I have become a Heel Dragger.
When natural convection was all the rage, I was touting fans. Once fan cooling became the status quo, I started dropping hints about heat pipes and thermoelectric coolers. The technology roadmap showed nothing but higher power ahead, and I would soon need those fancier cooling tools. Now everybody senses that this air cooling act is growing stale, and liquid cooling is in the wings, waiting to strut out onto the electronics cooling main stage. The buzz at the trade shows is that liquid is needed, liquid is ready, liquid is going to save the day. Only one thing is holding it back – the product developers are dragging their heels.
Until recently I was a liquid cooling proponent. “Soon,” I’d preach, “we won’t have any choice. Component power is going up, temperature limits are dropping! Fear not, but rejoice in all the wonderful things liquid will enable us to do with our new products!”
Then I bought a house with an “automatic” lawn sprinkler system.
Admittedly I have zero experience engineering liquid cooling for electronics. Perhaps that is why I was so gung-ho for it. But three months with sprinklers have worn the “ho” right out of my “gung”. Perhaps there are a few electronics cooling lessons to be learned from the sprinkler nightmare.
1. Some training will be required.
On the first day of summer I searched out the sprinkler control box in the garage and turned the knob to “RUN.” Nothing happened. Several days later, with still no spritzing, I consulted my neighbor. He stifled a snicker, then spent the next 90 minutes showing me all the hidden underground boxes, filled with valves that needed opening, then the software that tells which bank of sprinklers to come on when, and for how long. Not to mention the test program. And how to skip a day if it rains. Liquid cooling will force mechanical engineers (don’t think for a minute the electrical engineers will take responsibility for this!) to learn whole new disciplines – plumbing and industrial controls. Liquid is NOT just air that is a lot denser. The end user of the product will also need lots of additional training and/or customer support to learn how to operate and maintain the complex liquid cooling system. Even if you make it “automatic”.
2. Liquid carries solids.
The water supply was on and the controller programmed. At the appointed hour the sprinklers sputtered to life. With frustration I discovered that the sprinkler heads were not delivering the uniform spray pattern I had imagined. Some heads merely dribbled, some shot out ragged streams, only a few blossomed like silvery flowers. The nozzles were clogged with dirt. Almost every head had to be taken apart and cleaned. Only after that half-day chore was complete could I sit back and watch the “automatic” system water my grass. Any plumbing system, no matter how clean it starts out, is going to get contamination. Eventually the pipes themselves corrode or break down, and those pieces floating around in the system end up stuck in the worst possible places – such as the micro-channels of your micro-channel heat sink. No flow, no cooling. Filters help – but filters need to be changed periodically, or they clog up, too. That is not something an electronics customer is used to dealing with. They think filters are capacitors, after all.
3. Balancing the flow is a tricky, never-ending effort.
Even after cleaning out the dirt, the spray pattern was not uniform from one head to the next. The water pressure in the line goes down after each sprinkler head as the flow gets farther away from the main line, as shown in Figure 1. The first sprinkler head has high pressure and high flow, then the flow of each successive sprinkler head is less.
Figure 1. If all the sprinklers on a single line have the same flow resistance, you end up with hardly any flow at the end of the line, and a big yellow patch in the lawn.
That’s why sprinkler heads have flow adjustment screws. You put on your swim trunks and spend an afternoon turning those little screws while the water sprays in your face until you get the desired flow pattern, (which may mean your desire for an exactly uniform flow pattern eventually goes away). It is an iterative process, because a change in the resistance of one sprinkler head changes the flow in all of them on the line. A liquid cooling system is bound to have similar problems. It will be difficult to make sure that each branch in the network of pipes has the desired flow rate. At least I can see the amount of water shooting onto my grass. How easy will it be to know the flow rate in each pipe in a manifold buried inside an electronic assembly? Will the customer have to tweak little adjustment screws every few months to balance the temperatures in his electronics?
4. Leaks.
You might think leaking pipes happen only when the system is first built, and that once they were spotted and fixed, you could forget about them. I discovered three major leaks in one month alone, in a system that had supposedly had the bugs worked out of it for over 10 years. At least sprinkler leaks are easy to find. The gushing water eventually erodes the soil around the break and conveniently digs the hole in the lawn for you so you can repair it. Unless it is digging a big hole under the driveway, in which case you find it when the concrete collapses. Or when the underground leak is next to a hairline crack in the foundation of the house, and your daughter discovers it when she opens her toy chest in the basement playroom, and finds mushrooms growing on her Tickle-Me Elmo. Nothing like a chassis full of corrosion and mildew to put the “ick” back into electronics.
Naturally, you won’t build your liquid cooling system out of thin-wall vinyl pipe stuck together with model airplane glue. You’ll use stainless steel tubing, sophisticated quick-disconnect fittings, precision-machined manifolds, and lots of check valves and pressure sensors and redundancy and ultra-pure, non-corrosive, non-freezing fluid filled with nano-particles. Of course, liquid cooling has been done successfully for many years in mainframe computers and military aircraft (but each fighter plane has a ground crew of 32 skilled technicians to keep it running.) Deep in my engineer’s heart I still know that sooner or later I won’t have a choice.
But for now I remain a Heel Dragger, or at least a secret one. I do have one use for liquid cooling, which requires me to keep my dragginess secret. Whenever the circuit designers get carried away stuffing too many high power components into a box, I threaten to hit them with a metal pipe. “Looks like we’ll have to go with liquid cooling on this baby!” I tell them, rubbing my hands in mock anticipation. That usually scares them enough that they find a way to cut the power by about 75%, just enough to allow air cooling to return to the stage for at least one encore.
Tony Kordyban is the author of the ASME Press books, Hot Air Rises and Heat Sinks: Everything You Know About Cooling Electronics Is Wrong and, its long-awaited sequel, More Hot Air. They tell stories similar to this one, including the thermal challenges in developing a telepathy-based telecom system. All of them are educational. Some are even true.