|
|
General Liquid/Water Cooling Discussion For discussion about Full Cooling System kits, or general cooling topics. Keep specific cooling items like pumps, radiators, etc... in their specific forums. |
Thread Tools |
10-13-2002, 10:43 PM | #126 |
Cooling Neophyte
Join Date: Mar 2002
Posts: 20
|
Hey Cathar -
The point to the flow rate question was to support a point I had made long ago. That is, you don't need "allot" to work "well". I did similar testing and when posted my thoughts and early results, it of course, was translated into you don't need good flow. Which of course, ppl took and extrapolated to the n'th degree, Bill was one of those ppl. So Bill is right, it is a bit of an axe. Basically flow IS important, however you can obtain reasonable and even good results with lower flows as well... The point to the rate question is NOT to rehash what was said earlier. I understand the addition thing, I've accepted that. Since the backpressure is a function of velocity, once the 1st reduction is placed in line the velocity reduces significantly. Now when a 2nd restriction is added, the velocity drops of non-linearly. Hence, how close do we need to be? Is saying the 1st drop is a "reasonable" approximation in some situations? I suspect yes. |
10-13-2002, 11:25 PM | #127 | |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Quote:
The flow rate does not change non-linearly at different restriction points in the system. |
|
10-14-2002, 07:42 AM | #128 |
Cooling Neophyte
Join Date: Mar 2002
Posts: 20
|
Cathar -
This 12hr time change is a killer. Reading my post this morning, I agree it is very unclear, and I see where you have been mislead. The Velocity through both fittings in a closed loop system is the same, ok sorry didn't mean to imply that it would be different. I'm a visual person, so let me try to draw a picture in words (please be patient). 1" hose -> 1/4" fitting. --> some overall flow rate drop (FRD) 1" hose -> 1/4" tube --> the fitting is stretched so now it is more than one spot but rather a 6" long stretch or something. The FRD should be "about" the same. 1" hose -> 1/4" opening -> bulge the middle of the tube -> 1/4" opening. Ok when the last scenario is entered, the FRD over the "system" should still be "about" the same, no? Taking this thinking further, across the 1st reduction, if it was the only in the system you'd have the full FRD. Now the 2nd reduction is added. The flow rate will not change as significantly as when the 1st was added. So, in order for there to be equal drops across each reduction, the drop seen across the 1st reducer must be LESS than it was originally. Hence, I'm asking/stating that there is an apparent reduction in the first fittings resistance, as the second fitting is added, which seems to be logical if flow is not significantly altered. I guess my ultimate question here, is what are the real numbers. I'm not saying the resistances don't add. I'm asking, when they add, is the second restriction significant given the apparent reduction in resistance of the 1st fitting? The question that remains unanswered is, HOW MUCH? Furthermore, if this holds true, was the absolute bashing another forum goer received from some members warranted? I believe in 'partial credit'. You can be 'basically right' in life but you can't be 'basically right' when stating annoying in these forums. I've taken from this that I am pretty thin skinned and take to heart what is said. I think it is personal slander when a respected member lamb bastes anyone for a statement that is 'basically right'. |
10-14-2002, 07:58 AM | #129 | |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Quote:
Velocity does not equal flow rate. Flow rate = volume / time Velocity = distance / time For a set given flow rate, the velocity will be increased as the same volume of water per unit of time is forced through a smaller opening. Thus velocity will only be equal through a system of uniform orifice sizes. In summary: The sum of resistances defines a fixed flow rate when those resistances are applied against a pump's pumping pressure. It is perfectly valid to make comparisons of a system of a certain flow rate regardless of how that particular flow rate was achieved, whether that be through squeezing the hose, using a smaller capacity pump, or using a powerful pump but with lots of restrictions (note - we are ignoring pump heat in this scenario). With respect to your variety of scenarios, each scenario you presented represents a particular flow rate due to the restrictions that those scenarios offered against the pump. The end result would bea particular flow rate unique to each scenario. The water-block's performance is dictated purely by flow rate (given a set coolant temperature and a set heat load) regardless of how that particular flow rate was arrived at. You presented a set of slightly different restriction scenarios, but the exact details of those scenario's is unimportant. The only important detail is the resultant flow rate as a result of those restrictions. I hope that clears it up. |
|
10-14-2002, 08:36 AM | #130 |
Cooling Neophyte
Join Date: Mar 2002
Posts: 20
|
Cathar -
yes yes.. point taken about terms. You are correct, velocity was used incorrectly, mass flow rate is should be substituted for velocity and was indeed what was intended. I know each scenario would result in different MFR, but how much different? Would they be relatively close? Close enough to call them "the same"? As noted over and over again, blocks work better with high flows. How much better is the question. If the reduction in MFR is assumed to be "the same" will the error introduced into the equation be "negligible" with regard to how it impacts the blocks performance? I suspect each block will have a different characteristic based on its design, and for some this would be a true statement. >> end goal for us As a manufacturer, can we make any assertions to guild would be customers given the above. Since, I believe, you may looking at manufacturing your block and offering them to the public, how will you address this question? If someone uses your block along with a chipset or GPU block, how badly will it impact the CPU block performance? Hence, if the CPU block is the "greatest" restriction, will it be the dominating factor and the other blocks be "negligible" wgt MFR impact and ultimately CPU block performance? I've been out of school a while now, so perhaps it's time for me to visit the library and actually check out a book to calculate and answer these questions for myself. Problem is as a manufacture, we're still in the red as far as time invested vs money's made. Truly this exercise would be a 'government time' project. I'll let this topic rest. |
10-14-2002, 08:46 AM | #131 | |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
The only way to address a customer's concerns with flow rate based performance would be to invest in a proper testing heat die setup (like BillA has) and measure the block's performance for a variety of flow rates.
The block's performance (C/W) will not change as the heat load changes (at least not in the range of typical heat loads that CPU's put out being less than 200W) so we can basically ignore that aspect as a variable. ie. Setup a fixed heat die to push out something like 100W (which equates to around a 150-160W "Radiate" heat load) and then plot the block's performance for a set of flow rates (say from 1-12lpm in 1lpm steps). It's then up to the customer to determine what the resultant flow rate would be through their system with their pump with whatever restrictions they add. A manufacturer simply cannot answer that for the customer. All a manufacturer can do is say "Our testing indicates that for satisfactory block performance that the actual flow rate through the waterblock should be kept above x lpm". In fact, you won't even need to invest in a heat die to make that statement. This can be fairly easily derived even with a few system tests with a motherboard and CPU. It will be fairly easy to pick the point where lower flow rates are starting to cause the block to seriously underperform. Quote:
|
|
10-14-2002, 08:52 AM | #132 |
Responsible for 2%
of all the posts here. Join Date: May 2002
Location: Texas, U.S.A.
Posts: 8,302
|
Cathar:
I was reading over the OCAU thread last weekend and noticed one thing: You stated that your nozzle opening was 5 x 15 mm, for a total area of 40mm^2. However, 5 x 15mm is 75 mm^2, but I understand that you have 8 channels that are 5 x 1 mm, which totals 40mm^2. The thing is: you didn't nozzle the flow, you merely tried to distribute it better among all the channels. I was under the impression that you nozzled it, where the opening that you created was smaller that 40mm^2, which would have created an increase in coolant speed. Did you consider reducing the opening to say, 2 x 15mm? |
10-14-2002, 09:08 AM | #133 |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Initially it was 5 x 15.
In subsequent tests I resolved flow rate vs velocity with respect to cutting back the "5", down to 4, then 3, then 2. Basically 3mm x 15mm (really 8 of 1x3mm openings) was the cross-over point with the qualification that as the orifice was narrowed further, the resultant drop in flow rates outweighed the increase in nozzle velocity. Actually one could use 2.5mm, but now I also want to produce a block that isn't going to kill flow rates for other blocks that a user might have in their system. As it stands, then 3x15 (8 of 1x3 for 24mm^2 of "nozzle" orifices) presented the best trade-off of performance vs pump back-pressure in my mind. Oh, the actual "un-nozzled" channel area is 80mm^2, not 40mm^2. It's 40mm^2 but the flow splits into two directions from the central inlet so the effective channel orifice area is 80mm^2. Running from side-to-side is indeed 40mm^2. |
10-14-2002, 09:28 AM | #134 | |
Responsible for 2%
of all the posts here. Join Date: May 2002
Location: Texas, U.S.A.
Posts: 8,302
|
Quote:
Nice work. |
|
10-14-2002, 10:59 AM | #135 |
Cooling Neophyte
Join Date: Mar 2002
Posts: 20
|
wow Cathar! 1 lpm? Very impressive, what is that about 16gph or so? I've ROUGHLY measured mine down to about 25gph and they do quite well, but I've never dropped it down that far.
.. bits tough and stops rambling. |
10-14-2002, 05:33 PM | #136 | |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Quote:
I gave the performance results for this flow rate further back in this thread. I saw 2C higher temps at 1lpm than at 6.8lpm on a Duron @ 2.15v/1200MHz (~81W Computernerd Thermal Calculator) |
|
10-25-2002, 04:06 AM | #137 | |
Cooling Savant
Join Date: Oct 2001
Location: Wigan UK
Posts: 929
|
Quote:
Comparing predictions for a large area /thick based wb to your small area/thin based in the "Side-to-Side" configuration. . Some observed temperatures for the Les block (as shown on Page 2 of this thread): Morgan @ 2.03v, 1364Mhz (Computernerd 81.1W) @ 2.03v,1100Mhz (Computernerd 67.7W) @2.03v, 730Mhz (Computernerd 47W) Last edited by Les; 10-25-2002 at 04:57 AM. |
|
10-25-2002, 04:40 AM | #138 |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
It's a real shame that we can't bring the impact of the impingement cooling into the mix with your predictions Les.
In side-to-side mode, my block does indeed behave much like what you're predicting, but as soon as we go into middle-in impingement mode, the behavior of the block is very different. The final production block has substantially altered dimensions (relatively speaking). If you're interested in the measurements Les, then PM me. |
10-25-2002, 04:43 AM | #139 |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
The final block looks like this:
Note that the nozzle plate does look a touch different to that pictured. The one in the picture is a prototype plate. |
10-25-2002, 04:51 AM | #140 | |
Cooling Savant
Join Date: Oct 2001
Location: Wigan UK
Posts: 929
|
Quote:
2)Thanks but will not. NDAs can be rather restrictiveand already have sums for various configurations |
|
10-25-2002, 05:01 AM | #141 | |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Quote:
|
|
10-26-2002, 04:59 AM | #142 |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Well, there's some new data about the place. Divi8, who's a member at OCAU picked up a "production" block from me yesterday.
He is seeing a 6C drop over a Silverprop Cyclone 5 w/6.5mm base when running Folding@Home on an 1850MHz/2.00v AthlonXP. My own results saw a 6C drop for an AthlonXP @ 1925MHz/2.15v between my production block and a Silverprop Cyclone 5 w/6.5mm base. I have the Cyclone 5/3mm based block here and it performs about 0.5C better than the 6.5mm based model for the same test. The C5/3mm block is one that is being sold most predominantly now. Divi8 is seeing slightly better results than I expected, but he's using a different motherboard. His thread with his results is here: http://forums.overclockers.com.au/sh...hreadid=106836 |
10-26-2002, 12:35 PM | #143 |
Cooling Neophyte
Join Date: Aug 2002
Location: Atacama desert, Chile
Posts: 43
|
Les:
what is your block? I've looked several times not only the second page but also the first one and i don't find any reference to it. |
10-26-2002, 01:22 PM | #144 |
Cooling Savant
Join Date: Jul 2002
Location:
Posts: 264
|
What is the highest pressure you have tested these blocks to? It seems kinda gutsy having 2 O-rings, just wondering much pressure they can take before they burst.
I am sure a lot of people like myself are eager to run your block with high-pressure pumps, that is if you will ship to the US. Just curious, how many pre-orders have you gotten so far? Seems like there is a lot of interest in your block, just hope you can make it profitable! |
10-26-2002, 02:34 PM | #145 | |
Cooling Savant
Join Date: Oct 2001
Location: Wigan UK
Posts: 929
|
Quote:
dream caster. Some photos of Les 's block Reported: http://forums.procooling.com/vbb/sho...5&pagenumber=4 http://www.coolhardware.co.uk/module...121&highlight= Note. The K7Burn tests are on a different Morgan(and obviously seating) to the Jouni Vuorio's Stabilitity test results. Although the two CPUs are Nominally the same I got large differences in temperatures.Could be for many reasons but ,unless I can raise the dead, will remain an enigma to me.Reported: http://forums.overclockers.com.au/sh...threadid=67325 Last edited by Les; 10-26-2002 at 02:45 PM. |
|
10-26-2002, 06:00 PM | #146 | |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Quote:
I've gotten enough pre-orders to warrant the first batch to be made. They'll be ready within a week. I won't be making a profit from this batch, but merely meeting costs. Maybe later... If anyone wants one, PM me, or send me email via my profile link. |
|
10-26-2002, 06:11 PM | #147 |
CoolingWorks Tech Guy Formerly "Unregistered"
Join Date: Dec 2000
Location: Posts: 2,371.493,106
Posts: 4,440
|
re the o-rings:
given the matl thickness and assuming reasonable compression (I don't have any of that info), the wb should have little difficulty well beyond 300psi the o-rings are a non-issue (assuming as above) the threaded barb connections will leak first |
10-26-2002, 08:41 PM | #148 |
Cooling Savant
Join Date: Sep 2002
Location: Santiago, Chile
Posts: 403
|
You`ve done a great work with your block Cathar, congratulations!
The next thing you need for the big profit is a name for it and a nice looking webpage! There`s one question I have: Your intake nozzle is slightly off center to match hte center of the cpu core, but how can you tell the right mounting side when assembled? |
10-26-2002, 09:35 PM | #149 | |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Quote:
Failing that, look down the central barb and see which way the nozzle is and mount it with the nozzle furthest away from the socket A mounting arm. I have a name "Little River - Rapids". A web-page is on the way. |
|
10-30-2002, 05:16 AM | #150 | |
Cooling Savant
Join Date: Oct 2001
Location: Wigan UK
Posts: 929
|
Quote:
Silly me. Had intentions of look-see at Impingement on my 6mm bped block Have only just realised that I am probably on the upper bp thickness limit to get any improvement from impingement cooling. Playing with data from http://www.electronics-cooling.com/h...01_may_a2.html (Ta Bill) and http://widget.ecn.purdue.edu/~eclweb/jet_benchmark/ (Ta g-f) suggests: Speculative Performance Intention 3mm noozle .Impingement Cooling Diameter 15mm (max) with h ~ 100,000W/m*m*c for 200LPH flow. Think I need a thinner bit of Copper. |
|
Currently Active Users Viewing This Thread: 1 (0 members and 1 guests) | |
|
|