diyAudio (https://www.diyaudio.com/forums/index.php)
-   Everything Else (https://www.diyaudio.com/forums/everything-else/)
-   -   Heat sinks - going the other way (https://www.diyaudio.com/forums/everything-else/14348-heat-sinks.html)

 jackinnj 27th April 2003 02:41 AM

Heat sinks - going the other way

you go through the math on calculating the optimal heat sink for a design knowing how much power (heat, calories) it will to have to dissipate -- then look in the Thermalloy listings and choose the sink you need.

I am just ruminating about how to go the other way -- given a surplus heatsink, how much heat can it dissipate-- time to get out a big Dewar, some fluid of known dissipation parameters (i.e. salad oil), perhaps a '3055 or '2955 transistor and figure this out -- anyone out there done this?

 PassFan 27th April 2003 02:52 AM

Re: Heat sinks - going the other way

Quote:
 Originally posted by jackinnj you go through the math on calculating the optimal heat sink for a design knowing how much power (heat, calories) it will to have to dissipate -- then look in the Thermalloy listings and choose the sink you need. I am just ruminating about how to go the other way -- given a surplus heatsink, how much heat can it dissipate-- time to get out a big Dewar, some fluid of known dissipation parameters (i.e. salad oil), perhaps a '3055 or '2955 transistor and figure this out -- anyone out there done this?
Why not just match the surplus extrusion to one in the thermalloy catalog and see what it is? Unless you absolutely must...

:bigeyes: ;)

 jackinnj 27th April 2003 03:23 AM

well different alloys

have different dissipative characteristics

 stokessd 27th April 2003 03:48 AM

This isn't totally without error, but the most simple way I can think of, is to bolt a power resistor to the heat sink, and dump a known amount of power into the resistor.

This will also show you the real world heat sinking capacity of the unit at the point where you want to mount your transistors.

A multimeter and a thermometer should be all the measurement gear you need. You'll either need a DC power supply or a variac to supply the BTUs.

hope that helps.
Sheldon

 Dark Shadow 27th April 2003 05:19 AM

A more accurate way would be to use a peltier as your heat source. This is the way that a lot of internet review site proceed to benchmark CPU heat-sink.

 jackinnj 27th April 2003 01:55 PM

so here's what I am thinking

2 Attachment(s)
I bolt a couple of these onto the heatsink and cook in a dewar with non-conducting liquid of known thermal transfer properties, taking measurements of temeperature over time. I guess that at some point the curve will show saturation.

 cocolino 27th April 2003 03:32 PM

Heat sinks - going the other way

Quote:
 originally posted by jackinnj you go through the math on calculating the optimal heat sink for a design knowing how much power (heat, calories) it will to have to dissipate -- then look in the Thermalloy listings and choose the sink you need. I am just ruminating about how to go the other way -- given a surplus heatsink, how much heat can it dissipate-- time to get out a big Dewar, some fluid of known dissipation parameters (i.e. salad oil), perhaps a '3055 or '2955 transistor and figure this out -- anyone out there done this?
As heatsinks are THE major cost factor (many times even more costly than transformers and caps) when building high power dissipation amps, I think it`s a good idea to do the "other way round" (and actually that´s what I´m doing).
With Class AB designs I believe it`s not such a big issue when in practice comes out that Your surplus sinks are somewhat underestimated - just lower the bias somewhat. Assumed You are not way off Your calculations I doubt this would matter much.
When it turns out that You still have some reserve - even better (in contrary to an amp manufactorer who has to calculate right to the point, for a DIYer it`s acceptable having overspecified somewhat ).
It`s a bit trickier with Class-A as You don`t have the choice to vary bias with a given design.

Nevertheless I´m sure it`s possible to calculate unknown surplus heatsinks realistically and coming close into the ballpark, when comparing to known datas from similar sized heatsinks. With similar sized I mean proportions and surface area (don`t compare a 50cm length/10cm wide piece with one that is 25cm/20cm), thickness of baseplate and length of fins.

different alloys
Quote:
 originally posted by jackinnj have different dissipative characteristics
The following statement I found on a website from a heatsink producer (R-Theta ):
"It has become common practice in the electronics industry to specify the aluminum alloy 6063 or 6061 for finned heatsinks."

This information about thermal conductivity of aluminium alloys I found on another website:

Aluminum (6061)
conductivityW/m*K: 171
density: g/cm(3): 2.6-2.9

Aluminum (6063)
conductivityW/m*K: 193
density: g/cm(3): 2.6-2.9

From that it`s easy to figure that the differences between the commonly used heatsink aluminium alloys are between 10 to 15% of conductivity.
In regard to other (partly more unforseeable) variables (proportions/mounting/arrangement of heatsinks/semiconductors - material of insulation pads - forced convection or not) that`s not too bad and 15% variaton as worst case can be considered when dealing with aluminium heatsinks made from unknown alloy.

Quote:
 originally posted by stokessd This isn't totally without error, but the most simple way I can think of, is to bolt a power resistor to the heat sink, and dump a known amount of power into the resistor. This will also show you the real world heat sinking capacity of the unit at the point where you want to mount your transistors. A multimeter and a thermometer should be all the measurement gear you need. You'll either need a DC power supply or a variac to supply the BTUs.
This is exactly what I thought I should do after I got some samples of this unknown surplus heatsinks for inspection (spam, spam...):

After I have studied and compared datas from various similar heatsinks I came to the conclusion that the effort is not worthwile (moreover I did not want to spoil one and drilling holes for mounting resistors) as in regard to all the other variables which have influence on the ACTUAL thermal dissipation capability of a heatsink/transistor/enclosure arrangement my estimation for the thermal properties of that heatsink is reasonable and accurate enough.

After all even Nelson Pass seems not to do exact calculations from what he said here (when he was not kidding):

Calculations? We don't need no stinkin calculations! We just make them really big!

 cocolino 27th April 2003 05:32 PM

Quote:
 I bolt a couple of these onto the heatsink and cook in a dewar with non-conducting liquid of known thermal transfer properties, taking measurements of temeperature over time. I guess that at some point the curve will show saturation.
jackinnj,

I can follow that it makes perfectly sense to bolt a couple of resistors onto the heatsink and to apply power from a DC power supply (or AC with a Variac) to figure if a heatsinks power dissipation capability suits ones needs.

What I can`t quite understand why You`d want to put that thing into non-conducting liquid?

Why not just measure the surface temperature (after sufficient time until the temperature settled and the heat spreaded evenly) under conditions which come closer to the real application?
To say - in the position in that You want to use it actually and in free air (or with fans attached if You want to use).

 Dark Shadow 27th April 2003 05:53 PM

See the follwing link for a stup to test CPU heat-sink:

http://www.overclockers.com/tips263/index02.asp

 rikkitikkitavi 27th April 2003 06:01 PM

actually placing the heatsink in a liquid is going to give large measurement errors.

Air is a much worse conductor than any liquid at STP. This means that much of the thermal resistance lies in the boundary air-heatsing. (unless the heatsink is very long, or the heating component is placed in one corner) , but placing it in a liquid , this resistance decreases dramatically-

best is of course test it as close as to the real setup, same contact surface on the heating resistors, same number e t c.

a surface thermometer is also useful, but a small chip temp like LM335 taped to the surface is also useful, but it might be necessary with a small heatinsulating cover insulating it from the surrounding air to prevent measurement errors due to convective cooling from the chip.

I used a thermocouple which gives a fast and accurate response for testing my heatsinks (found datasheets later)

/rickard

 All times are GMT. The time now is 06:42 PM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.