Bulkhead Penetrators (Getting a good vacuum seal)

This part is needed for my clock, with a pendulum is intended to swing in a low vacuum, as low pressure as I can manage.

The challenge is even a tiny leak will spoil the vacuum, and it needs to last for years because evacuating the chamber disturbs the pendulum and spoils the clock's accuracy.

The penetrator takes 6 1mm diameter bare copper wires from the electrical gubbins inside the vacuum out to an external microcontroller. The wires will fit through an 8 or 9mm diameter hole, at the moment an 8.5mm hole drilled through an M12 penetrator.

My design:

The 4mm groove in the flange is designed to take a 13 x 3.0mm O-ring. The exact position of the groove depends on from which side the pressure comes from. When the O-ring is squished down, the 1mm gap is on the high-pressure side, so the o-ring is pushed tighter by air pressure into the opposite wall.

My first question, is should the o-ring be on the low-pressure side, in the vacuum, or on the high-pressure side:

My first thought was flange outside, but I'm veering to flange inside. Reason being I plan to use a thread sealer, which, being gooey, could be pushed up the thread and out into the vacuum container - there's nothing mechanical to stop it flowing. Other way round, with the flange inside, I guess the sealant is still pushed up the thread, but will be physically blocked at the top by the O-ring and flange.

Which side of the bulkhead the flange is mounted alters the design in two ways. The groove moves sideways by 1mm, and there's an internal step inside the penetrator to support the insulating compound against outside pressure.

Secondly, although I've assumed the penetrator should be made from Brass, is there any advantage in Aluminium, steel or plastic?

Thirdly, what should the insulator be made of? Something that can be poured into the hole around the wires. It needs to be electrically good, physically strong and non-gassy. Araldite is a contender but it's messy and might gas. Silicone sealant is weak and gassy - it emits Acetic Acid as it sets. I'm tempted to use Apiezon Wax W, which is a purpose made hard high temperature vacuum sealing wax. Only problem, £60 for a kilogram block.

What do the team think?

Dave

Search for 'vacuum feedthrough' - you'll see stuff from a range of manufacturers, and hopefully some outline information on seal types, materials etc.

I'd probably be looking to buy a commercial multi-pin feedthrough and solder wires onto both sides. It gets a bit more interesting if you need coaxial cables to be brought through, but that is possible. My own knowledge is with high pressure, so I'll avoid giving direct recommendations.

I'd use a dowty washer, you'd probably not need sealant on the threads then

The classic high vacuum sealant AND high voltage electrical insulation is sealing wax. So ideal for this. The type used on old fashioned documents. Just be wary of black wax, it might be conductive. There is actually very few modern materials that are better.
I'm a bit wary of rubber O-rings for high vacuum. copper O-rings (ideally hollow) or pure aluminium gaskets will provide a better seal.

Robert.

Edited By Robert Atkinson 2 on 11/05/2023 16:56:56

Robert, Dave.

You are right to be wary of rubber o-rings for high vacuum applications.

ALL natural and synthetic rubbers exhibit gas permeability to some degree. That is, the rubber allows molecules of a gas (air) to slowly migrate through the rubber given time. More so in high vacuum and long duration applications.

Go metal to metal if you can and minimise the number of penetrations.

Mike

Dowty washers WILL NOT seal a thread, good between 2 faces ! The vacuum is stated as low and though the OP does not say how he intends to pull the vacuum a refrigeration vac pump will not be a HIGH vacuum, just a good one. Noel.

Hi Noel,
It's clear the OP is not going for true high vacuum. He is however looking for the vacuum to be maintained for extended periods without pumping. To do this requires high vacuum techniques.

Robert.

Edited By Robert Atkinson 2 on 11/05/2023 18:55:21

This part is needed for my clock, with a pendulum is intended to swing in a low vacuum, as low pressure as I can manage.

I think SOD means he actually wants a high vacuum. It is still only about 14psi, but permanency is the hard bit.

The part has a head on, the dowty washer goes between the flat face of the plug and the flat thing it screws into. I see no reason why it won't seal. When I worked on vacuum systems which were a lot higher vac than SOD is proposing we used O ring seals. No problem. We did have continuous pumping of the process fluid, but air inleakage would have been seriously embarrassing.

O-Ring sealed fittings should get you down to about 10-5 torr but it won´t stay there. A lot depends on the pumping, a water jet pump is limited by the vapour pressure of water, a two stage rotary vane pump 10-3 - 10-4 torr, add a cold trap and or diffusion pump 10-6 torr and lower, then you get into the realm of turbo molecular and ion pumps.Several careers ago I ran UHV systems, on one occasion we baked the machine to get the best vacuum for a customer demonstration, upon reconnecting the ion pump the electrical feed-through shattered., 10-9 torr to atmospheric pressure in a twinkling. I could have cried..

Before you get too carried, away think about the vacuum chamber and containment if any thing goes wrong. One of the scariest chemical practicals I had to do involved producing a metal sandwich compound which involved evaporating a metal and combining is with an aromatic compound in a large glass flask in a cooling bath (Timms Synthesis) , What could possibly go wrong?.

I also ran grease free vacuum lines using a compression fitting sealing (Swaglock)on O-Rings. For what you want O-Rings (Viton) should do.. Conductors embedded in epoxy should also work.

Flange and seal on outside so that the pressure is compressing it not trying to push it out of the gap..
Can't see why you need a thread sealant. You should design out any threads that might in other lives need sealing. The sealant is likely to outgas for months after which it will be dried out and crack.
When I worked on satellites traces of silicone furniture or car polish were out demon but this might only apply to ultra high vacs.

Hmm, glad I asked. Maintaining a vacuum is going to cause endless trouble I'm sure!

Apologies for sloppy terminology about high and low vacuum. I'm after a high vacuum, that is low pressure, as close to zero as my pump can manage. It's an old Edwards EB3A, which Robert RPI found a spec of 26" Hg / 134mB / 2psi . I'd describe this as a good medium vacuum, better than most pumps, but with a fair way to go. The pressure is plenty low enough to cause gassing and to expose minor leaks. This assumes the EB3A works properly: if it doesn't the pendulum will work in whatever reduced pressure it can manage. Some high-precision clocks used mild vacuums (800mB) deliberately, I think to stop the period varying with air-pressure (weather), whilst not requiring super-good seals.

Going to ponder the comments - my faith in O-rings is battered! Nonetheless I'll probably try one as the simplest solution first: it might be 'good enough', especially if my vacuum pump is knackered.

There's a second potent source of leakage in the design. The container is a 4" diameter PVC soil pipe,, The top is sealed with a glued top-cap, effectively welded shut, but the other end sits in a shallow groove. Although the pipe is plenty strong enough to hold a 2psi vacuum, it will flex as the air is pumped out. I fear any movement in the tube is likely to damage the groove seal. Proper clocks used soldered copper or thick precision ground glass containers, not cheap plastic plumbing!

Another fine mess I've got myself into, but maybe the design is tolerable if I run the pendulum at 8 or 9 psi. I expect it's obvious how little I know about vacuums! What would I do without the forum?

Many thanks,

Dave

You also need to pay attention to 'virtual leaks' if you aspire to high vacuum, or to maintaining a lower one without pumping:

Essentially you need to avoid pockets where air can be trapped: Take for example the option of O-ring on the outside and clamping nut on the inside. The nut will form an imperfect seal with the inside of the chamber. While it probably isn't good enough to hold a vacuum, it slows down the rate at which the pocket of gas between the nut and the O-ring can escape. This will provide a steady stream of gas at low pressure until the pressure in the pocket equalises with the chamber.

The preferred arrangement of the high vacuum chambers where I used to work was a flange with sealing O-ring inside the chamber with a stud through an over-sized hole secured with a nut and washer on the outside. The washers had holes drilled in them to connect the space around the stud to the outside air.

I'm not sure that PVC is a good choice if you want to hold a vacuum (it will probably outgas and also allow gasses to diffuse through it).

I second the suggestion to buy a surplus vacuum feedthrough.

(The Apezion black wax is no less messy than araldite!)

28" of mercury isn't high vacuum, it's what the brakes on steam trains worked at. At the risk of repeating myself, an air tight enclosure without vacuum will largely isolate the pendulum from changes in barometric pressure without being subject to steady drift from loss of vac.

I used to help assemble marine equipment that was analogous in having electrical connectors between low pressure inside and high outside. Though of air at 1bar internally and water at 1Bar per 10m of depth outside, so much bigger differences.

It may worth considering our approach, but it does need quite a thick wall.

The connectors were proprietory items for such purposes, with the internal conductors embedded in epoxy-resin, I think.

The salient point was that the larger connectors were not screwed into the cover, so thread-sealing was not a problem. They had a flanged boss that fitted into a close, but not tight, hole in the cover.

The flanges had 4 holes to take ordinary screws into blind, tapped holes in the cover. A thinner cover would need a doubler-plate or blind bushes perhaps soldered or glued in.

The seal consisted of two O-rings on the wet (high-pressure) side. One occupied the corner between the flange face and boss, and worked in a fully-machined countersink round the cover hole. The other was external to that again, in an annular groove in the flange, within the screws' boundary. I think for air pressure <1 Bar you could replace the countersink O-ring with one in a groove round the boss.

The outer connector, for the cable, used pins and sockets in rubber mouldings that effectively squeezed around the conductors.

The cover itself still needs seal to the rest of the case, of course.

I suspect your chances of holding a static vacuum for any significant time in a PVC pipe are small. Even UPVC is significantly permeable to small molecules like O2 and N2, even setting aside outgassing.

Well I said "I'd describe this as a good medium vacuum, better than most pumps, but with a fair way to go."

However, The main issue with a vacuum isn't the pressure though, it's holding it for a long time without further pumping. My container is only about 4 litres so any gassing or leakage will soon have an affect.

Andy G (and now Norman B too) could well be right about PVC gassing, it's made with a solvent plasticiser that's likely to boil off at 134mB. At that pressure water boils at 58C, so it too will vaporise rapidly. I've no idea how gas tight PVC is, and, although used for natural gas, air molecules are smaller than Methane.

Two reasons for pumping air out - it eliminates barometric error, and increases Q. Not too worried about barometric error because the clock can already compensate for that. High Q is desirable though, and Q rises as pressure drops. My Q is about 15000 in air, miles behind the pendulum in a Shortt-Synchronome - about 100,000.

Worth me trying to hold a deepish vacuum for a year, but the more I think about practicalities, the more difficult it seems.

Dave

Edited By SillyOldDuffer on 12/05/2023 15:46:12