What exactly is the deal with "protected" metal halide lamps?

Why are they "protected"? Did they measure a tendency of explosion?

Why are UN-protected metal halides then, sold as well?

Is there a difference in the technology or can they explode, too?

Many thanks for the info,

Best to all,
--
Ioannis@

"Protected" O-rated lamps have a quartz shroud around the arctube that serves two purposes, it acts as a blast shield in the event of arctube explosion/rupture, the shield is also usually doped with UV Stop to block the strong UV emitted if the lamp's outer envelope becomes broken.

Un-protected (Type R, and type S) lamps have to be operated in completely enclosed fixtures with a lens, in the event of a arctube rupture, all the fragments are contained within the fixture. Also if the outer envelope breaks off and the arc tube continues to operate, the fixture lens blocks the UV. The only exception here is type "S" lamps, which are unprotected but are permitted to operate in open fixtures as long as they are operated base-up, as the likelihood of explosion is almost nil

The "protected" halides contain a measure designed to contain the eventual arctube explosion without breaking the outer bulb.
This is usually in the form of a thick quartz tube around the main arctube. Once the arctube explodes, this shroud then absorbs nearly all the kinetic energy coming from the flying remains of the arctube and the pressure wave from it's filling. The remaining energy is then low enough to be contained by the outer.

Why not all lamps are not protected? There are few reasons:
Many mainly outdoor lanterns are designed so, they safely contain the pieces of an exploding bulb, so there the protection onsode of the lamp is not needed.
The protection is obviously not for free, so that extra cost is not needed when the lantern design needs the extra protection due to other reasons anyway (protecting the hot lamp surface from water causing thermal shocks or dust blackening the outer surface and so on).
Other reason is the size: The shroud is of quite significant size around the arctube. With large elliptical bulbs that is no problem, but with compact formats for the compact fixtures needing really small format lamp that would be problematic.

For using lamp alternatives you may use a protected lamp, if it fit's into the lantern even if the lantern is designed for unprotected lamps - it mneans just an extra protection layer. It is just not needed, but it is of no problem.
But putting an unprotected lamp into a lantern designed for only protected lamps makes that lantern dangerous: If the arctube explodes and the outer bulb won't contain it, the lantern may not stop the flying hot pieces to escape and e.g. set something on fire or so.
Therefore the US code requires the unprotected lamps to have wide necks (the protected are supposed to have narrow neck) and the lanterns designed for protected lamps narrow hole (the ones designed for unprotected have then wide hole), not allowing to insert the unprotected lamps.

But with the shroud was made one observation: The quartz protected lamps tend to degrade slower than their unprotected counterparts. The reason is, the shroud hosts a charged layer (charged by the electron photo emission), where this charge causes an electrical field keeping the Sodium (and other such metals if used) atoms further away from the arctube wall, so slowing down the arctube sodium corrosion (blackening). This is the reason, why many makers do not offer unprotected quartz lamps in the Elliptical format (where the protection shroud fits well).

Thanks to both for the info.

So, is it a bad idea to use a small, elliptical, unprotected metal halide (say, a PowerStar OSRAM-HQI-70W) on my desk light?

The desk light's luminaire is the classic ellipsoid metal container and from my point of view sitting on the desk, I cannot see the lamp. If anything blows, things will go down.

I also plan on slightly under-powering it, @~ 40-50W, because I prefer slightly mercury color-shifted light.

Does this setup hide any obvious dangers?

Thanks again,
--
Ioannis@

Although more expensive and risky in the shipping, I'd try to get a 100 w clear mercury from the US eBay and use that instead.

I do not, under any circumstances, use unprotected lamps in open luminaires or setups. Others do, and have no problems doing it and are probably lucky nothing has happened, though. The risk is quite low, but still there. In your desk lamp, if a rupture was to occur, it could set paper on fire if you have some on your desk at that time. It could also burn your hands, depending on what you're doing and where on your desk.

Metal halide arc tube explosions are quite rare, others on this site may tell you otherwise, but from my experience ive seen far more halogen capsules pop than MH lamps. MH arctubes are most prone to rupture in the situation where a universal burn rated lamp is operated horizontally and ran until it completely fails.
I agree with Sol on the desklamp however, a 50w or 100w MV lamp (clear or DX) would probably make a better choice for a desklight as theres no high voltage involved (all 70w MH lamps are pulse start) and desklamp incandescent lampholders are not rated for HV ignition pulses like HID lampholders are. Also, unless you like quite a bit of light on your desk, the 70w MH may be a bit powerful. Though if you do choose the 70w MH, as far as I know, all Sylvania/Osram medium base E17 70w MH lamps are "O" rated (protected) so perfectly safe for an open fixture

Thank you for the information and your advice.

Has there been done any research for the calculation of actual propabilities of violent failure under normal usage or do the big ones keep constructing halides despite having not one iota of information about the phenomenon?

Any studies done? Any results published? By anyone?

Are the probabilities of this phenomenon the same in today's technology as they were during the first halides?
Was there any worsening? Any bettering?

What's going on?

Seems I've missed a lot of history...
--
Ioannis@

I personally would like a 50W MH for a desk lamp. I don't particularly like the colour of a clear mercury lamp. However, if you prefer that MV colour, it would be safer to use a clear mercury lamp than under driving a MH lamp (and it would be better for lamp (and ballast) life).

I never worry about MH lamp explosions seeing they are very rare and more likely to happen upon lamp start which makes them less violent. I am, however, the type of person that likes to "wear suspenders as well as a belt" !  You should be aware that even mercury lamps have an explosion risk although it is far less common than the not-really-common MH lamp explosions.

Otherwise if you're planning on building a fixture from scratch, you could always enclose the lamp completely in a glass jar.

I personally would like a 50W MH for a desk lamp. I don't particularly like the colour of a clear mercury lamp. However, if you prefer that MV colour, it would be safer to use a clear mercury lamp than under driving a MH lamp (and it would be better for lamp (and ballast) life).

I never worry about MH lamp explosions seeing they are very rare and more likely to happen upon lamp start which makes them less violent. I am, however, the type of person that likes to "wear suspenders as well as a belt" !  You should be aware that even mercury lamps have an explosion risk although it is far less common than the not-really-common MH lamp explosions.

Otherwise if you're planning on building a fixture from scratch, you could always enclose the lamp completely in a glass jar.

Many thanks for the advice sol. Allow me to disagree on the mercury lamp. It is indeed the optimal choice for my experiments, but unfortunately I have yet to see a small clear that's completely free of UVA. These little things pass through a lot of actinic radiation, in particular they do not attenuate the 365nm Hg line and overexposure to this line can cause corneal aging and cataracts. It's not wise to use actinic lamps for reading. Otherwise, I love them, too 

--
Ioannis@

If you use a mercury lamp with uv blocking glass, would that be a solution ?

Regarding the MH versus MV colour, that is my opinion and I am not trying to impose it on anyone as I wouldn't like to be imposed a lamp type. I am sorry if that is what you understood as it is not what I meant.

Don't get fooled by the MV's not exploding: They indeed do not explode on their own (unlike MH's), but still may explode violently when the ballast fails (winding get fried into low impedance and so heavily overloads the arctube; speaking about European series reactor).

The US CWA and/or HX ballasts are safeer in that respect, as the winding short circuit always means either lamp current decrease (primary insulation breakdown) and/or lowering OCV with missing inductance (so the lamp extinguish; in case of secondary breakdown).

So for a lantern designed for incandescents really use only a protected lamp, the protection takes care of the explosion regardless what causes it to happen.

Underpowering the lamp will shift it's parameters and may shorten it's life, but still won't prevent an eventual explosion, so I won't go that way...


For the high voltage: Usually the cable length between the ignitor and ballast is limited to max 2m, so most likely the ignitor will have to be put close to the lamp socket. In that way the cable to the ballast won't be exposed to the high voltage at all.

If you use a mercury lamp with uv blocking glass, would that be a solution ?

Of course. But then I would need a very special glass or plastic. Hard borosilicate glass and regular soda lime glass do not block this line. I'd need a glass that blocks anything below 390-400nm. Or something like translucent/transparent acrylic plastic, some spectacle lenses are made of.

Regarding the MH versus MV colour, that is my opinion and I am not trying to impose it on anyone as I wouldn't like to be imposed a lamp type. I am sorry if that is what you understood as it is not what I meant.

I did not misunderstand you, so you don't need to apologize. Neither does it make sense to apologize, about me misunderstanding, even if I did, so no worries  

Best,
--
Ioannis@

Don't get fooled by the MV's not exploding: They indeed do not explode on their own (unlike MH's), but still may explode violently when the ballast fails (winding get fried into low impedance and so heavily overloads the arctube; speaking about European series reactor).

I pushed my first mercury lamp to oblivion when I was 14. To reach total destruction, I ended up powering it with the load of a welding machine, at around 20-40V and 135A. The burner simply got bulged at the center, found a weak point and eventually formed a bubble, which burst quietly, not even with a loud pop.

In all other cases, i.e. when the current got very high, such as in accidental short circuits on 220V mains, the only thing that blew was the molybdenum seals of the burner. Sometimes even the seals were saved, because my house panel's safety electric fuse went first.

Anyway, it's not my intention to make this a debate on the robustness of the mercs. MHs exploding is an entirely different phenomenon and I was hoping that since it's been observed (I've seen some videos in James' website as early as 1990-2000), I thought there'd be some study of the phenomenon, mainly by the companies themselves, to substanciate opinions about it.

Without relevant studies and actual failure rates, statements such as "MHs do explode sometimes" or "can explode other times" or "rarely explode" are fairly moot. I am a Mathematician and Logician. Not even an Engineer. I only believe numbers and probabilities coming from scientifically correct conducted experiments, therefore I find it is attrocious that the big companies continue to produce lines of MHs without having investigated the phenomenon. Before you build a nuclear reactor, you examine the phenomenon of a possible melt-down and back-engineer it so as to avoid such an event in advance. You don't go on, building the reactor, saying, let's build it and deal with the possibility of meltdown by building a good shield around it, even if you can calculate the power of the leak. In Chernobyl the reactor had a shield. Didn't stop it from melting everything in its path and burrowing in the ground.

No offense intended, of course. It's more of a philosophical question of good Engineering practices rather than anything else.

Underpowering the lamp will shift it's parameters and may shorten it's life, but still won't prevent an eventual explosion, so I won't go that way...

Here I am in the blind again, because of conflicting data. My old catalogs state that sometimes underpowering mercs is favourable, because of the reduced sputtering from the thermionic emissions (lower temp of electrodes), hence a longer useful life. Overpowering mercs may be also be justified in some rare cases, such as running 1000W lamps at 1500W in stadium spotlights, because this is counterbalanced by the economic inpact factor of using them only for a few hours at a time, so you only expect a reduction to useful life.

Now you are saying that underpowering a MH reduces useful life, so this seems to be opposite to mercs. At a lower power density, the temperature of the arc and the pressure in the tube are lower, as a result some of the halides stay in the form of condensates and do not participate in the discharge, meaning they stay away from the tube walls altogether, so how exactly does this affect lamp life negatively?

The explosion statistics:
The problem was mainly observed with US probe start MH's on a CWA ballasts. There two factors are of quite high significance:
First (responsible mainly for the violence of the explosion) the arctube volume is rather high, so if it get pressured, there is a lot of accumulated energy inside.
Second big contributor (to the probability at the EOL) is the CWA characteristic: As the lamp gets older, the voltage drop rises. The CWA is a constant current source, capable to maintain the arc at nearly full current even when the arc voltage approaches the OCV. That means nearly tripple real power transfer to the lamp (so a 400W lamp may get nearly 1200W). That is tremondous amount of extra heat, usually yielding the arctube to fail when way overpressurized.

With European style installations I've never seen an MH lamp exploding so, the outer get smashed directly by the explosion. The series choke ballast has one property: If the arc voltage exceeds about half of the OCV (so the mains voltage), the arc usually can not hold anymore and so extinguishes itself. That means the EOL MH mainly starts cycling. After few cycles the quartz develop fatigue cracks, so finally rupture. But it does so at the time of the greatest stress, which is just few seconds after restrike. And at that time the lamp is rather cold, without much internal pressure, so the explosion does not have much force, so usually the outer just contains it pretty well.

All that above is valid for normal lamp EOL.
But the explosion problems had triggered a safety analysis and the result was a finding, than when the high pressure lamps get gradually overpowered (so they heat up and pressurize sooner than the electrodes fail because of a huge current; due to ballast failing and/or mains overvoltage), there is a possibility for an uncontained explosion.
Other possibility could be just a fault in the arctube itself, so it ruptures at normal pressure. Again, the MV operating pressure of about 10 atmospheres and a rather large arctube makes it still quite powerful.
And because such failure mode was theoretically not excluded from happening on any high pressure lamp, that recommendation was added to the safety rules, just as a safety precaution. The laws require every such thing can fail only in a safe manner and it has to be proven the only possible failure modes are the safe ones. With HID it was proven to be safe only when there is some measure to contain the eventual arctube explosion.
The law just does not allow any device, which reasonably may fail in an unsafe manner. And the "reasonably" means any single independent fault. So ballast insulation failure is a single point failure, so it is "reasonable". Mains overvoltage is another single point failure (even if it causes the ballast to overheat and fail as well - still it starts as just an overvoltage). Or a manufacturing defect in the arctube is again a single point failure. Even when no one have ever seen such failure to cause any explosion, if the first would ever happen and because of the lacking protection measure cause any damage, the maker (of the device) will be held 100% liable. Don't forget the lamps are designed with an assumption such measures are in place, so e.g. the MV's operate on way higher pressures (to get higher efficacy) than it was common 50 years ago...
The videos James was publishing are from tests, where the lamps were intentionally overpowered to get maximum force explosion (just to demonstrate it and to find out the efficiency of the different containment measures), that were in no way any real life cases.



Underpowering:
Indeed, when the lamp fill is inert towards all the components, lower loading means longer quartz life. But:
There are two problems:
- Electrodes are designed to operate at optimum temperature. Higher temperature means they evaporate too much. Lower temperature means the electron emission is not that good, so there is higher voltage buildup at the cathode, accelerating the ions to higher velocities and these then sandblast the electrodes faster. End result is similar to the evaporation: Material gets transferred from the electrode onto the arctube waqll and there is blocking the light and heat (that heat does not help the electrode, but alters the thermal balance). So that means there is an optimum, where the electrodes still do not evaporate that much, as well as are not that fast sputtered away. This is well known for the lamp design, so the electrode thermal balance is designed so, the operating temperature is exactly at that level when operated at rated current. Any deviation in any direction means the degradation gets only faster.
- Second is related to the halide pool. Many lamp designs have the arctube fill corrosive towards the material of the arctube or it's components. Normally the lamp is designed so, the corrosion does not cause failure or too much performance degradation before the end of the rated life.
Mainly with the CMH the pool is aggresive towards the PCA when liquid (it just dissolves the PCA), so larger liquid pool means faster arctube corrosion
Generally the lamps are expected to not last the full power rating when dimmed, the dimming is done only to reduce the power requirement when the full output is not required.

And there is yet another aspect when underpowering: The lamp starts with rather low pressure, then the heat warms it up and so increases the pressure to the operating one. But the lo pressure means low voltage drop. And low voltage drop means low power transfer (ballast to lamp).
Normally, with rated ballast, the lamps are designed so, the ballast delivers in any stage of the warmup sufficiently more power than required to maintain the temperature of that stage, so the lamp temperature increases. But when you use lower current ballast, the power may become insufficient in some stage and so the setup will thermally lock there.
This is the reason, why with dimming systems makers require to start the lamps always at full power for the first 15minutes and only then reduce the power to the desired level.
Mainly lower power mercs are known to be quite sensitive to this phenomenon.

I've dealt with way too many MHs that have exploded in enclosed fixtures. (thankfully not while i was around) I've slowly been converting those fixtures over to HPS and eventually LED once the prices come down enough to justify it.