Recently I was given the opportunity to test an Aquajoe waterblock in “real world” conditions. What follows are my observations and thoughts on the block and how it compares to my other block, a modified Swiftech MCW462.
Prologue: First, my expectations about the block prior to receiving it:
I had read a bit about the block prior to getting it – much of what I read had emphasis on the difficulties that the mounting hardware provided. Also, a question was raised about the testing of the block done at another site, which showed that while it didn’t beat Cathar’s block, it did come very close to matching it. Given that Cathar’s block was the king of the hill at the time, that is impressive, although I had some questions in my mind about the conditions surrounding this test.
So at this point, I suppose you might be thinking that I expect this block to fail and fail miserably, right? Wrong. While I do find some aspects of this other review troubling, I still nonetheless expect his block to perform better than mine. Why?
Well the block I currently have is a modified Swiftech MCW462:
It’s a great block but it’s one of Swiftech’s older models – they don’t even sell it anymore (unless you get the one with the TEC attachment). Aquajoe has had time to look around the market, see what products are out there by other companies and search forums to see what kind of innovations people working away in their garages/shops are tinkering with. They’ve had time to gauge the flow of what works and what doesn’t.
The Block Arrives: I found a slip of paper in my mailbox letting me know that I have a package waiting for me at my local post office. So I head on down and get the block. I get it home and open it up, thanks to JoeC my collection of packing peanuts is getting bigger. 😉
So now I have the block but have yet to test it. The initial impression I have is that this block is a diamond in the rough. The base looks as if it has been hand lapped to about 300-400 grit, and not lapped for very long. I can easily see, and feel, the scratches in the base.
By comparison, my Swiftech block is flat. Swiftech’s site claims that it is “lapped to 8 micro-finish or better” flat. Now I don’t have anything to measure that, but from my own personal observation it looks as if God waived his hand over the block and said “And I shall call this, the industry standard benchmark of what is flat!” and it was so.
Something that Aquajoe has done to their block that Swiftech didn’t with my older one was to pre-drill a hole for a sensor in the base of the block. That’s a piece of detail attention that I like.
On to the Mounting Hardware: The block comes with 8 steel washers (four big and four small) and 8 white plastic washers. I think the plastic washers are all supposed to be of the same size but mine weren’t. One of them has a smaller diameter (which shouldn’t affect mounting in any way) and was slightly thicker, which could affect mounting the block, if the washers are not all the same thickness and I were to tighten the nuts an equal number of turns to try to get a balance then I would get uneven mounting pressure.
All of the mounting equipment that I got with my Swiftech block was uniform and even. And finally, the paperwork. There was no paperwork. Perhaps people who get their blocks directly from Aquajoe will get some kind of an instruction sheet, but I didn’t get one when I got my block from JoeC.
Aquajoe’s site does have mounting instructions and they can be printed out, but there are times in everyone’s life when you aren’t going to be able to get on the net. Hopefully one of those times won’t be when you are trying to install or print out instructions for the Aquajoe block. The Swiftech site has downloadable instructions for mounting their block but they also ship a printed version of those documents out with the block.
Preparing the CPU: Up until I got the Aquajoe block, I did not have an accurate temperature sensor for my system, but since I wasn’t putting together a review of a product for someone else, I didn’t really need one. Now I’m not doing a super-scientific test like BillA or JoeC would but it would be irresponsible of me not to try to provide some tangible numbers for everyone, so I went out and got a Hardcano II hard drive cooling system, complete with two temp sensors.
I said earlier that the Aquajoe block has a pre-drilled hole in its base for a sensor but the Swiftech block doesn’t. Since I don’t have a drill press and am not too confident in my ability to hand drill a sensor hole in my Swiftech block, I won’t be using the sensor hole in the Aquajoe block. Instead, I put one of the Hardcano sensors on the underside of the CPU and taped in place with the doublesided tape that came with the Hardcano. The other sensor I left out in the open to measure ambient temps.
The Mounting Hardware: So now the Chip’s ready, now to get into the mounting hardware for the two blocks.
The Swiftech block comes with four aluminum motherboard standoffs and several different sets of washers and nuts to cover all of the possible motherboard mounting hole sizes. These are easy to work with if you glue the nuts onto the back of the motherboard or have exposed access to the back of the board to hold them in place, otherwise you have no way of keeping the nuts from spinning around when you try to tighten the block down.
That’s one drawback of the Swiftech mounting system, the other is the standoffs themselves. I have an older motherboard, the CPU socket has a plastic arm that needs to be moved out from the socket a bit before lifting the arm up. This is to clear the little plastic “nub” that is used to keep the arm locked down.
The Swiftech standoff that is closest to the end of the socket arm interferes with moving the arm out before lifting it up. Many of the newer motherboards that I’ve seen have a straight metal arm, so the Swiftech mounting system shouldn’t be a problem for them. I dealt with the first problem by crazy gluing the nuts to the back of the motherboard – man, were they a chore to get off without gouging any trace lines!
The second problem I fixed by cutting away part of the length of the arm; it’s still long enough to lock down around that plastic nub but not long enough to hit the motherboard standoff when I lift it out and up.
By contrast to the Swiftech block, the Aquajoe block is easier to install. It uses two arms that run the length of the socket sides and have two threaded bolts that protrude out and down through the mounting holes. On the back side of the motherboard, each of the four bolts is locked in by first putting one of the big white plastic washers over it, then putting the smaller metal washer over that and finally socketing down the end nut.
I’d like to thank Mr. B for mentioning that a 3/16″ socket is the perfect size for tightening down those nuts, it saved me a bit of fumbling. The Aquajoe site does explain how to connect all of these little parts, but I’d prefer some more shots of the installation process to go with the text.
The site says to first hand tighten the small nuts to the bolts on both ends of the retainer, then line up the metal washer so that it’s centered and tighten them fully. Centering the metal washer is important but it’s not as easy as it may seem.
While it was not difficult for me, many of the bolts and nuts that I use to hold case fans in place are of the same small size, so perhaps I just have more experience with them than others. I did take Mr. B’s advice and use a 3/16th socket to tighten the nuts down, but the socket overlaps the entire size of the metal washer (and then some), so it is impossible to tell if I had the washer centered when tightening.
Hand tightening makes controlling the location of the metal washer much easier, but the tightening itself becomes more difficult. Given the small size of the parts, it’s something of a damned-if-you-do, damned-if-you-don’t proposition if you try either, but if you do both (as the website implies) hand tightening to make sure it’s all lined up, then switch to the socket to tighten it down, it should all go fine.
One good thing about having two bars as the mounting hardware instead of four separate standoffs is that once they are locked into the motherboard, you don’t have to glue the backs of them down to the motherboard, as they won’t keep spinning around and around like the Swiftech standoffs. The bad thing about this style is that if you have a motherboard like mine (with the plastic socket arm that must be moved out before being lifted up), you have to completely remove one attachment to remove the CPU. This shouldn’t be a problem for those with the metal socket arm.
Hosing the Blocks: At this point I’m ready to hose the blocks.
Before going any further, I should explain the specific mod done to my Swiftech block. I really didn’t like the small size of the quick connects and no hardware store in my area had anything that was remotely helpful. So I turned to my friend Harry (forum name Joachim) for help.
I told him what I wanted and, as an engineer, he took my ideas and tweaked them. My block went from looking as shown above (straight from Swiftech’s site), to being gutted
as shown above, to being threaded
as shown here (The green circle shows where Joachim had to epoxy a hole that widening the outlet opened, keep this in mind if you plan on doing a similar mod to your MCW462) and finally to having home-made, true 1/2″ ID barbs threaded into the inlet and outlet.
The barbs can be seen here
compared to an “off the rack” 1/2″ barb. Sorry I don’t have pics of the barbs installed in the block – Joachim’s the one with the digital camera.
So now that you know what’s on my Swifty, let’s get into hosing both blocks.
I’m using generic vinyl 1/2″ ID hose with 1/8″ thickness, it’s good and sturdy hose. The Swiftech was easy to get the hose on. Very easy, ’nuff said.
The Aquajoe block was appreciably more difficult. It wasn’t the size of the barbs that was difficult, it was their proximity to one another. The barbs are so close that using 1/8″ thick tubing means the inlet and outlet tubes are snugged together, very snug.
If you aren’t planning on using clamps (and from the looks of these barbs it doesn’t seem insanely unreasonable) then you should have a bit of trouble putting the second hose on. If you are going with clamps (which I always do) then you are going to either have one hell of a time putting the metal screw-in clamps around the hose or you are going to give up trying to put those plastic alligator type clamps with the locking teeth on.
Mounting the Blocks: So now the block is hosed, what’s next? Mount it on the CPU.
Because I’m checking out a waterblock, not thermal grease, I used Radio Shack generic off-the-shelf white goop. I wanted something that didn’t have a “settle in period”. In the case of the Swiftech, it’s a breeze in every way except one, the movement of the standoffs.
If you glue the nuts to the back of the motherboard as I did, then mounting it is easy; if not, then you could get into the position where you keep twisting the screwdriver and the whole assembly rotates, instead of the base staying in place and the screw tightening itself down.
Now this is more likely to happen when untightening the block to remove it than when tightening, but either way it’s irritating. Beyond that, the Swifty presents no problem. The screws that hold the block in place cannot be tightened beyond the depth of the threading in the aluminum standoffs, so you can never (assuming Swiftech did a good job in measuring the lengths of the standoffs and screws) tighten it down too tight. This makes mounting the Swiftech block a joe-sixpack’s dream, it’s generally fool proof.
The Aquajoe block is a different matter…
The mounting bolts that protrude from the two arms at either side of the CPU socket have no way to limit them to prevent severe overtightening – you could even tighten them down completely without the springs, which could be disastrous. Removing the springs from the Swiftech block would also be disastrous, but only because it would prevent proper tightening, not allow excessive tightening.
While being just as destructive, in the case of the Swiftech block, tightening down the screws without the springs would be an obvious error – the block would rattle around on the bolts not sit tightly on the CPU, that’s not a mistake that could be overlooked, such as tightening the Aquajoe block down until the core is crushed.
The Aquajoe site shows pictures of nuts used to tighten the block down – this looks like an uneasy chore. I say looks like, because my block didn’t come with nuts, it came with wing nuts, which while being easier to hand tighten than regular nuts, they did present a new problem.
The “wings”on the wing nut closest to the outlet barb brush up against the tubing on the barb – it’s another case of too close for comfort, but unlike trying to get hose clamps squeezed on, this isn’t the same degree of closeness. The wing nut brushes up against the hose but it doesn’t actually block it from being tightened down.
I expect that 1/16″ thick hose would not present any impediment at all. Since the Aquajoe block has such a long range of potential tightness (unlike the Swiftech block, which is limited to the depth of threading within the standoffs), it becomes much easier to tighten the four wing nuts down unevenly.
This is a Joe Sixpack’s nightmare.
I myself went out of my way to measure the depth to which I tightened each nut, just to make sure I had tightened the block down evenly, and even then I had no way of knowing if I had tightened all four nuts down way too much or not nearly enough.
Fill ‘er up, let’s hit the highway:
So now that everything’s mounted, it’s time to fill the system and put it through it’s paces.
In the past I used a mixture of de-ionized water and DEX COOL; in the time between then and now, I’ve learned that de-ionized and distilled are not the same thing and that, by most people’s view, distilled is better. So I went with a mix of 80% distilled water and 20% DEX COOL.
Well by this point I’ve done most of what I set out to do, I evaluated this block from the perspective of an end user, how easily it can be mounted into a system, the quality of the finish and looked for any minor glitches that could be tweaked.
Now comes time to see how it performs:
I’m not going to pretend that the numbers I’ve come up with are lab quality measurements like BillA or JoeC could produce – the reason I was given this block to test was so it could be evaluated in “real world” conditions. Nonetheless I am going to give numbers that are equally “real world” for both blocks so that you can see how well it performs.
My system is an AMD and socket thermistors are notorious for their inaccuracy, especially when you factor in that water cooling removes the CPU fan that blows air around the CPU socket, so I didn’t even bother to look at what ASUS PROBE was going to tell me; instead, I relied on the numbers from my two Hardcano sensors. As a result, I do not have any screen shots of my temps. So here are the numbers, I may not have used the same test that you would use, but I used the same tests for each block:
- AMD Athlon 1.2 gig
- Underclocked to 12.0 x 100 from 9.0 x 133 (stupid KT133 non-a chipset!)
- Stock voltage 1.75v
- ASUS A7V-E motherboard
- Radio Shack generic white thermal paste
- Open case, no doors on either side
- Eheim 1250 pump
- 1/2″ ID (3/4″ OD) vinyl tubing
- 1/2″ ID copper “T” fill line
- 1993 Mustang heater core
- Half shroud with one NMB, 120 mm, 84 cfm fan (push only, no pull on other side) down volted to 10.5v using diodes
The ambient sensor was placed on my desk about 2-3 feet from the system case.
- Ambient = 25.1°C
- Idle temp = 38.0°C
- Peak temp during 3D Mark 2001SE = 43.3°C (it is so boring sitting there for 8 minutes just counting temps!)
- SiSoft SANDRA Burn-in 10 minutes max temp = 43.9°C
- Ambient = 25.0°C
- Idle temp = 37.1°C
- Peak temp during 3D Mark 2001SE = 40.4°C
- SiSoft SANDRA Burn-in 10 minutes max temp = 40.7°C
So there you have it – the ambient temp is roughly equal when testing the Swiftech MCW462 and the Aquajoe block, but the Aquajoe load temps are about three degrees lower. The pin array format beats the “big hollow chamber with rough baseplate to scatter water” design, so my initial performance expectations held true.
My end conclusion, having set up and tested the block, is that it is a good performer, but if I were unfamiliar with water cooling or otherwise had little to no hands-on experience monkeying around inside a computer case, I would be hesitant to use it.
It’s great for those who already know what they’re doing, but for those members who start threads with: “I’m a noob to watercooling and I was wondering which kit you recommend….” I think the built-in safety of the Swiftech block could be a better choice. If Aquajoe can come up with some form of limiter that will prevent catastrophic overtightening AND do a better job lapping the bottom of their blocks, then I think they could have a winner all around.
My thanks to Skip for having the best compuer related site online, JoeC for giving me this opportunity to test out a new block, to Aquajoe for providing the block to JoeC for me to test out, all of the members of Overclockers.com who watercooled before me for the inspiration/ideas they gave me and finally to my cat for proof reading this document.