Prototype KMPKT Dynamo DC-12V

[LjSpike]

Oh sweet, just came across this.

One observation, an equilateral triangle pattern would provide better cooling (it'd still look pretty badass too, maybe not as badass, but still pretty badass). It'd retain the omnidirectional airflow, but would be providing more channels. Alternatively, you could make tubes going through the hexagons that aren't visible from the top, allowing slightly improved airflow without detracting from the badassery.

 

[Kmpkt]

Kinda committed to hexes at this point, but I'll keep the triangles in mind for the future as better cooling is always welcome. The issue with the tubes through the hexagons, comes down to manufacturing costs to be honest (although it is a really good idea). For the first run I am having these CNC'd on a per unit basis which is already quite expensive. The only way I can see to make airflow tubes that run through the hexes would be the 3D print aluminum, which I've done and it is prohibitively expensive.

 

[LjSpike]

Kinda committed to hexes at this point, but I'll keep the triangles in mind for the future as better cooling is always welcome. The issue with the tubes through the hexagons, comes down to manufacturing costs to be honest (although it is a really good idea). For the first run I am having these CNC'd on a per unit basis which is already quite expensive. The only way I can see to make airflow tubes that run through the hexes would be the 3D print aluminum, which I've done and it is prohibitively expensive.

True, perhaps doing it in two layers, the first being the base layer, consisting of the solid level of aluminium then hexagons with the grooves, then via some joining technique attaching solid hexagons on top. I'm not sure which joining method would be best though, perhaps a chemical agents that could act a bit like acrylic cement, but for aluminium, if one such chemical exists.

 

[jtd871]

CNCing a triangular pattern could be cheaper (opens up other ways to manufacture it, as well) as the tool paths are now all straight lines across the entire workpiece.

I would guess that hex or triangle makes little difference in the cooling.

The biggest 2 factors for sink design are likely mass and radiative surface area per unit mass. I would also doubt that the current hex pattern is the most thermally efficient for the overall sink volume, even if it does look badass. Deep fins with cross-slots are probably close to optimal for an unknown orientation of airflow, otherwise everybody would be using another pattern.

 

[Kmpkt]

The hexes are as much aesthetic as anything else. I had considered a triangle pattern, and in the end felt that the hexes just presented a better aesthetic. If it's not well received, I'd obviously consider changing it. I could have gone with a more simple extruded deep fin array, but I really wanted to keep the unit around 25mm thick or less. The biggest thing for me is that there is more surface area to cool through than existing heatsinks which I believe I've managed to do. Again if we run into issues with dissipation then I would obviously change the setup for the desired functionality.

 

[LjSpike]

What tools for manufacturing the heatsink do you have access to Kmpkt...
It's be really nice if you could keep the hex pattern, If I know what tools you have access to I might be able to work out some ways to try to increase the surface area.

Jtd871 triangles have more surface area then volume as are further from a sphere then a hexagon. The less spherical an object, the greater its surface area to volume, an equilateral triangle is therefore has better thermal performance then a regular hexagon.

One Idea for a variation on the pattern would be this:

Hexagons with fins running down the middle of each groove. The fins would provide some additional heat spreading, while keeping it still looking a lot like a hexagon.
EDIT: Another alternative design, although this one begins to be a bit less obvious that its a hexagonal pattern, is this:

A third way I quick came up with as well would be to use smaller hexagons in nearly-hexagonal clusters with each cluster being separated by a larger groove then.

Just some ideas for you.

 

[Kmpkt]

That's a pretty cool idea with the fins. I'm also now playing with the idea of having the hexagons made of very small fins as well. To give you an idea of what my constraints are, I can CNC or stamp my heatsinks and will be having these all done through HDPlex's manufacturer in China. For my initial run of 200 units, CNC makes sense as the cost of making the die to stamp is quite steep. Depending on the reception of this product, I may try something different in the future.

 

[LjSpike]

That's a pretty cool idea with the fins. I'm also now playing with the idea of having the hexagons made of very small fins as well. To give you an idea of what my constraints are, I can CNC or stamp my heatsinks and will be having these all done through HDPlex's manufacturer in China. For my initial run of 200 units, CNC makes sense as the cost of making the die to stamp is quite steep. Depending on the reception of this product, I may try something different in the future.

Interesting.
I don't think I have any more ideas left, hexagons made from fins sounds like quite a nice idea, perhaps it might be nice if the fins protruded from the original solid hexagonal pattern, to retain a good bit of the definition of the pattern?

I hope your successful with these, I'd love to see a 500W model which can take an AC input so it can act as a total power supply, because the current market of PSU's are all a bit bland and poor performing.

 

[Kmpkt]

Well the Dynamo and Dynamo mini combo will be rated for 460W and capable of significantly more than that.

 

[ricochet]

Well the Dynamo and Dynamo mini combo will be rated for 460W and capable of significantly more than that.

Outstanding progress @Kmpkt ! Very impressed and really liking your overall design and styling. How big is the 300W AC-DC adapter or is that still hush-hush?

What steps have you and Larry made in regards to minimizing and/or eradicating coil whine that some HDPlex units have... and please don't get me wrong as I truly love my 160W DC-ATX + 80W AC-DC Combo but the coil whine could be bothersome to some especially with SFF builds sitting on a desk right by the user.

Well the Dynamo and Dynamo mini combo will be rated for 460W and capable of significantly more than that.

Wow!

 

[zovc]

These are DC-DC units. I made the same mistake.

 

[Kmpkt]

How big is the 300W AC-DC adapter or is that still hush-hush?

The most recent information I have, which I believe is the release information has the size of the unit pegged at 136mm x 62mm x 41mm. With the L-bracket at either end, the length increases to 151.4mm. I am not certain if these are removable, but hopefully they are.

What steps have you and Larry made in regards to minimizing and/or eradicating coil whine that some HDPlex units have

To be honest I am not exactly certain of the exact changes that have been made, but I do know that this is an issue of which he is acutely aware and wants to rectify. He will be releasing the newest revision of his HDPlex 300W DC-ATX soon and I know the changes he's made to that will also apply to the Dynamo. When I last mentioned it to him, he said his design goal for the Dynamo was to achieve no coil while running the 160W DC-ATX off the voltage flow through and powering a GTX 1080ti.

One other fun fact is with better heat sinking the rated wattage of the HDPlex 300W AC-DC will theoretically go up. The case I am designing will largely consist of 5-7mm thick aluminum plate for exactly this purpose : )

Also if you don't mind purchasing a 160W AC-DC as well as a 300W AC-DC to go with it, the Dynamo combo will run 450W comfortably with the capacity to handle significant power spikes on the GPU end of things.

 

[ricochet]

The most recent information I have, which I believe is the release information has the size of the unit pegged at 136mm x 62mm x 41mm. With the L-bracket at either end, the length increases to 151.4mm. I am not certain if these are removable, but hopefully they are.

Amazing that the 300W AC-DC Adapter is not much larger than the 160W unit:

300W AC-DC: 151.4 mm x 62 mm x 41 mm
160W AC-DC: 150.0 mm x 52 mm x 40 mm
80W AC-DC: 121.4 mm x 30 mm x 40 mm

 

[ricochet]

These are DC-DC units. I made the same mistake.

???

 

[LjSpike]

The most recent information I have, which I believe is the release information has the size of the unit pegged at 136mm x 62mm x 41mm. With the L-bracket at either end, the length increases to 151.4mm. I am not certain if these are removable, but hopefully they are.



To be honest I am not exactly certain of the exact changes that have been made, but I do know that this is an issue of which he is acutely aware and wants to rectify. He will be releasing the newest revision of his HDPlex 300W DC-ATX soon and I know the changes he's made to that will also apply to the Dynamo. When I last mentioned it to him, he said his design goal for the Dynamo was to achieve no coil while running the 160W DC-ATX off the voltage flow through and powering a GTX 1080ti.

One other fun fact is with better heat sinking the rated wattage of the HDPlex 300W AC-DC will theoretically go up. The case I am designing will largely consist of 5-7mm thick aluminum plate for exactly this purpose : )

Also if you don't mind purchasing a 160W AC-DC as well as a 300W AC-DC to go with it, the Dynamo combo will run 450W comfortably with the capacity to handle significant power spikes on the GPU end of things.

So if I were to get this and the AC-DC unit it'd function essentially as a complete 450W PSU? Would be nice to see a slight increase in wattage to perhaps 500W when combined, roughly how expensive are the two units going to be?

 

[Kmpkt]

You'd need to get the 300W AC-DC and hook it up to the Dynamo and then get the 160W AC-DC and hook it up to the Dynamo Mini. There is a sync cable from one unit to the other so that when you power on the motherboard, everything powers up together like a traditional ATX PSU.

 

[LjSpike]

You'd need to get the 300W AC-DC and hook it up to the Dynamo and then get the 160W AC-DC and hook it up to the Dynamo Mini. There is a sync cable from one unit to the other so that when you power on the motherboard, everything powers up together like a traditional ATX PSU.

So 4 units you have to get in total. That's quite a few :c
It'd be nice to see perhaps a later version cut down to just needing 1 of the AC-DC's and either a Dynamo or Dynamo mini to help simplify the setup for builders...

 

[Kmpkt]

Yeah I totally get that, and the issue is there really isn't much readily available above 330-350W in the market. Most of this I believe comes down to how much heat can be dissipated before active cooling is required. As far as the Dynamo and Dynamo Mini, the mini is basically a custom and fully modular version of the 160W DC-ATX and will be able to be purchased separately if you'd like. The Dynamo will function as an independent unit if you want to use a device with a non ATX 24 pin power input (Thin ITX, Mini STX, Micro STX, Intel NuC, etc.) The only situation in which you'd need both DC-DC units is for use with an ATX solution with a draw > 175W.

 

[LjSpike]

Yeah I totally get that, and the issue is there really isn't much readily available above 330-350W in the market. Most of this I believe comes down to how much heat can be dissipated before active cooling is required. As far as the Dynamo and Dynamo Mini, the mini is basically a custom and fully modular version of the 160W DC-ATX and will be able to be purchased separately if you'd like. The Dynamo will function as an independent unit if you want to use a device with a non ATX 24 pin power input (Thin ITX, Mini STX, Micro STX, Intel NuC, etc.) The only situation in which you'd need both DC-DC units is for use with an ATX solution with a draw > 175W.

I wonder if a later version of the dynamo might be able to employ active cooling, but just at low RPM's to make it not silent but just really quiet instead? I have had one thought, as a possible solution to dissipating heat better passively for PSU's. Rectangles that are long and thin have a higher surface area to volume ratio then just a square, and increasing that ratio is what passive cooling always needs. Perhaps if you had a slightly more spread out, long, but really flat PSU along with some form of heatsink applied across lots of it, if that could perform better for passive cooling?

The only other methods I could imagine would be to make all the cables and such a bit chunkier to decrease resistance...

EDIT/EXTENSION: Actually, heat pipes might be an option? to try and carry heat from the hottest components... Some sort of interesting implementation of evaporation cooling with heat pipes and a small chamber might also be an interesting possibility although I'm not sure how effective it'd prove.

 

[Kmpkt]

I wonder if a later version of the dynamo might be able to employ active cooling

It's not the Dynamo that would need active cooling, it's the AC-DC portion of the equation. I've talked to Larry about going bigger with the AC-DC, but 300W is the largest supply he feels he can make in a passive enclosure. The holy grail for this sort of application would be to integrate GaN into the equation which would increase efficiency from 90 to 94%, substantially reducing heat produced. Unfortunately this appears to be quite far off on the horizon for the time being. As far as heatsinking the Dynamo, my heatsink is already 2-3 times the size of what is required, so going with a higher dissipation heatsink is moot.