Thermoengine Heatsinks : Black or white? Here are the results !

[MGTasco]

Hi guys,

Since there is a big debate in the past on hollow/solid core or white/black thermoengine, I decided to put both onto a test and find out myself. Here are the results :

http://forums.reviewmakers.com/viewtopic.php?topic=138&forum=1&4

enjoy...

 

[Colin]

Thank your for your efforts. Two points.

Thermosonic continues to send out hollow core HS for reviews despite the fact that the solid core model is the only one availble to the public. Thermosonic's own documents reveal they knew of the performance difference last December, and yet the continue with this deceptive practice.

Socket thermistors are known to compress temperature readings due to the secondary heat paths. In addition, the Thermoengine cools the socket thermistor through secondary heat paths. The only reliable way to measure the difference between the two Thermoengines is with a CPU that has an in die thermistor. The Palomino will do this and put egg on the face of almost all of the hardware Sites. History in the making!

 

[MGTasco]

Colin (Jul 20, 2001 12:54 a.m.):
Socket thermistors are known to compress temperature readings due to the secondary heat paths. In addition, the Thermoengine cools the socket thermistor through secondary heat paths. The only reliable way to measure the difference between the two Thermoengines is with a CPU that has an in die thermistor. The Palomino will do this and put egg on the face of almost all of the hardware Sites. History in the making!

But I did not use socket thermistors. My readings are from thermocouple attached to the core, and is connected to a digital thermometer.

 

[Nevin]

MGTasco,

All my measurements of hollow core vs. solid core ThermoEngines have resulted in a 20-gram weight difference, plus or minus two grams. In your write up, you state the weight difference is 11 grams.
Did you visually verify that the ThermoEngine you believed had the hollow core did indeed have the cap with writing on it under the attachment clip. If there was no writing on top of the unit and it does not have a cap, then your comparison was actually solid core to solid core and the weight difference was due to other design changes.

Also, many sites reported early on that the hollow core black version performed significantly better than the hollow core silver version. Since 99% of the tests on the web are of the hollow core black version, and your comparison of the new solid core version is to a hollow core silver version is it possible that the true differences between the new and old versions are minimized?

And while measurements by thermocouples touching the side of the core may not exhibit quite as much compression as in-socket thermistor measurements, they are still measurements taken external to the heat source in a secondary heat path and are subject to unavoidable natural compression. For example, if you are trying to measure the internal temperature in a wood-burning stove, it doesn't matter whether you use a thermocouple or a thermistor nor whether you attach it to the bottom or the side of the stove, you will never get an accurate measurement.

Nevin

 

[Colin]

MGTasco (Jul 20, 2001 03:38 a.m.):
But I did not use socket thermistors. My readings are from thermocouple attached to the core, and is connected to a digital thermometer.

Sorry, I just skimmed the post you referenced. Thanks for pointing this out. What Nevin wrote is correct. You will still get compression of temperature measurements with a themistor at the side of the core.

 

[MGTasco]

Nevin (Jul 20, 2001 12:17 p.m.):
MGTasco,

All my measurements of hollow core vs. solid core ThermoEngines have resulted in a 20-gram weight difference, plus or minus two grams. In your write up, you state the weight difference is 11 grams.
Did you visually verify that the ThermoEngine you believed had the hollow core did indeed have the cap with writing on it under the attachment clip. If there was no writing on top of the unit and it does not have a cap, then your comparison was actually solid core to solid core and the weight difference was due to other design changes.

Also, many sites reported early on that the hollow core black version performed significantly better than the hollow core silver version. Since 99% of the tests on the web are of the hollow core black version, and your comparison of the new solid core version is to a hollow core silver version is it possible that the true differences between the new and old versions are minimized?

And while measurements by thermocouples touching the side of the core may not exhibit quite as much compression as in-socket thermistor measurements, they are still measurements taken external to the heat source in a secondary heat path and are subject to unavoidable natural compression. For example, if you are trying to measure the internal temperature in a wood-burning stove, it doesn't matter whether you use a thermocouple or a thermistor nor whether you attach it to the bottom or the side of the stove, you will never get an accurate measurement.

Nevin

Hi Nevin, thanks for pointing out a couple of things. For the silver version, there was indeed writings below the attachment clip. But there ain't any words for the black one.

I have check and use a few weighing machines, and it was indeed 11 grams lighter. The hollow silver was gotten from Taiwan and the black solid one was from Japan, so I am not sure if Thermosonic produces different specs heatsinks for different market.

For the use of thermocouples, I know that this method is not the most accurate possible, but I really couldn't think if there is really any other more accurate methods. Since my test are meant for just comparison, I thought the method should be ok to a certain extend.

 

[Kingslayer]

Accuracy shouldn't be the issue. You were comparing two basically same items. If the thermocouple is a few degress of on one, it will be a few off on the other. But it will still show the difference. And that was what you were looking for.

Good Job. Now if we can get a straight answer from the manufacturer.

 

[Nevin]

Kingslayer,

The problem with external measurements is not only accuracy; it is that a measurement downstream from a heat source in a secondary heat path will always exhibit compressive errors.

Using my wood burning stove example from above, when the inside of the stove is 500F, a thermistor attached to the outside wall may read only 250F. Raise the temperature inside the stove 100F to 600F and the thermistor measurement will only go up 20F to 25F.

If you could accurately measure the actual internal temperature of the CPU core in this way, then you could measure the temperature inside an oven by touching a thermocouple to the front of the door, you could measure the temperature inside a house by touching a thermocouple to the outside of a window or door, and you could measure the temperature of coffee in a cup by touching a thermocouple to the outside of the cup.

In any of these cases, the thermocouple is measuring the temperature down stream from the heat source in a secondary heat path. Raise the temperature of the oven, house or coffee 20F and you will only measure a change of 5F to 7F. Thermocouples touching the side of a CPU core are subject to exactly the same type of error. The temperature change compression is a natural phenomenon of secondary flow path when there is "both" heat flow and thermal resistance from the heat source to the measurement point. This is the normal/natural case and cannot be avoided.

You also have to remember when making temperature measurements, that a thermocouple or thermistor does not measure the temperature of what it is touching, it measures the temperature of itself. In this particular test, when 5% of the thermocouple area is touching the side of the core and 95% is touching either the ceramic of the chip or the air between the chip and the heatsink, then the temperature of the thermocouple will be an average of the side of the core, the top of the ceramic and the air temperature.

Nevin

 

[MGTasco]

up

 

[Kingslayer]

Sure, but he said that he had the probe in the same place, and the heatsink is the same design. There for the heatpaths should be identical. And if the heatpaths are identical, then accuracy of the thermocouple is null and void because it will be inaccurate by the same amount for the next HSF.

If your heat paths are identical and you do the test.

Heatsink #1 show 75 degrees.
Thermal couple is 10 degress innacurate.
Actual temp is 65 degress.

Heatsink #2 show 85 degrees.
Themal couple is still 10 degrees innacurate.
Acutal temp is 75 dgrees.

There is still a ten degree difference in #1 and #2 becuase the heatsinks will create the same heat path, measured from the same point, with the same inaccurate TC. At this point you aren't measuring heat, you are measuring difference in readings.