different colored filters I figured I could just use a blue one instead since blue + red = purple.
So I cut out a piece and installed it and tried again:

kIcvrck.jpg

Fail.
Now the light of the LED barely shone through and there was some bleeding at the edges where the red shone through.

I tried all sorts of different combinations of blues and layers of blue. Here's a couple more.

iwtibhZ.jpg

2Flgpdp.jpg

I finally got so fed up with cutting out pieces of the filters that I simply plugged in the LED and put the filters on top just to see what it might look like. Nothing looked good.
If I got a good purple hue then the light was too dimmed. If the light was good then the hue was bad. Nothing worked.

This is the last one I did today. I might keep it, I might change it. It depends on my mood in the coming days:

2Flgpdp.jpg


I also promised a while ago to show you my custom cables and some of the pitfalls you can quite easily fall into.
Here they are, all done, far from perfect and in some cases absolutely messy:
39OiCdQ.jpg

The 24-pin "looks" good, but only because I've made two bends on the cables. There is still quite a bit of gap in between the wires but if I only had done one bend it would be way worse:
hM3RqJG.jpg

This is the 4-pin 12VATX cable. This one illustrates this problem even more. With just one bend the gap between the wires is humongous:
gWFD1ad.jpg

With two bends it's decent, but far from great:
Y2K0OKN.jpg

So what is the reason for this?
Wire length. Basically you have to adjust your wire length depending on how you are going to bend the finished cable. The wires on the outside of the bend needs to be longer than the inner ones, if not then you get this sort of issue.
There are two factors to wire length, the wire itself and the crimp. If you strip the wires unevenly you might place the crimp at a different spot on each wire which will affect the overall length of your wires.

Moving on... This is the 8-pin 12VATX cable:
kXckswD.jpg

Once again we have the wire length issue, but not quite as severe since I started paying attention to this at this point. But do you notice the twisting of the connectors/wires?

This, again, has two causes(from what I can tell) and the obvious one is that the crimps at either end of the wire aren't aligned. There is some leeway with this alignment but the more one of them are twisted compared to the other, the more twisting of the wires/connectors you get.
The second cause is the sleeving itself. If you have different amounts of slack on the sleeving you get different amounts of tension on the wires which will cause them to twist.
I really did not think that this would be such a major issue but it sort of is. The amount of slack I have on the sleeving varies a lot between individual wires.

These are my GPU cables, with basically all of the above issues exemplified:
BiXyo2B.jpg

But it's not all bad because if you just think about it a little when you are assembling the cables and if you use good cable combs you can make the "problems" appear at only one end of the cables. In my case the cable ends above will not be visible since they'll be at the PSU end, hidden behind the motherboard tray/wall.

This is what will be visible, and it looks sort of neat, even though the cable training isn't quite done yet:
wQ5uYF0.jpg
 
Thanks all.

Not so surprising TechMinerUK because the filters all have different transparancy values. They range from 6% up to about 65%. The lower the number the less light can pass through. With a high value transparancy the LED is really bright.

Ah, that would explain it. Thanks Cenedd!
I'll have to take a closer look at the LED construction but from what I saw earlier today it won't be easy to add or change the LED. :(
 
Cenedd, Yeah there's plenty of room to add another LED but I'm unsure of how I'd be able to power it.
I did take a closer look at the construction and the LED part is just two tiny LEDS(about 2mm wide) that are directed to another piece of plexi that then spreads the light.
I see no solder joints for these LEDS though so I think it's out of my league to swap them out but that might be an option as well(or just swap one out for a blue LED).
(Might show an image later today)

Thanks Zeeflyboy!
You are quite correct, those bends with multiple direction changes in tight spaces are quite troublesome but with freehand bending and some patience it ain't as bad as I suspected. :)


As of now the loop is running trying to cycle out the air which is a major PITA.
There is one issue though... The inlet port on the res is at the top and there is no pipe from this inlet that goes a little bit into the res. That means that the water goes straight from the fitting into the res, creating bubbles(at least with pump at full speed).
I could fill the res entirely but that leaves no room for air or expansion when I finally seal the loop so I'm not sure that is such a good idea.
Any advice on this part?
 
Good idea about the G1/4 die. I'm hunting high and low right now to see if there's a place in town that has one in stock.

As for the LED. Yes there is a diffuser but the LEDs are mounted by it's side and I see no solder joints on them. But yes, they can probably be swapped out.
I'll try to show an image of them later today.
 
When life gives you lemons...

I bought a whole set of dies just in case... and it turns out I got it all wrong and none of them works. *sigh*
So it's either ordering a new die, the EK internal tube, or going full ghetto on this.
I struck me that I could fit a fitting inside the reservoir and fasten the tube with an o-ring and not install the compression cap. That way I won't get any "bad" metal in my loop and I can have it done today. But I'm really not sure it's a wise move.

Ah, the greatness of water cooling :D


Or well, there is yet another solution. Make a new inlet tube that goes to the bottom of the reservoir. It can be done but it will be a major, major PITA.
 
Setbacks are only troublesome when they happen. Brush it off and just go at it again, right? :)

So, the mistake I made yesterday was that I was unaware of that there are different dies depending on what they should be used for. The ones I bought were not for pipes but for rods.

Thinking about various solutions yesterday I finally decided on trying out the idea of changing inlet port on the reservoir. I was a bit worried that I might just move the problem though since I might end up with the vortex effect by doing this.
After I woke up and had some morning coffee I got to bending. It's not perfect, there is a slight angle to it, but I'm happy with it for now:
pT3J5Gn.jpg

Filled the loop again and have had it running for 20 minutes now with no apparant leaks. No vortex either which was a major surprise.
Now I'm trying to vent the loop which seems even harder now than last time. I'll get it done though.


And I promised some LED pictures yesterday.
This is the basic construction of it:
cvp4C7O.jpg

L1 and L2 are leds and upon closer examination there might be small tacks of solder at the edges. Either way I will not bother with trying to swap one out at this point. I might down the road though.

This is what it looks like naked but lit up:
D01wIsM.jpg
It is more red in real life, way more red.
 
Thanks all.

Good tips regarding the LEDS. I'm sure I'll swap them out at a later point but right now I just want the build finished. :)

Zeeflyboy, I only have my own stupidity to blame but I didn't stop to reflect that G1/4 is vastly different than 1/4 :rolleyes: :(
So when I researched this a bit later most sources said that the G[measurement] dies are mainly meant for tubes/pipes/etc and mainly for plumbing use.



Yesterday was mostly spent trying to vent the loop. I think it's as close to vented as it can get. The top radiator was really stubborn with letting the air pockets out. It probably took a couple of hours just to vent that part. The rest of it seemed quite fine from the get-go though.
There are some tiny bubbles stuck on the reservoir wall that refuse to move, but apart from aesthetics I don't think they're an issue.

Speaking of venting by the way, what I haven't mentioned is that my drain port works a charm. I partly drained the system yesterday so I could install the new tube and it went really smoothly.

While the system was venting yesterday I decided to mount a couple of things. To do that I had to drill a couple of holes:
eh2ze3C.jpg

Installed in an unorthodox spot:
OHaay8D.jpg

I also needed a place to mount the fan hub where it wouldn't be visible and the cables wouldn't be as well. Said and done:
MSOb2U5.jpg

Late last night I started to do the cabling, I still have some left to do but most of the tricky ones are done.
Getting the 8- & 4-pin 12VATX connectors in behind the Monsta rad was a pain but I still managed by just being patient.
The 24-pin was extremely tricky to get in, mostly because of the tight spaces and the PCIe cables were a breeze after that. :D
Here's a little sneak-peak:
bKOhPDj.jpg

There should hopefully be an update tonight with some finished shots and perhaps some benchmarks.
 
Zeeflyboy, Thanks.
Well the "G" isn't specific to imperial but used for metric as well. But either way, I was unaware and I'm definitely not comfortable with the imperial system which didn't make things easier ;)

TechMinerUK, I'm sure it would fit the standard DVD-mounts but I didn't mount it on a 5.25" bay but rather on the side of the 5.25" cage. Unnecessary maybe but I like it.

And no, I had done no cable routing earlier, simply because I did not want to risk the cables getting wet upon a disastrous leak. In hindsight it would've made things so much easier if I had at least installed my custom extension cables before installing the water cooling. Another lesson learned ;)


Back to the cables I go :D :D
 
Of course Cenedd. These are the measurements I got:

From the mounting surface to the top of the power connector: 2 cm.
From the mounting surface to the top of the plugged in connector: 3.5 cm

I would imagine you would need about 4 cm of clearance from the mounting surface to accomodate the cables(because a 90 degree angle of the cables directly from the connector might be hard to achieve).

If you remove the standoffs and have a thin isolating material between the aquaero and the mounting surface you could gain a few millimeters as well.


In other news...
Cabling is done, photos have been taken with everything assembled apart from the case exterior. Photos will be shown tonight. :)
Going to boot it up in a minute and hopefully all goes well.
 
Let me know if you need any other measurements Cenedd :)


I'm not going to say these are the final images of this build but at least they will show it completed apart from the case exterior. I'm going to start doing some overclocking before putting the case together and once that's done I'll do final build shots for you guys :D

The SLI bridge sticks out when powered off:
mtxIjYP.jpg

Pn4ln4a.jpg

XP0GwDm.jpg

I hadn't noticed that my cable combs had moved after I did the final cabling. This will be fixed:
I5pwVxH.jpg

tBfED1J.jpg

5XtCjeJ.jpg

kCwa8Og.jpg

qDRUwTX.jpg

Don't know why but I really like this image:
7Sp6aYi.jpg

WFsvziS.jpg

C4XpbbL.jpg

Bubbles for everyone!
e1gauUF.jpg

When you make short extension cables and don't install them early you might have to improvise "a bit":
Q2IoFgr.jpg

Not quite a rat's nest but pretty close. Still the best cable management job I've ever done(mostly because I never cared before):
cf9AuEO.jpg

Who's got fans? I got fans! 16 of them in fact ;)
Aw0qmZu.jpg

Rear window? Not with this cabling :D
lydnjL9.jpg

I'll have to do better in the final photos when it comes to showing the system under power, but at least this will give you an idea:
CWsvzwP.jpg
I noticed later on that one of the GPU leds had come loose so I have to fix that as well. The yellow LEDS match the fittings really well, but in turn it makes the yellow cables a bit too bright.
The reservoir is lit with white LEDS so it is purple, not blue as it looks here.


First boot went fine but for some reason it took three reboots before I could get the computer to recognize my monitor. Really strange but it works fine now.

Temps at idle and fans at 300 RPM and pump running at 50%(ambient temp is 24 degrees).
NK6YpOP.jpg

Firestrike Extreme test:
8tUhw4V.jpg

I have no idea if that is good or not, but 3DMark says it is. :D
Temperatures peaked at roughly 46 degrees during the test. I thought it was a little high until I realized that fans were still running at 300 RPM :rolleyes:
After doing a quick setup of the Aquaero I redid the test and peak temperature was roughly 38 degrees(pump still at 50% and fans at slightly less than 800 RPM(I think)).

Did the normal Firestrike test as well
bkhrNze.jpg


Next up for this build is overclocking. Something I haven't done in 15 years so much have changed and I've honestly forgotten most of it.
I've got some reading to do, and then system tests before I'll start, but I'll keep you updated. :D
 
Thanks Bossman & Cenedd!

As for your questions Cenedd, I bought the F4-120ER Vardars which have a greater PWM span. So they can be controlled from 300 RPM up to the full 2200 RPM through PWM control. If you use voltage control you might be able to make them go even slower if you wanted.
The non ER Vardars should be around 40% as you mention.

As for sound, I can't hear the fans at 300 RPM. In fact, when I ran the first Firestrike Extreme test I could hear the GPUs working over the fan noise.
Even at 800 RPM I consider them to be fairly inaudible.

I have connected all fans to a Phanteks PWM hub(with splitters) and then I connect that PWM hub to the Aquaero. The pump is connected to a PWM header on the motherboard.
 
Thank you everyone for your kind words! :D

Been tweaking the cooling with custom fan & pump controller curves which has made the cooling slightly better while still being virtually silent.
I've tried two other benchmarks, Valley & Heaven, but don't really know if my results are good or not(I'll post screenies in a day or two). I ran these mostly to have a reference value when I start overclocking.

And speaking of overclocking, I've been reading up and am beginning to get the hang of it. As soon as I can dedicate half a day(or a whole) I'll start the OC adventure. First with the GPUs and then the CPU & memory.
 
Thank you for the kind comments.


So I've slowly started my OC adventure and it sort of follows the rest of this build... unforseen issues. :(

The GPU's are sort of okay with a 150 core clock and 300 memory clock without any added voltage. Anything above this and Heaven crashes during benchmark.

The CPU is a whole different story. Unless I'm missing something obvious I think I've really lost out on the silicon lottery.
I keep getting stuck on loading Windows, no matter what sort of multiplier and vCore I'm using. I tried up to 1.26 vCore and 38x multiplier for 3800 clock and it just won't get past the Windows loading screen.
I tried the built in overclock settings on my motherboard and same story.
I also played around with LLC but it did not help either.

Just for giggles I tried to keep vCore at auto and then using a 38x multiplier and I actually get into Windows then and can run Intel Extreme Tuning Utility but it crashes after 7 minutes. The really funny thing is that my reported vCore during this was 1.075.
I guess I'll have to keep reading and see what I'm missing... or just keep it at stock speeds :rolleyes:
 
Thanks.

I'll have to look into DIP and optimization software. I don't recall seeing either tbh.
I'm not really sure what is wrong with my OC attempts and it's sort of frustrating :(
 
Back
Top Bottom