Surprise photocell circuit.

[schmitt trigger]

For many years now, as my mind ages and my attention span wanes, I have decided to attach automatic photocell on/off circuits to the lights outside the home.
Otherwise I would forget and they would remain on all day long....

One such photocell started behaving erratically, meaning that it really required a lot of sunlight to turn the light off. I replaced it, but instead of trashing it, decided to tear it down.
I've learned a few habits from lurking in this forum. 

The circuit really surprised me! I kind of expected of cheap and dirty electronic circuit, with at least a pair of semiconductor devices.
To my surprise, the system is electro-thermal-mechanic. Not a single silicon device!

The photo, which I have annotated, is self explanatory.

Question: are all street light photocells built the same way?

[T3sl4co1l]

Heh, that might explain some lamps turning on and off too (though AFAIK that's also a self-protection mechanism if they don't start properly?).

Tim

[james_s]

That's the standard way those small photocells are made, I'm not sure I've ever seen one that was solid state.

They're pretty clever, a CdS cell in series with a heater wound around a bimetal strip. Light drops the resistance of the CdS which increases the current through the resistance wire which heats the bimetal strip which snaps open the contacts and turns off the light. This gives you a built in delay with some hysteresis, it's simple, inexpensive and robust, able to tolerate power surges, current surges from incandescent lamps burning out and inductive loads.

[james_s]

Heh, that might explain some lamps turning on and off too (though AFAIK that's also a self-protection mechanism if they don't start properly?).

Tim

If you're referring to HPS lamps cycling, that happens because the arc voltage increases as HPS lamps age and also the arc tube runs hotter as it blackens and absorbs more energy. A hotter arc tube has a higher pressure which also increases the voltage required to sustain a discharge. At some point the voltage needed exceeds what the ballast can supply and the lamp extinguishes, cools off and restrikes, then the cycle repeats. Left unchecked it's not uncommon for the repeated ignition pulses to eventually break down the insulation on the ballast windings or burn out the ignitor.

Some crazy people actually believe that their bodies interfere with streetlights and cause them to extinguish. The phenomenon is not too surprising when you consider how many thousands of streetlights are in a typical city with a percentage of them likely to be cycling with tired lamps at any given time.

[mikerj]

They're pretty clever, a CdS cell in series with a heater wound around a bimetal strip. Light drops the resistance of the CdS which increases the current through the resistance wire which heats the bimetal strip which snaps open the contacts and turns off the light. This gives you a built in delay with some hysteresis, it's simple, inexpensive and robust, able to tolerate power surges, current surges from incandescent lamps burning out and inductive loads.

And also makes the trip point sensitive to ambient temperature...

[schmitt trigger]

That was my thought also. How does this simple mechanism compensate for the wide ambient fluctuations found in streetlight service?
Unless it is tripping somehow, with a temperature differential.

All in all a very clever device. 

[Zero999]

That's the standard way those small photocells are made, I'm not sure I've ever seen one that was solid state.

They're pretty clever, a CdS cell in series with a heater wound around a bimetal strip. Light drops the resistance of the CdS which increases the current through the resistance wire which heats the bimetal strip which snaps open the contacts and turns off the light. This gives you a built in delay with some hysteresis, it's simple, inexpensive and robust, able to tolerate power surges, current surges from incandescent lamps burning out and inductive loads.

That seems an odd way of doing it. My first thought was why not use a relay? But that wouldn't include the delay, given by a bimetal strip due to the thermal inertia.

[SiliconWizard]

It's simple and (relatively) rugged, but I guess not that frugal in terms of power draw? Wouldn't it need constant heating for one of the states?

[schmitt trigger]

It's simple and (relatively) rugged, but I guess not that frugal in terms of power draw? Wouldn't it need constant heating for one of the states?

It does draw some parasite power for sure.
I measured the photocell's resistance with strong sunlight (above 5000 lux), and the resistance was about 400 ohms. With the resistor in series, it would be roughly 10k, and in a 120 volt line, that means 1.44 watts.
But an electronic supply, at least for an unsophisticated low cost one, I would suspect that the parasite power would be in the same ballpark.

[SiliconWizard]

1.44W is certainly significant. Would be interesting to measure it exactly.

Note that this could violate some local regulations.

Eg: https://en.wikipedia.org/wiki/EC_Regulation_No._1275/2008

If it really draws that much power in "off" mode, it could not be legally sold in the EU (and probably other parts of the world).

[Gyro]

That was my thought also. How does this simple mechanism compensate for the wide ambient fluctuations found in streetlight service?
Unless it is tripping somehow, with a temperature differential.

All in all a very clever device. 

I can confirm that these are used on UK traffic signals (they used to be used on street lamps too, not sure about the newer LED ones). They can be seen as a translucent conical head on one of the signals (or top of the streetlamp). They are actually socketed - a twist and pull action, with a substantial neoprene gasket for easy replacement (once you're on the top of the cherry picker!).

In terms of ambient temperature compensation... very simple, there are bimetal strips on both contact actuator arms, only the moving one has the thick film ceramic heating element attached to it.

The big advantage of these switches is long thermal inertia. They are resistant to false triggering by passing dark clouds, heavy rain storms etc. The other advantage very high reliability. The design of the microswitch style tensioned contact spring gives a very good snap action and hysteresis.

Calibration of the street furniture ones is/was crude - a coating of Aluminium paint on the front of the LDR, scraped away to achieve switching at the correct luminance level. Not great for even power dissipation over the LDR element surface but these are big and conservatively rated.

Yes, very simple, reliable and clever.

[T3sl4co1l]

Incidentally, even though an LDR is made of semiconductor (CdS) -- they are resistive (linear) to an amazingly good degree.  It's not surprising that they can be used to control real power (~watts) at mains voltage.

About the only shenanigans you can play with them is to put an electric field (100s V) on top of the lacquer coating, making a really shitty MOSFET...

Pretty good on-off ratio too, considering.  I mean, to be fair, sunlight is rather intense, which helps a lot.  But even just with LEDs, you can make a very passable UJT* with them!

(Their time dependence however, is some kind of wicked awfulness.  They aren't practical to use for, say, communication.  Weird time constants and diffusion effects in the microseconds to seconds range.)

*Unijunction transistor: an identical component, but with the free charges (the semiconductor magically becoming conductive) provided by a forward-biased diode junction, rather than by light.

Tim

[Zero999]

1.44W is certainly significant. Would be interesting to measure it exactly.

Note that this could violate some local regulations.

Eg: https://en.wikipedia.org/wiki/EC_Regulation_No._1275/2008

If it really draws that much power in "off" mode, it could not be legally sold in the EU (and probably other parts of the world).

CdS cells are alsobanned under RoSH.

I imagine it would be fairly easy to design a circuit using a TRIAC, which will use a tiny fraction of the power.

[jimmc]

I had a similar one, the photocell was covered in paint and the light level to operate was set by scraping areas of the paint off.

Jim

[Gyro]

CdS cells are alsobanned under RoSH.

Are you sure about that? New stock is available in quantity from all the major distributors. I know the magic word Cadmium is there but I wonder. 

[SeanB]

Only manufacturer I am aware that is making them is Matsushita, but the majority of the photocontrols on street lights here are made by Royce Thompson, and have a small integrated circuit inside, encased in clear epoxy, that has the photodiode and all the electronics integrated into it. This does light level sensing, does a 5 minute timed on 30 seconds after power is applied for daytime lamp testing, and also tracks ambient light variations, so the lights come on at a constant light level, integrating even things like tree shadows, car headlights and such, and making the lights stay on all night.

You can program them as well, as in the UK there are many areas where the street lights are only powered from sunset to midnight, turn off till 4AM, and come on again till daylight. They also include a very large 480VAC VDR across the incoming mains supply, and have a very robust capacitive dropper circuit in them, that will survive being run on 400VAC for long periods. The relay they use is around 48VDC, and is rated to switch 16A, as the photocell can be used for either individual lamps ( plugged into the socket on top of the fitting) or to be a group control, switching a whole row of lights, which then just have a shorting cell plugged in, which only has a 16A non replaceable fuse and a 480VAC VDR in it, or just a shorting link.

They do fail, mostly from lightning impulse, or from water ingress, often caused by bird attack, as the common Ibis here loves to congregate on the street lights, and also has an impressive armoured beak. Dave should be familiar with them, they are also common in Australia, and are a pest.

Residential there are smaller versions, typically a phototransistor, capacitive dropper, 48VDC Songle relay rated at 250VAC 10A, and a simple circuit, often using the venerable 555 timer as comparator and hysteresis. They go bang nicely when they fail, or the capacitive dropper goes low value from self heating. I use one to drive general lighting, using a very overkill contactor, 3 phase 60A with all 3 phases in parallel, as that was very cheap at the scrapyard, and came in a box already. Load is around 10A, but lower now with LED lighting, but will never fail, even with the old wiring shorting out like it did this week, having me replacing 60 year old cabling at night.

[floobydust]

I've tried to sense outdoor light intensity and it's actually very difficult.
Getting a wide angle view of the sky, not the siding on the neighbor's house, is tough.

CdS is superior for having wide angle, slow response (filters flicker), and spectral response matching the human eye. Silicon phototransistors are closer on the IR end which makes mistakes under cloudy skies.

A better method is sensorless, where you use an "astronomical" timer or light switch.
Knowing date and time and latitude and longitude, and from that you can estimate twilight hours by calculating the sun's position. This is then used to turn on lights.
In reality it is a lookup table in an MCU due to the very complex math of planet's elliptical orbits.
It will be with a few minutes of dusk, assuming it's not too overcast.

[james_s]

That was my thought also. How does this simple mechanism compensate for the wide ambient fluctuations found in streetlight service?
Unless it is tripping somehow, with a temperature differential.

All in all a very clever device. 

It doesn't, but in practice that isn't really an issue. Photocells for switching outside lights are not precision devices, and the resistance changes pretty dramatically from light to dark so there really are only two states that matter, 'hot' and 'cold'. You generally want the light to come on when it is almost completely dark, and turn off just after dawn when there is enough natural light to see.

Yes they draw some power all the time, this is one reason modern photocells are usually electronic for the larger ones, but the small very low cost type with CdS and bimetal is still quite common. It's one of those things that was developed many decades ago and just kept being made the same way because it was inexpensive to produce and works well. The earliest photocells for street lighting (in the US anyway) used a phototube with vacuum tube electronics housed in the same glass domed plug-in module as our kWh meters. Being bulky and expensive they were usually used to switch groups of lights. The small thermal CdS type enabled inexpensive photocells on the individual luminaires and compared to the 175W-1kW lamps they typically controlled the additional power consumption was minor. I first started seeing electronic photocells in the mid 80s, I think most of the NEMA twist lock type have been that way since then.

[Gyro]

That was my thought also. How does this simple mechanism compensate for the wide ambient fluctuations found in streetlight service?
Unless it is tripping somehow, with a temperature differential.

All in all a very clever device. 

It doesn't, but in practice that isn't really an issue. Photocells for switching outside lights are not precision devices, and the resistance changes pretty dramatically from light to dark so there really are only two states that matter, 'hot' and 'cold'. You generally want the light to come on when it is almost completely dark, and turn off just after dawn when there is enough natural light to see.

As I said, a few posts up, they are temperature compensated anyway.

[schmitt trigger]

Replaced the faulty unit.
Before installing the new one, decided to measure it.

Not surprisingly, the highest consumption is under bright light conditions, see photo.

At dark is only a few milliwatts.

[jonpaul]

In 1950s-1960s General Electric in US and Saulter in Switzerland made astronomical time switches, a synchronous clock motor driving a 1 rev per day and one per year cam gear réduction.

The cam was cut to the equation of sunshine hours at the desired latitudes.

These were in every New York city light post and the Saulter unis on top of the luminaries, where later  the photo controls mentioned above replaced the astronomical time switch.

After a power outage all the  time switches had to be manually reset!

Just Reminiscing, from an old New Yorker

Jon

[james_s]

Perhaps an interesting comparison, next to the several Watts likely consumed by the timer motor the draw from the heater in the older photocontrolls was likely an improvement.

Those old astronomical timers sound interesting, I have a few electronic astronomical timers and they're great compared to the standard kind that you have to reset all the time.