Features


Overclockers Forum

Shopping Search



Top Products


Motherboards

Intel

Abit

Gigabyte

Asus

EPoX

Iwill

MSI

Shuttle

Tyan

Soyo


Processors

AMD

Intel

Compaq


Cases

Antec

ATX

Enlight


Graphic Cards

Leadtek

ATI

Creative Labs

Hercules

MSI

VisionTek

nVidia


Memory

SDRAM

RDRAM

DDRAM

DDR


Sound Cards

Creative Labs

Jaton

SIIG

SB Live

Guillemot


Hard Drives

Fujitsu

IBM

Maxtor

Quantum


Networking

Linksys Lan Cards

Ethernet Cards

FDDI Cards

Networking Kits


Misc.

Monitors

CD Roms/Burners

Printers

Scanners

Software

REVIEWS & TIPS
Over 1000 Topics

ACCESSORIES
ADAPTERS
BEGINNER GUIDES
BENCHMARKING
BIOS
BUYING ADVICE
BX COOLER
CASE COOLING
CASE REVIEWS
CONTESTS
CONTROLLERS
CPU REVIEWS AND TIPS
HARD DRIVES
HEATSINK REVWS AND TIPS
HUMAN INTEREST
MEMORY MISCELLANEOUS
MOTHERBOARD REVIEWS
OVERCLOCKING EXPERIENCES AND TIPS
PELTIERS
PROBLEM PRODUCTS
REPAIRS
SYSTEM COOLERS
SYSTEM REVIEWS
THERMAL GREASE
UTILITIES
VENDOR REVIEWS
VIDEO CARD COOLING AND PERFORMANCE
VIDEO CARD REVIEWS
WATER COOLING


Please read EMAIL FAQs first: Comments, suggestions, and questions to Joe Citarella, Skip MacWilliam, or Ed Stroligo

"Swiftech MCW50-T GPU Waterblock"
Joe Citarella - 3/11/04

page 3

Using a TEC

Effectively using a TEC involves more than simply slapping one on and running it. I strongly advise anyone interested in using this product to carefully read all the items HERE. In particular, Swiftech notes that:

  • "The combined amount of heat (TEC + GPU) added to the circuit by the MCW50-T is approximately 115 watts (392 BTU/hr).
  • The thermoeletric module pulls approximately 6A at 12 volts. When connecting the MCW50-T module to an existing power supply, users should verify their power supply current rating for the +12 volts line.
  • [The MCW50-T] will dissipate 53 Watts continuous thermal load to ambient temperature."

As I tested this unit, I noted that the heat load was considerably higher and I had to cool much more aggressively than for a waterblock without a TEC (no surprise). I also tested with a separate 12 volt power supply - a practice I highly recommend when using a TEC in any system.

I also want to point out that the load you can place on any TEC is limited. If you exceed its capacity, you will find that the TEC will run wild; that is, temps will continue to increase in a destructive heat loop. A safe, conservative operating guidline is to not exceed 50% of the TEC's rating - in this case, 40 watts. In this instance, once I went over 50 watts, I started to see temps continue to creep up. At 70 watts, it started to get scary.

THE TEST

The Swiftech MCW50-T was tested using the CPU Die Simulator and Waterblock Test Rig. In this case, I will not report a C/W as it has no meaning when using a TEC, but show the delta T I achieved at 40 and 50 watts.

WATERBLOCK TEST RESULTS

Test Conditions: Inlet Water Temp: 28.5 C; Mounting force: 15 pounds.

Waterblock @ watts, gpm
Die Temp Diff to Water
Swiftech MCW50-T @ 40 watts, 1 gpm
-8.1 C
Swiftech MCW50-T @ 50 watts, 1 gpm
1.6 C
Swiftech MCW50-T @ 50 watts, 1.5 gpm
-2.8 C

It was interesting to see the increase in performance by increasing waterflow through the waterblock; high efficiencies are key to extracting maximum performance.

CONCLUSIONS

Swiftech's MCW50-T Waterblock is a fine choice for very aggressive GPU cooling, but I caution that TECs are tricky animals to tame. As GPU waterblocks are usually fitted in series with a CPU waterblock, in this case low flow resistance is VERY important; maximum performance may require a separate cooling loop.

Thanks again to Swiftech for sending this our way.

Email Joe