Need help with an odd request.

Hello Everyone,

I found this forum and decided to go straight to the pros for the answers I'm looking for. I'll try to briefly explain what it is that I do and what I am looking for.

I work for a company that offers solutions for dimming office building lights. I am personally in charge of producing a sensor that will detect light levels (outside) for the system to use to dim/adjust the office lighting, nearest the windows, accordingly. The production of such a sensor is easy. The calibration of the sensor with the rest of the system is a little bit tricky.

I have to calibrate these sensors before they are installed using varying degrees false but consistant "sunlight" (ie. flashlight or spotlight). Basically, I'm looking for something that I can put right up against my light meter and sensors.

The idea is that I will simulate the light of a bright day (using my light meter to get the appropriate output) and then use that to get the output of my sensor at that light level. Then I'd repeat the same process with a cloudy day level, evening light level, direct sunlight level, indirect sunlight level, etc etc. By the end I would have a little table of how the sensor would react at varying levels. It would then be inserted into the software look up tables.

So I think you can see what it is that I'm looking for:

1. A flashlight that can dim and simulate various light levels with an adjustable switch of some sort. The closest I found to this was the Blackhawk Night-Ops but the problem is that it dims rather quickly making it really hard for me to control and repeat the process over and over. I would prefer a little more discrete control over the dimming.

OR

2. A simple bright light in unison with various degrees of tinted light filters (ie. tinted plastic). But does anyone know where I might be able to find such tinted pieces? I think there is a place here that can do such a thing but it would have to be custom job (thus time consuming) and there is no guarantee that it would be the correct tinting that I would need for each level.

Keep in mind that I am not against building something as long as its fairly portable. However, I would prefer to be able to grab something from off the shelf just to make things easier. Money is really not a concern here either.

Any suggestions? Other ideas?

Thanks!
Eric.

I'm not sure an LED light is the answer, considering LED light is probably the least like sunlight, but others may differ in opinion.

If an LED light would work, then the only light that comes to mind, is the HDS EDC. Since it has (I think) 20 different brightness levels, you could probably always find one close to the one you needed. I need to read up more on the Night-Ops light, since I don't know much about it. The Surefire U2 has 6 brightness levels, and is LED as well.

I'm sure some others with more experience than I will chime in.

Good luck, sounds like a cool project. Are the dimmers computer controlled, or are you using some kind of microcontroller to provide outputs to the lights and take the sensor data?

If you can use a LED, what about a Elitemax?
You can find it here http://www.techass.com/el/elm4z/index.php

Eric,

The HDS lights can do many repeatable regulated dimming levels. It's an LED, so depending on where your sensor responds in the visible spectrum that may or may not be an issue.

The Surefire L2 does 2 levels of light and the Surefire L1 also does two levels. The Surefire U2 has 6 levels of light output. All of these are LED lights.

Be aware that the LED multi-level lights -may- have a spectral shift between their various levels.

And also remember that a flashlight beam has varying brightness levels within the beam, varying from the intense center hotspot to the dimmer outer corona of light. Depending on the sensitivity of your sensor, you might have repeatability problems due to positioning of the flashlight beam onto the sensor. Also remember that distance between the light source and the sensor is another variable that can affect your sensor readouts.

Or you could use various photographic ND (neutral density) filters to cut down on the light emitted from your light source or received by your sensor.

Money no object, you say? I think you really will need to make a calibration fixture for your sensor, incorporating a voltage/current/power regulator for your light source, the light source itself (probably a bulb), a diffuser to even out the beam and eliminate hotspots, a slot or mechanism to put the various ND filters into the light path between the light source and the sensor, and a housing to contain everything -and- give you the constant stand-off between the light and the sensor. You might want to include a built-in sensor into the calibration fixture to help insure consistency because light sources (bulb or LED or whatever) can drift or vary over their life and with environmental conditions.

I should note that the sensor is not spectrum specific at all.
Infact, its just a simple photoconductor so it responds well in all regions but i guess that it does favor the green/blue..... but its so insignificant, it shouldnt matter.

Thanks for the quick replies!

Ahh, Mike, you bring up a good point. It would be best to stay away from the use of batteries. A constant power supply will be best.

I wasn't even aware of all the other things you mentioned. Ugh, this is what happens when someone like myself, who really has no experience or knowledge of the topic, has to undertake such a project..... difficulties.

I will have to re-consider everything now. I'm on such a tight deadline and this only makes matters a little more complicated.


For the time being just to get the job done, albeit very inaccurately, I will look into a quick and dirty method to meet my deadline. In the long run, I will move towards somethign a little more custom.

Ok, if I were to go the route of the calibration fixture with all the extras, do you think you could point me in the right direction? What type of bulb should I be looking for? It would have to be pretty bright. Any idea where I might be able to get some diffusers and ND filters?

Thanks so much for the insight. This is really very helpful so far.
Everyone else, thanks for your choices of lights. If you know of anymore that are better suited, please feel free to reply. I like the idea of the one with 20 levels so far.

Geogecko,

You hit it pretty much straight on!

Completely solidstate computers (no fans/harddrives) that we've coined "floor controllers" simply cause they only control the lights on their particular floor, communicate with each individual dimmer. They also receive all the sensor feedback. There is a master computer that keeps track of all the floor controllers.

Microcontrollers (PICs) are used in each dimmer. Each dimmer can control one magnetic ballast....the key word being magnetic, as no such thing exists on the market for magnetics.

The company I work for is called Fifth Light (www.fifthlight.com) and we have just began our first installation here in Toronto. We mainly deal with retrofits on buildings with existing magnetics as new buildings usually install the more expense electronic ballasts by default. Both are dimmable in the end.

As a side note, I mainly deal with the PICs and the actual coding. While its something I'm still fairly new to, I'm pretty comfortable doing it. These sensors on the other hand are a whole new animal to me. Should be interesting!

Eric

Think this thread fits better in the "Flashlight Electronics Forum", so I moved it!

Interesting ideas. I used PIC's in school, and haven't used them since, at least, at work. I use them at home, for little home projects. They are great little devices, as long as you don't need something too fast (great for sensor/control type applications though!). I do wish I had a job where I could use them, but oh, well, whatever pays the bills, I guess.

I personally like using the assembly language to program them, it just seems more simple to me, than using a C compilier, plus, it's probably more efficient in code.

Good luck with your project. If it's just a standard photoelectric sensor, they can be tricky. I tried using one in a backlight for an LCD, and had problems, of course, I was doing everything analog, with opamps and everything, probably should have used an on-board ADC instead...sure would have been easier to figure out the hysteresis part of it...

So are you using serial interfaces for each dimmer? I wouldn't think you'd be able to go very far with USB. That would be a lot of serial ports needed, if so, so you'd probably need one of those Octopus PCI cards that have 8 serial ports on them. Unless there is a way to have them all run off of the same bus somehow, maybe a floor controller, controller, another PIC that runs a bus interface between all the dimmers, and then converts to a serial interface, perhaps. Many possible solutions...that's what is cool about engineering.

The chief eng (and the inventor) developed his own communication protocol which runs alongside the dimming code on the same chip. It's both insane and clever..... and the dimming aspect of it is kept secret for obvious reasons.

As for the communication: He uses the 60 Hz of the ac power grid to time the protocol so that all units on the bus are in sync at all times. We have an output module that can provide enough signal to reach far off units on the floor (and a repeater if its not possible). The output module is also set to use the 60hz to transmission. PICs in each unit is set up to "read" one particular pin level at the 1/60s interrupt. 2 line cable is all that is needed to run from each unit to each unit and finally back to the controller. A 3rd line is going to be installed for future provisions so we will be able to receive answer bakcs from the units and provide a more 2-way communication. Its quite simple really...but effective.

It's a good form of transmission provided that you don't need to send a lot of data quickly. We are really only able to send 3 commands to 3 independent units per second with this protocol so you can see it is quite slow...
however, we are able to group units and entire floors and send commands to all or some at the same time.

Ouch, the EDC Ultimate 85 GT is slated for first week october shipment at the earliest. I was hoping to get one by next week, heh. Oh well.

Right now we are using a filter over our sensors that dampens the light that is tranmitted to the photoconductors because they were getting saturated far too easily without it. The filter only allows a 10% transmission of light and it allows me to better differentiate between the various gains of the various light levels.

How bright is the 60 GT? I would certainly be able to simulate VERY bright with it? I will buy it instead if this is the case and if i need to, buy the 80 later on.

Hello Eric,

I would go about it a little bit differently...

I just measured direct sunlight and got in the neighborhood of 110000 lux, I then went to an area that was not in direct sunlight and measured around 3500 lux.

Since light falls off with the square of distance, I then went and checked a halogen desk lamp I have and found that when I place my meter 2 inches from the lamp I get over 100000 lux. Moving the meter about a foot away gives me around 3500 lux.

I would pick up a halogen desk lamp (or borrow one from the office) and use a light meter to adjust the distance you need from your sensor to correlate to bright, cloudy, and dusk readings.

Tom

Eric,

First off, I think you're working too hard on characterizing the photocells. If you look at the curves a single point (offset) is all that's needed. You don't need a look up table, you need to set a cal constant to standardize the detector.

I sent an email to your company, did you get it? In it I discussed a source for quite cheap filter material (used by the square yard by the movie industry) and offered some 'half value' material I have left from my 'Poor man's run time meter' project (you can no doubt find the thread with the search function).

After a bit of thought, I suggest that a good test rig would be a source (probably a high power LED) with a photocell 'watching' it and a regulator circuit driving the source to a series of preset levels based on the photocell. A diffuser in front of the source should make the light uniform. In use the rig would be held over the sensor (perhaps with a gasket to keep out other light).

Should be pretty simple, and a fun project to boot.

Doug Owen

This is honestly the best forum for help that I've been on outside of the microchip.com forum. I really appreciate all the effort!

Tom:
That is good to know. With a decent halogen light, I can just vary the distance. This is definitely a viable option. Thank you!

Doug:
I did get the email and I will definitely use your advice when I begin building the custom fixture.

BTW, the photoconductors are very nasty little buggers. Besides being highly non-linear, they are highly inconsistent between each other. Two photoconductors might both measure 50kO at 400lux and then measure completely different resistances at 4000lux. I shouldn't even say they might because they definitely will. There is nothing constant for me to measure. I can't imagine trying to make calibration constant.... but I've been wrong before and I'm sure something fairly accurate is possible.

On top of that, the very act of exposing them to too much heat will change their operating characteristics. Apply too much solder heat and the resistor can change pretty drastically.

This is the problem with photoconductors. I'm not even sure that they used very often in practice for measuring distinct levels of light as much as they are for just measuring the presence of light. The reason why we are using then instead of other photodetectors is simply because they don't require a 3rd line to be run to supply power. 2 leads and that's it. Our housing fixture that will be mounted near windows basically just consists of one of these with 2 leads. All the circuitry (op amps, filters, etc) is located in an "input module" with all the other circuitry that controls push buttons and other types of sensors. This sits right next to the "output module" that I spoke of earlier that controls all the dimmers.

The good news is that we have decided to detect for only 4 light levels (bright/semibright/dim/dark) and these photoconductors should suffice. That's all we really need because we intend to dim all the office lights at all times by at least 20-30%. Our system moves in increments of 5% and the usable light really becomes unpratical around 60-70% dimming. So that doesn't leave very much room in between these regions for all sorts of daylight harvesting settings. If its a bright day, the lights nearest a window might be dimmed by 60% and the next row of lights, a little less... and so on. If its semi bright, it wont be a big change. Everything needn't be too complex.

Anyways, thanks everyone. I will keep you all informed on my progress.