Mounting Pressure vs performance

[freeloadingbum]

I was rereading some of pH's block test reviews and was wondering if I could get some clarification on the actual mounting pressure used. The article states to use amd's recommended pressure but no number is mentioned.

Amd states a recommended load of 16lbs with a minimum of 12lbs and a max of 30 lbs using all 6 prongs on the socket. I couldn't find any load specs for using mounting holes but I'm sure using the socket specs would be fine.

The mcw5000a is stated as using a 30lb mounting load but the mcw6000 mounting load isn't mentioned. Is it using a 30lb load as well? Also, what was the load of the mcw5000a's original springs?

The Aquajoe review has a chart showing performance vs spring compression. Looking at the chart, the performance at 7 turns vs 13 turns shows about a 1.3C difference (as best as I could make out). Now if 7 turns equals 16lbs load then 13 turns should equal about 30lbs. Maybe joemac could clarify the load since this test was done using his springs.

[pHaestus]

The best answer I can give you is that all the 4 hole wbs are tested with the same springs at the same compression. The hardware used was from the Little River Cascade wb. For blocks that use different mounting methods I'm at the mercy of the mfgrs more or less. You can see that changing the springs on the MCW5000-A made a pretty substantial difference:



For the aquajoe wb full compression was recommended for Intel P4s, so presumably that's closer to 100lb force?

[freeloadingbum]

I have a mcw5000a and I seen about a 2c improvement switching to the stiffer springs as well but this was also only a single mount comparison (too lazy to do more).

Hopefully cathar can clarify the spring load at 5 turns

I don't think the aquajoe springs are applying 100 lbs at 14 turns as that would mean that at 7 turns the spring load would be near 50lbs. Hopefully joemac will also be able to clarify the spring load.

Personally, I think it would be nice if all the block manufactures would specify the spring load instead of forcing consumers to play a guessing game as to how much pressure they are actually applying.

I recently decided to measure the spring rate with my maze3 and found that it took 8 lbs to fully compress one spring (14 turns). From what I read, coil spring rates are linear from 20% to 80% compression with the first 20% having lower and the last 20% having higher rates.

[HaloJones]

I've given up using springs. With my SlitEdge, I just tighten the finger nuts down to "finger tight" withut any springs. With my LRWW, I bought some thin tubing to place over the bolts and then tightened the wing nuts to "finger tight" again. I just never feel that the springs with all the weight of the tubing at right angles can really hold the block on firmly enough.

[Hoot]

It's even more of a "dice toss" with an A64 that has the IHS removed. The two screws down the centerline provide good retention in that axis, but suck royally in the other axis with the weight and tubing torque causing it to vary all over the place. It's probably the number one problem trying to get reproduceable results from one mount to the next. When you can reach in and press each of the four quadrants of the block and get four different temps, that's aggrivating, to say the least.

Hoot

[BillA]

Originally Posted by freeloadingbum

. . . . .
Personally, I think it would be nice if all the block manufactures would specify the spring load instead of forcing consumers to play a guessing game as to how much pressure they are actually applying.
. . . . . .

are you guys asleep ?
mounting pressures are socket specific, VERY socket specific
-> as in LIGHTER force on some sockets
NOT wb specific, eh ?

guess as you wish, following the Swiftech instructions will yield the mfgr's recommended clamping force for that specific mounting system

Swiftech does not use a 'universal' spring

[freeloadingbum]

Originally Posted by unregistered

guess as you wish, following the Swiftech instructions will yield the mfgr's recommended clamping force for that specific mounting system

Which is what? Are you saying that the mounting pressure for the mcw6000 is going to vary between 12 and 30 lbs depending on the motherboard

[pHaestus]

I think he means for each platform (Socket A, Socket 478, socket 754, socket 939, etc). The sockets are all the same and the CPU mfgr recs are the same for a particular socket.

Yea Swiftech does this and as a result it's quite a bit more expensive for them to support aging platforms like Socket A. They engineer mounting systems based upon CPU mfgr specs. It sounds like a good thing but it's why we didnt ever see a MCW5000-T for Socket A I would guess?

Most people (myself included) crank down on springs more than is recommended to get that last 0.5C. It isn't a big deal unless the force is substantial enough to seriously warp the mobo.

[BillA]

the Intel and AMD thermal design guides define the limits
we follow those guidelines

so an A64 has one setup, 775 another, etc

this is not difficult,
one wb is used across a number of systems; but each mounting system is specific to that processor/socket
- as I said, if you are using a Swiftech product just follow the directions

if you wish to design your own clamping system, feel free
some just screw down the nuts - can't get any more basic, eh ?

bending the mobo can be viewed differently as well; short term and long term
as a mfgr we can only make recommendations based on long term suitability
so if we do not know, we say no

[pHaestus]

You have to cut out space for the socket a cam box though; that means you have to have some way to estimate the market or else make in small batches and have delays.

When did demand for the -A blocks take a big hit? I am just now moving to A64.

[BillA]

the whole step issue is a pain, which Foxcon added to by making some 775 levers 'too high'
so now the outlet cannot always be positioned on the high side, etc

[freeloadingbum]

I guess I should have made the title more specific like "Mounting pressure vs Performance using socket A". So my question for BillA is what is the mounting load for the mcw6000 on a socket A motherboard (min/max if it varies) ? Also, what was the mounting load of the original springs for the mcw5000a (as tested by pH)?

[BillA]

as I recall,
the AMD MCW5000 had a 24lbf design basis using 2 lugs, and when 3 lugs were used it went to 30lbf
- a 25% increase