The View Up Here

Random scribblings about kites, photography, machining, and anything else

  • Flickr Gallery

Potting Vacuum Feed-Throughs – Take Two

Posted by Tom Benedict on 01/02/2013

In the middle of all the other recent disasters at work, I had a personal disaster of my own: I made a feed-through that leaked. This may not sound like much, but it brought a project to a screeching halt while we all tried to figure out what went wrong.

Just to back up a little, I’ve been designing and making vacuum feed-throughs since 2002. I’ve lost count of how many I’ve made. Most were electrical. One was fiber optic. One was mechanical. And until this month, not one ever leaked. Not one. So I thought I could reasonably say I knew what I was doing. After this failure? I wasn’t so sure any more.

The good thing about working with a bunch of really sharp people is that as long as you’re willing to set your own ego aside, you have almost limitless opportunities to learn. This was one of those times. Everyone involved brainstormed, scrambled to learn more about the materials and techniques we were using, and joined in to pool their ideas and see what we could come up with. We talked, we tested, and we repeated our tests until we were confident we could move forward. All of us learned something new.

I think I have a reasonable handle on what I did wrong, but a lot of other good changes came out of the process as well. We wound up incorporating quite a few of these into our procedure, so I thought it was worth posting it. So without further ado, here’s the re-write of my earlier article, Potting Vacuum Feed-Throughs:


A vacuum feed-through consists of three parts:

First is the thing you’re feeding through the vessel wall. In most cases this is an electrical signal. In others it’s optical. It’s typical to have a connector on one or both sides of the feed-through, though this isn’t always the case. If a connector is used, the easiest way to do this is by using a hermetic vacuum-rated connector. Most of the time these contain the provision for o-rings, so this may be the extent of the feed through since it supplies the other two components I’m about to go into.

The second part of a feed-through is some sort of mechanical shell that will actually connect to the vacuum vessel. In the case of a hermetic connector, this can be the connector itself. In the case of a non-hermetic connector, or a connector that doesn’t readily mount to the outside of the vacuum vessel, it’s typical to make a custom machined part.

The third part of a feed-through is to have some means of making a gas-tight seal between all the bits. In the case of a hermetic connector, that’s the o-ring. In the case of a custom feed-through shell, typically it’ll involve an o-ring in a groove that will bear on a flat surface on the cryovessel, or vice-versa, and some means of sealing the wires or fiber optics that pass through it.

The easiest way to accomplish that last part is to use some sort of encapsulant like a resin. Resin won’t work with extremely high vacuum situations, which often use glass as the encapsulant, but it’s good for most instrumentation purposes. The resin we use is Stycast 2850FT, which is black, optically opaque, a good electrical insulator, and a decent thermal insulator as well. We use a Stycast 24LV catalyst, mixed at 7% to the 2850FT by weight. 24LV has good curing characteristics, and is a low-volatile catalyst, meaning it won’t continue to ooze organic vapors into your vacuum system once the potting is complete.

Stycast is neat stuff to work with, but in order for it to make a good potting encapsulant there are some procedural details that need to be followed closely:

  • Surface prep is king. Let me repeat that: Surface prep is KING! We degrease everything using an ultrasonic cleaner and a solvent appropriate to the part, and give every part a final alcohol clean just prior to assembly. Dirty parts make crappy feed-throughs.
  • We store our Stycast in a dorm fridge. This gives it a longer shelf life, but it also makes it hard as tar. Stycast needs to be heated to room temp or higher in order for it to mix with the catalyst. We cure our parts at 45C, so we pre-heat our Stycast to 45C as well, prior to mixing.
  • Once the Stycast is heated and combined with 7% 24LV catalyst, it must be mixed for five full minutes in order to ensure a homogeneous solution. The pot life on this stuff is close to an hour, so you have the time. Take it.
  • After mixing the solution must be degassed. This is a vacuum application, after all, and virtual leaks are the bane of a vacuum system. We use a small dessicator hooked up to a two-stage roughing pump. The Stycast 2850FT datasheet says to degas at 1-5 torr for 3-10 minutes. We degas for five minutes. The Stycast will foam like nuts when you pump on it, so you need to keep a steady hand on both the vacuum and the bleed valves. The trick is to foam it up and slump it a couple of times to get the bubbles on the top, then foam it up and hold it there for the remaining time.
  • There are tricks you can do with pouring resins to minimize trapped gas as well, but I won’t go into those. Look for a good manual on casting plastic parts using resin. We almost exclusively inject our feed-throughs using syringes loaded with Stycast, so almost none of the pouring techniques apply to our procedure.
  • When preparing the feed-through for encapsulation, it’s important to protect any sensitive surfaces from accidental spillage. Blue painter’s tape does a great job of covering o-ring grooves, screw holes, or other places where an accidental blob of Stycast could be a real headache to remove later.
  • If the geometry of the feed-through cavity is complex, it’s best to fill the small voids first using a syringe loaded with the warm Stycast. This avoids trapped volumes of air. It also provides an easily controlled flow of resin, which is why we started using a syringe for all our feed-throughs rather than a direct pour.
  • Finally, fill the cavity until you’re satisfied with the level inside. If you use idea in the next step, it’s important not to over-fill. I now leave at least 3mm of head space below the top of the cavity or dam that I’m filling.
  • At this point you need to pop it back in the oven, and heat to 45C until cured. But there’s one more idea we picked up: Careful mixing and a thorough vacuum degas should remove any trapped air in the Stycast. But any errors in pouring can re-introduce air bubbles that can provide leak paths through the cryovessel wall. Vibrating the part while curing can help release trapped bubbles. We built a shaker table that fits into our oven to vibrate our parts while they cure. This was nothing more than a quarter inch aluminum plate, mounted on spring legs, with a motor mounted to it. The motor has an offset weight attached to its output shaft. A variable power supply and the ability to change the offset weight lets us dial in the frequency and amplitude of the vibration.
  • I like to make two witness samples, one that goes inside the oven and one that stays outside. In the past I’ve used small dabs of Stycast on something flat like a chunk of metal. But cure time depends on the depth of the Stycast. It’s better to use a witness that at least approximates the rough shape and size of the feed-through. When both the inside and the outside witness samples have hardened, you know your feed-through is completely cured.

Don’t rush the curing process, if you can help it. A feed-through should last essentially forever. Rushing it and flexing semi-hardened resin so that you create a trapped volume or worse yet create an air path through the thing ruins a lot of hard work. Be patient!

That’s it. Be clean, be careful, and enjoy your vacuum.

– Tom

P.S. No, I never mentioned what went wrong with my recent failed feed-through. Any guesses? It was step #1: surface prep. The shell had been cleaned thoroughly in an ultrasonic cleaner, but the electronics had not. That was enough to cause an adhesion problem between the Stycast and the insulation on the wires. All it takes is one hole a hundred time smaller than a human hair to kill a vacuum. Surface prep really is king.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: