...
Printable View
...
Good guys :) I have bought some time ago stuff for my eshop.
I tested the older ybris cpu block with a prescott and it worked ok.
Of course i dont remember temps now...
the system is:
lapped X6800
Asus Pk5
Galaxy 1000
i'm not able to translate the thread (it is too technical for my english)
it is important to know the method to measure temperature
HERE the article
it's a new way to wc better the cpu...
if you can see the water enter from the center and jump up-down the sphere...
a sim:
http://www.pctuner.net/img_wb/acs/sc02.jpg
every time the water goes down take part oh heat...
it's very hard to imagine how it work because have pins to receive the water...
http://www.pctuner.net/img_wb/acs/pict8688.jpg
the wafer you can see have pins for the water
there's the original th:
http://www.pctuner.net/forum/showthread.php?t=88158
keep in mind that Y-one (it's in the comparative between acs) is about 1,5° worser than fuzion...
btw (sorry for english) don't try this @homer...
Unfortunely I don't have the Fuzion...
I can't say ACS is better than fuzion but
I can say that ACS is better than an Y-one by ~3.5°
That's it
Hi everybody, I'm italian and I own 2 PCs cooled by Ybris.
To show you the professionalism of Ybris WB builder, i decided to traslate the METHODOLOGIES OF TEST used by him to test WBs.
I repeat, I'm italian, so my english couldn't be perfect, but I think you will understand the most of what I'll write.
PCTUNER’S METHODOLOGIES OF TEST
The following methodologies of test are currently used in the laboratory of PcTuner for the execution of the comparative tests among waterblocks of different brands. It deals with some simple rules that we are given to ourselves to assuring us reliability and the possibility of repeat measures, they don't contain inapplicable conditions and, with a minimum of equipment, they are easily repeatable from whoever is interested to this type of verifications.
THE PLACE:
The first important thing is the absolutely necessary to do so that the air temperature remains constant during the whole test. It needs to do so that there is not current of air provoked by open windows or direct throws of cold air.
If a air conditioner is used to maintain constant the temperature of the place it will need to avoid that the area of test isn’t directly run over by the fresh air coming from the air conditioner.
The tests must possibly be effected in an ample room. This will make him that the heating of the air isn’t of any influence to the goals of the maintenance of the air temperature.
To keep in mind that if radiators of various type will be used for the cooling of the water the issued heat in the atmosphere is able to influence the result of the measure.
You will owe therefore to keep in mind at the most of this detail and threshold the influence of the exhausted air on the radiator.
In every case a variation of 1,0° in the air temperature involves the annulment of the test and the remaking of the test.
Gives the tendency of the air to create zones of temperatures stratification different to the various heights (there can be also 2° of difference among the temperature to 20 earth centimeters and that taking on the table of job) a probe environment of precision is positioned to the same height of the plan of job and non superior distance to the linear meter.
The probe is in such position to result not invested from appreciable current of air.
If a radiator is used, the probe is positioned near it to its same height, so that to notice the values of the temperature of the air that invests the radiator. During the execution of the measure it needs to avoid assemblages of personnel around the table of test because they can easily provoke some variations of temperature difficult to check.
THE SYSTEM USED FOR TEST:
Currently two systems are used for our tests:
• AMD based system for the measurements of precision
• motherboard: DFI NF3 250 UT
• processor: a couple of CPU AMD A64 socket 754 series ADA2800AEP4AX FC0 DH7-CG from 89 W of power to the frequency of 1800 MHz to 1,5 volts.
To one of the two CPU, the HIS of protection has been removed in copper so that to guarantee more precise measures without tied up problem list to the contact between base of the water block and processor.
The other one is maintained under standard conditions to simulate the normal assemblages.
In the image, the chart of reference of the processors used A64:
***original image cut***
Mother board Abit Fatality AA8X + serious P4 660.
P4 775 system used for less precise measurements because it demands only an indicative verification of WB’s behavior.
This because of the fact that on socket 775 are impossible to place asensor of precision and it need to exclusively submit him to those presents on the motherboard. (Not still correct!!! Refer to this thread: omissis)
The tests are done with the motherboard completely outside the case and put in plain.
If the motherboard is not provided of autonomous cooler in the zone of the mosfets, a 80mm of diameter fan is used to cool them, powered by the same motherboard and estate on the vertical one of the mosfets so that to create a flow of air that invests the whole section of feeding of the CPU and the components around the socket. You remembers that under load such components reach temperatures higher than 100°C if not adequately cooled and they can provoke instability in the system of test as well as breakups of the hardware. On some motherboards for base P4 and A64 of last generation such shrewdness is useless because the same cards are provided of opportune ventilation canalizations. With these the ventilation of the mosfets is can to be omitted.
The CPUs used for the tests of comparison have to be always the same and they must be used only to this purpose in combining with their own motherboards.
In fact, because of the tolerances in the feedings, different thermal behaviors of the same processor are found on different motherboard, also of the same type and model.
This fact forces to use always the same hardware and the same processors to the purpose to guarantee that the measure is repeatable in case of the need of another verification.
The frequency of operation and the tensions of feeding are decided time for time and recorded for opportune verification in case of repetition of the measure.
Fixing of the waterblock:
Currently the tendency is that to use the stirrups furnished by the builder even if it’s often preferred to primarily use a system of fixing that exploits the holes of the motherboard and that allows to slightly get best pressures of contact without risks of breakup of the hooks of retention.
USED SOFTWARE:
The system tested has installed Windows XP and the more common bench normally used.
For the tests of maximum load a program called Toast is used to bring the processor to the maximum real power.
Test of stress as S.Pi or 3dmark200x are not meaningful because they don't hock in constant way the CPU and they are not used for the tests of full load.
THE RESERVE OF WATER:
If is necessary to know the values of power put into the water, a thermal tank by 15 liters it is used or, in some cases, a 2 liters tank, always thermal isolated by the external temperature.
It deals with a normal refrigerator from camping, dressed in plastics.
It has a good mechanical estate and it allows to maintain the water inside to a temperature practically constant for some hours without dispersions toward the outside. The tank is completely closed, except the exits of the two pipes and the hole for the entry of the probe of the thermometer.
It is filled according to the intent of tests with a varying quantity from 7 to 15 liters.
In alternative, if there is not the necessity to verify the indeed absorbed power from the WB during the test, a normal plastic tank of 7 liters is used.
We use a discreet quantity of water because doing so we have a good thermal inactivity of the system and at the same time it’s insensitive to possible small sudden variations of the surrounding temperature .
According to the tests that must be done we use either a refrigerator to maintain constant the temperature of the liquid of cooling or a system composed by radiators, able to maintain a difference of T between air and water not higher that the 3°C, cooled by low volts fans.
This avoids to have some components (external walls of the WB and pipelines) with too different temperature from the surrounding air baiting phenomenons of natural dissipation that can alter the measure in progress.
In the case that a radiator has to be used the test is done in an ample room and with a big volume to minimize the variations of air temperature caused by the heat released by the same radiator.
In every case a variation in the air temperature of 1°C involves the annulment of the test and the remaking of it.
WHAT WE MEASURE:
Air Temperature °C
Water's Temperature °C
CPU’s Temperature °C
Liters / hour in the WB
TOOLS OF MEASURE:
We normally use a triple probes thermometer LM35 endowed with a circuit of correction of the errors and set with a referring tool of precision. The tool has entirely been realized by Omega3 and subsequently set with a thermometer in a range of temperature of + - 20° from the threshold of reference of the 25°C air temperature.
We use it to measure the air temperature and the temperature of water used in the cooling system.
(for anyone that didn't know it, these are the links to the characteristics of the tool built by Paul Omega)
omissis
omissis
***original image cut***
On motherboards with AMD socket A we use the third probe for the temperature of the CPU and it is positioned inside the socket, placed at the center of the core.
On motherboards with P4 478 socket (not largely used today), because of the presence of component smd on the back of the processor there is not the possibility to guarantee a contact suitable to the probes of lm35 and we opt for of the miniaturized sensor mod. pt100 class 1/3, used in combining to a double thermometer of precision Delta Ohm HD9117 , placed under the core of the cpu. :
omissis
All the tools are set to 1/10 of degree so that to guarantee precision and possibility to repeat measures.
***original image cut***
Why we use a system based on a 754 socket?
Simple, it is currently the only one (besides the old socket for A32 and p4 478) that allows to place some micro sensors of precision inside the socket and it allows to directly glue the probe in contact with the CPU.
If someone is interest in , here is showed the sequence to climb on a 2 mms sensor of precision pt100 on A64 754.
The socket is in contact with the pcb of the MB and is impossible to pass the wire rap wires below it, but.... with a tap of dremel...... ;)
***original images cut***
The images refer to the assemblage done on mother board Abit KV8 but obviously the discourse is worth for any MB 754.
The tool, equipped by its probe, was subsequently set and verified with a thermometer of reference mod. HD8601P also used for the verifications done preceding the tests and used to check the alignment of all the thermometers before beginning the tests.
It is not possible to use the sensors without an opportune compensation’s circuit for the errors, otherwise we will have a non linear answer in the execution of the measure and some errors will be done.
Using the sensors positioned under the socket we have the possibility to change the WBs without taking the risk of undue moves of the probes, guaranteeing a greater possibility to repeat measures already done.
Combined to the Triple thermometer of Omega3, the temperature of the water is also noticed with a thermometer Delta Ohm HD8601P with probe made by platinum, set and verified at least once a year.
It is not the best existing thermometer, let's intend us, it has a precision of 0,2C° but it guarantees a great stability in the time and it succeeds in maintaining the setting for a long time without the necessity to be retouched.
METHODOLOGY OF TEST:
The thermal cream used is always of good quality, generally AS5 or Dow Corning 340.
The quantity used is equal to the head of a match and it’s stretched only on the surface of the WB with circular movements up to form a circle of around 30 mms of diameter.
The surface of the processor is perfectly cleaned every assemblage / dismantlement.
The fund of the WB is photographed before and after the assemblage so that to be able to verify the imprint and to be able to check its correct assemblage.
2 types of test are done varying frequencies and Vcore, for a range of power substantially included between the 90 and the 180 W:
1) Varying Lt/H and rigorously constant water’s temperature:
It is the test usually used overseas.
It gives a graph that shows the behavior of the WB, varying Lt/h, generally in a range among the 150 and the 700 Lt/h. For this particular type of test a special pump is used (working with CC, for example 12v) and served by a special varying feeder.
With this pump (a SANSO Mod. PD31) is possible to vary exactly the Lt/h independently from the WB used and we have the certainty to do the comparison to the same Lt/h. The pump has a nominal course of 800 Lt/h with a prevalence of 8 meters to 24 Vs but it is also possible push it over the 30 Vs, increasing the course and the prevalence in a considerable way.
This test is useful to understand the behavior of the WB once you notice the values of course of your own cooling system.
2) With constant temperature of the water and typical course with a series of different pumps.
This test is very interesting because it allows to see the WB’s behavior in comparison to another, feeding it of time in time with different pumps and allowing to it working to the typical course given by the used pumps.
the following pumps are used for this tests:
• NJ1200
• NJ1700
• Soda L30
• Swiftech MCP 650
• Sanso PD 31
• Other possible pumps according to the evolution of the market.
The pc is turned on, not working and without active idle. The real test begins after about twenty minutes from the switching on.
The WB must have climbed on from at least 6 hours because it needs to attend the arrangement of the thermal cream, otherwise non reliable results and poorly repeatable will be gotten, with errors that can also arrive to over 3°C.
We launch the program of burn and we leave the PC at work for at least 30 minutes. At this point we take all the interested greatness, air T, water T and T of the processor.
Once the test is finished, we turn off the system and we notice the course that the WB had with that specific pump.
All data we got are represented in a graph that shows the behavior of every single WB with all the available pumps.
For an inside rule, if during the test the air T varies more than 1,0°C the test is suspended and it is restarted afresh.
This because, where is necessary to cool the mosfets with a fan, the variations of air T can influence the reading of the temperature on the CPU.
The problem in reality doesn’t exist on P4 and A64 based system with motherboards that have a canalized cooling system of the sections of feeding but for an our decision this aspect doesn’t change the rule.
The probe that notices the air T is positioned to around 1 mt of distance by the motherboard and placed at the same height of the CPU or, in the cases in which a system of radiators is used for the maintenance of the temperatures, at the same height of this system.
Also in this case, as in all the other tests, the room has an air conditioner set among 23° (during the winter) and the 25° (during the summer) because these are easily obtainable values without too much effort.
At least 3 tests are performed and one at least 2 hours later than the other. If differences are found in the superior readings are higher than 2/10 of °C on the CPU it means that the period of arrangement of the thermal cream is not finished yet and the test isn’t valid.
In this case the test is postponed of 12 hours, otherwise the test is considered acceptable and recorded.
The whole procedure is repeated thin to get a series of at least three complete tests for every WB, after that an average of the recorded values is made and the final graph is drawn.
We however say that these are not the only measures that we notice. According to necessities, some other different specific tests can be developed, anyway all general conditions will be respected. The possibility to repeat these other tests (verified with enough reliability) will be respected too.
You can find the original text in PCTuner forum>Tuner/Waterblock>metodologie di test by Alex Ta.
That's all, bye
Too tecnical for my english too, but let my try... (I'm italian, bla bla bla..., you already know it:) )Quote:
Originally Posted by stoner
Ybris ACS
The history of ACS begins in the first half of 2005, during the phase of development of another waterblock, the K11 Leviathan.
In that period the most of WB builders were developing the fins technologies but, at the same time, we began to also look for possible alternatives, well knowing that very soon the limits would be reached and that it was not possible to subsequently develop some complex fin system, able to improve the performances of the waterblock.
The first prototype of the waterblock had been realized building a rectangular matrix of needles directly elicited through a series of close perforations.
The structure that results had an interesting characteristic: the fund was not flat but it was formed from so many small half-spheres and in the points of union the needles departed working as the cooling element.
***Image 1, 2 & 3***
The practical realization however wasn’t simple: too much time spent in CNC cutter but above all a notable quantity of dribbles to be eliminated after the work.
Here we see the only prototype immediately realized after cleaning with sand.
***Image 4***
The waterblock received the water through a shower diffuser realized in delrin and, beyond the difficulties in the realization, it was very effective during the tests.
***Image 5 & 6***
The principal problem of this structure however was represented by the excessive thinness of the plate in copper of the base.
If the performances were wanted it needed to be near to the core but in this case the simple pressure of fixing of the grapevines was sufficient to deform the plan of contact and to immediately worsen the temperatures of job. Increasing the thickness, the matrix began to lose in performances for which we needed to look for the correct compromise.
The matrix, anyway, had the tendency to assemble too much dissipation to the center; it was fantastic on the processors with naked core but it lost a lot of its characteristics when it needed to cool the new cpu that were arriving in the meantime on the market: the new Dual Core Intel.
So the project was radically modified.
The initial idea (the structure of the fund made with half-spheres), were maintained eliminating the needles in relief and, this time, we studied a different distribution of the water.
It was necessary to get an elevated speed of slide, great strength of impact on the copper but, above all, it needed to extend the effect on a very ample surface, equal at least to that a square from 29 mms of side.
So we thought the idea to realize a double circular matrix of contrasted half-spheres: the core of Ybris A.C.S was born.
***Image 7***
In the image the superior half-spheres have been cut to let you see the underlying part but imagine them completely closed.
***Image 8***
As you can see, departing from any point of the cell, every time that a fluid fillet enters an inferior half-spheres (water goes down) it finds at least three possible streets of exit (water climbs) exploiting the passage created by the bewilderment with the superior half-spheres.
In this way water was forced to do a series of casual rebounds before finding the street of exit.
We built a hypothetical external form: the matrix was closed in a wrap and we started a series of simulations to study the fluid-dynamic behavior inside.
In the image: one of the so many simulations
***Image 9***
That was the moment to try a live the structure: it came so a first prototype and was tested in the laboratory.
***Image 10, 11 & 12***
The resultant effect of this structure was impressive: the turbulence was inclined to a levels that the waterblock start "to blow" but the performances allowed to intend that the way was correct.
Found defects: general courses too limited (therefore necessity of powerful pumps) and a mechanical shape not so good looking at it.
It was the case to slightly reduce the effect "puff" trying to maintain unchanged the performances.
So the matrix was modified again.
In the actual version every cell is composed of 3 half-spheres, every half-spheres is surrounded by 6 spheres (therefore the cells melt him among them because they have some common spheres) and water is sent in a point that breaks the flow in three parts.
***Image 13***
From that moment, we were able to exploit the rebounds among the "domes" and the "cave" and water looked for a casual passage crossing the matrix up to reach the peripheral zone of harvest.
This is what happens in ONE elementary cell. Now expand the same concept to the whole matrix: every cave will receive a part of a direct throw and at least two parts of throw coming from the adjacent caves.
We increase therefore the number of the throws and we pass from the initial idea to unique central collector to a series of holes predisposed in some specific points of the structure:
This is a hypothesis of the inside distribution calculated on a limited number of fillets and a distribution of 7 holes (they are the red zones of acceleration)
***Image 14***
To be continued…