Lookie what I just got

[Can O' Beans]

Nice

Now to polish it up?

[Volenti]

Originally posted by Can O' Beans
Nice

Now to polish it up?

bah! testing first, polishing second

now this is really bizzare, I don't get it at all.

what I was using previously was my 1/2'' fitting "carved out" block (which is my best performing block to date) with a small 700L/H pump (around 500L/H system flow) this is cooling a P4 1.8a@2.64, 1.8v, 11 degrees above water temp at idle and 14 degrees above under load.

now this new block, with my big 2800L/H pump up it's butt, pushing 3 times the flowrate of the previous setup, is getting exactly the same temps! (+- a degree or so)

I still have to muck around with the exact mounting of the block, but the P4's a far more forgiving in this area than the athlons...

Oh I'm using a pretty large volume of water in the system, so the larger pump putting more heat into the water isn't an issue at this stage.

Mabie the P4 simply isn't generating enough heat to show a difference? :shrug:

[Les]

Interesting .
Maybe the flow is borderline laminar/turbulent.
For example:
50mm wide plate with 58fins and flow rate of 1600l/h:
1) 0.1mm thick x 8mm high :-- :Individual Channel size 0.78x8mm Flow Xsectional Area 354 sq.mm ,Velocity 1.26m/s,Characteristic dimension 1.5mm, Reynolds No 1841.
2) 0.2mm thick x 8mm high:-- Individual Channel size 0.67x8mm,Flow Xsectional Area 307 sq.mm ,Velocity 1.45m/s,Characteristic dimension 1.3mm, Reynolds No 1852

Reducing the fin height to 5mm for the 0.2mm case would increase the Reynolds to 2893 at a flow of 1600l/h but unfortunately with an increase in 3-4fold increase in "pump delivery head" requirements,

[BillA]

yup, certain paths only go so far
consider the thermal gradients involved
shorten the upstream length a bunch, downstream 1/2 as much
then boost the pressure/velocity

even so your gains will be quite slight

with a P4 you can also thin the bp, that will help more

BTW, you have a 'low flow' design; you need pressure, not volume

[Volenti]

Originally posted by unregistered
yup, certain paths only go so far
consider the thermal gradients involved
shorten the upstream length a bunch, downstream 1/2 as much
then boost the pressure/velocity

even so your gains will be quite slight

with a P4 you can also thin the bp, that will help more

BTW, you have a 'low flow' design; you need pressure, not volume

so you mean I should trim the fins more so they are mainly just over the core, say the size of the intel heatspreader, and force the water through that?

the bp is 4mm thick, little bit thinner now after I lapped it (had to lap quite a bit)

I can easily change the design, just add heat and it'll all fall apart...

and do you mean that I have a low flow volecity design?

[BillA]

how far 'out' do you think the heat is propagating ?
and my asymmetrical suggestion is a bit extreme as the temp rise of the coolant is not significant
- but surely it cannot go as far parallel to the flow as transversely

no, you always want high velocity
your design is predicated on high surface area,
hence you will have lower volume (for a given pump head)

[Volenti]

Originally posted by unregistered
how far 'out' do you think the heat is propagating ?

probably not very far at all, here's something else to chew on;

I have a collection of pumps here, the biggest rated at 3800L/H, 4m head, 150w, the smallest rated at 700L/H, .5m head, 12w.

I've tried both with this block, and there's one degree difference between them (+ - .5 degree)

If I try that with my 1/2'' carved out block the difference is a little larger, approching 2 degrees.

Am I simply approaching the limits of the TIM? (using AS3 , applied as per instructions ect)

I'll have to knock up a "micro block" with a 1.5mm thick bp or something similar, though I think I'm just smashing headlong into the brick wall of diminishing returns...

[BillA]

as you said, "diminishing returns..."

but do note that this wb is better than your carved- out one as it can cool similarly on less flow
(if I understand you correctly)

the TIM joint is not a limit, simply a series (thermal) impedance (there is a temp offset across it)

to reduce the thermal gradient across the wb bp you will need to do different things

ahhh . . . .
(its all been posted, dig a bit)

[Can O' Beans]

Well, for initial design, it matched your "best" block already, so it's off to a good start.

Now to just find out how to tweak more performance out of it.

Ideas:
The "channels" between the fins are straight & long... I was thinking about taking a small drill bit and running it down vertically, in between the fins, and continue on down the length of the fin:

<=============> fin with sharp edge
OOOOOOOOOOOO drill bit "holes"
<=============> fin with sharp edge

This would add some turbulence reducing the laminar flow (if any).
I would also try to knife-edge, or at least chamfer, the ends of the fins to help water flow into the channels better.

I'm just brainstorming here at work, so any positive/negative comments are welcome (as I'm learning)

[Volenti]

Originally posted by Can O' Beans
Well, for initial design, it matched your "best" block already, so it's off to a good start.

Now to just find out how to tweak more performance out of it.

Ideas:
The "channels" between the fins are straight & long... I was thinking about taking a small drill bit and running it down vertically, in between the fins, and continue on down the length of the fin:

<=============> fin with sharp edge
OOOOOOOOOOOO drill bit "holes"
<=============> fin with sharp edge

This would add some turbulence reducing the laminar flow (if any).
I would also try to knife-edge, or at least chamfer, the ends of the fins to help water flow into the channels better.

I'm just brainstorming here at work, so any positive/negative comments are welcome (as I'm learning)

oh I'm always open to suggestions, I don't feel that I'm suffering from laminar flow too much, after all the fins are quite close together, and the water has no choice but to flow between them.

I can't imagine the boundary layer being very thick, there just isn't room for a thick boundary layer as well as the water flow, something has to give. (though I have no idea what constitutes a "thick" boundary layer, thermally)

Had I had a problem with laminar flow it should have shown up in my testing with the small v large pump, but it didn't.

The fins are already very thin, they only take up 25% of the volume of the flow channel, flow rate (volume) isn't a problem, but as billa said, I can always do with more flow volecity.

[Les]

Originally posted by Volenti
o I don't feel that I'm suffering from laminar flow too much, after all the fins are quite close together, and the water has no choice but to flow between them.

According to my calculations(in previous post), using Kryotherm ,you are likely to have some laminar flow at a flow rate of 1600l/h..
Remember there are 58 parallel channels( ~0.8x8mm) giving an average flow of only ~28l/h I suggest any Reynolds Number caculations will give the same answer - that the flow is laminar.

[Volenti]

Originally posted by Les
According to my calculations(in previous post), using Kryotherm ,you are likely to have some laminar flow at a flow rate of 1600l/h..
Remember there are 58 parallel channels( ~0.8x8mm) giving an average flow of only ~28l/h I suggest any Reynolds Number caculations will give the same answer - that the flow is laminar.

ok ,according to theory the flow's laminar, I have no problem seeing laminar flow being bad in big round pipe suitation. The block however has 58 very thin rectangular channels, with less than a milimeter seperating the 2 walls of the channel, forget about the maths for a second and visualize the water flowing through the channels, and how thick the boundary layer can be.