I'm anxious to see how high your GPU clock (and how low your temp) can go with that 80W pelt. I'm in the design stage of an air cooled (gasp!) pelt setup for my eVGA FX5900. I should be able to do an 80W pelt with my standard 350 watt PSU as it has a 16A 12v rail. I have only the FX5900 (max 5-6 amp draw at the clock speeds I'm hoping for), 1 40GB single platter Maxtor (max 1.3 amp), and a 40x IDE-CDROM. All the fans in the system draw less than 2A total. Motherboards typically don't use much +12v at all, except for serial ports, which I'm not using. Under my present load, the 12v rail hovers around 11.8v. An 80w pelt from DangerDen draws 8A max, and I would assume that's at it's max input of 16v. It may turn out that this PSU doesn't like running close to full load on its 12v rail, but I'll find out soon. I'll be running an AC muffin fan to cool the TEC heatsink, so no 12V draw there. I just don't know if an 80 will be enough, considering the FX59xx series cards suck more total power than any GPU before or since:
http://www.xbitlabs.com/articles/video/display/ati-vs-nv-power_9.html
(these figures are for the entire vid card, not just the GPU. I'd assume the GPU is about 85-90% of the total).
On another note, to help keep your case from "cramping" your style
, something you might want to look at is this chassis:
http://www.chenbro.com.tw/usa/product/product_preview.php?pid=74
Follow the chenbro link and take a look at the internal layout.
I bought one of these back in 2000 for $300 (dealer cost) with a 300W Sparkle PSU, and it's served me very well as a server case. In fact, it's serving the images in this post, as it's my web server, among other duties. They've always been somewhat of a rare product, as there is but a single U.S. distributor that handles them. They are typically purchased direct from this distributor (although I can't recall the name) by niche whitebox integrators that do custom server work. I designed and built custom servers while working at one such company at the time, however most of my projects were 48 VDC NEBS compliant telco rack servers and SCSI RAID chasses. I bought the two-tone blue model before it was discontinued. I just google'd for it, and there are some U.S. e-tailers that carry it for around $200 without a PSU. Only one of them showed a pic of it, so make sure the manufacturer part number is listed, and correct, which is SR10303, typically followed by a two letter color code, should you decide to order one. They apparently only come in black or gray now.
I forsee in the not to distant future I'll probably go to some form of water or phase change system, and at this point I will use this case for my workstation. The reasons are obvious, from a functionality standpoint, if not necessarily an aesthetic one. I like the aesthetics, but that's just my taste. Anything that looks like a powerful server, to me, is cool. You can put neon fans in and get a pretty cool effect thru the front of the mobo half of the chassis, although you'll have to do without a transparent side panel, unless you chose to mod one in. I can put everthing I need inside this case, and have maximum airflow and velocity due to the lack of turbulence (like that found in most cramped mini tower cases). The right side of the chassis is one large open cavity. 8 drives, a large radiator, a pump, and 2 power supplies fit in this space easily (and anything else your imagination can come up with). This chassis is simply cavernous.
One other huge advantage of this chassis is that you can eliminate the need for multiple fans to cool the radiator. The rear panel of the drive bay half of the chassis has an integral mount for a 120mm fan. Why would this be an advantage, since most 120mm fans top out at ~110 cf/m max? Because there is enough area on the back panel that the fan hole can be enlarged to fit a 160mm wide industrial AC fan. One advantage of going this single fan route is static pressure. The reason you didn't seem to "feel" any air coming out of your radiator with 3 120mm fans mounted to the front is the drop in static pressure caused by the restrictive nature of the fins, and the low static pressure of the DC PanaFlows you're using. Check out these two AC fans from Sanyo Denki and Comair Rotron:
http://sanyodb.colle.co.jp/product_db_e/coolingfan/acfan/ac_fan_detail.php?master_id=440
http://www.comairrotron.com/ac_family.asp?FamilyID=39
They each flow 250-300 cf/m, but more importantly, have MUCH higher static pressure than all the various fans that are peddled to PC modders/enthusiasts; we're talking 3-6 times the pressure. Pressure is required to overcome resistance. Heatsinks and radiators present a huge resistive load to fans. Some good info in this thread regarding airflow resistance and fan pressures:
http://www.ocforums.com/showthread.php?t=327785
There is a direct relationship between flow, pressure, and power consumption. If you don't have the resources to find out the pressure rating of a fan you already own, look at the amperage draw rating on the tag on the back of the motor. Multiply that by the voltage to find the wattage consumption. Few 12VDC "PC type" fans use more than 10 watts, thus indicating they produce low pressure, and fairly low flow. Look at product line cards, pick two fans the same size (diameter x depth). The higher consumption model with have higher RPM, pressure, and flow, but the pressure increase will be greater than that of the RPM or flow, typically 3 times the pressure, but less than double the RPM and flow. Take a quick peek at this
http://www.eminebea.com/content/pdf/TMP00123/M/2410ml.pdf to see the relationships between power, RPM, pressure, and flow.
The two ~160mm AC fans listed above consume ~40W. So, yet another advantage of going with a single high pressure AC fan, obviously, is that all the power required for it to do its job comes from the wall, not your power supply. With relays and/or toggle switches, a 110VAC input plug cannibalized from an old junk AT PSU, some 16 AWG wire and some soldering, this is a no-brainer to do. Pushing the air "harder" requires more work. Very simplified, watts = work.
The noise level of each of these two AC fans at full boogie is ~60 dB. This chassis is most likely not going to be sitting on your desk, but on the floor, so 60 dB is going to get muffled quite a bit due to this alone, not to mention the fact that this case is fairly sonically "dead", as it's very solid and well built, with zero panel resonance. I'd guess that your HDs would be hiding most of this fan's noise.
I wish that the "neweggs" and/or the more specialized modder e-tailers of the world carried these products that are much better suited to the overclocker/modder scene that the low end "junk" that's available now. This is one reason I won't use colored fans, as they are all lowball, low quality junk. You'd have to pay me to take a Sunon, and then I'd toss it in the can and keep the cash. At least there is some PanaFlow, Delta, and Pabst stuff available thru the main outlets, but these are only the low end of each company's respective product line. The other "Top shelf" manufacturers are Sanyo Denki, NMB, Nidec, Comair Rotron, JMC (did I miss any?). These are the fans, impellers, and blowers you'll find in Sun's 72 CPU servers, SGI and Cray supercomputers, HPs 64 CPU servers, specialized ruggedized military electronics...you get the idea. My apologies for my zealotry here, but I really like top quality fans, and have a fair bit of experience with them and fairly high end electronics air cooling designs. Maybe now you have an idea why I'm going the air cooled route for my first TEC.
Join me on the Dark Side. Best place to pick up these "industrial strength" fans is at "electronics exchange" type places. There are at least two in St. Louis that I frequent, and I'm sure there are in most major cities. These places have alll kinds of great stuff. For instance, today I found 5 110 VAC-12 VDC power supplies rated from 3 to 20A, that had been cannibalized from some kind of industrial device--top price? $25 And this stuff is built to be abused. Compare that to the hundreds for new "specialty" supplies from various "overclocker" e-tailers
You could mount one of either of the two 160mm AC fans I recommended (or an equivalent) to the back panel/wall of the Chenbro, obviously after cutting a larger fan hole and drilling new mounting holes. I recommend a square cut, not circular, as it's easier to do and comes out looking better. If required due to the dimensions of the radiator, cut some kind of sheetmetal "frame" to mount the radiator in the chassis, eliminating air leakage around the frame of the radiator. Attach some kind of foam or weather stripping around the edges for a tight seal. Mount the radiator just far enough away from the back panel (and thus the fan, which alone is about 2 inches deep) to allow room for your PSU(s). Seal up any small holes/gaps in the back panel (there are a few by deisgn). Additionally you may have to stuff the "cable pass through" holes at the top and bottom (power, SCSI/SATA/etc from mobo side to disk bay side) as well. The goal here is making sure all airflow through this half the chassis comes in the front door and passes through the radiator, with no air bypassing it. Don't be tempted to relocate the PSUs and mount the radiator directly in front of the fan "mouth". I say this assuming the radiator is at least a few inches taller than the fan diameter. We need to make sure the fan is pulling air somewhat evenly through all the fins, top to bottom, so the radiator can't be in extreme proximity to the fan. If it is, the top half of the radiator will get great airflow, but the bottom half will suffer.
HOWEVER, if you choose to go with 2 of these AC monsters (probably not needed--strictly for "wow" factor and bragging rights?) instead of one, *DO* mount the radiator directly to the interior side of the back panel. Then mount the two fans on the exterior side. If you do this, you'll obviously need to completely cut all the panel out except the half inch or so on each side required for the screw/bolt holes. You'll also have to relocate your PSUs to the middle.
In case you're wondering, static pressure is a measured differential between two environments, so there is no functional difference if the fan is in front or back of the radiator. With heatsinks it can make a difference, but there are all kinds of dynamics that come into play because of angular momentum, surface deflections, swirling, etc. These things don't come into play with a radiator--in the front and out the back--no angles or deflections to deal with.
If you choose to accept this misson, you will find yourself pleased and amazed at the fantastic design and endless possibilities of this chassis. Upon completion of this project, you will power up the system. You will then proceed to open the front door and stick your hand in front of the radiator. You will be amazed at the amount of suction you feel, assuring you that you're getting maximum capacity from that radiator.
Good god this post got long....manic evening I guess.