Heatsink Position, Temp, and Vgs

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I am working an Aleph-5 Monoblock that will have 2 large heatsinks with 3 IRFP244's per heatsink.

However, due to the design of these heatsinks, the devices will be mounted vertically; i.e., one on the bottom third of the heatsink, one in the middle, and one near the top.

I am a little concerned that the upper device might be a little warmer than the bottom device due to natural convection. I know that Vgs is temperature dependent, but I'm not sure if being warmer will make the device run harder or easier.

So... Which of the following am I looking at?

  1. The top device, being warmer, runs harder, making it even warmer - a runaway effect. Bad.
  2. The top device, being warmer, runs easier, making it cooler - a self-correcting effect. Good.
  3. Because they are all on the same heatsink, the convection effect is minimal, so don't worry about it. Good
    [/list=1]

    Any ideas?
 
Member
Joined 2003
Paid Member
A picture might be good, or some dimensions, maybe a drawing. That way we could help you better.


john


edit: If you're using these, using 2 per monoblock, it may run a little too hot for comfort. If your using those sinks, I would use 6 per monoblock and mount 1 transistor per sink, that way you could probably raise the bias as high as you want, but then again, you havent specified the size of your sinks.
 
MattM,

>> 1)- The top device, being warmer, runs harder, making it even warmer -
>> a runaway effect. Bad.
>> 2)- The top device, being warmer, runs easier, making it cooler -
>> a self-correcting effect. Good.
>> 3)- Because they are all on the same heatsink, the convection effect is minimal,
>> so don't worry about it. Good

First, I assume your heatsinks are mounted vertically, and the fins are also vertical...
Also there is space under the bottom of the heatsink ...
Plus these heatsinks were wisely calculated, so :

Q1: On my A75 no more than 1or 2 degree celcius between the bottom and the top
on 8" inches heatsinks ... after 20 minutes
Q2: Wrong, the Mosfets must share the same temp if possible otherwise
the sound will suffer....
Q3: The convection effect is the big principle of a vertical heatsink ; air when flowing
on the surface of the fins charges with heat, and so removes heat from the heatsink.

The 24 * TO3 Fets reach a case temperature of 70 degrees on each side ;
the angle that support the fets reach about 62 degrees and the heatsinks 52 degrees.

If too much difference between lower and upper mosfet case is > 3 to 5 degrees
it could be wise to install a support sharing the 3 mosfets attached to the heatsink
copper or brass to help share the same temp.

Without a complete schematic, models and dimensions of your heatsinks, fixation used
for the MosFets ; Bias etc, can't help you further...

PS:
Not so easy without any technical specifications... :)

Alain.
 
Alain,
Certainly the air would be hotter as it travels up the heatsink, but I am not sure about the inside surface of the heatsink itself...

You could run a simple test to see if there is much difference in the heat sink temp from top to bottom. The thermal conduction of the aluminum may be great enough so that the transistor mounting surface temp is fairly even.

If you do conduct a test, let us know what you find.

Jeff
 
Thanks for the replies!

Yes, I am using these.

I plan to mount 3 devices per heatsink (two heatsinks per monoblock). The heatsinks will be mounted with the fins running vertically. In that orientation, they are 12" tall, 5.8" front to back, and the fins are 3" wide. Each heatsink weighs 5-1/2 pounds and has 16 fins. They are Wakefield Engineering part number MD387A, but I can't find them on their web site so they may have been a custom item. They are closest to extrusion profile 8415 - here is a link with some specs.

They will have air space under them, so I am expecting natural convection to create quite a healthy air flow across the 12" height.

Maybe the answer is simply to mount the devices at the same level. I had originally planned to spread them out across the heatsink (as depicted in the "wrong" half of the diagram below). While mounting them close together may or may not make them run a little hotter, at least they will all be at the same temp.
 

Attachments

  • heatsink.jpg
    heatsink.jpg
    7.7 KB · Views: 448
My take is that you shouldn't worry so much about bias problems as reliability. The top-most device will run hotter, which will stress it more than the lower devices. The smaller the heatsink (hence hotter, all other factors assumed equal), the more pronounced the effect.

Grey
 
Hi,
the layout of the heatsink and mounting devices is covered in detail in a Wakefield paper.

The devices are usually spaced out more at the top and closer together at the bottom.

I seem to recall that for three, the bottom device is only 15% up from the bottom edge, the second is about 40% up from the bottom and the third device is about 60% up from the bottom.

Doing something along these lines will nearly keep the devices at similar temperatures.

Similar temps are critical for temperature derating.

The thickness of the main plate should be no less than 10% of the maximum radius from device centre to the extreme corner of the sink.

The worst location for this thickness to radius ratio is the top device to top corner.
 
MattM said:
Thanks for the replies!

Yes, I am using these.

I plan to mount 3 devices per heatsink (two heatsinks per monoblock). The heatsinks will be mounted with the fins running vertically. In that orientation, they are 12" tall, 5.8" front to back, and the fins are 3" wide. Each heatsink weighs 5-1/2 pounds and has 16 fins. They are Wakefield Engineering part number MD387A, but I can't find them on their web site so they may have been a custom item. They are closest to extrusion profile 8415 - here is a link with some specs.




They will have air space under them, so I am expecting natural convection to create quite a healthy air flow across the 12" height.

Maybe the answer is simply to mount the devices at the same level. I had originally planned to spread them out across the heatsink (as depicted in the "wrong" half of the diagram below). While mounting them close together may or may not make them run a little hotter, at least they will all be at the same temp.


You got a great deal with those babies.

The heat sink will operate most effectively with the devices spread out. Aavid Thermalloy tells you why on their website -- it's that the devices in close proximity act like a point source so the heat takes a bit of time to migrate across the sink.

You can try this experiment to determine the thermal impedance of a heat sink -- really easy -- get a chasis mounting power resistor of (perhaps) 50 Watt rating and mount it on the sink -- use thermal grease -- measure the ambient temperature -- then cook at 25 Watts for one hour and measure the new temperature -- take the delta, divide by the power and you're good to go.

If you have 3 such resistors you can experiment with their position -- and the thermal impedance of the device. Prove it for yourself.

Try mounting the fins horizontally, then compare the result when you mount the fins vertically...

Natural convection is only some tens of feet per minute.
 
AndrewT said:
The thickness of the main plate should be no less than 10% of the maximum radius from device centre to the extreme corner of the sink.

The worst location for this thickness to radius ratio is the top device to top corner.
remember that guidance?
I see the sink you plan to use is VERY thin. You will run into uneven temperatures across the backlate and that will reduce the effectiveness of the sink to well below the manufacturers specification.

If you were to bolt a thick copper plate between each device and the sink to bring the heat out to nearer the edges then you will get better performance. Still sticking to that 10 to 1 rule a 60mm square plate needs to be at least 3mm thick. Conversely if you can lay your hands on 5mm copper then you could bolt a long strip 100mm wide from near bottom to near top. BTW. chip manufacturers do this to spread the heat out along the mounting plate to improve Rth j-c.

Mounting the plate with the fins horizontal will cook your devices. It really is that bad! Maybe worse for your deep, thin fins.
 
Formerly "jh6you". R.I.P.
Joined 2006
MattM said:
the base plate is only 4mm thick


If heat sinks have the same surface area, I prefer the one having thinner base plate because the thinner base plate make it possible to reach faster thermal balance (or steady-state) with the air temp surrounding the sinks. This means that we can more quickly arrive the final amp's warm operational condition from the cold start. Just my preference . . .
 
post script-- there is nothing linear about this discussion.

rule #301 about heat sinks -- a device doesn't necessarily get rid of its heat by the path that you intended -- a thick slab of metal stores energy -- nice if you like heating hysterisis in your system as it will heat up the junction when the device is dissipating less power as heat.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.