Archive for February 27, 2012

DT 5 NOZ CPU Block Preview

Posted: February 27, 2012 in Blocks
Tags: , , ,

Welcome to my preview of the all new DT 5NOZ CPU waterblock.  He asked that I keep the internal details private at the moment, but I will attest to the internal design being something very different.  The name of the block itself says quite a bit, but I’ll have to leave you with your imagination on what’s going on inside.

For now I’ll just share with you some photos, DT’s product announcement, and my pressure drop test as a teaser.

I would like to give special thanks to Erik from http://www.dtwaterblocks.com/  for the preview sample.

PHOTOS

DETROIT AC’s PRODUCT ANNOUNCEMENT

Xtremesystems’s announcement

DT 5Noz waterblock

Hello, let me present my new waterblock. I call it the 5Noz, as there is an unique nozzle inside the block with 5 openings.Martin was good enough to take these excellent pictures for me, thank you Martin!

I machine this block myself on my CNC.

I use copolymer acetal for the top and the spider. With each, I flycut the top with a slow speed and fast feed which gives it checkering like a gun grip. Because it’s a circular cutter, the pattern changes over the top and spider, very coarse near the center and fine near the edges. G 1/4 ports, of course, spaced very widely apart, so I don’t think there will be any fitting issues. There is a spotface at each fitting for proper O-ring sealing (can’t seal on the textured surface). Internally, the top has a couple of horseshoe shaped features that poke-yoke the nozzle so you can only install it in the correct orientation. The O-ring is kept round, which means super short cycle time for me to machine, but it’s also not fiddly to deal with, easy to reinstall if you need to.

The top and spider are also available in white acetal, I have parts finished but no good pictures yet.

The spider is removable and is available in Intel pattern, AMD pattern shortly, and this could be replaced for future socket designs. The spider has a pocket that the springs go down into. This hides the springs a little bit, but also the thickness of the material underneath the springs will allow a metal spider to be used instead of acetal.

The base is 110 alloy copper, 44 microchannels are machined in two steps 0.020 inches and then 0.012 inches. The base contact surface is machined flat and smooth, and is not polished. M4 female threads are in the base, which means that during disassembly for cleaning you only deal with metal threads, so less likely to strip something. The M4 stainless steel cap screws use the full head, so the hex wrench is a nice big 3mm size, less likely to be rounded out.

The nozzle is machined from 360 alloy brass and allows a unique flow path through this block. The nozzle is one of the best looking parts of this block, I wish I could share it, but it’s good and it’ll likely be…adopted…:rolleyes: The nozzle allows a flow path that contacts more fin leading edges, parallel flow paths, and provides a lower pressure drop. The base is bowed by the nozzle. One side of the nozzle is flat, and the base contact face of the nozzle is machined into a convex surface. Close tolerances must be held on the base, top and nozzle to get the bow to the 0.005 inches it’s designed for. To do this, I have to match the parts as they are built, as there are a few thou’ that parts when combined can drift. This means that it will be possible to custom assemble to any amount of bow. I think 0.005 bow is about right, but but I also think bigger IHS processors might do better with less bow, and smaller might do better with more :shrug: In any case, any amount of bow within reason can be specified.

The backplate is machined from 6061-T6 aluminum, and has machined chamfers to accept the heads of flathead M4 screws. This keeps the backplate and screws from slipping around when installing it to the mobo. There is a chamfer for LGA775, one for 115x, and one for 1366/2011. LGA2011 hardware is included, all hardware is stainless steel except for the springs which are zinc coated steel.

I’ve tested the 5Noz and the Koolance CPU-370 in my test loop.

D5 Pump with DT top, setting 5, 12V
Single 120mm radiator
Single Yate Loon D12SM-12 fan, 12V
Q6600 die simulator (copper block machined to Q6600 die sizes, Q6600 IHS soldered to these die)
200W applied to cartridge heaters in the die simulator

5Noz => 31.9C temperature delta between die simulator thermocouple and air entering the radiator
CPU-370 => 33.4C delta

5Noz => 2.55 GPM
CPU-370 => 1.27 GPM

Samples are with a few excellent, non-biased reviewers. Hopefully you can get some independent reviews of this block shortly.

Price will be $87.95 USD.

PRESSURE DROP TESTING

While my CPU thermal test rig is currently apart from my recent LAN event rebuilding, I do have the ability to test pressure drop very quickly.  So…I did and…
We have a new KING of low restriction in my 2011-2012 round of CPU blocks tested!
This block is a good 50%+ less restrictive than my previous next best grouping utilizing microchannel designs….wow!  It is the least restrictive microchannel block I’ve ever tested and nearly as low in restriction as large open pin designs of several years ago.That 5NOZzle is quite impressive in reducing restriction.
I should also comment on how excellent the machining quality on this block is.  Unlike some of the mass produced blocks which have minor irregularities, this one is excellence in precision and machining quality is as good as it gets.  All the parts come together with high precision and a lot of thought has gone into the design.  It is one of those blocks that comes apart and goes together as if you had a swiss watch in your hands, just very well machined with extremely tight tolerances and quality.While I don’t have thermal results yet, the restriction is exceptionally low, the machining quality is exceptionally high, and the design is something very different..:)
Cheers!
Martin
Thermal Preview 3-11-12
While I don’t have anything to compare yet, I did finally get my new 3930K processor up and running:
I replaced the standard M4 nuts with DD brass M4 nuts and 35lb springs which I plan to use for all of my 3930K block testing for mounting pressure consistency.
I figured as a placeholder, I’d give you a quick screenshot of my very first 3930K mount:
Still in the process of getting all my software installed, but I’m thinking I’ll hold back a bit on the the overclock at 4.5Ghz for some extra stability.  I’m pretty sure I could easily see 4.8+ out of this processor, but my old 2600K was pushing a bit hard and testing is hard enough without having to deal with occasional stability issue.  This overclock is extremely stable and the way I like it..:)  This is also pushing considerably more heat over the 2600K as can be seen in the water/air delta numbers hitting nearly 6C despite the strong full speed d12SM12 fans on a quad radiator.  I’m not quite sure what the actual Watt load is, but it’s very very high.  According to Tom’sHardware here, at 4.5Ghz it should be pushing roughly 175W. I wonder if that’s even enough myself considering system power consumption goes from 360W at stock settings loaded clear up to 516W (+156W).  Bottom line the 3930K is throwing out tons of heat, so this is a pretty good cooling test regardless of the final overclock used.
This is the first block I’m testing with the new processor, so I don’t have anything to compare with just yet.
Despite the very large LGA2011 6 cores and extra heat load, 4.5Ghz is still running fairly cool and a bit cooler than my 2600K at 4.8.
This is a new core distribution chart showing the deviation between core temperatures.  The first two mounts were done with the block lettering vertical to the left, where the last three mounts were done horizontal in line with the graphics card and motherboard lettering.  While the differences are within standard deviation error, my best mount #4 was with the block oriented horizontally.  My hottest #5 core and differences between cores was also better using the horizontal orientation.
I really like this block, it flows like crazy and looks good in my new 3930K based system.  I kind of like the white block with black spider option myself.  I did utilize my own mounting hardware for testing consistency and to make mounting a bit easier.  My only reservation would be the shipped standard M4 nuts that came in my package and stiff springs, but I understand the latest version is coming with large black thumb-nuts to improve the hardware package from what I recieved.  I measured the springs compression force and came up with roughly 20lbs per spring fully compressed which is a bit more stiff than I prefer using myself.  I used 8lb springs and brass thumb-nuts from the M6 which I plan to use for all my 3930K testing since it allows fully compressing the spring and feedback when complete so I can get a good measured installation force.  I also did not use the supplied aluminum back-plate since the 2011 platform has one built in but it looks well made.  In general the system is still very much a loose parts type system which takes a fair bit more patience and time to install than some of the more refined mounting systems out there that have reduced the loose parts down to only a few.
More to come as I get a few more 3930K based block tests done.

WHERE TO BUY