Finding leaks in a furnace or chamber that has been used for several years takes some careful “sniffing” around using a helium leak detector. However, it’s easy to get sidetracked and frustrated if you are not careful.


This article is intended to assist readers with some tips I have developed over the years in the vacuum business to make the job a bit easier.

Getting Started

If your leak detector has an external standard leak, it makes sense to fabricate an inlet manifold that will allow you to valve in the standard leak and connect the detector to your system without mounting and dismounting the standard leak from the detector.

If your leak detector has an internal standard, it is still wise to mount a manual isolation valve to the leak-detector inlet. To run a calibration (cal) check with an internal standard, the leak detector needs to be blanked off, and the isolation valve easily provides this feature.

Most standard leaks have a dual diameter of 1.125 inch and .750 inch. An adapter can be fabricated, as follows:

  • Use a Swagelok ¾-inch Ultratorr gland, SS-12-UT-A-6 and weld it onto a KF25 tube flange, Nor Cal NW-25B-100 using a TIG process with E308L filler and argon purge on the ID.
  • Make up the inlet manifold using KF25 components, including a KF25 manual isolation valve. Mating the manifold to the leak detector is easy if it already has a KF inlet flange. For Varian Porta Tests, the inlet is 1.12 inch diameter, and this will require a special stub to connect KF25 to 1.125-inch tube OD. Varian should be able to supply this fitting (I hope).
  • The inlet hose can be two KF25 x 36-inch hoses connected with a centering ring and a clamp. Use the thin-wall, super-flexible hoses. The equipment to be leak checked will also need either a blanked-off port or another KF25 manual valve.

Another good idea is to attach a metering valve onto the equipment to be leak checked. This will allow a small amount of helium to be bled into the system to check for leak response time. The Swagelok SS-4MG2 metering valve works fine and has 1/4 NPT male connectors on each end. This can be adapted with appropriate KF fittings to attach to the main chamber on the equipment to be leak checked.

A good helium probe needs to be made. Use ¼-ID hose from the helium tank regulator to a Swagelok B-4JN4 1/4 NPT needle valve. Attach this to a Coilhose 600S-DL blowgun (1/4 NPT). Remove the blowgun tip and tap 1/8-27 NPT. Thread in a Swagelok SS-200-1-4 male connector, 1/8-inch tube x 1/8 NPT male. The wand is a piece of 1/8-inch OD x 0.030-inch wall 304 stainless tube about 18 inches long.

Enlarged Image


Fig. 1. Helium leak detector – Varian 959 Porta Test


There are 31.5 x 106 seconds in a year. Multiplying the leak value in cc/second by this factor gives the equivalent yearly leak rate. Therefore, a 5x10-8 std. cc/sec leak equals 1.57 cc/year. Likewise, a 1.5 x 10-7 std. cc/sec leak equals 4.73 cc/year. (These small leaks are made by enclosing a pressurized helium glass tube in a suitable housing, which when evacuated allows the helium to diffuse through the glass – a so-called permeation leak. They are fragile and do not like being dropped. Also, store with the isolation valve open)

PVD-type coating systems should be leak checked using the lower 5 x 10-8 cc/second standard leak to calibrate the leak detector. Other metallurgical systems like vacuum heat-treating furnaces can use the higher “seven range” standard leak. An exception might be a vacuum brazing furnace operating in the 10-5 torr range in which the lower leak-rate standard would be dictated.

Most of the leak detectors out there are Varian Porta Test models (Fig. 1) that use an external leak with manual tuning. This discussion is biased toward these installations.

Install the standard leak on the inlet port of the leak detector (or use the manifold mentioned in the first paragraph). Let the machine run for 20-30 minutes to stabilize the electronics. With the filament on, set the coarse and fine zero to roughly 1.0 on the 0-10 scale with the standard leak closed and the range switch set at the 10-7 or 10-8 scale. Then open the standard leak. There will be a rapid upscale rise indicating a large leak. Wait until the output settles. Then adjust the focus and ion controls for a peak indication. The emission pot can be tweaked to give 1.00 volts DC between TP4 (ground) and TP5.

Ideally, you want a 10-division indication (full scale) with the 5 x 10-8 leak open and the “cal” pot full CW (max gain). The “cal” pot can then be turned down to make the output agree with the value of the standard leak. The leak detector is now calibrated and can “see” the helium standard leak.

After tuning, test the leak detector for leaks by flowing helium around the spectrometer tube and valve manifold. The last thing you need is a leak on the leak detector, which will pick up helium “drift” from your leak-check activity!

Enlarged Image


Fig. 2. Typical vacuum system with a diffusion pumping system

Connecting to the System

Where to connect the detector is a trade-off between sensitivity and low background. Connecting to the main chamber will reduce sensitivity but will generally have the lowest background. Best sensitivity will be at the mechanical pump inlet, but the background will be highest there. I generally use the blower inlet. Per previous discussions, install a KF25 isolation valve at the blower inlet with suitable adapters. This way you do not have to dump the system to air while you attach the leak-detector hose. The last thing you want to do is dump the blower inlet to air while it is backing the diffusion pump – a good way to set off an explosion if you are using “Convoil 20” oil in the diffusion pump!

Also attach the SS-4MG2 metering valve on the main chamber. (An unused thermocouple feedthrough port works well.) Make sure the metering valve is closed.

Enlarged Image


Fig. 3. Typical vacuum system with a blower pumping system

Leak-Checking Techniques

Pump down the chamber/furnace, and get it on the diffusion pump if applicable. Then valve-in the leak detector. Note the background leak indication. It must be less than the standard leak, preferably a decade below the standard.

High background probably indicates some gross leaks, in the 10-4 or 10-3 cc/second range, so these should be repaired in order to get the background down to a reasonable level. Adjust the helium flow on the wand using your tongue as a sensor to sense a low flow. You can later use a coffee cup with water and adjust the flow for several bubbles per second, but for the initial leak check, the tongue method is good enough.

Start at the top and work down. Check the following:

  • Power feedthroughs
  • Door seals
  • Viewport seals
  • All tapered pipe-thread connections
  • Vacuum/pressure switches (remove the switch covers and bleed helium in)
  • Electrical feedthrough on quench blower motor (if applicable)
  • Glands on internal cooling loops
  • Nitrogen/inert-gas bleed valves (will require opening the upstream connection and introducing helium into the valve)
  • Any Dresser-type couplings
  • Oil drain/fill connection on the diffusion pump (flood bottom of diffusion pump to see if there are leaks on the boiler plate)
  • Cold cap glands on the diffusion pump (if applicable)
  • Butterfly valve shafts
  • High-vacuum valve piston rod seal (may require introducing helium into the lower port of the high-vacuum cylinder)
  • Mechanical seal, end plate and oil line leaks on the roughing pump
  • Mechanical seal, end plate and oil-fill/drain leaks on the blower

Helium “drift” is the major problem in locating the site of the actual leak. The tendency is to use too large a flow of helium, which results in a leak indication that is not in the location probed. When you are on top of the leak, the response time will be very quick (a second or two), and you will be able to repeat the leak indication at least three times successively at the same location.

If no leaks are found, note the vacuum level and slowly open the SS-4MG2 needle valve on the main chamber to give a small upscale deflection on the vacuum level. Now introduce a puff of helium into the metering valve. You should see a large leak indication in 1-2 seconds. Close the metering valve.

Here’s another quick test: Probe helium into the gas ballast check valve on the Stokes backing pump. Put your thumb over the port to seal it off and then quickly open and close the gas ballast valve to introduce a small amount of helium into the pump. It will “back diffuse” into the blower, and you should see a large leak indication quickly.

Leaks that remain could be due to water leakage on the furnace shell or gas-cooling finned heat exchangers. The gas-cooling heat exchangers will have to be blown out to get rid of water and helium introduced using a large helium tank/ regulator. We have seen water leaks on the diffusion-pump cold cap. Check this by blowing out the cap water line and introducing helium.

With all leaks repaired, you want a background in the low 10-9 cc/second range. You may have to vent and repump several times to get a low enough background.

At the conclusion of leak checking, get in the habit to valve off the leak detector so that it won’t be dumped to air on the next roughing cycle.

Other Issues

Check the roughing valve by closing it and introducing helium on the upstream side. This normally requires disconnecting piping or introducing a port as outlined here.

The high-vacuum poppet can be checked by opening the foreline valve and spraying helium around the high-vacuum valve poppet plate. Again, this is very difficult to do because access to the poppet is nearly impossible with most vacuum-furnace designs.

One solution is to make up a mild steel boss with a -8 SAE port machined in. This can be TIG welded onto the diffusion-pump plenum and a small hole drilled through so that helium can be introduced into the poppet area. The boss is then blanked off with a -8 SAE port plug and “O” ring.

The ideas presented here are not meant to be all-inclusive, as each furnace or vacuum system has a unique design. We have found many variations over the years, but I hope these tips will make your leak checking a bit easier. IH

For more information: Contact Geoff Humberstone, president, Metallurgical High Vacuum, 6708 124th Avenue, Fennville, MI 49408; tel: 269-543-4291; fax: 269-543-4750; e-mail:; web: