Exploding Bulbs?

Hello all:

I've been running in to a peculiar problem at work the last few days. We have a set of heating bulbs in one of our machines that keep exploding during our industrial process.

The heater bank itself consists of 16 halogen heat lamps wired in parallel. The bulb in question is GE lighting's QH1200T3/CL bulb, rated for 144V, 1200W, 350 deg C.

We are running the bank with a Sorrenson 160/158 power supply in voltage control mode, at about 130V and 128A total, or 1040W per bulb.

The machine itself is a high vaccuum machine in the 10^-6 torr range doing long throw sputter deposition. We cannot see inside to look at the bulbs as they break, but we can observe the change in resistivity of the load as bulbs go out.

We also occassionally see melted wires... despite the fact that we have a good safety factor on our lead wires going to the bank. We are using 12 AWG nichrome coated copper multi stranded wire, with fiberglass insulation rated to 500+ deg C. On top of that, we have added insulating ceramic beads.

Any thoughts as to what could cause something like this?

are they exposed to the high vacuum ? they look like standard heat lamps and i doubt they are designed for such a hostile environment if they are

I have no real idea but it won't stop me from guessing:

I assume that the power supply has been tested. A quick way to check for intermittent high voltage across a bad supply would be a TVS rated just above the output voltage, the smaller the better. If it's damaged that would point to a bad supply.

Is it possible that something from the sputter deposition is getting on the bulbs? I've heard - possibly urban myth - that even oil from greasy fingers can cause halogen bulbs to break. It seems to me that the melted wires could occur if the filament shorts when it breaks, though the wires should last much longer than the filament. In that case they're a secondary effect not related to the initial problem.

years ago i worked as an onsite service engineer for a manufacturer of halogen heaters, in my experience finger prints are a sure fire way for tubes to fail. melted wires close to the tube are caused by loose connections / connection springs loose etc causing arcing and excess heat. The arcing can also add what i can only describe as a "harmonic" which causes the filament to vibrate excessively and fail prematurely although this shouldnt cause them to explode. but excessive heating from a loose connection could especially if the heat generated is enough to melt 500deg wire thermal expansion et-al

I actually had a similar thought. It came to my attention that the bulbs get hot enough in some machines we have to start to sag over the months of use we get out of them... and we are talking a quartz outer shell here. Thats no small amount of heat, to be releasing the stress in quartz based material (definitely above the operating temp spec). It's entirely likely that the pressure differential is enough to pop them inside the machine, given that they are a little weakened by the heat already... Thanks!

We actually have a meter across the load on this machine, complete with datalogger. Looking back at the system there is very little over-shoot in the voltage, with a decent amount of overshoot in the current. Sometimes as high as 20A over the equilibrium point.

As to the sputtering: It is out of the line of sight, so I don't expect we would see anything significant in terms of sputtered material on the bulbs. We might get some occasional debris when we vent to atmosphere, but nothing too out of the ordinary.

But I didn't know that about the greasy fingers... I wonder if the bulbs got touched in the process of installation, I definitely wouldn't doubt it. I know I've touched one or two with my fingers in my time there... Oops! I'll make sure to let everyone know that one.

Also consider that in the partial vacuum of the chamber, there is a lose of convection cooling which means the bulbs rely more on radiant cooling to draw heat away from them. Try running the lamps at a lower voltage and see if they stop exploding. Also, if you can get your hands on some wireless temperature probes and use them to monitor the temperature of the bulbs both while at normal atmosphere and at 10^-6 torr and see if there is a significant difference.

If running them at a lower voltage does the trick, then consider adding a few more lamps to make up the difference in heat volume.

I was thinking about this (sometimes it even works) and realized you said the problem occurs in one of your machines, which I think means that the other machines are OK. In that case all the general questions, such as whether the bulbs will tolerate a vacuum, are answered by the bulbs working in the other machines. The answer lies in what makes this machine different. It could even be something as simple as a single blown bulb, followed by replacing the bulb without being careful about fingerprints. This would create an ongoing problem particular to that machine.

Indeed! They do work in other machines, so in some capacity they are definitely OK in vacuum.

The main difference in this machine is that we are running them considerably hotter than in other machines. We are running at a similar operating point, but for significantly longer. Unfortunately we do not currently have a way of measuring the temp at the bulbs directly... we only measure temp at the points we are interested in from a product point of view. Installing something to monitor and limit bulb temperature will be my next step, I think.

Very silly question perhaps, but I noticed this on design using halogen bulbs on would you believe a light scanner and couldnt work out why there was an initial voltage present on the bulbs before switch on, although strictly speaking they were biased on, but not enough to see light emmision

I came to the conclusion that what they were doing was preheating the elements inside the bulb reason being that from cold for example, If they were turned on fast the elements inside the bulbs would fracture "heating to fast sudden extreme changes in temperature"

Also on cooling down you are bound to get condensation forming which would also aspire to blowing bulbs You come across this quite often in security lamps this is a common problem No preheat of filament to bring halogen bulbs up to starting temperature before firing

Examples Street lighting uses this preheat cycle

Very similar to zero cross mains detectors to ensure filament switched on at the correct phase i.e just above or below zero volts rather than suddenly halfway through the cycle

examples light dimmer circuits no zero crossing detectors implemented especially old types of dimmer circuits

I wonder if this is related to your problem that you are experiancing

Just a suggestion which might answer this

Respectfully
.
Mark Harrington