It's mercury lamps, the starting reliability is not affected by distance

I have a wild guess, but i dont think this was the reason, or even a consideration. A pole in the median is more likely to be hit by cars, and not having a ballast can up there is one less thing that can fall down into a car in case of an accident (into the windshield, etc) and cause an injury

The outstanding performance of HPS for outdoor lighting does not come down only to Lumens, but to our eyes and brain ability to complete the full image from an otherwise very unbalanced input (a scene lit by non uniform, near monochrome, orange light)

(Incidentally, those is the very same reason why LED outdoor lighting achieves the exact opposite for our eyes, but might actually perform well for a video camera)



Dynamic range :

Our eyes have very wide dynamic range. Every time you look from an outdoors location under bright sunlight (100000 lux and higher) into a dimly lit lobby of a building through an entrance door (can be as low as 100 lux), you are seeing at the same time two areas with 1000x difference in the average light levels, and able to perceive what is it that you see in both of them

Under HPS (or really pretty much any) outdoor lighting, there are areas under the light which are brightly lit, and areas away from the light that get dim lighting. HPS does not ruin night vision, therefore it allows us to make use of huge dynamic range

Video cameras of the 80s have awful dynamic range. Even today's cameras are a far cry from the capabilities of our eyes. It is possible that filming under HPS would capture a scene which is well lit or even clipping directly under the light, but have an abrupt cutoff area beyond which it is unable to capture

Under Mercury light, the light in the area to be filmed have possibly higher uniformity and lower intensity, allowing the camera to see a wider area correctly within a single brightness setting



Color rendering :

The tiny amount of blue and green in HPS spectrum is all it takes to provide us with full color vision

Our brain understands that the light itself is of orange color, and extrapolates from the little blue/green the eyes see to understand what each surface in the scene would look like under natural light. Then, it is matched to things we are familiar with as reference

A video camera does not do any of this. The image is quantified into 3 color channels of an image sensor (which sensitivity spectrum is completely different from any cells in the eye), each of them is stored on and read from media (with or without correctly matched gain), and finally displayed on 3 color channels of a screen (with light spectrum that have nothing in common with the original spectrum at the scene, or even with the camera sensor)

Now, instead of the 589nm peak of Sodium light + little more info in other areas of the spectrum, we are presented with a scene with "the same colors" but is actually recreated either with wide emission spectra of Green and Red phosphors (CRT), or is filtered down from the emission of Fluorescent or LED (LCD), or UHP Mercury lamp (in the cinema), made from data already corrupted by the camera

Our brain won't know what to make of it, and either won't, or won't successfully, do what it can with the real life scene

And provided the awful dynamic range, the blue and green data will probably be nearly fully lost already at the camera anyway

With Mercury light which is white to begin with, there is much less to go wrong here

@Caroline
That is quite the system! Very impressive. How low is your typical wall voltage (if you don't mind me asking)? There has to be some sort of rule in place to govern how low it can legally be in a supply.

Nine months without power is outrageous! That must have been fun...

Your power situation sounds like a nightmare!

I experience it pretty much constantly, brownouts, not drunkards crashing against utility poles, that's another fun story.

Had to come up with a way to step-up the power for the lights because they're all incandescent (regular and halogen) and choke fluorescents for the kitchen, I ended up adding a subpanel next to the fusebox that takes care of that by using autotransformers and a custom control board, there's an override contactor that's supposed to close and bypass the circuit when the voltage gets past 230, but in the almost 4 years that system has worked it never happened, the xfrms can only bump 20% of the input V, they're made with an absolute low of 200V in mind, but ours can go much lower, to around 170 on a hot summer day, that if nothing blows up on the supply side.


The only limitation is it's only for the lighting circuits, the sockets remain at low voltages so that's why all of our desk/table lamps are low voltage driven by electronic power supplies, they can tank the low input and output stable 12V, if they were straight to mains most lamps would be really dim and flickery.


The only crash incident involved a drunkard smashing his truck at 180km/h against an utility pole nearby, it ripped the wiring on both directions and we were out of power and phone for 9 months. Generator can't run constantly so it was only to keep the fridge working twice a day, no landline = no internet so I didn't even use the computer. No such thing as same-day fix here, and power is state-owned.

Snappy strip itself has higher stresses, so higher risk of fatigue cracking...

Some well-made glow starters actually snap because of momentarily sticking the electrodes, giving a well defined preheat time on a first try.
 

AFAIK, LPS lamps were not especially popular in USA compared to Europe. May be no one really bothered to optimize ballasting for somewhat niche product. You can also consider reduced service costs vs saved electricity cost.

Sure, poor ballast efficiency very much likely driven system efficiency of American LPS luminaries below of what is achievable with HPS lamps.

@LightBulbFun
Very interesting, I don't think any choke ballasts were ever made for LPS in the US but I may be wrong. Are there any US-based ballasts for LPS lamps that use ignitors that you know of?

@RRK
But isn't the whole point of LPS to have higher efficiency than HPS? Unless we are talking about the pre-HPS years here. I thought that was the whole LPS mission statement, to be the most efficient lamp. Were they still more efficient than HPS even when considering these high ballast losses?

Lamps like HPS and PSMH have relatively low open-circuit voltages because they use ignitors to strike the lamp.

MV lamps use an OCV of around 240V to strike, which is around as low as you can get considering the running voltage of ~130V. I hear that the OCV has to be around double the running voltage for an arc to be stable. MH is very similar (around 300VOC).

But with LPS this doesn't seem to be the case. For example, a 90W SOX lamp has an arc drop of 112V. This means that this lamp can run with an OCV of around 230V. But it can't strike on voltages that low, which is why the ballasts output ~480VOC. But isn't this wasteful? Why not just have a ballast with 230VOC with a ~1kV ignitor? I feel like this would be easiest to manufacture and also more efficient.

Any thoughts?

Almost certain, since electricity is relatively inexpensive in USA, manufacturers preferred extremely robust, but really inefficient high-OCV leak transformers. Back in 1970-1980s semiconductor circuits were still considered a big step back in terms of reliability.

@LightBulbFun
Yes! Exactly like that, I didn't know those existed. Why aren't they all like that?

you mean like the

WRTL Thryactor?

https://www.lighting-gallery.net/gallery/displayimage.php?pos=-184846

or the Venture SZ090K245

https://ledspares.co.uk/products/sz090k245

or the


ok ill stop now


most modern SOX ballasts *are* like that.... (for those with arc voltages compatible with 240V OCV, those of higher arc voltages are run on lower-loss CWA setups sometimes)