Water blocks for amd 64

[Joe]

This post was reported to me, but after reading through it, I dont see anything THAT wrong.

Just some dudes looking for answers, and the people who have the knowledge/experience maybe not spelling it out clear enough.

such is life I guess... overall this thread doesnt warrant a "mod report" since its quite civil, and productive.

[Brians256]

I agree, Joe. There have been some posts in this thread which seem negative and without value to this discussion. Not terrifically obnoxious, all things speaking, but not nice. In other words, kind of like real life, and not worth getting all excited and putting on the super-mod hat. People are showing whether they can be adult and polite. Or not.

We all know that there are terrifically smart and knowledgeable people here. What is most important to remember is that lots of knowledge does not mean that some beginner doesn't deserve an explaination geared to his level. If someone does not understand your first explaination but still wants to learn (and is not simply a troll), I suggest trying again (with patience) using different analogies etc. Someone may simply be new, or may not have understood previous explainations. Remember: everyone learns in a different way and one explaination may not work for everyone who is listening.

Now for the obligatory "Adding something worthwhile to this thread" portion of my post:

Does anyone other than BillA understand why a peltier would behave differently than a radiator in terms of the flow-rate vs. CpuTemp graph? I can understand BillA not wanting to leak all his company proprietary info to public domain.

My thought is that an active device is fundamentally different than a passive device. Having the water in the pelt device longer results in more heat extracted per unit time because the pelt can deal with a much higher DeltaT on the hot side. So, as the flow rate is decreased, the hot side deltaT goes way up, causing higher efficiencies in dumping heat. At the same time, the pelt is removing similar amounts of heat (calories) from less water. So, a low-end maxima in the SYSTEM efficiency curve is seen.

The sweet spot would therefore be where the heat moving efficiencies in hot-side deltaT best combine with the efficiencies of the low-side. There should be a sweet spot simply because both functions are monotonically increasing in different directions on the X axis. There would not, of course, be any maxima if both functions were perfect inverses, but there definitely is an observable maxima.

Note that none of this has any mathematical rigor, so feel free to tell me I don't have a clue as long as you attempt to explain yourself. Because of the fact that I do not have good training in this area, I have kept my mouth shut in this thread until now.

Flames away!

[pHaestus]

I find it amusing that Bill gets reported to the moderators for not giving up his company's trade secrets. In another thread he is asked who mfgrs the new pumps so Swiftech can be undercut on price. Laughable.

The important point here is that Bill pointed out that lower flow rates coupled with the pin fin design on the inner chamber are how the swiftech pelt chiller succeeds while other setups have failed. That seems to be useful information. Also consistent with Lytron's choice of pumps in their chillers:

http://www.lytron.com/standard/cs_th...be_perform.htm

This may or may not be relevant though; Lytron uses incredibly restrictive waterblock designs, don't they?

Now it seems clear to me that I should test the Swiftech chiller with a valve in the loop to adjust flow rates way down. The smallest pump (and lowest power) I have handy is a Hydor L20 and so it should still be able to go from 0.3-1 GPM or so with my flowmeter present. The reasons why this works are probably something we should come up with ourselves. Develop a hypothesis that can be tested and I will be glad to do so if it's reasonable. I haven't had time to think too much about this because I am trying to get work stuff done

[BillA]

for those with a sense of humor, catch this thread
http://forums.procooling.com/vbb/sho...&threadid=7992
where you will soon see a demonstration of just what I'm talkin' about
note that there is nothing instructive about watching a fool be a fool, its purely for entertainment

yes pHaestus, I debated even contributing to this thread
and then that di*kheaded pump comment, which was made by wormywood also
makes me question what I'm doing, clearly I am in error
hmmm

[Althornin]

Originally posted by unregistered
vagueness ??
no, I would say you did not think through the scenario I posed
don't get your panties bunched
several of us here have been at this a long time, I simply do not recognize your 'handle' as a long term, ahem, contributor

perhaps Cathar will share his experiences with his late chiller project

Since87, see what happens when you give Les some slack ?
next he'll be pounding on me, lol

Hmm, ok. With some further thought, let me say:

You are simply saying that with the blocks you use, the limiting factor is heat "Extraction" at the chiller, and thus, you slow coolant flow to change this limit point great enough to offset the decreased efficiency of the waterblocks (both chiller blocks and cpu block).

Ok, i understand that.
Now, why must this be the only way? Why wont a high flow, high efficiency system work?

Oh yes: No, i am not a long time contributor, but i am a long time lurker. Just because i only registered a year ago and dont post much doesnt mean i am clueless/dont know history, etc.

I find it amusing that Bill gets reported to the moderators for not giving up his company's trade secrets.

No, i reported this thread because of Les. At the time, it really pissed me off - and in retrospect, i agree with Joe, but i think that rudeness is uncalled for and should not be accepted in polite company - which this should be.

[pHaestus]

I GREATLY prefer to see you dangle carrots a la this thread than not contribute. It's a good way to learn for those of us interested. We can't have it both ways; we don't heavily moderate here so we get some stupid questions and flames to go with reasonably unfettered discussion.

I spent an hour today trying to convince a modeler that an outlying point on a plot of data was just "the way it was in the lab" and didn't justify modifications to the underlying chemistry in said model. I finally took him to the lab and showed him all the steps between "idea" and "data point". Ben will learn eventually by trying and doing (over and over) perhaps? Expensive lessons when dealing with equipment purchases though...

[Since87]

Originally posted by pHaestus
I GREATLY prefer to see you dangle carrots a la this thread than not contribute.

Yes, I strongly agree. Besides, I like having the puzzle in front of me. Even if I don't figure out the answer, I'm likely to learn all kinds of things in the process of trying to figure it out.

I've had to give myself a refresher course in Physics in discussing things with Bill in the past. Doing so lead to me having more clarity wrt some things I deal with at work. With my present line of thinking, I may have to relearn Calculus. Hmm, rethinking whether I like this puzzle...

Les,

LOL, I think you're right.

[gone_fishin]

When a pelt setup is utilized directly on the die, the the heat goes from die-tim-coldplate-tim-pelt-tim-wbbp-water.

In waterchillers the heat is located in the water and goes water-through block interior geometry surface area-tim-pelt-tim-wbbp-water.

In all the former shitty attempts I have seen, heatsinkfans were used in an attempt to remove heat from the hotside pelt.

There is one less tim per the two heat exit paths(each side of the chiller) compared to direct die pelt usage which is good. Water is being asked to give up its heat instead of a solid material of much more highly concentrated heat. Water is better at absorbing heat energy than giving it up.

The fin arrangement is equivalent to expanding the heat exchanger size.
Slowing the flowrate of the heatsource (water) would be equivalent to expanding the heat exchanger size would it not? Also slowing the flowrate concentrates more heat per unit of water.

Just some observations, nothing new just didn't see them spelled out for those scratching their heads.

Bill why did you pull the pics, didn't get to see em, darn.

This has turned into an unexpected highly interesting thread.

[Cathar]

Originally posted by unregistered
perhaps Cathar will share his experiences with his late chiller project

Actually I only just got a hold of the last pieces of equipment that I need to do it, after all this time.

One of these days I'll give it a burl....

It's on my "To Do" list which is presently stretching out of sight.

[Boli]

Don't sleep.

Some night I can't sleep at all... but somehow I find the enargy and end up doing stuff I've being putting off for ages... washing my socks for instance.

~ Boli

[BillA]

Originally posted by gone_fishin
. . . .
Water is better at absorbing heat energy than giving it up.

The fin arrangement is equivalent to expanding the heat exchanger size.
Slowing the flowrate of the heatsource (water) would be equivalent to expanding the heat exchanger size would it not? Also slowing the flowrate concentrates more heat per unit of water.
. . . .

Bill why did you pull the pics, didn't get to see em, darn.
. . . .

"Water is better at absorbing heat energy than giving it up."
what is the basis for this statement ?
while I am sympathetic of this view, I know of NO reason why it is true - and plenty why it is not

yes, the 3 observations are valid

my old site is down, ???
I'll repost from another

[Zhentar]

"better at absorbing heat than giving it up" is a common misconception held for copper too. Its just the specific heat of the object; a higher specific heat means it can absorb more energy with the same temp increase. This leads to the absorbing heat part; it can absorb a lot of energy without a great increase in temps. This also however leads to the giving it up part- now that it has heated up, it holds a lot of energy so cooling it down takes a while to cool down.

[Brians256]

Dangit! I was hoping someone would tell me why my wacky theory was wrong.

[gone_fishin]

Originally posted by unregistered
"Water is better at absorbing heat energy than giving it up."
what is the basis for this statement ?
while I am sympathetic of this view, I know of NO reason why it is true - and plenty why it is not

yes, the 3 observations are valid

my old site is down, ???
I'll repost from another

Water has a high specific heat index. This means that water can absorb a lot of heat before it begins to get hot. Inversely does this mean that a lot of energy needs to be removed from the water before it begins to cool off?
Can't give a difinitive reason for my thinking this way and I wish I had a more clear understanding of it.

[Since87]

Originally posted by gone_fishin
Water has a high specific heat index. This means that water can absorb a lot of heat before it begins to get hot. Inversely does this mean that a lot of energy needs to be removed from the water before it begins to cool off?

Assuming we're ignoring quantum mechanics...

There is no sudden transition going on where water absorbs a certain amount of heat and then steps in temperature.

Water "begins to get hot" when the slightest amount of heat is added, and "begins to cool off" when the the slightest amount of heat is removed.

Temperature - A measure of the average kinetic energy of the particles in a sample of matter, expressed in terms of units or degrees designated on a standard scale. (From www.dictionary.com)

Simple conservation of energy dictates that if adding X amount of heat changes the temperature of a quanity of water from A to B, then removing the same amount of heat will change the temperature from B to A.

[BillA]

amen bro

image on pg 1 back up

[Since87]

Originally posted by Brians256
Dangit! I was hoping someone would tell me why my wacky theory was wrong.

I'm not really following your theory, but I'm wondering if the behavior of TEC's isn't playing a role as well.

As a further contribution to weird science... My current thinking involves:

1. the operating characteristic of TEC's (dT vs Q)
2. the behavior of crossflow heat exchangers
3. slowing the flowrate is somewhat analogous to stretching the length of the heat exchanger in particular (as compared to increasing the area of the heat exchanger.)

My thoughts are too nebulous right now for me to think this train (wreck?) of thought will amount to anything though.

[Zhentar]

Originally posted by Brians256
Dangit! I was hoping someone would tell me why my wacky theory was wrong.

Me too. This physics stuff is tricky and I want to learn exactly why this works how it does.

[Les]

All
I apologise for my non-constructive comments.

Bill,
I guess you are defining "key to chiller performance"as the Temp difference between inlet and outlet?

Dunno but have done some rough sums:
2x172w Peltier at 19v
2x wbs C/W~ 0.05 (40x40mm Heat source)
1 Cold chamber wb C/W~ 0.025 (Two(2) 40x40mm Heat sources)
2 x Radiators each with C/W~ 0.06
Insulated to same standard as Simulator Heat Die C/W~ 15

Kryotherm suggests that the 2 Peltiers can maintain a 25c differential between Hot Water and Cold Object whilst extracting ~ 140Watt .
Extraction of 140Watt will cool water ~10c at a flow rate of 0.2lpm.
This will dump ~500Watt in the Hot Water (140Watt + Peltier va (358Watt) + Insulation leaks(2Watt)) giving a Hot water Temp ~15c above Ambient.
So I guess the above Chiller would be capable of cooling ambient water to ~10c below ambient at a Flow Rate of 0.2LPM.

[gone_fishin]

I see what you mean about instantaneous response to temp changes.

Now I see them pics, thanks. A little different than what I thought from the discussion. Here's some paint slop to show you what I thought it would look like.



If you got two loops goin, why not three?


Edit, just noticed that tiny spout in the center, at first glance I thought it was for inserting a temp probe, but is that for a central water chamber?

[BillA]

Les
there is absolutely no reason that I am aware of that necessitates an apology from you

no, "key to chiller performance" is the attainable die temp at load
to me the actual cold side coolant temp is not the measure of performance

here is some grist for the mill:
2 250mm² TECs @ 226W (presumably Q max ?, V max and I max not known)
- @12.0VDC drawing 16.7A each
hot side coolant held @ 35.0°C with a flow rate of 340lph
cold loop using 25% antifreeze, no insulation on lines or wb - and LOTS of condensation
heat load applied through 100mm² heat die and a MCW5000 rev2 wb

some results:
Watts . coolant . die
39.4 ..... -1.9 ..... 8.1
49.8 ..... -1.0 ... 10.9
59.6 ...... 0.4 .... 14.1
69.9 ...... 1.8 .... 17.6
79.6 ...... 3.3 .... 21.2

steady state assumed, but may not be absolutely correct

can Kryrotherm back calculate the cold loop flow rate ?

g_f
that is what it looks like if the outside items are wbs
yes, the 1/4" barb is one side of the 'chillin' chamber'

[BrianW]

bill: 340lpm or lph?

Thanks,

Brian

[BillA]

340lph
thanks Brian

[Les]

Originally posted by unregistered
Les
can Kryrotherm back calculate the cold loop flow rate ?

No, and unfortunately it does not include the 220w Peltier.
Ta for the numbers,am playing but my brain only works slowly.
More data would be welcome.

[Brians256]

My theory is based upon steady state situation, where the goal is to move as much heat energy as possible out of the water. The goal is not to get the lowest temperature water.

So, I assume that there is a there is a point of flow rate where higher flows or lower flows through the pelt block move less heat out of the water. I assume that this measurement is made with a controlled inlet temperature (i.e. higher flow tests weren't made with a larger pump that would dump heat into the system and vice versa) and all other system variables mostly controlled. Knowing BillA, that is what I would assume, since he has a good track record of trying to minimize independent variables.

1. As flow rate drops, T(water) decreases. This seems intuitively obvious.

2. Does the temperature on the hot side of the TEC go up, down or stay the same? Assuming that heat movement is INCREASING as flow rate decreases, the TEC hot side must increase. If more heat is being moved, the only place to dump it is the hot side of the TEC, and the only way to dump more heat is to move it across a steeper gradient. A higher temp on the hot side of the TEC allows more heat to be dumped (to air, water, etc...) at higher efficiences FOR THAT JUNCTION. A higher hot-side temp of the TEC allows more heat to be dumped from the TEC (temporarily ignoring where the heat comes from: CPU, TEC innefficiencies, etc...). To me, this is where the negative term on system efficiency arises. As flow rate increases, hot side temps drop and the TEC is working less to move less heat.

3. TEC efficiency decreases as dT increases. So, as flow rate decreases, the efficiency at which electricity is used to move BTUs/Calories decreases. Does this reach a limit of zero efficiency where the maximum dT for the particular device is reached when trying to cool a perfectly insulated cold side? Probably not relevant for our discussion, but I'm still curious. In any case, this trend introduces a positive coefficient in favor or higher flow rates.

It still seems wierd to me that the decreased flow would actually help, since it would *seem* that the decreasing efficiency of the WattsMoved/WattsElectricityConsumed term would dominate. But experimental evidence trumps all theory.

Is any of this making sense? If I'm not adding to the discussion, let me know and I'll shut up.

Just trying to learn.