Silly surface mount dummy load module

[mcdn]

There's actually a separate use of heat sinking for stepped power measurements, simply to add thermal mass to keep short term temps under control. So that could be an interesting tradeoff where some 3mm aluminium strip is pasted on to help with short term power testing, but it actually might not help that much or even be a problem for long term power handling. I'll test it out.

 

[Old_School_Brad]

With stationary gaming PCs and PA amplifiers in mind, I’d suggest that a controlled airflow with one or two 80 mm fans should be adequate.

 

[mcdn]

With stationary gaming PCs and PA amplifiers in mind, I’d suggest that a controlled airflow with one or two 80 mm fans should be adequate.

Agreed, and the math bears it out. The bigger issue is getting the heat from the boards to the air.

 

[Old_School_Brad]

Agreed, and the math bears it out. The bigger issue is getting the heat from the boards to the air.

I'd construct a slight constricting funnel to get higher air speed towards the end where the air is hotter and need higher flow above the PCB to remove the same heat effect. Fan of course placed in the large opening.

 

[mcdn]

Here's the board as delivered. Bigger than I imagined!

They measure as close to 2ohms as my multimeters can tell. Putting 3A DC through them raises the temperature by about 40C above ambient. So that's about 18W continuous. That bodes well for cooling them with a fan to get to ~10x that, even in this early hack version.

 

[mcdn]

I'd construct a slight constricting funnel to get higher air speed towards the end where the air is hotter and need higher flow above the PCB to remove the same heat effect. Fan of course placed in the large opening.

Yes! I think an 80mm or 100mm fan at each end would do the job very well with the boards stacked. Based on the initial test I think a basic fan setup would be fine for 50W continuous per board, and for 200W per board for stepped power testing. Four boards therefore gets you up to 800W/2ohm burst without special measures beyond a fan.

 

[mcdn]

With stationary gaming PCs and PA amplifiers in mind, I’d suggest that a controlled airflow with one or two 80 mm fans should be adequate.

I agree. Most of the papers I found on this topic are concerned with reducing device size while maintaining good thermal management, e.g. moving from 0805 to 0603. In the PCIM paper referenced above they go so far as to delete the radiant heat transfer from the top surface of the component from their model because they assume no active air flow. Given much bigger devices with correspondingly bigger surface area that factor needs to be put back in, and then airflow can come into play. Based on my early tests at 18W in free air I'd say that 200W continuous per board might be possible with PC case fans.

You'd want to point the hot end away from you though.

 

[boXem]

Once you decide to push the thermal boundaries, pay attention to keep the board horizontal, with components on top. This will avoid the components to discover the joys of travel when the solder melting point will be reached.
Don't ask me how I know this...

 

[mcdn]

Once you decide to push the thermal boundaries, pay attention to keep the board horizontal, with components on top. This will avoid the components to discover the joys of travel when the solder melting point will be reached.
Don't ask me how I know this...

Inverted accidental reflow is never funny!

 

[wwenze]

You know how much heatsinking 350W needs?

Something like this

Even 100W is not easy, but at least a cheap single-tower single-fan heatsink would do it.

Computer CPUs cooling solutions give a good real-world perspective. They are also cheap for the amount of cooling you get per dollar, with many cheap towers cooling 100W ~ 150W at 90 degrees celsius.

Oh remember to cover the resistors with some electrically insulative material before slapping the heatsink onto it

 

[Old_School_Brad]

You know how much heatsinking 350W needs?

Something like this

Even 100W is not easy, but at least a cheap single-tower single-fan heatsink would do it.

Computer CPUs cooling solutions give a good real-world perspective. They are also cheap for the amount of cooling you get per dollar, with many cheap towers cooling 100W ~ 150W at 90 degrees.

The cooling systems for CPUs and GPUs are remarkable. I’ve always been amazed by the power density of GPU silicon, dissipating between 300 to 500 watts from an area smaller than a postage stamp. It's astounding that all this energy is used to perform calculations for pretty shapes and colors.

 

[egellings]

Instead of mounting a heat sink, I would just increase the spacing and make the board larger. But you seem to have a good idea of the air flow needed, and these are not semiconductors, so until you reach the solder reflow temp you should be ok.

You're basically making the board into a bespoke heatsink. High temp operation of electronic components shortens their lifespan and there is a possibility of fire if anything goes wrong. Operating a component at just below solder-melting temperature is just asking for trouble.

 

[mcdn]

You're basically making the board into a bespoke heatsink. High temp operation of electronic components shortens their lifespan and there is a possibility of fire if anything goes wrong. Operating a component at just below solder-melting temperature is just asking for trouble.

That's rather melodramatic! Used mounted sideways they have a built in fail-safe, as above 200C the resistors will all fall off

 

[wwenze]

The cooling systems for CPUs and GPUs are remarkable. I’ve always been amazed by the power density of GPU silicon, dissipating between 300 to 500 watts from an area smaller than a postage stamp. It's astounding that all this energy is used to perform calculations for pretty shapes and colors.

No different from shorting stuff together. A single transistor e.g. MJ15022 in a TO-3 package is rated for 250W.

The difference is that we are actually shorting things in a specific formula to make colors do stuff. (Everything digital just a lot of pull-up / shorting to VCC and pull-down / shorting to ground) Ideally we don't even want any power dissipation - The main reason for switching to MOSFET and smaller processes, but gate capacitance and RDS losses are real physical properties. We just kind of like kept increasing the hertz until right before failure. A single CPU core nowadays don't even dissipate that much power, but things are different when you put 192 cores into one package.

 

[egellings]

That's rather melodramatic! Used mounted sideways they have a built in fail-safe, as above 200C the resistors will all fall off

I 'spoze.

 

[Gruesome]

300W might be a bit optimistic. Your board seems to be about 6" by 4", so about 150 cm^2 or 0.015 m^2 per side. You are seeing 40°C for 18W, so a heat transfer coefficient of 18/(40*0.03)= 15 W/m^2K, which makes sense, maybe a bit on the high (=lucky) side without forced air flow. I think it would be hard to get ten times that with air. But why can't you run the resistors at say 120K above ambient? They are not semiconductors, and you are taking care of the derating by actually computing that the power can be removed.

 

[mcdn]

Rigged up 4 of the boards as a 2ohm setup in prep for testing the Fosi M03 sub amp. With a pedestal fan pointing at them from a few feet away temps are stable at ~58C with a 160W load. The Fosi won't do more than that with the stock supply but I have a 350W 48V Mean-Well one lying around which will stress things a bit further. With a more focused fan setup I'm still pretty confident of getting 100W/board continuous. They don't heat up too quickly either, so burst testing up to 200W/board is definitely doable, and maybe more.

 

[mcdn]

300W might be a bit optimistic. Your board seems to be about 6" by 4", so about 150 cm^2 or 0.015 m^2 per side. You are seeing 40°C for 18W, so a heat transfer coefficient of 18/(40*0.03)= 15 W/m^2K, which makes sense, maybe a bit on the high (=lucky) side without forced air flow. I think it would be hard to get ten times that with air. But why can't you run the resistors at say 120K above ambient? They are not semiconductors, and you are taking care of the derating by actually computing that the power can be removed.

Great points. Take a look at the papers linked in posts above - 125C is taken as a standard max design temp for SMD components on FR4, but the resistors begin to derate around 70C.

 

[mcdn]

With the fan a bit closer and the mean-well supply I'm getting 4C above ambient at 20W, and 40C above ambient at 200W. And the Fosi amp itself has gone into thermal shutdown for reasons that are obvious if you've seen a picture of the internals but I'll explain again in my review. Still confident of 400W @ 50C above ambient for the silly dummy load, but only with a closer/ducted fan setup.

 

[Old_School_Brad]

With the fan a bit closer and the mean-well supply I'm getting 4C above ambient at 20W, and 40C above ambient at 200W. And the Fosi amp itself has gone into thermal shutdown for reasons that are obvious if you've seen a picture of the internals but I'll explain again in my review. Still confident of 400W @ 50C above ambient for the silly dummy load, but only with a closer/ducted fan setup.

It's only a matter of air flow. -And that is easy to scale to need.

Will you be sharing any test results once you’re done experimenting with it?