The Overhead Streetlights of Local Streets, Part 1

Previously I wrote about streetlights, with a special emphasis on the freeways since these were about to be converted to LED. Now Xcel Energy has announced 100,000 local streetlights in its service area will be converted, so it is time to look at some of these.

The NEMA Lights

In the early days, every manufacturer had it’s own design. Here’s a beautiful example from Lexington, MA near Battle Green.

Street Light, Lexington MA

Streetlight near Battle Green, Lexington MA

The pole is decorative, but the luminaire itself, although beautiful, is simply designed to be functional. The white porcelain protects the metal, and the waves add strength to the reflector and help throw light to the sides. This was known as the “radial wave” design and these are highly desired by lighting collectors.

In 1941 all the manufactures got together and developed the NEMA (The National Electrical Manufacturers Association) standard. NEMA luminaires standardized the placement of two latches to mount the optics into the rest of the luminaire. This form had two advantages.

  1. An optical assembly for any manufacturer would work on the luminaire base for any manufacturer. If a globe got smashed, a city could just take a Westinghouse globe out and mount it to a GE luminaire, without needing a different stock for every different manufacturer on the streets.
  2. Optics could be changed, i.e., if a new more efficient reflector was developed, or more light was desired and thus a bigger globe was needed, they could just go out and pop in a new, bigger lamp (since incandescent lamps didn’t need specialized control gear for each different wattage) and new optics on the existing luminaire.

Originally there were a wide variety of NEMA optics– “Radial Wave”, “Half Moon”, “Admiral Hat”, “Gumdrop”, “Teardrop”, “Bucket” to name some. Here’s a NEMA “Gumdrop” I own. It has a 327 watt, 4000 hour streetlight lamp in it.


The “Teardrops” were the same idea, but the refractor extended down farther  to accommodate larger lamps for major streets.  This style is now very popular with those gaudy, atrocious “fake history” lights I’ve seen in Chicago, on Lake Street, the Minneapolis parkways, and numerous other places. I get that as much as I like the old mid-century modern lights the installation is now a maintenance nightmare and need to be replaced, but when the teardrops first went up it was the style of the time; they weren’t putting in fake chimneys to imitate Victorian or Federalist style lighting. There’s plenty of designs that while modern, are still decorative.

And an “Admiral Hat”, using an economical open reflector.


Streetlighting Technology

There were some differences in the lamps used in early streetlights. They cared more about life than efficiency, so they were under-driven to the point of having a decidedly yellow cast. And they cared about lumen output rather than watts, so they have even lumen ratings, like 2000 and 4000, and weird watt ratings, like 327.

Rather than line voltage, many incandescent streetlights, in a blocks-long version of Christmas lights, were run in series, the supply voltage was around 2000 volts with each lamp running at 50 volts. This controlled voltage drop for the enormously power hungry incandescent lamps. Early models had porcelain insulators but improved wire insulation later enabled them to be wired through the luminaire neck like paralleled lights. Obviously you didn’t want the street to go dark if a lamp burned out or was removed, so these had “Jones Sockets”, so 1) If the bulb was removed, a spring in the mogul socket shorted the contacts together, 2) the mogul socket itself fit into another special socket. The receptacle in the base would short if the socket were pulled out, and in between the prongs was a clay disc to short out if the lamp burned out.


To relamp, you would pull the socket out, replace the shunt, screw in a new lamp, and push the socket back in, all without affecting the rest of the street. My light was originally a series string so I pulled out the shunt and replaced it with a piece of credit card wrapped several times with electrical tape.

With the coming of High Intensity Discharge (HID) lamps (mercury vapor and it’s relative metal halide, and high-pressure sodium vapor) these were much more efficient, but required other components in or near the luminaire to operate. The problem is that they appear as a dead short when run directly off the mains voltage, so a device to limit that current and reduce the voltage is needed. Originally this was just a ballast, but with newer HID lamps also requiring ignitors, the term “control gear” is sometimes used.

The simplest ballast is a two-wire choke ballast, a coil of fine wire wrapped around an iron core many times. This both limits current and reduces the voltage by a substantial, fixed ratio. More complicated ballasts are 4 or more wire autotransformer types. They are more flexible in input vs output voltage and often have multiple taps for different input voltages. When designing the lamp plus ballast combination, there is a trade-off vs the optimal arc voltage for the lamp vs what voltage is obtainable from the mains with a simple choke ballast. Mercury vapor lamps were designed in the UK with 240 volt mains. With an arc voltage around 100 volts, they require an autotransformer in the US on 120 volts, although many older streetlights used chokes with a 240 volt supply. Here’s a 240 volt choke mercury ballast from my 1960s streetlight

240 Volt only choke type ballast

240 Volt only choke type ballast

High pressure sodium lamps were designed in the US and the smaller ones (150 watts and less) can use choke ballasts with an arc voltage of 55. That would be inefficient in Europe, where 120 volts is unavailable, so they designed their own sodium lamps for 100 volts (and for good measure called them “SON” lamps and used smaller, higher strength glass envelopes). Larger sodium lamps  are designed to use autotransformer ballasts only to have the most efficient arc voltage for each size lamp regardless of mains voltage and the additional cost of an autotransformer ballast is less of an issue.

Even though there’s little resistance when a lamp is hot, there is significant resistance when it is cold. Mercury and older metal halide lamps use a starting electrode; when the lamps starts, the arc forms between two electrodes very near each other on one end, and eventually the end heats up enough to form the arc from one end to the other.  Newer lamps use a several thousand volt zap to initialize the arc, generated by the ignitor.

Here’s the gear of an American Electric model 315 cobrahead. The  ballast is at lower left, the ignitor is the circuit board at the lower right. At top is a capacitor for power factor correction. This is important to utilities and businesses (who the utilities charge if it’s not corrected) but not strictly necessary for lamp operation and is omitted in inexpensive fixtures for home use.


The most modern ballasts use switching power supplies rather than heavy magnetics to produce the desired current and voltage. These are now standard for fluorescent lamps, but were never adopted for HID use in the US except for interior ceramic metal halide fixtures.

As a side note, Europe developed “White SON” for indoor use, which drives high pressure sodium hard enough that it produces a pleasant incandescent like white color common in shops, but it was never common in the US and even in Europe was less common than ceramic metal halide. White SON is more expensive, less efficient, and shorter lived then regular high pressure sodium so it was never used for street lighting.


White “SON” lamp, note European style lamp with small, high strength glass instead of the huge bulging shapes of US lamps.

NEMA Lights Today

Back to our discussion on NEMA lights. Some of the previous optics attempted to direct the light in specific directions, but the “open bucket” refractor, which blasts light equally in all directions, is the only one still in common use. However, Xcel has recently introduced a dark sky friendly full cutoff reflector for new installations.

Here’s a NEMA light in St. Paul. Uniquely, Xcel supplies the power and the mast, while St. Paul owns and maintains the luminaire. Since they blast light in all directions, they tend to be on very long poles to have them towards the center of the street. With the newer, more efficient lamps, St. Paul and many small towns still use NEMA heads for streetlighting the much smaller lamp size of smaller and medium sized HID lamps enabled them to still fit even with the required control gear.

St Paul NEMA "Bucket" light

St Paul NEMA “Bucket” light,

A pair of old and new Xcel NEMA area lights, these are a design by GE called the “Power Bracket”. The much shorter length of HID lamps opposed to the large incandescent lamps for which NEMA luminaires were designed enabled the ballast to be stacked on top in the conventional design, but these moved it into short, integrated arms, which eliminated heat rising and building up around it.  More or less the same model has been in production for many decades.

Old and New GE NEMA Area lights

Old and New GE NEMA Area lights. The one on the left is a 400 watt lamp which is the largest that the NEMA form factor could accommodate, which was enough for just about all purposes.

Also notable are what are pejoratively called “yardblasters” by lighting enthusiasts; cheap fixtures from big box stores that look somewhat like true NEMA heads equipped with a bucket, but the refractor is bolted on, they’re made of stamped metal,  and the ballast is usually sketchy; it’s the buzzing mercury light on your neighbor’s garage that’s shining in your window.

Fluorescent and Induction Street Lights

The ubiquitous fluorescent tube was also once used for streetlighting. Municipalities were eager to discard the old fashioned looking and inefficient acorn lanterns and “teardrop” overhead lights in their downtown districts in favor of these bright, streamlined luminaires. However their bulk, lack of control over the broad diffuse light sources, and poor cold weather performance eventually led to their extinction for streetlighting. All the major streets in Minneapolis were once lit by fluorescent luminaires but these were all gone by 1990 and no lighting enthusiast knows of any in service or in private collections.  The last ones I saw in the field were in Waconia in the late 1990s.

Fluorescent Streetlights in Spooner

Fluorescent Streetlights in Spooner

A modern development has been induction lights. These basically use a high frequency magnetic field to put energy into the lamp. They’re very efficient, more compact than standard fluorescents, and last a very, very long time because they lack a starting electrode, which is the usual failure point on conventional fluorescents. Like fluorescents, they give a very diffuse, omnidirectional light. They’re good for lanterns, good cold weather performance, less good for overhead lighting where you want it focused down. There’s a test installation in the lanterns on 46th street west of Hiawatha, and a few cobraheads near the river in Hudson. Had LED technology not progressed we would be seeing a lot more of these.


150 watt Jersey Series Induction Light- NJ made extensive use of them.

Although NEMA luminaires were able to accommodate HID lamps, soon came luminaires designed from the start to take advantage of them: the “clamshell” which evolved into the cobrahead. More about those and a spotter’s guide to cobraheads in the next article.


About Monte Castleman

Monte is a long time "roadgeek" who lives in Bloomington. He's interested in all aspects of roads and design, but particularly traffic signals, major bridges, and lighting. He works as an insurance adjuster, and likes to collect maps and traffic signals, travel, recreational bicycling, and visiting amusement parks.

11 thoughts on “The Overhead Streetlights of Local Streets, Part 1

  1. Matt SteeleMatthew Steele

    Wow thank you for writing this series. I love all the detail.

    When I was a young pup, I had a (probably unhealthy) fascination with NEMA lights. Growing up in the burbs, there was something about the aesthetic that seemed so industrial and legitimate compared to the plastic crap I was used to seeing. I used to ask my parents to drive different routes on road trips around the region, just so I could see the infrastructure of different towns. I kept a list of towns where I spotted NEMA lights, but I don’t know what happened to that list.

    I had another experience with these lights recently, as I was recently helping a family friend who had property damage at their cabin due to a F2 tornado that hit last month. They have a privately-owned NEMA light (nicer than a “yard blaster”) on its own breaker, and the reflector and luminaire was basically tugged off the housing and held on by the supply wires. It was surprisingly fixable and un-damaged after taking a hit from a tornado.

    Regarding Xcel’s plans to convert 100,000 local streetlights to LED… I assume this will be a new LED luminaire on the on the existing wooden pole and mast arm? I’ve been asking multiple people in the city about this, to see if there’s a way to get some sort of credit instead towards the Minneapolis-owned “pedestrian scale lighting.” They haven’t been able to figure that out, and I haven’t been able to find out more info from Xcel either. While conversion to LED luminaires is a drop in the bucket compared to the expense of ped scale lighting implementation, I still think it’s worth having a discussion to see if cost avoidance from one effort could be applied towards investment in the other.

  2. Monte Castleman Post author

    LED lights are outside of the scope of the series (for now), but last weekend I went down to take a ride on the Root River Trail and check out the last circular “Walk” indication in Minnesota in Austin, and noticed that the places that used NEMA area lights as streetlights are using LED area lights, rather than convert to full-cutoff streetlight pattern cobraheads. Either specialized LED area lights or the American Electric ATBS Autobahn lights (one of two models Mn/DOT uses, these are the round ones) equipped with an optional bucket refractor that converts them from streetlights to area lights.

    As far as Minneapolis cashing out, I’m not surprised Xcel hasn’t been to receptive. Besides the unusual nature of the request, if they invest the money in LED streetlights they’re going to make it back and more in the next few years. If they pay a percentage to Minneapolis, that money will not be recouped and what’s more they’re not getting revenue from streetlight revenue any more.

    The other problem is we’re not talking about a relatively large amount of money. With the scale of the project it’s only a couple of hundred dollars in materials and labor to slip on a new LED luminaire, so a percentage of that as a cash-out isn’t going to go far towards a nice lantern installation.

    1. Sean Hayford OlearySean Hayford Oleary

      Agree with this assessment. And if Xcel wants to do large-scale conversion, it would be an annoying maintenance burden to have one little section not converted in any otherwise all-LED area. Unless the neighborhood has the lighting project already funded and scheduled, it seems silly to delay the upgrades for what might happen in 2 or 5 or 10 years.

      And as Matt notes, a drop in the bucket compared to the cost of replacing with new post lights.

      Related idea for a future lighting post: why do those Minneapolis street light conversions cost so much? Is there a cheaper way to still get that quality of lighting?

      1. Matt SteeleMatthew Steele

        How exactly is the standard set for Xcel’s lighting?
        – How do we determine lighting coverage?
        – How do we determine aesthetic minimums?
        – Who determines the rates charged to local governments?

        As you can tell, I’m rather disappointed by the Xcel lighting in my neighborhood.

        1. Monte Castleman Post author

          There’s state and national standards for street lighting as far as foot-candles / lux and uniformity, which are essentially identical. The City of Minneapolis has it’s own, which are similar but add standards in regards to glare. There’s no enforcement so cities can choose to do what they want. The Minneapolis lantern installations meet their standards for foot-candles and uniformity. The newest “ghost lanterns” also meet their standards for glare. Xcel wood pole installations meet none of them.

          Richfield decided to do install lights based on what they though looked cute rather than what meets standards. At about 40 feet or so a pedestrian scale light can’t adequately light a street, (and I took my light meter out and confirmed they did not) so you need high mounted lights to supplement, like Minneapolis did on Penn Ave just north of Crosstown.

          It wouldn’t look very pretty, but I suspect adding a third wood pole per block as well as using LED luminaires would meet standards at an additional cost of of $12 a month per block

  3. robsk

    Great article.

    My dad used to call the orange (or pink?) sodium vapor street lights “Ghetto lights”. Said they were originally used to subdue rowdy neighborhoods. Does anyone have any insight about this?

    1. Monte Castleman Post author

      Nope. Orange lights were used because they are much more efficient than the old bluish mercury vapor lights. There has been an experiment with pure blue streetlights to reduce crime in Edinburgh; it works but it’s now known if it’s something intrinsic to the color or if it’s because the unusual nature makes criminals think someone is being observant.

  4. GlowBoy

    Very informative article. I learned quite a few things I didn’t know. I’ll be interested to see if the next story goes in depth on low vs high pressure sodium lights.

    And sorry I’m going to talk about LED streetlights too. Portland is rapidly converting all of its residential streets to these. My neighborhood there got switched over last year, shortly after I moved away. Although I appreciate the improved efficiency and maintenance costs, I think there are some concerns:
    – More light spill than the old cobraheads, much of which is going into people’s bedrooms at night.
    – The light going into people’s bedrooms is much more bluish-white than before. There’s increasing concern about disruption that higher color-temperature lighting may be causing to people’s sleep patterns (I have recently enabled Night Shift on my iPhone, by the way). I believe even the AMA is recommending streetlights have a color temperature of 3000k or lower (the Portland ones appear to be at least 4000k).
    – The fixtures are much brighter than before, which isn’t necessarily a good thing even outdoors, because it makes the spaces between the lights seem relatively darker. Walking down the street at night now, the dark areas seem REALLY dark, and I think it’s now more likely than ever that you’ll trip on an unseen sidewalk crack.
    – The light comes out of the fixture at 8 points (8 separate emitters or clusters of emitters, a couple inches apart) instead of a single larger source, which means everything casts 8 shadows instead of one. I can’t think of anything really wrong with this, but it’s kind of weird.

  5. Jim Terry


    Interesting info about street lighting. I am a streetlight geek, and I co-own a company that makes transformers for series lighting systems. I just wanted to make a few observations about your article that might be interesting.

    – Incandescent lamps and wattages. They were mostly 120V but there were (are?) 240V versions available as well. The oddball wattage is down to the lumen output. Incandescent lamps were specified based on output, not wattage. And, for example, your 4000L incandescent lamp at 327W has a relatively short life compared to a 6000 hour model, which was likely 405W. Which also contributed to slightly warmer color temperature.

    – Series circuits are constant current, not constant voltage. *Most* series circuits in the US are 6.6A, but there are also some 20A in parts of the Midwest. The voltage of a series circuit is dependent upon the size of the regulator (moving coil transformer that powers a series loop). Anyhow, a 30kW regulator, which is usually the largest you’ll find in the US, has an operating voltage of just over 4500V… so bigger systems were higher voltage. The voltage drop across loads on series circuits vary based on the size of the load… so the voltage drop might be different for different fixtures. And the highest voltages on the system would be seen at the lights closest to the regulators… the lowest voltage being at the middle of the system. Just curious, are you aware of any of these systems in your area?

    – Fluorescent street lights – the last place I am aware of fluorescents still being used as street lights (as they were in the 50s) is Waterloo, Wisconsin. I was through there in June and the lights were still there as of then. It’s worth a look! 🙂

    – Induction – it’s not “new” technology as it was marketed in the past 8 or 9 years. It’s actually been around and commercially viable since the early ’90s. Philips developed the QL lamp, which was shaped similar to a standard light bulb… and Osram came up with the “Icetron” design, which is what is pictured in the 150w Jersey fixture pictured, which is from US Lighting Tech. Manufacturers liberally claimed these were 80 lumen/watt (LPW) sources… but the reality is really in the low 70 LPW range.
    So what made induction feel so new in the mid-late 2000s? Osram and Philips’ patents expired. So anyone could copy the technology, and China jumped right on that. So how did induction companies pass off 73 LPW as being more efficient than HPS (90-125 LPW)? They jumped on some research, which indicated you needed lower levels of cooler “white” light vs. amber HPS light to aid peripheral vision. The factor for that was called out by the IES (Illuminating Engineering Society) in the document TM-12. The induction companies seized on this information and claimed that they could have equivalent light output vs HID with greatly reduced wattage. They lumped all HID together as well, when the research only indicated improvement when going from amber to “white” light. In reality, the net result was a reduction of actual light levels of roughly 70% in places induction was used to replace HPS. And I haven’t even touched on fixture efficiency, which puts even more nails in the coffin of induction lighting.

    To address one of the comments I noticed about the AMA report that’s been making the rounds – the AMA did not reach out to the industry and thus does not contain recommendations that have been endorsed by the IES. At best, the AMA report is an incomplete and unreviewed set of recommendations; at worst, it may be written with the interest of one specific company in mind – a company which is launching a new product that uses a different type of LED. While the AMA article is interesting, it’s got a lot of holes in it… I wouldn’t go quoting it as science – since they didn’t even bother to ask the standards bearer of lighting to review or comment prior to publishing it.

    Anyhow, thank you very much for your informative article. Look forward to seeing more!

Comments are closed.