Mounting base for rotating rectifiers - Aluminum or Copper? 2

[edison123]

I have seen many rotating rectifiers (3 phase, bridge type) for generators mounted on aluminum plates, which also act as heat sinks.

Given that copper has a better thermal conductivity, wouldn't it make more sense to use it than aluminum?

Muthu

http://www.edison.co.in

 

[ScottyUK]

If aluminium does the job why spend extra money on making it from copper?

 

[itsmoked]

Have you tried working with copper? Gads. Plus it's a lot heavier if you're talking rotating.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[ScottyUK]

Yes, the copper components will experience much higher centrifugal forces than the lighter aluminium, and (unalloyed) copper is less able to withstand those forces than aluminium. I'm no metallurgist but I think the mechanically strong copper alloys have markedly poorer thermal and electrical conductivity compared to pure copper, while the conductivity of many aluminium alloys remains comparable to the pure metal.

 

[cswilson]

I have seen power electronic designs where the electronic component is mounted on a small copper plate, and the copper plate is mounted on a large aluminum heat sink.

 

[edison123]

Setting aside weight and cost considerations (which are negligible in overall scheme of things), talking of physics, which one is better as a heat sink?

I have seen american machines with copper base plates while the europeans use aluminum.

Muthu

http://www.edison.co.in

 

[edison123]

My confusion is between thermal conductivity (Cu/Al = 401/205) and specific heat capacity (Cu/Al = 0.37/0.92).

Muthu

http://www.edison.co.in

 

[LionelHutz]

Now you've got it.

Long term cooling capacity of a heatsink is mostly determined by airflow and surface area. So, if you have the same surface area on the heatsink then the use of Cu vs Al really makes little difference.

The higher heat capacity of Cu means a smaller piece of copper heatsink can be used in a power electronics application subject to short term overloads or periods of high power output followed by cooling off time. The higher thermal conductivity of Cu is why you might see a heatsink with a Cu center and Al fins. The Cu center helps conduct the heat to the fins a little quicker. Size vs cost vs weight has to evaluated to determine if it's worthwhile in these cases.

 

[waross]

How about copper washers on an aluminum base plate.
If you can source suitable copper washers, that avoids machining costs and keeps the mass of copper purchased low.
This may give the best of both worlds.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[edison123]

Thanks guys. I am now experimenting in our shop floor with Cu and Al bases of identical sizes in static condition to see which one takes away the diode heat faster and better. Will post the results shortly. I will be monitoring both the diode and the base temps at the full rated current of the diode.

This being a rotating unit, no fins can be fixed and surface cooling should be good enough with diodes themselves acting somewhat as a fan pushing the air around.

Muthu

http://www.edison.co.in

 

[cswilson]

It is important to distinguish the cooling system's steady state power transfer capability (in Watts) from the steady state temperature differential required to achieve that power transfer.

As a very simple illustrative example, imagine geometrically identical copper and aluminum bars insulated along their length. On one end, each bar has a constant power source (our power electronic device). On the other end, the bar is in contact with a large reservoir of ice water at 0C.

With the thermal conductivity of copper being twice that of aluminum, the power source device on the aluminum bar will be at a steady state temperature twice as high over 0C as the device on the copper bar providing the same amount of power.

The comparsion is not much different when dissipating to ambient air at, say, 25C.

Curt Wilson
Omron Delta Tau

 

[edison123]

Here are the test results. The diodes, rated 70 A, were loaded up to 60 A (heater load) for 30 minutes for each base in static condition with just natural cooling.

With 4 mm thick copper, the final diode temps were from 58 to 68 deg c and the copper temps (in 4 places) were from 57 to 67 deg C. Cu = 570 gm

With 5 mm thick aluminum, the final diode temps were from 59 to 77 deg c and the aluminum temps (in 4 places) were from 60 to 73 deg C. Al = 198 gm

With 3 mm thick aluminum, the final diode temps were from 59 to 70 deg c and the aluminum temps (in 4 places) were from 57 to 65 deg C. Al = 132 gm

All the diode temps were measured at the point where they were bolted to the base plates. The ambient was around 32 deg C.

Basically, not a great difference between the three. I am going with 3 mm thick Al mainly because of lower weight.

Thank you all for your tips.

Muthu

http://www.edison.co.in

 

[waross]

Thank you Muthu for sharing your interesting test results. lps

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[edison123]

Thanks Bill. I was kinda surprised that 3 mm Al had lower temp rise as compared to the 5 mm Al despite a 3/5th cross sectional area reduction. Guess being thinner, 3 mm Al was dissipating heat better.

Muthu

http://www.edison.co.in

 

[ScottyUK]

Yes, a slightly puzzling result. Same alloy, same surface preparation, similar surface finish, same diode package? Interesting that there's not a huge variance in results for copper or aluminium, although perhaps that can be explained by the metal-air interface being the dominant factor in power dissipation.

Thanks for posting the results of your experiment.

 

[edison123]

Did a DC resistance check of the plates

Copper 3 mm - Plate 1 & 2 - 0.070 & 0.071 mΩ
Aluminum 5 mm - Plate 1 & 2 - 0.071 & 0.071 mΩ
Aluminum 3 mm - Plate 1 & 2 - 0.10 & 0.10 mΩ

With such low values, the I^2R is minuscule in mW.

Muthu

http://www.edison.co.in

 

[LionelHutz]

Not too much a surprise when using a flat plate as a heatsink. The heatsink to air interface is the most dominating thermal resistance by far. Small differences in the test configuration are all that are necessary to account for the difference between the 2 pieces of aluminum plate.

I'm curious why you can't fit any fins? Space? I've seen quite a few rotating rectifiers with fins over the years.

 

[waross]

Woth the subtle differences that we are seeing, it may be something as simple as a different routing of the conductors to the diodes acting as a slightly more efficient fan.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[edison123]

Lionel

A pic of the test set up.

Except for the material thickness, all the test parameters are almost identical.

Bill

Yes, the diodes themselves will act like a fan.

Muthu

http://www.edison.co.in

 

[itsmoked]

I agree that the temps you're seeing, edison123, are (in the scope of this test) trivial differences in a static test so not much can be said about the impact of the different heatsinks.

Keith Cress
kcress -

http://www.flaminsystems.com