As I saw in Hawaii (unfortunately only through Google street view, I cannot afford that in person ) some streetlights are on during daylight hours.
I suppose it's because of the photocells on the top of each lantern (?) - some of them are broken.

In our country, the streetlights are usually started centrally (per street or per block etc.)
I wonder what are the advantages of those separately started streetlights? Preventing the "power shock" thanks to the fact that each light starts in a different time?

I can imagine disadvantages (photocells dirty or covered by birds) - i.e. unwanted starts.
And how electricians check if HPS do not cycle? I suppose they don't wait at each light with photocell covered/bridged if it starts cycling or not... 

The advantage of the individual photo control is not in the lantern itself, but in the infrastructure behind:
In most (older) installations the fixtures share poles with electricity distribution wiring.
Centralized control (like most common in the Czech republic) require separate cabling to feed those fixtures.
But if each lantern has it's own control, it is then hooked to a permanent supply. Well, the permanent supply is on the poles without any extra wiring: The fixtures just get hooked onto the wires of the electricity distribution system, co no separate wiring needed. So there was quite huge investment savings by doing that: The extra wires cost way more than the photo control cells (usually just a CdS photoresistor controlling power to either thermal or electromagnetic NC relay).
The "centralized" control has it's advantage as well: You may easily control the operation over the night, so e.g. switch it OFF after midnight to save power, when practically all people are already home (in many villages, unless they experienced problems with crime at that time). Somewhere this eases the installation of dimming systems: You suffice with just one dimmer unit (an autotransformer with many outputs allowing to reduce the voltage by small steps to about 80%) for the long stretch of lanterns and the lanterns are then equipped by just ordinary ballasts (quite common in Brno on main roads - after ~10PM or midnight the light is dimmed, about 4AM..6AM the brightness is eventually returned back to full power), so allows an economic way to save power, when the road lights do not have to overpower the car headlights from heavy traffic.
And the central control was used here for yet another purpose: The streetlights are not a negligible load, so their exact turn ON/OFF moments were used to shape the mains network load balance, easing the network management.

For the electricity consumption calculation are not much used the real electricity meters, but just a fixed payment based on estimated energy consumption (calculated from the number of lights, their power input and average operating time). With individual controlled lamps hooked to the main distribution network it is even not possible in an other way (unless you will place a meter to each lantern - quite ridiculous, as the meter alone consume about 5..10W), the central controlled systems get their power from many places as well (the signals are spread via control relays), so the real measurement would need many meters as well, while the estimation could be made with comparable accuracy without the need for any meter at all.

Our local power company has a number you can call to report a nonfunctional streetlamp. I reckon it never gets used, though, as I always see a handful of cycling lamps at night.

Re : cycling HPS, here it is the same as MaritimeMan said, you call the street light hotline and report cycling fixtures. They rely on the public to notify of cycling lamps. When they come, they usually replace the whole assembly, including the mounting arm, with a LED fixture 

About five years ago, the power company bought a bad batch of photocells. That meant a lot of day burners and a lot of "night darkers" !  They're all gone now and day burners are quite rare nowadays.

The main weapon against cycling lamps in the field is to use quality lamps and fixtures and properly scheduled group relamping. In this way the lamps get replaced before they start to cycle and the lamp quality shall ensure it won't be so early in their life.
It look like wasting still good lamps (and many people used to replace lamps just at their home), but you should not forget the lamp has only small part of it's life left and most important, the cost (and environmental impact) of a single trip of the bucket truck cost way more than a single lamp, even when still new. So with earlier group relamping you suffice with just single trip to replace dozens of lamps, mostly even during the regular inspections of the system. So it mean you need just very few trips to have even large installation "cycling free". The "cyclers" are then only a few, in fact defective, lamps, that didn't make the relamping period.
The relamping schedule is then designed according to the lamp mortality curve so, only 2% (and that is, what nearly all road lighting standards allow in the installation) of the lamps are expected to fail before the relamping happen. And it is only for these quality escapes, what is supposed to be covered by the "hot line" information from the public and eventually spot replaced.

In Israel, we use central operated streetlighting, either by a photocell, timer switch or a wireless controlled streetlighting controllers (Like are used in Haifa, Nesher, Jerusalem and Tzfat).
We have american lanterns with photocell sockets, but their photocell sockets are unused.
We have also lanterns that are installed on existing power poles, but with a seperate cabling, so they started with the other streetlighting.
As far as I know, roadlighting systems where lanterns are seperately started with a photocell on the lantern, exists mainly in the US/Canada, but also in the UK and Australia.

The "centralized" control has it's advantage as well: You may easily control the operation over the night, so e.g. switch it OFF after midnight to save power, when practically all people are already home (in many villages, unless they experienced problems with crime at that time). Somewhere this eases the installation of dimming systems: You suffice with just one dimmer unit (an autotransformer with many outputs allowing to reduce the voltage by small steps to about 80%) for the long stretch of lanterns and the lanterns are then equipped by just ordinary ballasts (quite common in Brno on main roads - after ~10PM or midnight the light is dimmed, about 4AM..6AM the brightness is eventually returned back to full power), so allows an economic way to save power, when the road lights do not have to overpower the car headlights from heavy traffic.
And the central control was used here for yet another purpose: The streetlights are not a negligible load, so their exact turn ON/OFF moments were used to shape the mains network load balance, easing the network management.
while the estimation could be made with comparable accuracy without the need for any meter at all.

I found there are electronic ballasts such as "LUMTEC ECOLUM" (and I suppose other manufacturers have something similar to that) where you can preset dimming down to 60% using the switches directly on the ballast. You can set the range (such as 2 hours after mains on and 3 hours before mains off) for the full power and the rest is dimmed. It has to remember the previous day on/off time to do so. I think it could also work with photocell started lights provided the photocell is working correctly.
This electronic ballast can also start HPS in a "softer" way by reducing start-up current and it can compensate voltage fluctuations. But I don't believe much in the declared 20 years durability in outdoor conditions.

With electronic ballastst you need the dimming to be part of the ballast, as the mains variations are normally suppressed by the electronic, so the mains variation won't work.

But the system I'm talking about came from 80's, when the electronic was practically not available.

The "soft starting" feature I strongly doubt, as for the lamp performance is quite critical to run up the lamp at full current and keep at full power for some time, otherwise the lamp won't be able to correctly settle and so then degrade too quickly. Therefore most lamp makers do require the lamps to be operated at normal full power for the first 15 minutes after turn ON or even restart (the "after restart" is not fulfilled by the group dimming system here, as it does not have any means to detect some lamp extinguished and restarted), only afterwards the dimming is allowed. So gradual increase of the power after turn ON would be straight against this recommendation.

In the contrary, the ballasts boost the current upon start (they keep constant power, so as the cold lamp has lower voltage drop, it causes the ballast to deliver higher current), so the stabilization is quicker and more thorough. But for the really initial phase, when the arc voltage is really low, the current is limited, usually to 1.5..2x the rated current, usually given by the maximum ballast current (higher current is better, but more demanding on the ballast).

But you should not be mistaken by the input current: As in the initial phase the current is limited, it mean lower power than  normally, what mean lower input current from the mains. With magnetic ballasts this lower power transfer is usually offset by the higher losses in the choke and not optimum phase compensation (that is designed for steady operation).

@Medved: Yes, I perfectly understand that centralised regulation by lowering the voltage and electronic ballasts are mutually exclusive. What I wanted to say is that even with totally "dumb" mains (i.e. two wires without any kind of control) and lights turned on by a photocell you can dim by means of this programmable ballast.

But thanks for the pointing out on the HPS starting recommendations. The "LUMTEC" manufacturer seems to be Spanish but you've probably found this page in Czech . What confuses me there is the almost 100% luminous flux in within 30 seconds (moreover with that limited startup current(!!!)). I don't know whether it is somehow possible or it's just another marketing BS.


@dor123: According to Wikipedia you have 230V/50Hz in Israel (the same as in our country). In the USA there is 120V/60Hz system. So it seems there are non-US versions of those lanterns.

@Medved: Yes, I perfectly understand that centralised regulation by lowering the voltage and electronic ballasts are mutually exclusive. What I wanted to say is that even with totally "dumb" mains (i.e. two wires without any kind of control) and lights turned on by a photocell you can dim by means of this programmable ballast.

For that purpose there are photo switches on the market, which (after synchronizing with day/night cycle) are able to switch OFF the light for the given part of the night

But thanks for the pointing out on the HPS starting recommendations. The "LUMTEC" manufacturer seems to be Spanish but you've probably found this page in Czech . What confuses me there is the almost 100% luminous flux in within 30 seconds (moreover with that limited startup current(!!!)). I don't know whether it is somehow possible or it's just another marketing BS.

The "controlled startup current" means the current from the mains. Normally, with an inductive ballast, there is quite long turn ON transition (few mains cycles), when the core of the ballast saturate, so cause current spikes. These spikes then could trigger the circuit breakers, if it is too sensitive (so with small margin).
The maker of these ballasts claim, they do not have such inrush current from the mains, so are able to operate with fusing with very little margin over the operating current (so 20..30x 70W fixtures on a 15A breaker, instead of just 5..10 with the classic series choke)

@dor123: According to Wikipedia you have 230V/50Hz in Israel (the same as in our country). In the USA there is 120V/60Hz system. So it seems there are non-US versions of those lanterns.

The individual photo control is used not only in the US, but it is very common e.g. in the UK. As a consequence, most fixture makers mold the cases so, you can easily fit the photocontrol (after snapping out a piece of plastic on the top, or make a variant, where the hole is already present...)

We have several areas here where a circuit of lights will be on 1 centralized photo-cell.  It seems to work quite well.  Most of the streetlights in my area have individual photocells for them and I've found that the photocells are lasting about 10 years before failure as a result of weather damage (UV and moisture) The HPS bulbs are lasting about 9-10 years as well, so I've noticed that the power company's contractor who does all the street lighting stuff for them are replacing the photocells along with the lamps at the same time. 

In Australia there are mixtures of both individual lights with daylight switches and one daylight switch controlling several lights in groups. Sometimes you see individual lights come on then others in groups. Also you see quite a few "dayburners" with cheap Chinese made switches.

@Medved: Yes, I perfectly understand that centralised regulation by lowering the voltage and electronic ballasts are mutually exclusive. What I wanted to say is that even with totally "dumb" mains (i.e. two wires without any kind of control) and lights turned on by a photocell you can dim by means of this programmable ballast.

But thanks for the pointing out on the HPS starting recommendations. The "LUMTEC" manufacturer seems to be Spanish but you've probably found this page in Czech . What confuses me there is the almost 100% luminous flux in within 30 seconds (moreover with that limited startup current(!!!)). I don't know whether it is somehow possible or it's just another marketing BS.


@dor123: According to Wikipedia you have 230V/50Hz in Israel (the same as in our country). In the USA there is 120V/60Hz system. So it seems there are non-US versions of those lanterns.

Merc: We have american lanterns, but with 230V/50hz european and local made ballasts, ignitors and european lamps. Also, Eltam Ein-Hashofet, a local ballasts company, manufactures 230V/50hz ballasts for american lamps.

Most people are also unaware of another advantage of central control. Central control allows the street lighting circuits to be cut in emergency circumstances such as Air Raids. During an air raid the idea is to reduce and eliminate the amount of light visible from offshore and from the air to increase the difficulty in enemy targeting. With local photocell control and shared distribution the entire power grid would have to be shut down to provide an air raid blackout and this is undesirable.

Practically all streetlights here have individual photocontrols, and it's exactly for the reason Medved mentioned: they are fed from the existing transformers that power homes and businesses. Not sure if it's done this way in other parts of the world, but here, the lights are connected directly to the 120v secondaries of the transformers with no metering whatsoever. The power company charges the city/county an appropriate amount of money to cover electricity and maintenance costs (the power company maintains the lights too). The only exception I've seen to this method is on the interstate; in some areas, sections of lights are turned on and off simultaneously. Other areas still use lights with dedicated controls.

There is, however, a relatively recent system called GE Lightgrid (and probably others, too) that uses a wireless connection to operate streetlights connected to different, "always-on" power sources. Besides turning lights on and off and dimming them on a set schedule, it can monitor the condition of streetlights and send maintenance alerts. I wonder how long it will take for this sort of thing to start showing up everywhere...I give it 10-20 years. One thing that's scary with stuff like this is the potential for hacking: imagine you're driving down a large road at night and suddenly every streetlight in sight goes dark. While it's not an enormous safety threat, it's still a concern.