The Wild West of Thermal Engineering

Brent Cullimore

Let me start by confessing that I don’t know much about Bitcoin, much less Bitcoin mining. I’m not even sure you can refer to them as ‘Bitcoins’ if you have two of them. You certainly can’t rub them together and make sparks. (They are sparks, if you think about it!)

The evening news programs make it sound as though Bitcoin is a nefarious tool for drug dealers and organized crime. Then you hear that the inventor would be nominated to win the Nobel Prize in economics … if only we could be sure who that inventor is.

I’m an engineer. When I gamble, I want to know the safety factor. So guess who doesn’t gamble, much less own any Bitcoinage (if that is a word)?

But when an article in Electronics Cooling came out about the same time as the Noble Prize news, I started reading up on that world. I found out how you can “mine” something that was never buried.

Whatever you think of Bitcoin, the upshot is that Bitcoin mining is pushing the frontiers of thermal management, as summarized by Alex Kampl of Allied Controls. Phil Tuma of 3M also posted a useful presentation on YouTube, and if you search you can find other cool videos of PCB boards submerged and boiling away.

Many of these presentations and videos reference the Immersion-2 system for cooling a Bitcoin mining operation in Hong Kong.

Source: https://bitcointalk.org/index.php?topic=346134.0

If watching bare PCB boards boiling doesn’t give you the shivers, you are either (1) not working in the electronics industry or (2) are working in the electronics industry and are farther ahead of the tech curve than I am.

Assuming you aren’t into electronics cooling, you should note that the term “liquid cooling” refers to any introduction of liquid into the system. Not just a sealed liquid coolant loop or an evaporator in a vapor compression (refrigeration) cycle, but also a heat pipe or vapor chamber fin. Reading between the lines: liquid is scary. “Get that liquid away from my electronics! What are you, insane?” You can see why that attitude exists, even when the liquid is a dielectric.

But just like e-money itself, Bitcoin mining is at the frontier. It isn’t just that new rules apply, it is that there is one overriding rule due to its competitive nature: “There is money to be made if this works!”

Of course, once new ground is broken … and presuming that this technology proves itself … you can count on it being adopted elsewhere. That will be true of any new technology that Bitcoin mining uncovers. Expect it to be exploring all types of solutions that everyone else is hesitant to adopt. And as a technologist, that’s very exciting!

So I was inspired to make a thermal/fluid model, despite not having a lot of detailed information due to the proprietary nature of systems like Immersion-2. (As you’ll soon see, “I want to make a model of that!” is my way of expressing enthusiasm and participating, even if remotely.)

Below you can see a screen shot of a tank that occupies 1/3rd of a 19” rack (about 24U), devoted to about 8kW of power generation. I used 3M’s NOVEC® 649 fluid, which has a saturation temperature of 49°C (the tank is exposed to open air). I set back the water-cooled (32.5°C inlet) hermetically-sealed condenser coils. That placement allows the boards to be lifted from the pool without the condenser getting in the way. I added a sloping tray to return the condensate to the pool. There should actually be almost twice the number of boards present, but I deleted half of them (and stepped up the power to compensate) purely for better visibility. As a result, the boards are probably twice as tall as they should be.

The model and write-up are available on CRTech’s user forum.

So the next time you virtually pick up a Bitcoin, turn it over to see where it was e-minted. If it was made in Hong Kong, perhaps you can still smell the dielectric fluid. That’s the smell of progress!