[Soul Surfer]

Street light control was done using cycle modulation and rythmatic relays woking at a few cycles per second around 50 years ago. About 28 years ago I invented and patented a spread spectrum communications system for electricity gas and water utilities http://worldwide.espacenet.com/publicationDetails/biblio?DB=EPODOC&II=3&adjacent=true&locale=en_gb&FT=D&date=19830911&CC=DK&NR=116383A&KC=A  that went as far as a full scale trial but the british world beating technology was killed stone dead when the government decided to privatise the utilities.

DCC chips are easily available an used for lots of model railways but they are only good for low voltages.

[Geezer]

Street light control was done using cycle modulation and rythmatic relays woking at a few cycles per second around 50 years ago. About 28 years ago I invented and patented a spread spectrum communications system for electricity gas and water utilities

Oh yes! EMI, I remember them. Didn't they make gramaphone records?  []

[AlmostHuman (OP)]

To (try to) cut long story short:
Let's say that right now I need off-the-shelf solution for testing and marketing purpose, and I think I've managed to find solution that we could try (there's a lot space for improvement in this phase, but it's not necessary).
As I said, that solution is going to be for marketing purposes only (to animate politician to install some level of energy saving street light system). On the other side, when the politicians take the bait and we get green light for this project, we will implement the solution that we find optimal.
As far as my planning goes, I think that we will take few development platforms for power line communication and develop system that will fit our needs when we are done with marketing (if our marketing is successful).

System implemented now is done with mercury bulbs and without any smart control. By some Kyoto agreement, mercury bulbs are to be replaced with something eco-friendly by 2013. So, our goal is to implement power saving system (we're not responsible for bulbs, but just for their control). New bulbs will have two modes: power saving and full power. Solution (that will be implemented) needed is one that could control any particular bulb at any time and (optional) have some feedback (is bulb turned on or off).

Every block has its own transformer (nothing else but the street lights is connected to that transformer). One power cable (one wire for each phase (R,S,T) and one for neutral (N) = 4 wires) goes from transformer to the end of the block and all lamps are connected to that power line (first bulb connected to R, second to S, third to T, fourth to R...) everything is weatherproof and dry. Control device should be able to communicate with main control via RS232 and to transmit commands to modules that control particular bulbs and to collect and dispatch data considering status of each bulb to main control. Size of any block isn't over 1000 meters, and there's 40 lamps max per block. So there's max of 80 outputs to control per block.

Project leaders are eager to experiment and find optimal solution (price per lamp should not be over 120 USD). IT support is probably the strongest link in our team, so there should be no problems as far as software is considered.

[Geezer]

To (try to) cut long story short:
Let's say that right now I need off-the-shelf solution for testing and marketing purpose, and I think I've managed to find solution that we could try (there's a lot space for improvement in this phase, but it's not necessary).
As I said, that solution is going to be for marketing purposes only (to animate politician to install some level of energy saving street light system). On the other side, when the politicians take the bait and we get green light for this project, we will implement the solution that we find optimal.
As far as my planning goes, I think that we will take few development platforms for power line communication and develop system that will fit our needs when we are done with marketing (if our marketing is successful).

System implemented now is done with mercury bulbs and without any smart control. By some Kyoto agreement, mercury bulbs are to be replaced with something eco-friendly by 2013. So, our goal is to implement power saving system (we're not responsible for bulbs, but just for their control). New bulbs will have two modes: power saving and full power. Solution (that will be implemented) needed is one that could control any particular bulb at any time and (optional) have some feedback (is bulb turned on or off).

Every block has its own transformer (nothing else but the street lights is connected to that transformer). One power cable (one wire for each phase (R,S,T) and one for neutral (N) = 4 wires) goes from transformer to the end of the block and all lamps are connected to that power line (first bulb connected to R, second to S, third to T, fourth to R...) everything is weatherproof and dry. Control device should be able to communicate with main control via RS232 and to transmit commands to modules that control particular bulbs and to collect and dispatch data considering status of each bulb to main control. Size of any block isn't over 1000 meters, and there's 40 lamps max per block. So there's max of 80 outputs to control per block.

Project leaders are eager to experiment and find optimal solution (price per lamp should not be over 120 USD). IT support is probably the strongest link in our team, so there should be no problems as far as software is considered.

What?! You mean you're actually looking for a practical solution? We're scientists!  [] (er, well, some of us anyway)

[AlmostHuman (OP)]

What?! You mean you're actually looking for a practical solution? We're scientists!  [] (er, well, some of us anyway)

Not really. I am looking for hybrid solution (half practical, and half "unpractical"  [] ). I am looking for development platforms on which solution to this "problem" could be based.
Thanks for help though!

[Geezer]

If the poles are all "line of sight", could you use IR communication (although it can be influenced by weather).

[Soul Surfer]

Almost human.  For a one off  there are probably lots of basic digital communication chips that can be used to implement your system as you suggest simple microcontrollers or PGA can be used to do the logic.  Any single cable has a reasonable characteristic impedence and a simple capacitance resistance circuit can be used to isolate the high voltage and protect the input.  you might need a slightly more poweful driver than a simple communications line but they are available.

When I did the mainsborne telecontrol project we had only just started to use the simplest microprocessors. and the range of chips was very limited an mostly basic logic so you had to design most of it yourself!

[CliffordK]

At this point the networking would be nice.  However, I think I would avoid the temptation to do networking to every individual lamp.

If several lamps are wired together, then just treat the cluster as a single lamp, and design a dimming approach that can turn the string on and off, as well as dim them.  The Cold Cathode Compact Florescent Lamps are low power, and dimmable, so you would just set the desired voltage at the source.  I know the street lamps are bright...  can the CCFL lamps put out enough light on high power?  Mercury might be an issue, but the lamps should be able to be recycled, especially if you have a lot of them.

It is a bit confusing, but there are both PLC (programmable logic controllers), and PIC Controllers which are essentially a PLC on a single chip.  You should be able to program a PIC to run independently based on the specified algorithm.  It would take the inputs from a photo sensor to determine dawn and dusk.  I would calculate noon and midnight based on halfway between dawn and dusk (good enough for the purposes of the system) to prevent time-drift.  Then program each controller with the desired schedule for the individual lamp.  If necessary, it should be able to guess within a week of the date based on the length of day.

Maintenance would have to be done at each lamp, or each PIC.  However, it should have relatively lower power requirements, and would be unaffected by systemic issues.  Did someone suggest that Fog & rain would interrupt IR?  Is it possible to at least mount your control box at ground level?  Anyway, if you create a good algorithm, then systemic upgrades should be rare.  Keep in mind that with a networking approach, there are still needs for the initial installs, and periodic network repairs and upgrades.

[SeanB]

The dimming photocell has already been invented, and is quite common in Switzerland AFAIK, where you find a lot of fixtures that have 2 lamps, one high power and one low power. Both lamps are switched on at sunset, and at midnight the high power one is switched off, only coming on 2 hours before sunrise. The low power lamp is on all night. Generally implemented as a mercury lamp and a LPS lamp, as both have the distinct advantage of a lifetime measured in decades ( not like the current lighting that might make 5 years at best, more like 2 in most cases) and giving a good quality light when people are around ( MV lighted) and a lot of light the rest of the time ( LPS is really efficient at making light, though it is red) when there are not many people around. Otherwise 2 HPS lamps of different power in more recent installations.

They compensate for day length, and set themselves up automatically to switch at midnight.

Dimming is really only going to be feasable if you have high power LED lamp units, and you can integrate the controller into the power supply for the unit, along with the power line interface. Do not forget to make the power supply capable of surviving 5 years at 125C, and to be able to survive multiple simulated lightning spikes applied to the power supply terminals, along with being able to survive prolonged overvoltage ( at least capable to survive lost neutral on a 3 phase system without blowing up or popping a fuse) along with reliable startup at any point from -55 to 125C and any input voltage in the range at the same time. Self resetting thermal protection would also be good.

MV lamps are dimmable, simply by using a multitap ballast and dropping the current of the lit lamp to a lower level. Not going to give a big dimming range, but will offer savings without causing problems for the lamp in terms of life and poor operation. Most sodium and metal halide lamps are not dimmable, unless you use an electronic ballast to run them, and these are not going to survive in street lamp use.

[Geezer]

....and LED lights might not be such a good idea as someone is likely to purloin them []

[CliffordK]

I did look up LED street lights, and found some 100% LED street light bulbs, as well as complete lamp units with lower overall power consumption than the low-light versions of the standard bulbs.  Since the LEDs use multiple bulbs, it should be easy to shut off 50% or more to run at half power.