Yesterday was an important day in Xymo’s development. It started out with a bullet train ride inland to a city called Dongguan, and when I disembarked I was greeted by three members from a prominent heatsink manufacturing company. My hosts were very accommodating. They filled me with Chinese cuisine before shuttling me to their factory for a tour and a chat about business.

This meeting didn’t happen spontaneously, it was a culmination of numerous emails, Skype chats and iterations of 3D CAD designs to reach a design that would meet Xymo’s technical requirements while also being manufacturable at a reasonable cost.

Here’s a picture of our current heatsink design.

Xymo Heatsink Closeup

A closeup of Xymo’s heatsink fins

Xymo's heatsink

Xymo’s heatsink

Overall I’m very happy with our heatsink design, I’m sure it will perform adequately and we ended the day with plans to have two samples manufactured for us in the coming weeks.

Why all the fuss about heatsinks?

So first, a little background. Xymo’s primary goal is to make brewing great tasting beer easy and repeatable by tackling the biggest variable in home brewing – fermentation. At his heart, Xymo is a computerised fermentation controller and his primary function is to maintain wort at an ideal temperature throughout the different stages of fermentation, all thanks to the heating and cooling abilities of thermoelectric coolers (often referred to as the Peltier effect).

Xymo’s heatsink is one of its most important components because in order to cool beer, we need to dispose of a lot of heat (up to 300 watts of heat!) and the better we can dissipate that heat, the more cooling capacity Xymo will have, and the more efficiently Xymo will operate.

The heatsink we’ve come up with has a number of cool features and this was largely thanks to the design input from the experts at our heatsink manufacturing company.

The Fins

One of the main awesome things about our heatsink is that it is being built with a manufacturing process called “Skived Fins” which is very different to your typical heatsink.

Most heatsinks are produced with an “extrusion” process where metal (Aluminium or Copper) is heated and forced through a die that is cut the the shape of the heatsink’s final profile. While extrusion is the cheapest method for mass production, the disadvantage with extrusion is that you’re limited in the height and number of fins that can be stacked in a given space, meaning you need a relatively large heatsink for any given amount of power to dissipate.

Taking things a step up, a common solution to obtain higher fin densities and performance is to use a “bonded fins” manufacturing process where small channels are cut into a base block, and vertical fins are glued into these channels. While this process can yield a better heatsink with a much larger surface area for cooling, the trouble is that the glue isn’t great at conducting heat, so the fins are slightly thermally isolated from the base of the heatsink reducing performance.

For Xymo, we’re using a process called “skived fins” where the fins are actually sliced from the base of the heatsink and bent up at 90 degrees to it. This process requires some very interesting machines but it will give us the best performance possible for a moderately sized heatsink.

Heat Pipes

Heat pipes are truly amazing devices for distributing heat. They contain a liquid that’s under vacuum such that it is in a constant state of being partially evaporated, along with a wicking coating on the inner walls that can distribute the moisture inside through capillary action. We’re having prototypes made both with and without Heat Pipes so that we can evaluate their benefits for Xymo. On one hand, they will distribute the heat away from the thermoelectric coolers to the extremities of our heatsink, but testing will show if this is necessary.

Parallel TECs

This is a cool one (no pun intended). Thermoelectric coolers can be pretty complicated to model. When you pick up a TEC that says Qc 100watts, that doesn’t mean it will cool 100 watts, nor heat 100 watts. On its own it doesn’t mean much at all for a real life scenario. My first prototype for Xymo was a single-Peltier affair, and I found that while I could get a pretty sweet delta T of over 10 degrees celsius below ambient, it took about 24 hours to get there so I knew that more than one TEC would be required. Rather than building more prototypes, I then began running some thermal modelling with 2 or more thermoelectric coolers. From the software simulations I found that running three devices at a slightly lower voltage I found my coefficient of performance to be higher meaning less overall heat being produced for the same cooling effect. This has relaxed our heatsink requirements and lowers Xymo’s power requirements a little.

The curved base

This post would be incomplete without talking about the curved base that’s in the pictures as well. This concave block of machined aluminium will contact the side of Xymo’s stainless barrel to transfer heat. We are working with some stainless steel manufacturers to have our fermenters manufactured with nuts welded to them (more on that in another post) and it is very important that the two shapes mate up well. You can see one of the grey nuts at the top of the picture, this will actually be welded to the fermenter rather than being a part of the heatsink.

Xymo's Spacer

Xymo’s Aluminium spacer.

The size of this spacer is no accident either. Xymo will have the technical ability to heat at almost 300 watts, however this is far too much for single-batch home brewing needs and will be limited in software to around 30 watts like a regular heater belt. To size the contact patch with the fermenting barrel I calculated the contact patch of a regular heating belt and made sure that Xymo’s contact batch was significantly larger. Given the larger contact patch and the ability to apply only the amount of heating power that is required I believe Xymo will be able to heat a brew much more gently (and of course accurately) than a traditional heater belt.

Another interesting fact you might realise is that the curved heat spreader block is physically smaller than the heatsink. It is in fact only 80mm wide and 200mm high whereas the heatsink is 94mm wide and 240mm high. This is actually due to the Coefficient of Performance of the Thermoelectric coolers being far less than 1.0 during normal operation. E.g. When we are cooling, if we are drawing 20 watts out of the fermentor, we may we using 80 watts of power do do this, meaning although the cold side has to move 20 watts, the hot side has to move 100 watts total. This is why the Hot side is larger than the cold side.

Well that’s all for today. If you have any questions about the progress of this design, please feel free to post.


Michael Burton.



Share this with your friends: