When large industrial vacuum brazing furnaces begin to produce customer parts that show discoloration instead of a pristine stainless steel finish or joints where the brazing material has refused to flow properly, production is quickly halted. Easy-to-find-and-fix leaks can take the furnace down for a half of a day, and more downtime is likely. The topics presented in this article will help make your next leak-checking search a quick and successful one.

Vacuum brazing furnaces, whether continuous in-line or batch machines, are large complicated systems that must be able to pump down to the required vacuum level to produce high-quality heat-treated parts (Fig. 1). It is important that these machines be vacuum-leak-free if they are to produce parts with void-free braze joints and flawless surface finishes. Ideally, these requirements would suggest that routine leak detection be a part of the regular maintenance schedule on these machines. Yet real-world production demands force some users to occasionally delay regularly scheduled preventive-maintenance tasks and instead operate their equipment to failure.

A poorly maintained vacuum heating furnace can have multiple problems, making troubleshooting that much more difficult. These concerns and operational issues apply to all types of vacuum heating furnaces, whether used for vacuum melting, sintering, quenching, annealing or tempering work.

 

Different Characteristics Between In-line vs. Batch Systems

An in-line vacuum furnace like the type shown in Figure 2 can have five individually pumped sections.

The preparation chamber is the entrance load-lock section of the in-line furnace. It contains the same vacuum hardware found in single-chamber batch systems (Fig.  3). Like a batch system, the preparation chamber is pumped from atmosphere to a required vacuum level each time a set of carriers is loaded. However, the three central chambers – preheating, heating and radiation – are able to remain under vacuum (and high temperatures) continuously. The gas-cooling chamber also functions as the exit load-lock chamber of the system, which means it is pumped from atmosphere to vacuum and vented back to atmosphere (just like the preparation chamber) for each set of carriers that are processed through the machine.

In a batch vacuum heating furnace (Fig. 3), this single chamber is pumped, heated, cooled and vented all during the course of each run. As such, the mechanical and thermal stresses and strains on these single-chamber systems can be even greater than those experienced by larger in-line systems.

During periods of heavy production usage, it is quite likely that certain somewhat-obvious preventive-maintenance tasks – like a noisy bearing in a vacuum pump, leaking oil or a questionably operating water-flow switch, or an intermittently operating vacuum gauge – will be ignored in favor of “keeping the machine running production.” This mindset can be risky because production can come to an absolute halt when parts begin to come out of the machine with defects. This unscheduled downtime becomes totally disruptive to factory output and causes a lot of unwanted excitement for the maintenance engineer.

 

Things that Can Go Wrong that Might Seem like a Vacuum Leak

Starting at the beginning of an industrial heating process, let’s consider one troubleshooting example where the furnace will not pump down to the required vacuum level. In this case, the operator reports to the maintenance engineer that the machine may have a leak because it is either pumping too slowly or simply won’t reach the required starting vacuum level. The maintenance engineer knows the mechanical pumps have been noisier than usual and begins to wonder if the oil in the rotary pump needs changing or possibly the mechanical booster pump is failing. What does the maintenance engineer do first?

The first thing he should do is change the rough vacuum gauge sensor head. Typically, these are Pirani-gauge sensors that have thin wire filaments inside that are subject to contamination and can break easily. Replacing this gauge is the single-fastest, least-expensive thing a maintenance engineer can do to try to rule out the possibility that the system is working fine and the problem is only that the vacuum gauge has become defective. If the system pumps properly with the replacement vacuum gauge installed, then the problem has been solved. The “bad vacuum” problem was a bad vacuum gauge – simple. If the replacement vacuum gauge doesn’t fix the problem, however, it’s time to look elsewhere.

The next step before bringing out the leak detector could be to perform a simple rate-of-rise test on the vacuum chamber. The procedure is easy. Just pump the unit as low as it will go, record the vacuum level and then close the valve to the pump and begin to record the chamber pressure (vacuum level) versus time from the moment the chamber is isolated from the pumps. The idea is to see if the pressure rises quickly, which could indicate the presence of a real leak.

A very useful idea is to benchmark the rate of rise of the vacuum chamber when the system is working normally. The rate of rise formula is Q = (P1-P0) x V/t , where:

P0 (Torr) is the vacuum gauge reading when the valve to the pumps has closed.
P1 (Torr) is the vacuum gauge reading after “t” seconds has elapsed after closing the valve to the pumps.
V (liters) is the volume of the vacuum chamber.
Q (Torr-liters/sec) is the rate of rise in the chamber that is caused by leaks and outgassing inside the vacuum chamber. This number is big if there is a leak and smaller if the measurement is mostly outgassing.

For example, a vacuum chamber with a 2,500-liter volume is evacuated to 3 x 10-3 Torr. The valve to the pump is closed. After one hour, the chamber pressure rises to a level of 1 x 10-2 Torr. What’s the rate of rise?

Calculation: Q = (1 x 10-2 Torr – 3 x 10-3 Torr) x 2,500 liters/3,600 seconds
                   Q = 4.86 x 10-3 Torr-liters/second

Because industrial furnaces are loaded with heating electrodes, thermocouples and thermal insulation – all of which outgas at considerable rates – there is no standard value of what an acceptably small leak rate is using this rate-of-rise test. The acceptable rate-of-rise value must be determined in advance for each individual chamber when it is known to be leak-tight. It is useful to run this test weekly to track the degree of outgassing and buildup of debris inside the heating chamber that can lead to poor vacuum levels without the presence of a real atmospheric vacuum leak.

It is important to mention the accepted truth that “everything leaks.” Even leak-tight vacuum chambers have permeation leaks, possibly some virtual leaks and definitely a considerable degree of outgassing that limit the ultimate pressure and slow the pumping speed of vacuum systems.

To limit outgassing (i.e., the release of contaminants on surfaces inside the vacuum chamber, such as cleaning solvents or moisture from the air), it is advisable to always keep the chamber door closed and the system pumped down whenever possible. The thermal insulation inside these chambers can be very hydroscopic, and pumping the humidity out of those materials can look as if there is a vacuum leak.

Be especially aware of this situation on hot, humid days in facilities that are not air-conditioned. These different types of “virtual leaks” are annoying in that they slow pumping times. However, the damaging leaks are the real atmospheric leaks that introduce oxygen and water vapor inside the heating chamber that can spoil the product in a heating run.

 

What to do After the System Passes the Rate-of-Rise Test

If the rate-of-rise test shows that the pressure in the chamber remains stable and low (but just not low enough), then there is a possibility the vacuum pump(s) could be faulty, which would also not be a vacuum leak but might look like one to the system operator. It’s also possible the problem could be a leak in the vacuum piping (foreline) between the pumps or somewhere in the suction line leading from the pumps to the isolation valve.

To troubleshoot this section of the furnace (the pumps, vacuum lines and valves leading to the chamber), it helps if there are extra connection ports in those sections where a vacuum gauge or helium leak tester can be connected for testing. First, connect a vacuum gauge in the foreline and read the pressure. If the vacuum reading is bad in this line, you’re close to finding the problem. Examine the pumps. Check their oil level and color.

Referring to the color chart in Figure 4, if the oil is dark, change it. If the oil appears to be the color of coffee with cream, there is very likely a lot of water in the oil. Water can accumulate in the pump oil over time when pumping atmosphere that is very humid.

During the hot, humid summer months, customers who don’t regularly change their oil frequently report “pumping problems” because of the buildup of water in the oil. Simply draining the oil and replacing it with fresh oil will immediately solve this problem (which is also not a leak!). If the oil is fine in the pumps and if all the flange bolts, clamps and wingnut fittings are tight, then it is finally time to get out the helium leak detector.

 

Quick Tour of Helium Leak-Detector Operations

When a helium leak detector is connected to a vacuum chamber or line that is pumped by a large pump, the apparent sensitivity of the leak detector can be dramatically reduced because a lot of the helium that enters through the leak is pumped away by the large pumps before it’s able to reach the leak detector (Fig. 5).

It’s helpful to know the pumping speed of the system pumps compared with the pumping speed of the leak detector itself if the pumps can’t be valved off (isolated) from the equipment being leak checked. If the ratio of pumping speed of the system pumps versus the leak-detector pumps is 10:1, then 90% of the helium tracer gas entering a leak will be removed by the system pumps, and only 10% of that helium will make its way into the leak detector, registering a leak. The apparent size of the leak will likewise be comparatively reduced by the same amount due to this reduction in signal-strength intensity. For this reason, it is a good idea to design a furnace system with a lot of test-port flanges and isolation valves throughout because building them in will greatly help when it is necessary to find a vacuum leak with a helium mass spectrometer leak detector.

Let’s look at this relationship in a little more detail. The leak detector has a pumping speed at the test port of 2.5 liters per second, whereas the vacuum pumping system (mechanical-pump package) has a pumping speed of 2,400 cubic meters per hour, which is 40,000 liters/minute (667 liters per second). That ratio of 667 liters/sec to 2.5 liters/sec is 267:1.

This means 99.6% of the helium tracer gas that makes its way through the leak will be swallowed by the pumps, while only 0.4% of that helium tracer gas will find its way into the leak detector and get recorded as a “leak” signal. Connecting the leak detector to the wrong spot in the vacuum system could cause the operator to believe that a large leak is a much smaller one. For the leak detector to only record 0.4% of a 2 x 10-4 atm-cc/sec leak (a big leak) means it would display that leak size as being approximately 267 times smaller, or just 7.5 x 10-7 atm-cc/sec.

A better place to connect the leak detector would be between the booster pump and the oil rotary pump, if possible (Fig. 6). In that location, the pumping speed is 420 m3/hour, or 7,000 l/minute or 117 l/sec. Compared with 2.5 l/sec speed from the leak detector, that ratio of pumping speeds is 47:1. This means a 2 x 10-4 atm-cc/sec leak would measure 4.3 x 10-6 atm-cc/sec. All these numbers mean that connecting the leak detector between the booster and oil rotary pump will give 5.7 times more sensitivity than connecting it up in front of the booster pump.

Of course, the best-case scenario is to have an isolation valve located just before the booster pump so the vacuum lines can be tested by the leak detector by itself, but few systems are so equipped. It turns out that most leaks do not occur in the vacuum piping but rather in the main vacuum chamber because there are so many more potential leak paths.

Vacuum leaks in chambers typically occur in areas such as feedthroughs of all sorts: water-cooling lines, heater feedthroughs, thermocouple feedthroughs and gas-inlet or rotary-drive feedthroughs. Other locations include but are not limited to the chamber door seal, valves, welds, bellows joints, threaded joints and other O-rings.

Big leaks can sometimes be found using alcohol and a spray bottle. Just spray the suspected leak area with alcohol and have the operator watch the Pirani gauge reading on the chamber. With luck, if the leak is large enough, the alcohol will weep into the vacuum chamber and instantly vaporize, causing a sudden and immediate jump in the rough vacuum gauge reading, indicating the location of the leak.

 

Knowing Your System Speeds Up Leak Detection

When attempting to locate a vacuum-chamber leak using a helium mass spectrometer leak detector, you should have an idea about the response time of the system, which will qualify that the leak detector is “working” and is able to sense the leak from where you have the detector installed on the system. Let’s assume the leak detector is installed at the near end of the chamber.

At the far end of the system (or the far end of the chamber being tested), you should install a helium-calibrated standard leak with a valve. This way, with the chamber pumped down as far as it will go, the helium leak detector connected in test mode and all other valves closed, open the valve to the calibrated-standard leak, which introduces a known helium leak signal into the system.

It is important to understand the time lag required for the leak detector to register this leak (count the number of seconds it takes from opening the valve until the leak detector “sees” the leak). This way, you will know how long to wait between spraying suspect areas of the chamber until the leak detector should respond. Spray helium on the outside of the chamber.

Using a low-pressure spray pistol, begin spraying at the top of the system and work methodically down from there. It is very important to begin leak checking at the top of the chamber because helium floats. You might be testing joints that are leak tight if you were to start at the bottom, but the leak detector could start registering a leak, or an increase in the background leak signal, because the helium sprayed on the bottom of the system rises and can get pulled into a leak that is located higher up on the chamber. Set the spray pistol to deliver just a barely detectable puff of helium.

Less is more when leak checking with helium to avoid filling the room with helium, which will increase the background signal recording on the leak detector and reduce the chances of being able to detect very small leaks. Ideally, the response time of your system should be no more than a few seconds. Connecting the leak detector in the line between the booster and rotary pumps will very often shorten response time, making it the recommended location. Furnaces are full of insulation, which can slow the response time of the leak detector. Therefore, performing this check with a standard leak can be very useful. Also, compare the recorded leak rate with the value stamped on the standard leak to be sure these two values are known. They are almost always different.

 

Parts with Unexpected Colors can Provide Information

Industrial heating furnaces are unique in that the surface appearance of the parts they heat treat (or their carriers/fixtures/tooling) can be changed if the part is located in the system near the vicinity of an atmospheric air leak. Air leaks in furnace chambers admit oxygen, nitrogen gas and water vapor in high concentrations in the vicinity of the leak, which can color the parts that are located in that area of the furnace. These unwanted gas inlets (leaks) can lead to the formation of oxides, nitrides or hydroxides on the surface of the parts in that area of the furnace.

The spatial positioning of your off-color parts inside the system will give you a head start as to where to begin looking for a potential air leak in the vacuum chamber. Typically, blue staining of parts indicates an atmospheric air leak. The larger that leak, the darker the blue staining, which can range from very light to almost black in color. Yellow-stained parts can also indicate small air leaks. Greenish-stained parts tend to be caused not by air leaks but by excessive moisture in the chamber, either from leaving the chamber door open too long between runs on humid days (outgassing) or loading wet parts into the system. Of course, water leaks from cooling lines into the vacuum chamber would be another very real concern.

 

Find Leaks Fast, Keep Production Running

Leak testing of large, complicated industrial heat-treating machines can be a task very much like playing a challenging game of hide and seek. The question is, “Where’s the leak?” The leak is often well hidden, and it is the maintenance engineer’s job is to find it. Because the production clock is always ticking, it is imperative that you find the leak quickly.

By knowing the straightforward and simple interrelationships of vacuum-system performance and the vital signs of these systems – including chamber cleanliness, preventive-maintenance upkeep, pump-oil changes, vacuum-gauge condition, rate-of-rise measurements and part color or carrier condition – you can get a significant head start on finding any vacuum-system leak before you must resort to helium leak testing, which should always be your last option.

Unless the maintenance engineer is well-trained and experienced and understands the science and art of helium leak testing, it might be wise to contract a professional helium leak-testing company. It may be your most efficient way to find a tough leak quickly and keep production running.


 

For more information:  Contact Evan Sohm, Director of Advanced Vacuum Technology for ULVAC Technologies, Inc., 401 Griffin Brook Drive, Methuen, MA 01844; tel: 978-686-7550 x244; fax: 978-689-6300; e-mail: esohm@us.ulvac.com; web: www.ulvac.com. Special thanks to Frank Mason of T.RAD North America for assistance with content of this article.