Allrigth! :-)
this is pretty interesting… how close to a finished product is this?
Probably a few months…
I was playing with one of my big honkin’ Bridgelux arrays. My 16” sphere normally maxes out at just over 6000 lumens with the light sensor mounted inside the sphere. I was driving the Bridgelux to over 12,000 lumens, so I mounted the sensor on the outside of the sphere to use the sphere wall as an attenuator. It is around 100 times less sensitive that way, so should be good to around 600,000 lumens. Should hold me for a while…
Here’s the plot of the Bridgelux C8000 data. Ignore the LED numbers, those sensors are not connected. The Batt V / Batt I / Batt W numbers are the LED voltage, current, and watts. Peaked at 12,300 lumens at 146 watts to the LED. The aluminum CPU heatsink got to 60 degrees C (fan cooled). The color sensor was rather saturated so the CCT value is bogus.
The driver was a $6 Chinese 150W DC-DC converter, upconverting 12V to 31.7V at 89% efficiency. It is a constant voltage converter. I need to mod it for constant current before trying it with the C9000 arrays on a copper heat sink.
I got the converter modified for constant current output and hooked up the C9000 array onto a copper heatsink/fan. See 15,000 Lumen Bridgelux C9000 Light Engine
I have been playing some more with the using the Taos TCS3210 color sensor to calculate color temperature. I take the ratio of the RED and BLUE channels and use a table lookup/interpolation to calculate color temperature.
I now have a pretty good selection of LEDs with semi-known color temps and some more on the way. A BIG problem with LEDs is that their color temperature is very loosely specified. A tint bin can cover 1000K. You never know what you are going to get. Anyway, a picture is worth a kiloword. Plotting the color temp vs the sensor ratio gives a nice curve. I’m pretty confident that the results are pretty darn good. It sure beats spending $20,000+ on a spectrophotometer. Now, if I could just calculate CRI…
Well, after downing a few beers and digging into the deepest, darkest realms of bogophysics, I have it calculating a bogoCRI value.
Using the ever popular Planck equation, I calculate the energy emitted by a black body radiator over the wavelengths covered by each of the three color sensor channels.
I then drink some more beer and calculate the RMS distance between the Planck energy values and the energy values reported by the color sensor.
I then drink some more beer and apply a bogofudgefactor and come up with the bogoCRI value.
This is kinda sorta how a real CRI value is calculated, but that requires many measurements at closely spaced wavelengths. The color sensor chip basically has three wide (and somewhat overlapping) wavelength channels, so don’t expect miracles (those cost beaucoup extra). Or anything even remotely rooted in reality for that matter.
I find your project very interesting!
Have you checked out the Mired unit that takes the reciproc of CCT and multiplies by 1 million?
I think you could get a nice linear curve with your numbers.
Woo hoo! Prototype circuit boards have been laid out and are on order!
The main board is 4x5 inches. It has the processor and sensor interfaces.
There are also two different sensor boards. They are 24mm in diameter. They connect to the main board via 10 pin ribbon cables. The first one can be built up as the lux or color sensor. The second one is for the thermocouples/IR thermometer/PWM sensor.
Lack of USB, quite a design flaw IMHO. Everything works via USB now, RS-232 is dead.
Plug it into computer, installs drivers itself if necessary, done. No dongle, no serial drivers and what not.
Not interested.
I can't wait to see YOUR version. When will it be available ?
A USB to RS-232 converter cable costs like $2 If you can’t handle using that, you couldn’t handle using the device anyway.
The main reason for not adding a USB chip to the board is that they are notorious for being picky about what machines they will work with. They are also notorious for getting fried/zapped. By using an external converter cable you try different ones if you have a problem.
Oh, and there are set of pads on the board for installing these: http://www.ebay.com/itm/1pc-PL2303-USB-To-RS232-TTL-Converter-Adapter-Module-F-Arduino-CAR-Detection-GPS-/130807016972?pt=LH_DefaultDomain_0&hash=item1e74b4ae0c
Texas, will this still be around the same price point?
Yes!
An image of the main board:
Wow! Nice work, and I'm very excited to see the completed thing.
I built up one of the prototype boards. It needed a little hacking to fix a problem with the power MOSFET used to switch the battery on and off… I’ll need to order some more fixed prototype boards.
The battery switch FET is connected to one of the microcontroller PWM timer/counters. This lets you do all sorts of neat things… like implementing a constant-current load. Here is a plot of the device PWMing the power input to a XML T6 P60 drop in to produce a constant 1 amp load on a battery. I was rather surprised that the drop-in worked just fine while being fed pulsed 16 kHz power… I was expecting to have to use a load resistor to do this test.
I let it run until the battery voltage hit the programmed 3.0V low voltage cutoff. Voila! Battery capacity tester. It can also measure cell internal resistance by looking at the battery voltage change when a load is applied to/removed from the battery. The battery voltage is show in red since it hit the LVC cutoff and caused the battery to be turned off. The yellow numbers say that battery has been turned off and those numbers are probably not useful.
if you’re tracking # of seriously interested buyers, count me IN
popcorn. 