3D mechanical integration and optimal thermal dissipation placement questions

[AlfBaz]

I just wanted to see what others would do in the following situation

I have created a 3D model of a mechanical assembly involving heat sinks, fan and a bunch of TO-247's in an external CAD package

My question is would you leave the TO-247 models attached to the mechanical assembly and then move the modeless footprints to line up with the assembly or have the TO-247 model as part of the footprint and attempt to align it with the heatsink assembly within altium

The second question, referencing the attached picture, would it be more thermally efficient to place the TO-247's further up (in the middle) than where they currently sit, ignoring assembly difficulties?






 

[DerekG]

would it be more thermally efficient to place the TO-247's further up (in the middle) than where they currently sit, ignoring assembly difficulties?

No, keep the leads of the TO-247 as short as possible.

[T3sl4co1l]

Well, short leads depends on assembly constraints (nevermind the specific request to ignore them ) and frequency range... if you're trying to get that much power out of 'em, your switcher must suck; I might just as well guess it's a linear application, which might benefit from using the heatsinks better.  But yeah, we'd have to know to be sure.

The heatsinks don't seem all that thick on the face, suggesting they'll be somewhat cool at the top, which would be suboptimal.  A thicker face would better carry heat to all fins.  Alternately, a shorter heatsink has fewer fins to carry heat to, but doesn't dissipate as much per length -- necessitating a higher air velocity from a smaller fan (and then you get air pressure and perhaps noise concerns), or more heatsinks (which need more devices* to deliver the heat in the first place).

*Which isn't always a bad thing; TO-220s handle about half the power of TO-247s, but they are cheaper too.  But if you factor in assembly time (if it matters), it's probably not worth it.

Anyway, regarding the placement, yeah I'd probably do that: place the model, locate snap points, and move the footprints to match.

Tim

[AlfBaz]

Well, short leads depends on assembly constraints (nevermind the specific request to ignore them ) and frequency range... if you're trying to get that much power out of 'em, your switcher must suck; I might just as well guess it's a linear application, which might benefit from using the heatsinks better.  But yeah, we'd have to know to be sure.

Aah I see now. Was a bit confused about "keeping the leads short" comment. My fault for not stipulating a linear application.
I've been meaning to build this high power eload for some time out of a tube of IRFP450's and the heatsinks pictured above that I've had laying around for ages

The heatsinks don't seem all that thick on the face, suggesting they'll be somewhat cool at the top, which would be suboptimal.  A thicker face would better carry heat to all fins.  Alternately, a shorter heatsink has fewer fins to carry heat to, but doesn't dissipate as much per length -- necessitating a higher air velocity from a smaller fan (and then you get air pressure and perhaps noise concerns), or more heatsinks (which need more devices* to deliver the heat in the first place).

*Which isn't always a bad thing; TO-220s handle about half the power of TO-247s, but they are cheaper too.  But if you factor in assembly time (if it matters), it's probably not worth it.

So the optimal positioning would be in the centre, given what I have. I guess the remaining question is the gain worth the effort. I might literally wire up the thing and take some measurements with the TO-247's in both positions and see what the difference is

Anyway, regarding the placement, yeah I'd probably do that: place the model, locate snap points, and move the footprints to match.

I went down that road, it did however, require some selective DRC altering to avoid collision errors regardless of the fact that I set the footprints height to 0

[hagster]

Possible things to consider. I'm not thermal engineer so don't take this as gospel.

The fan is very close to the first TO-247. This might result in pockets of stagnant air near the first package?

The air will get hotter as it passes each package. The last TO-247 will probably get hotter than the first.

The airflow might become laminar as it travels down the tube. This will result in hot air clinging to the fins while the cooler air travels down the central cavity without doing too much work. You could look at methods to cause a bit of turbulence to mix the air a bit.

[free_electron]

the to247 package should be placed in the to247 footprint.
place the step model for the mechanical assembly, then place the transistors. line em up in 3d view. you can use the body placement tool to create snap points on the step models.

once that is done export the step data back to your mech cad

[AlfBaz]

Possible things to consider. I'm not thermal engineer so don't take this as gospel.

The fan is very close to the first TO-247. This might result in pockets of stagnant air near the first package?

The air will get hotter as it passes each package. The last TO-247 will probably get hotter than the first.

The airflow might become laminar as it travels down the tube. This will result in hot air clinging to the fins while the cooler air travels down the central cavity without doing too much work. You could look at methods to cause a bit of turbulence to mix the air a bit.

Interesting observations. I'll be sure to look into them when I wire up a test jig... Starting to think a thermal imaging camera may come in handy here

[AlfBaz]

the to247 package should be placed in the to247 footprint.
place the step model for the mechanical assembly, then place the transistors. line em up in 3d view. you can use the body placement tool to create snap points on the step models.

once that is done export the step data back to your mech cad

I tend to like this idea the most, 3D model with the footprint. Aside from placing 3D models on their associated foot prints in my libs I haven't done any real mechanical manipulations in altium. It's probably high time I learnt