Water Block Deconstructed
XSPC designed the Razor GTX680 water block with four distinct layers. The top brushed-aluminum plate and the acrylic layer are held to the block by the five inside black screws, sandwiching the acrylic layer between the aluminum and steel plates. The four screws on the outside left and right sides of the top layer hold the top plate to the acrylic layer only and do not pass through to the steel plate. Both the top plates have cut-outs for the multi-port connect, which is bolted directly to the block. The acrylic layer also has a cutout in the lower right corner to make room for capacitors and power components on the PCB.
The multi-port connector is directly bolted to the block with four screws. The inlet and outlet holes for water channels are defined by the holes in the steel plate. The steel plate is held to the copper under plate by nine steel screws. From the under-view of the block, you can see that those nine steel screws are the only screws going in to the copper plate. The five screws from the upper layers are affixed to the steel plate only. The steel plate has a cut-out in the upper center-right quadrant to make room for capacitors and power circuitry on the PCB.
With the steel plate removed, you can see how the steel screens and the multi-port connector screws work in concert to create a a water tight seal between the steel and copper plates. The copper plate contains two o-rings, an outer o-ring sealing the block and an inner o-ring preventing water into the screw hole located in the center of the block. Both o-rings site in pre-formed channels directly milled into the block by the fabrication process, ensuring a perfect seal. While some of the screw holes look to be in close proximity to the outer o-ring channel, XSPC engineered the block to exacting specifications to negate any overlap between the two.
The block's inlet is the channel to the left. The water enters the block and is forced down over the first set of memory chips. Then the water flows through the micro-channels sitting on top of the GPU chipset. The micro-channels have two functions. First, they increase the surface area over the hottest part of the video card, ensuring optimal heat transfer from the chipset. Second, the micro-channels force an increase in the water velocity because of the pressure increase caused by entering the channels and the decrease on the other side. The increase in ther coolant velocity ensures better heat transfer to the coolant and dissipation from the GPU. While a pin matrix-style channel design may have netted an increase in heat transfer and dissipation, the micro-channel design does its job flawlessly.
The only downside with using a micro-channel design is the increase in flow restriction inherent in its use – pressure builds up on the inlet side because water flow is restricted by the micro-channels. While the water velocity through the micro-channels is increased, the flow rate over the entire system decreases because of the micro-channels.
Once the water exits the micro-channels over the CPU, the coolant flows over the remaining memory chips and the VRM circuitry. Notice that XSPC designed the Razor GTX680 block with a pass-through for the water, splitting the flow. In splitting the flow, XSPC may have found a way to increase the coolant flow rate, combating the negative effects of the micro-channels.
The coolant outlet channel is the upper right channel on the block, just above the inner o-ring.
The multi-port connector sits at the top of the block, held to the steel and copper plates by four screws. The connector contains three inlet ports on its left side and four outlet ports on its right. The three inlet ports are in a line on the top, outside top, and bottom of the connector with the top and bottom ports forming a pass-through. Similarly, three of the outlet ports are arranged in a line on the top, outside top, and bottom with the forth outlet port on the outside right of the connector. The variety of inlet and outlet ports available allow for better flexibility when linking the blocking into the reset of the system water loop.
The under side of the connector contains the contact points between the connector and the block. The connector contains a built in guide to ensure proper construction when the block is disassembled for cleaning. Further, both of the channel holes have o-rings sitting in pre-formed channels, ensuring a water-tight seal between the connector and the steel plate.
The acrylic middle layer contains LED ports along the outer left and right sides of the plate, capable of housing 3mm LEDs. If you choose to use different LEDs than those included with the block, make sure that the selected LEDs have wire leads that can accommodate the 12 inch span between the left and right side LED mounts on the block.
Is there a reason we you only
Is there a reason we you only show the temp over ambient vs actual temp? Most households would have ambient temps between 22-26. I just think actual temp would have been easier for an idiot like me to comprehend.
Of course, there is ambient
Of course, there is ambient temp within the case, which is probably what should be used… but still confusing. Just give me temps! TEMPS!!!
We report delta temps instead
We report delta temps instead of actual temps to allow you to calculate what your temps would be based on your house or office ambient. The delta temps are the card gpu temps minus ambient temperature measured at the time of testing. We moved to reporting delta temps rather than measured temps based on user feedback from previous cooler reviews.
For example, typically my home office temp ambient is between 26-27C. For stock temps with the ambient at 26C, the GPU temp should idle at 28C and go up to 47C under load temps. If the ambient were 21C, the idle temp should be 23C and load at 42C.
add the numbers in the graphs
add the numbers in the graphs to your ambient. There’s your temp.
your ambient: 22
card delta: 10
22 + 10 = 32 degrees
math = gud
Yea, I just didn’t realize
Yea, I just didn’t realize what I was looking at initially. Many times I elect to just jump straight to the graphs, and they didn’t state anywhere that they were based on ambient… had to actually read. Knowledge is power!
Finding it hard to justify
Finding it hard to justify $99 for this waterblock.
The 10% overclock is not unrealistic for an air cooler, especially an aftermarket cooler.
Aftermarket coolers are typically very quiet.
Real-world usage will not push a 6xx/7xx series card with an aftermarket cooler beyond 70C, perfectly acceptable ultimate/maximum-use temperature.
Waterblocks are typically made for reference model designs, which fail to match the benefits of aftermarket designs.
If you can find an aftermarket model for the same/near the same price as a reference model, why bother with watercooling?
Keep in mind that the EVGA
Keep in mind that the EVGA GTX 670 FTW card used for the review came overclocked from the factory, so a 10% overclock with that card equates to a more than 20% overclock on a vanilla GTX 670. On a non-factory o/c'd card, the boost clock speed for the GPU typically hits 980MHz.
The choice between using an aftermarket cooler vs a water cooler is a long running debate that really has no right answer. I've been water cooling my rigs for a long time, and prefer that to pure air cooling mainly because of the fan noise associated with air coolers if you want to push the performance limits. It really comes down to the preference of the inidividual enthusiast…
Steel touching copper with a
Steel touching copper with a fluid running between them. I’ve had some issues with this before.
As long as your coolant has
As long as your coolant has some type of corrosion inhibiter in it (which most good coolant's should have), you should not run into any cross-metal type problems like galvanic corrosion…
The coolant is supposed to be
The coolant is supposed to be “non conductive”, so electrolytic/galvanic corrosion is a non-issue. Especially since the steel used is stainless and the copper is very pure and the lifespan of the cooler is much much shorter than it would take for the electrolysis to damage it so badly as to cause a malfunction. Would it be Aluminium instead of steel, it would “pee” in a few weeks.
Most of the good pumps used have a stainless steel part touching the coolant and there are no horror stories floating around.
Mild steel is a completely different story, that much is true.
Morry – i love your reviews,
Morry – i love your reviews, just one question: do things work when you put them back together? jk. keep up the awesome work!
Thanks. And yes, in most
Thanks. And yes, in most cases they do work after putting them back together. There was one time in the recent past were teh take apart result in disaster – but we won't discuss the H80i dissasembly here…
I like to see both because
I like to see both because what many heatsink reviews fail to show is the cooling performance curve with ambient temperature, eg CPU temperature will not scale linearly with ambient temperature, it tapers off slightly as ambient gets higher.
so temperature over ambient may not be the best indicator if you have some test done in a hot environment, where a 5 degree rise in ambient may only cause a 3-4 degree increase in CPU or GPU temperature.
I notice this with my current air cooler.
Good point. I will continue
Good point. I will continue to report in deltas in future cooling reviews but will include ambient temperature at the time of testing as an additional data point either in the graphs or test description.
Typically, my office ambient runs between 25-27C which is most likely typical or a bit hotter than most homes / home offices.
Thanks for the feedback…