Two butterflies powered by rocket fuel
Tue, 10/02/2007 - 22:57 — Lubomir Kuzman
I have to write some words for better understanding my methodology of thermal qualities of the coolers, I think this will be constant in future. The heaters for my tester are two 50 watts power resistors of a values 470Ohm parallel-connected. ( these resistors are also suitable for long-time overloading ) screwed to the copper base. Base plate is at the second side milled to the dimensions of the IHS of the CPU.
Both of the surfaces are well wroughted to almost the ideal flatness and well polished too. I don't want to write the recipe for fabrication of the thermal simulator of the CPU, but for the first time it is the needing to specificate what, of what and how. The copper block have installed three thermal sensors ( next TCS ). Hole, thermal paste and thermo-couple wire type J ( thermo-electrical cell based of the constantan and the iron ) First is situated between the power resistors, approximately 10 mm deep. The others are situated in the holes at the sides of the copper block. The result temperature of the surface of our CPU is averaged of these three TCSs. The three TCSs are enough to have exact and non-floating results. :)
At these pictures is shown work-flow as I did for achievement the accurate flatness. As the flatness of a copper block after previous miling wasn't so accurate, I worked on it for three hours. As the "ideal" base I used the glass, for grinding I used standard sand-paper ( 600, 1000, 2000 ) and of course the water.
Some water-adhesion tests of the copper block with glass - what can I say? Need to see! The result of my work was good reflection and almost ideal flatness of a copper block.
Here are some photos where exactly at the block are the TCS situated
Two side views of the tester. Here are so good visible how the power wires and the TCSs wires are secured.
This picture shows the complete tester. Only to need to clean the copper surface by rubber, to apply the thermal paste, screw the cooler, connect the cables to the power supply and we can do the measurements. :)
40-channel temperature meter and datalogger YOKOGAWA DX2040 - "the top of the measuring instruments". This meter is very acurate. It contains also LAN and USB2 connector and have 80MB internal memory for measured data.
By the LAN connection to the PC, I can see on-line when the temperatures are stabilized. Stability in this case means that the temperature fluctuation is less than 0,1oC in last 15 minutes. At this picture are temperatures of ambient air of cooler that runs in passive operation.
As the power supply for resistors is used Hewlet-Packard 6811B power supply and analyser. I simply set up the voltage and measure the "heat" power. Temperature leakages of resistors are almost the same as the leakages through the CPU pins. So I simply didn't no corrections of the measured data.
So that was measurements, now something about methdology.
At first at the beginning of this review, I have to make a little correction of the head-line. The rocket powered cooler from the ZEROtherm company is only the one of these both good-lookers. It's name is BTF80 and it is thanking for this to it's fan. BTF95-Fanless by my mind losed his fan during the supersonic speed under water. I certainly don't know where else this could be possibble. Be sure that this cooler is dedicated rather for a passive PC. While the butterfly named as BTF80 will be better to use in PC cases, where the circulation of the air needn't to be so ideal. And what are the commons of these coolers? You really don't know? Well, the side panel at the PC case have to be made of clear plexi glass because both of the butterflies are really looking very good. Well, if You accidentaly don't have this, so chop-chop and run to buy some discs for your angle grinder, or buy the fret-saw, next some paper-glue and piece of glass, that will be no problem for everyone to scrabble-up from some old windows because of "plastic windows madness". Well, but this activity sequence is my patent, so you have to ask me for my approval :) . This modification is for sure really profitable. Only like that the beauty of one of these butterflies would be shown roundly.
I have to write some words for better understanding my methodology of thermal qualities of the coolers, I think this will be constant in future. The heaters for my tester are two 50 watts power resistors of a values 470Ohm parallel-connected. ( these resistors are also suitable for long-time overloading ) screwed to the copper base. Base plate is at the second side milled to the dimensions of the IHS of the CPU.
Both of the surfaces are well wroughted to almost the ideal flatness and well polished too. I don't want to write the recipe for fabrication of the thermal simulator of the CPU, but for the first time it is the needing to specificate what, of what and how. The copper block have installed three thermal sensors ( next TCS ). Hole, thermal paste and thermo-couple wire type J ( thermo-electrical cell based of the constantan and the iron ) First is situated between the power resistors, approximately 10 mm deep. The others are situated in the holes at the sides of the copper block. The result temperature of the surface of our CPU is averaged of these three TCSs. The three TCSs are enough to have exact and non-floating results. :)
At these pictures is shown work-flow as I did for achievement the accurate flatness. As the flatness of a copper block after previous miling wasn't so accurate, I worked on it for three hours. As the "ideal" base I used the glass, for grinding I used standard sand-paper ( 600, 1000, 2000 ) and of course the water.
Some water-adhesion tests of the copper block with glass - what can I say? Need to see! The result of my work was good reflection and almost ideal flatness of a copper block.
Here are some photos where exactly at the block are the TCS situated
Two side views of the tester. Here are so good visible how the power wires and the TCSs wires are secured.
This picture shows the complete tester. Only to need to clean the copper surface by rubber, to apply the thermal paste, screw the cooler, connect the cables to the power supply and we can do the measurements. :)
40-channel temperature meter and datalogger YOKOGAWA DX2040 - "the top of the measuring instruments". This meter is very acurate. It contains also LAN and USB2 connector and have 80MB internal memory for measured data.
By the LAN connection to the PC, I can see on-line when the temperatures are stabilized. Stability in this case means that the temperature fluctuation is less than 0,1oC in last 15 minutes. At this picture are temperatures of ambient air of cooler that runs in passive operation.
As the power supply for resistors is used Hewlet-Packard 6811B power supply and analyser. I simply set up the voltage and measure the "heat" power. Temperature leakages of resistors are almost the same as the leakages through the CPU pins. So I simply didn't no corrections of the measured data.
So that was measurements, now something about methdology.
- Used input powers of resistors are: 35W, 65W, 95W, 125W, 155W, 185W, 215W.
- I don't measure the rpm of fans, but I set the voltages for fans by regulated power supply to: 5V, 7V, 12V.
- For valuation of quality of cooler will be used the table, where will be typed measured values of the temperature of the "processor" ( next as CPU ) , ribbs of the cooler during the conditions, that are typed thereinbefore
- For quick mutual comparison of efectivity of coolers will be presentated in every review also comparison database of coolers - as table. For the active coolers will be beside the name of cooler typed number, that means the difference between the temperature of the CPU and t a @ input power of 155W and the voltages of fan: 5V, 7V and 12V. So if the CPU will have the temperature durring these conditions of 85 oC and t a will be 24,5 oC, the result number will be 60,5oC. For the passive coolers will be this progress detto, but measure conditions will be static air and power of 65W and the worst position of the cooler. Of the mathematics view is this database no nexact, but it have to be used mainly for quick comparism. For detailed comparism have to be used detailed table for every cooler.
- As the thermal conductive paste will be used only the OEM paste delivered with cooler. For a subjective description of a quality ( Q ) of the thermal conductive paste will be the quality number, that means the thermal difference between the t a and the CPU, typed in each review. For the active coolers will be used the power of 215W and the standard voltages 5V, 7V, 12V and for the passive coolers will be used the power of 65W and the worst possition of cooler.
- The screws for the fixation of the cooler's board to the tester's board are screwed so that between the all screwed boards have to be space of 10mm in corners.
- Tested cooler will be "burned" due to the thermal conductive paste for 1 hour at 35W in a passive mode.
- All measurements are made at first with minimal power and with various voltages for fan. If the temperature of the CPU will drop to cca 35oC, for next measurement have to be increased the power to resistors and not voltage for the fan. If the temperature of the CPU will overruns 90oC, for next measurement have to be as first increased the voltage of the fan. For the power of 155W and more will be setted voltage for the fan only to 12V
- For the active coolers will be used only one position - tester horizontally. Fan blows air to the center of the room, or like as BOX coolers, the air blows to the boards of the tester.
- Ambient temperature ( t a ) is measured cca 10 cm above the ground and approximately cca 60 cm under the cooler. The position of the thermal sensor is for all thermal measurements always identical. The thermal fluctuation of t a don't have to be the correction number for all measured values, that means that t a would be always different, becouse the calculation of our measured thermal values to the constant t a is uselessly and too difficult
- Temperature of the tabular air is used only for a orientation. It can closer say something about the thermal transfer efectivity of cooler. The tabular air is measured by using of 9 thermal sensors matrix 5 cm behind the ribs of cooler and it have to be centered to the shape of the measured cooler. For the BOX coolers will be this matrix paralelly situated.to the least one plain shape of the measured cooler. Thermal sensors matrix has dimmensions 10 x 10 cm and have 9 thermal sensors.
- For the passive coolers pays the same rules as thereinbefore, but there are used all 3 orientations of the cooler. Because of I want to find the best position of the cooler according to the best efficiency of the natural draught Thermal sensors matrix is placed the best centered to the shape of the cooler and cca 5 cm above it.
- To the tested cooler are glued-on the thermal sensors. At the base of the cooler will be glued-on one sensor, and at the ribs next 6 to 12 sensors. This depends of the size of the cooler. If the cooler is large and complex like Macs Triumph MA-7131-A Deluxe it would be better to use at least 18 sensors.
I think, that I named all the basis of exact measurement, and I will try to observe this terms for all of the reviews that I'll do.
