High Intensity Discharge lamps 101

[Vince]

The lighting world of today is highly complex. This topic explains how the different HID sources work, their pros and cons, etc.

MERCURY VAPOR

Overview

The mercury vapor lamp is the oldest, and probably the most-known HID lamp. This type of lamps uses a special quartz glass arctube (not true quartz as we could think), which contains mercury in an argon atmosphere. Unlit, the mercury is easy to see in the arctube, but once lit, all the mercury evaporates and emits light. Without any color-correcting phosphors, a mercury lamp is bluish-white with poor CRI, especially with warm colors such as yellow, orange...

Many phosphors were developed. Most of them aren't used anymore. Here's a little list:

/A: Clear. Mercury lamps were marketed first with no phosphors. These are known today with the /A suffix. Natural mercury spectrum and low CRI (15).
/DX: Deluxe White. The only coating still available today. Pinkish white light with moderate CRI (50) and a little less lumens.
/C: Standard White. Last manufactuer of them was Philips, which discontinued them in 1986. Yellowish-white light with higher CRI (60), but less lumens than /DX.
/W: White. Bluish-white light (less bluish than clear) with more light output and a little higher CRI (~35)
/N: Neutral white. Similar to incandescents and Soft White fluorescents, widely used in commercial applications up to the early 90s. GE manufactured them as /WDX, or Warm Deluxe White, which is the same.

There are also some even rarer coatings:

/Y: Caution Yellow. Yellow-tinted lamp, only produced up the early 70s. HPS lamps replaced them in their application.
/R: Beauty-Tone mercury (to confirm). This was a coating exclusive to Westinghouse, if I'm right.
/X: Predecessor of the /DX. It was a lacquer of red-tinted transparent coating. /DX replaced it in 1964, the year /DX was invented.

How they work

Here's a picture to explain it:



The starting probe allows the lamp to be started with less than 300V open circuit (except for 700W & 1000W lamps which need ~450V). The arctube contains mercury in a argon atmosphere. This simple mix in the arctube leads to lamps that are chemically very stable, hence their very long lifespan. (In the other hand, sodium in HPS lamps is an alkaline, and metal halide lamps use a mix of different salts).

Mercury vapor lamps take up to 7 minutes to warm up completely, their hot restrike time is about 4-5 minutes. The lifespan specified on mercury lamps is most of the time 24,000 hrs. In many case, lamps way exceed the lifespan.

However, a problem comes out as the lamp ages, especially when approaching or overpassing the 24,000 hrs. Mercury lamps are prone to lose light output, so after 40,000 hrs, the lamp could still work, but only with a lot less light output. For example, a 175W mercury lamp usually emits 8,000 lumens. At 40,000 hrs. and 40% of light output, the lamp would only emit 3,200 lumens. These values are just an example. Some lamps, such as the Westinghouse LifeGuards, can still have their full output after 100,000 hrs working!

HIGH PRESSURE SODIUM

High pressure sodium lamps, often contracted "HPS", are the most used HID light sources in the world today. In most North-American and European countries, they replaced the mercury vapor lighting. HPS lamps first came out in the early-mid 60s.

How they work

They have a structure that differs from the mercury lamp:

End of life

Unlike mercury lamps, which get dimmer and dimmer, HPS lamps do what's called cycling. The lamps shuts off, then partially warms-up back, then shuts off again and so on. This phenomenon is caused by the voltage rise-up, which occurs as the lamp ages. The voltage rise-up is due the instability of sodium, which is a alkaline and reacts easily with other elements in the arctube, such as alumina of arctube, tungsten of electrode, emitter of electrodes or mercury (if any). When voltage is too high, the ballast cannot maintain the arc anymore and the lamp shuts off. As the lamp cools down, the voltage gets back to lower values that allow the ballast to strike the lamp again.

Extra part

The HPS lamp has one part that isn't present in the mercury lamp: the getter. What's a getter? It's a piece of a specific metal that will react with impurities in the vacuum of the outer bulb. Why impurities? Well, a HPS arctube can leak, letting some sodium go in the outer bulb. Like gaseous tungsten in incandescent lamps, gaseous sodium will deposit onto the outer bulb, which will blacken it. To avoid this blackening (and lumen loss), the getter will "catch" the sodium by chemically reacting with it. Smart isn't it?

METAL HALIDE

Metal halide lamps are probably the HID lamps that will replace all other HIDs in the future. In 2010, their development is not finished and better lamps are still being created. Metal halide lamps came out in the late 60s. They were created in order to have a efficient white light source. Basically it's a mercury vapor lamp with extra salts. The most basic mix used is Mercury-Sodium-Scandium, contracted as Hg-Na-Sc. It is still used today in most probe start MH lamps.

How they work

MH lamps are pretty similar to Mercury lamps



MH lamps use a more complex mix of different metal salts, many mixes exist. What differs is, first of all, the white paint at the tips of the arctube. That's a special paint used to keep the heat around the electrodes, it eliminates the "cold spot" around the electrodes. Also, arc tubes in MH lamps are shorter than MV arctubes of the same wattage.

Pulse start metal halide lamps

Pulse start lamps are very similar to their probe start counterparts, but lack the probe. An ignitor, the same as HPS lamps, is used to strike the lamp. Their minimal start temperature is -40C (-40F), instead of the -30C (-22F) of probe start lamps. Recently, pulse start MH lamps became the only other choice than HPS in streetlights. Fortunately, probe start MH ballasts are still available for replacement because of the UL safety violations encountered when converting a probe start fixture to pulse start.

New variety of MH lamps: Ceramic Metal Halide

Developed in the early 90s, these lamps use an arctube made of transparent ceramic, the same used with HPS lamps. CMH lamps are obviously pulse start, so they use the same ignitor than HPS lamps. Their salt mix can also be more complex. Rated life of CMH lamps is usually higher and their hot restrike time is much shorter. They strike back after 5 or 6 min, compared to the 15 min. of standard MH lamps.

*** BALLASTS ***

Now, let's talk about ballasts. Many types of ballasts exist and they are all different. Also, they have weird names, like CWA, CWI, HX, what are those? Well, You're going to see it in just a minute! ^_^

Before we begin, some basic things. First, a ballast is a transformer. It's very important to know, but I'm sure that most of you, even if you don't know anything about lighting, know this. ;-) Second thing, a basic transformer has two coils, one is a primary and the other, the secondary. That's another important point since ballasts differ by their coil configuration!

The HX - High reactance autotransformer



That ballast is an autotransformer, whose primary and secondary (if any) are electrically connected together.  Basically the secondary acts as a step-up to give higher voltage than line voltage.
HX ballasts generally give a voltage of up to 125 - 130V. They also give an higher voltage "kick", called the Open Circuit Voltage. The OCV is very important to start any gas discharge lamp (including HID, but also fluorescent lamps!). The OCV, for most mercury lamps, is generally of ~225V. Unlike chokes that need a circuit breaking to give a higher voltage (like those made by fluorescent starters), HX ballasts can generate a higher voltage by themselves in case of an open circuit. The voltage rise is generally sufficient to start a mercury or metal halide lamp.

In Europe, where line voltage is 230V, autotransformers are not necessary. Simple reactors can drive most HID lamps.

The CWA - Constant Wattage Autotransformer



This ballast type has been developped later than HX ballast. Physically the only change is the addition of a capacitor in series with the lamp. Electrically the differences are more important. First it raises the power factor from 50% to 90%. This reduces the current draw and allows more ballasts to be used on a circuit. It also reduces the effect of line voltage variations, voiding the X3 amplification of variations. Roughly a 10% voltage drop would lead to 30% voltage drop at lamp on a HX ballast.

The primary can be configured as an autotransformer with multiple voltage taps. and the secondary acts as a step-up as usual. On HPS CWA ballasts, the capacitor is sometimes placed between the autotransformer and the step-up. Due to the incompatibility of standard CWA ballasts with HPS lamps they have been modified to allow a more constant voltage.

== NEXT PART BELOW ==

[Lil'Cinnamon]

Wonderful info Vince! I learned lots of new things in this post!

[gailgrove]

Very nicely done, except it has nothing about LPS

[Jace the Gull]

LPS isn't a High Intensity Discharge Lamp ;-)  It's actually more in the fluorescent light category. (or low pressure mercury vapor)

[Vince]

Yeah, I could always add LPS though, or start another article about low pressure lamps

So, here's the next part:

The CWI - Constant Wattage Isolated



The CWI ballast existed even before CWA, it's actually the SECOND oldest ballast. While the CWA uses an autotransformer coupled with a step-up coil, the CWI ballast is a simpler transformer with isolated primary and secondary. A cap is wired in series as well. This type of ballast may have to be used in specific conditions to meet the Canadian Electrical Code.

The Magnetic Regulator, or Regulated Lag ballast



The Magnetic Regulator is probably the most complex ballast. On this type of ballast, the cap, instead of regulating the power electrically (directly), it regulates the power by the magnetic circuit of the ballast! This kind of power regulation leads to an exceptionally steady lamp voltage. This ballast is quite useful when heavy loads (like an A/C) are connected on the same circuit the HID lamp is connected on. Magnetic Regulator ballasts can easily correct line voltage variations, at the cost of higher watt losses.

[gailgrove]

Jace any where I have looked they say LPS is a HID, including the manufactures 

[Medved]

@gailgove: Marketing-wise it is, indeed, closer to HID's (due to their application - lot of lumens, but cheap in energy use, color quality not a factor).

But technically it is low pressure discharge lamp, exactly the same as fluorescents (large, long surface of low brightness, discharge occupy the whole cross-section of the arctube, touching it's walls, low current density in the discharge,...).
So from technical perspective, the LPS is not considered as HID.

[Vince]

*** CONCLUSION ***

I hope this article have answered to a couple of your questions about HID lamps. This article is mainly intended for beginners who just got into hobby lighting. There is of course a whole bunch more theory about lighting, and I could make other articles about other lighting sources in the future if members show interest to it.

But to conclude this very article, we could summarize the three major HID sources in this list:

MERCURY VAPOR

PROS:

- Simple gears, easy to use.
- Impressive reliability
- Choice of soft, white, non-glaring light (/DX, /C, etc.) or bright, crispy, high-contrast light (/A).
- Specialty applications (/Y, /BLB etc.)

CONS:

- Lamps are considered toxic waste (Hg), special disposal required.
- Lamps dim out over time, especially with newer, cheaper lamps.
- Relatively inefficient, compared to other HID sources, but still as efficient as fluorescent tubes and more than incandescents.

HIGH PRESSURE SODIUM

PROS

- High efficiency, reaching 140 lm/W with highest wattage lamps.
- Spectrum of the lamps raises contrast and overall visibility (perfect for hazardous roads or other similar applications)
- HPS systems start at lower temperatures.

CONS:

- More or less monochromatic light leads to very low CRI.
- Ignitor adds a part to the ballast, making it more likely to fail
- Cheap HPS lamps can fail much earlier than their cheap MV counterparts. They are prone to sodium leak or premature voltage rise (early cycling), those problems are inexistant with MVs.
- Cycling lamps left for too long can damage some ignitors.
- Lamps are more sensitives to line voltage variations.

METAL HALIDE

PROS:

- Exceptional quality of light.
- Various configurations available (probe start, pulse start, CMH).
- Wide choice of whites (2700K, 3500K, 4100K etc.)
- Development not yet done, improvements are still to come.

CONS:

- Ballasts prone to the same problems as HPS, though probe start ballasts can be as reliable as mercury ballasts.
- Some lamps can explode at EOL. Though rare, type E or S lamps can indeed burst is pressure rises due to various reasons.
- Cheap MH lamps can have poor last mixes, making their color ununiform from a lamp to another.
- Color shift through the lamps' life can appear annoying when a large number of lamps is used in a given area (tint difference from one lamp to another).
- Metal halide equipment is still kind of expensive.

[gailgrove]

Hang on Vince, HPS are not monochromatic your thinking of LPS, there is still colour rendering under HPS 

[Vince]

Edited I added "More or less" LOL.

[Medved]

I think there is one other mistake:
"Lamps sensitive to mains variation": It are not lamps, what are sensitive, but their ballasts. On regulated ballast they would be fine as well, same as e.g. MV. The problem is, ten the cheapest regulated ballasts is a form of constant current source, what is fine with MV's, but make the HPS thermally unstable. However the "Mag-Reg" ballast style would be fine, as this type decouple the load characteristics from the line regulation property.
And these days electronic ballasts do the regulation job even better, however are more expensive (and i do not mean those cheap LOA-like high frequency lamp killers) and quite sensitive (overvoltage spikes, temperature)

[Jace the Gull]

I think there is one other mistake:
"Lamps sensitive to mains variation": It are not lamps, what are sensitive, but their ballasts. On regulated ballast they would be fine as well, same as e.g. MV. The problem is, ten the cheapest regulated ballasts is a form of constant current source, what is fine with MV's, but make the HPS thermally unstable. However the "Mag-Reg" ballast style would be fine, as this type decouple the load characteristics from the line regulation property.
And these days electronic ballasts do the regulation job even better, however are more expensive (and i do not mean those cheap LOA-like high frequency lamp killers) and quite sensitive (overvoltage spikes, temperature)

I am not exactly sure about electronic ballasts, They seem more prone to failures in a way....I have a 39 watt CMH and a 20 watt CMH ballasts that don't work... (the 39 watt worked before) and they are electronic....they seem pretty sensitive....

[Medved]

There are two different things:
Ballast reliability (so the risk, then it simply burn out), where electronic are generally lacking a lot (due to many reasons, but mainly the poor design and manufacturing quality - and effort to compensate the cost of higher complexity)

And it's supply regulation capability: Ability to compensate for mains voltage variation and keep lamp power not dependent on the mains voltage. Here electronic ballast excell, mainly due to the very high gain and accurate and strong regulation feedbacks.

Generally electronic ballast can easily handle quite wide voltage range (+/-30% or more, compare to +/-5% of the electromagnetic HX autotransformer for staying in the tolerance of output parameters) and are able to correctly handle the departure from this range.

What they are lacking is the immunity against really excessive overvoltage spikes, even short ones.
During such spikes magnetic ballasts send huge current into the lamp but as these are quite rare in the system, both lamp and ballast can handle it - as both have quite high mass to swallow the energy to not cause severe temperature rise. Moreover magnetic ballasts are quite lossy, what make the excessive power dissipation relatively not as high

But when you exceed breakdown voltage of a semiconductor device, their low impedances cause the current and mainly the local power dissipation to rise a lot. And as the electronic is normally very efficient, it is designed to handle only quite limited power dissipation, so the spike-dissipation is relatively very huge. Moreover most semiconductor devices are of very small physical size (e.g. 600V 10A FET is about 5x5mm die, where the active transistor is only ~100um thick), so heat up very quickly (in less then us it may reach the run-away temperature and then melt).

So it is not the 150V present in the mains for half a hour (instead of 120V; that would normally fry any magnetic ballast), what kill the electronic, but the 1kV 10us wide spike (no issue for magnetic at all). Such spikes originate from high current switching and arcing, mainly in some failure modes in the power network (e.g. fuse break an arc, that formed in failing incandescent lamp) and it is mainly their rarity, what causes troubles - as ballast seems to be working fine when testing, but then blow up "with no reason" after installed to final place...

So electronic need protection circuits, mainly against these pulses. Best protection is a filter (even out these short pulses, so these do not disturb the ballast operation) But such protection is a bit more complex to design (e.g. to not cause overvoltage by itself as a response to power ON; it is frequent design error), but usually these filters have quite high losses (in the order of 1% of the transferred power) - these are seemingly small, but as the whole elctronic is very efficient, it become quite significant part of overall ballast losses and eat-up from total power loss budget, so the main ballast has to be even more efficient. On good quality ballast design half of the total losses are on this filter - what mean for "90%" overall efficiency rating the main ballast itself can have only 5% losses, other 5% are reserved only for the input overvoltage spike filter.
Due to these reasons these filters have only limitted overstress energy capability, very often too low, as the 1/2 of the total loss budget seems to be too high (= they want to make the ballast cheaper, so reserve more then only 5% for the main ballast)

Then for higher energy pulses (what are very rare), the target is to limit the ballast damage only to the input fuse, so it is very easy to repair it even without extra knowledge (replace the fuse). Here the most frequent design error is insufficient breaking capacity of the used fuse (so the fuse keep arc in it, so the high current flow too long to damage the electronic). Usually the main cause (= severe mains overvoltage) is not taken into account for the fuse specification, so n case such energetic overvoltage appear, the whole front side of the ballast blow up, because the fuse is not able to protect it.

[Jace the Gull]

from where I used to work...I noticed the fixtures with programmed start ballasts seemed to have lower ballast failure rates than the instant start ones....plus the programmed start ones has EOL protection too....

The programmed start ones were motion detector controlled while the instant start (most of them) turns off and on when the store opens or closes....(some instant start were motioned detected (which the detector was added later) (and the lamps don't last long!)

[Medved]

I think this would be related to the fact, then programmed start are more expensive anyway, so there is more budget room to make more robust protection. Moreover the lamp EOL shutdown make quite huge difference in ballast component life, as the "permanent ignition mode" mean very severe stress on them.