New block (Hysterical post)

[Les]

One problem I have with the design is h(eff)

Originally Posted by Incoherent

............The problem is to maintain a high convection coefficient.
This is done in this case, as Jabo suggested, by reducing the height of the block structure clearance. I've done this progressively radiating out from the inlet. At the outer edge of the block the clearance between block and middle cap is 0.5mm.
This maintains the water velocity down the slope of the sphere and thereby a reasonable convection coefficient. The structure I have not optimized at all, the

With no furniture and flow@4LPM this, according to Kryotherm, equates to h(conv)=h(eff)=7547.6 w/m^2*c
Remaining about the same at 12.5mm radius with 0.65mm channel height
Introduction of 20 1.4mm Fins at 8.5mm radius and 1mm channel height gives h(conv)=14262.4 w/m^2*c and h(eff)=28495.7w/m^2*c
Apologise for the messy sums, but I hope it conveys the point that I am not convinced a high h(eff) is maintained.

However did note:-

Originally Posted by Incoherent

............
The structure I have not optimized at all, the object was to test the concept.
.

[pHaestus]

I wanted to interject that I am at this moment booting my test system with everything seemingly working again. Issue was some residual water in the top PCI slot (currently using a S3 Virge card). Dried board out a little more, used a difft PCI slot, and it's running again. Will take me a day or two to run some checks and make sure all is well, but nothing seems to be dead.

[UberBlue]

Can't remember where I found it, or if it's even applicable.

[Cathar]

Originally Posted by Les

One problem I have with the design is h(eff)

With no furniture and flow@4LPM this, according to Kryotherm, equates to h(conv)=h(eff)=7547.6 w/m^2*c
Remaining about the same at 12.5mm radius with 0.65mm channel height
Introduction of 20 1.4mm Fins at 8.5mm radius and 1mm channel height gives h(conv)=14262.4 w/m^2*c and h(eff)=28495.7w/m^2*c
Apologise for the messy sums, but I hope it conveys the point that I am not convinced a high h(eff) is maintained.

However did note:-

It must also be said that as one moves further away from the center that water velocity will drop quite rapidly unless the clearances are progressively altered, such as fitting a parabolic bowl atop the hemisphere.

Alternately, using 1mm cutters, one could form, say, a 12mm diameter pin grid on top and implement 4 or 6 channel that radiate out and descend, giving a constant mini-channel water velocity when the hemisphere is capped. In this instance, would probably want to make the hemisphere around a 15mm radius and carve 4mm deep channels/pins into it. Truly something that would require a full 3D CAD model, and a fair amount of trig. to accomplish.

[Incoherent]

OK, I'm stupid

I have pulled the comparison chart above because I don't think it is correct. I don't actually think the other one is correct either but it can stay because it doesn't make bold drunken claims involving other blocks.
I remembered yesterday that I had changed the thermometer power supply from battery to a mains adapter a while ago, after testing the crater block. And did not check the calibration of the thermometers.
I did a sanity check on my thermometry last night, measuring an old cross drilled block. Basically I am seeing an almost 0.01 C/W improvement for the cross drilled block. There a a few variables for this one, it has no mounting holes so it's position on on the die can move and I do not have extensive data on it, being only one flowrate, also it can read differently with water flow direction due to some enhancements done long ago. But it's enough for me to distrust my data at this point. I will wait until I have a more stable test setup before making anymore rash performance claims.
I am very sorry about this. Not to mention disappointed. It means that all the measured performance numbers I have stated should be offset by a not fully established 0.01CW.
The theory discussion is still relevant however.

[Les]

Commiserations
A testing experience.
Plus some progress on theory and data manipulation.
Have a beer .

[Cathar]

I think most of us have been there at some point or another.

Have learned that whenever I find something that seems exceptional, I then immediately throw an old known block on to retest the calibration. As a habit now I alternate tests between a known reference block and a new block.

[Incoherent]

Originally Posted by Les

Commiserations
A testing experience.
Plus some progress on theory and data manipulation.
Have a beer .

Originally Posted by Cather

I think most of us have been there at some point or another.

Have learned that whenever I find something that seems exceptional, I then immediately throw an old known block on to retest the calibration. As a habit now I alternate tests between a known reference block and a new block.

Well, I've had the day to sulk and now, beer in hand on the strength of what I am sure is good advice, I'm reflecting on the situation.
First step is to get the die sim up and running in a sensible, reliable fashion.
I am ordering some Betatherm miniature thermistors, I have the parts to assemble what will be a fairly high spec data aquisition/control unit to capture temperature, pressure and flowrate and control pumps. I have a Swissflow flowmeter on line already and the 2ohm heater resistor just needs a die of an established footprint instead of the Barton sized one I have been using to fit the Fluxblock:


When this is in place I will move on with the block.
Despite the debacle I find myself having generated, I think the fundamental principle for this block is sound.

Originally Posted by Les

... it conveys the point that I am not convinced a high h(eff) is maintained.

I believe you are probably right Les. However I am figuring that if I can get it up to 30000 the block will perform well. Heres a "to scale" drawing of the block as measured. You can probably refine your calculation from it if you don't feel it a waste of time under the circumstances.

Originally Posted by Cather

It must also be said that as one moves further away from the center that water velocity will drop quite rapidly unless the clearances are progressively altered, such as fitting a parabolic bowl atop the hemisphere.

Definitely Cather. The current design does do this, but a machining balls-up (I make quite a few of these) means it is not quite the way I intended, the clearances are a little bigger, 0.8 instead of 0.4mm at the outer rim.

[mwolfman]

Originally Posted by Incoherent

Oh Shite. I am so very sorry, what's the damage to your kit?
I was pretty sure that cap was sound, damn damn. Another polycarbonate victim?
I feel very bad about this.

I have had 2 polycarbonate tops that cracked after about a month of mounting (before installation in/on the computer).
The GPU-block is on hold while I have modded the CPU-block with a coppertop (heavy!)!!!!

[mwolfman]

Originally Posted by UberBlue

Can't remember where I found it, or if it's even applicable.

What software?
(Im useing cosmos works)

[Incoherent]

Originally Posted by mwolfman

I have had 2 polycarbonate tops that cracked after about a month of mounting (before installation in/on the computer).
The GPU-block is on hold while I have modded the CPU-block with a coppertop (heavy!)!!!!

This one doesn't seem to have cracked, it's just leaking despite the sealing gunk I'm using. For me the only option now is brass.

[SysCrusher]

Originally Posted by Incoherent

This one doesn't seem to have cracked, it's just leaking despite the sealing gunk I'm using. For me the only option now is brass.

Brass is going to add some weight to it so keep that in mind. Another option is delrin.

[killernoodle]

UHMW polyethylene (sp?) is good for waterblocks: it is easy to machine, it is does not absorb water at all, it is scratch resistant, and looks very nice on a block. Plus when you screw barbs in, it stretches slightly and can nearly florm a perfect seal by itself.

[Les]

Originally Posted by Incoherent

......... just needs a die of an established footprint instead of the Barton sized one I have been using to fit the Fluxblock:
......... .

Think I would standardise Die-size with Billa and pHaestus.
Unless, of course, you wish to vary Die-size and investigate these predictions.



Would like to see the continued use of an appropriately sized Flux -block.
Think gives the truest value for TIM resistance

[Incoherent]

Originally Posted by Les

Think I would standardise Die-size with Billa and pHaestus.
Unless, of course, you wish to vary Die-size and investigate these predictions.

Would like to see the continued use of an appropriately sized Flux -block.
Think gives the truest value for TIM resistance

Hmm, you want me to work.

The fluxblock will stay in the picture Les. Unless tests show that it is unviable. At this point my rough measurements show it agreeing with electrical input, but I will test a fluxblock with a third sensor to verify gradient linearity. I believe it is possible to quantify the shadowing effect and according my numbers, it's a small effect. (~0.1 deg C with 2 sensors @70W) I will want to get an exact number for this though.

I am planning to limit myself to two die sizes: 10x10mm and 12x12mm, with appropriate fluxblocks which are 12mm high, sensor spacing 10mm. In addition a die sensor 1mm below the surface, the die itself being 4-5mm deep before spreading to match the heater dimensions. Something like this:



At the moment I am wrapped up in the electronics side of things, and it is going to be a while before I am up and running. The metalwork I am leaving until the Data Aquisition side of things is done so I might change dimensions depending on input.

[mwolfman]

Originally Posted by SysCrusher

Brass is going to add some weight to it so keep that in mind. Another option is delrin.

and delrin cant crack?

Originally Posted by killernoodle

UHMW polyethylene (sp?) is good for waterblocks: it is easy to machine, it is does not absorb water at all, it is scratch resistant, and looks very nice on a block. Plus when you screw barbs in, it stretches slightly and can nearly florm a perfect seal by itself.

the same here.

There are an other problem when milling plastic, they flex = the size is not the same when U have fixed it in the mill.

[HammerSandwich]

Originally Posted by Les

Unless, of course, you wish to vary Die-size and investigate these predictions.

I expected a larger difference in C/W with those die size changes. Isn't heat transfer proportional to delta-T * area?

[Les]

Originally Posted by HammerSandwich

I expected a larger difference in C/W with those die size changes. Isn't heat transfer proportional to delta-T * area?

The Resistance("C/W") will be inversely proportional to the area, for Heat-Flow through a parallel sided plate when the dimensions(area) of heat source and heat sink are the same.
This is not the problem though.
Here the dimensions of the source vary and the sink remains constant.
I use Waterloo to solve the problem.
The Solution("C/W") includes the Convective Resistance and the Conductive Resistance(including TIM)

[bigben2k]

Delrin is a soft plastic, no it won't crack.

Les has the answer.