Optical Reflectivity Measurement System
It's all very well to be competent at hardware design, laying out printed circuit boards and writing firmware. What distinguishes an Embedded Systems Engineer is the ability to embrace other fields of knowledge and assess their impact on a product to ensure first-time success. This system for the measurement of optical reflectivity gave ECROS Technology the chance to demonstrate its breadth of engineering and scientific know-how by calling for a working knowledge of photometry and optics.
The customer's requirements were fairly simple. An opaque, rectangular box would have a hole in the top over which the sample would be placed. Inside the box, one of several LEDs would illuminate the sample from slightly to one side. One photodiode, placed at the same distance to the other side of the hole would measure specular (mirror-like) reflected light, while a second photodiode placed under the hold would measure diffuse reflected light. To prevent interference from ambient light sources, the whole apparatus would be inside a second opaque box with a black cloth curtain for access. The customer had published approximate dimensions for the inner box, restricting design choices so that the height of the box had to be six inches, making the sample rather more distant from the LEDs and photodiodes than would have been chosen without this limitation.
Selecting the Photo-Detector
With the usual tight budget and short timeframe, nothing too fancy was possible in the selection of a photodetector. The customer wanted to use illumination of wavelengths from ultra-violet to infra-red. Most photodiodes have a poor response to ultra-violet light, often in a deliberate attempt to match the sensitivity of the human eye. The Osram SFH 229 silicon PIN photodiode was found to be readily available and has a response at 400 nm that is 20% of its peak response at 860 nm. The 3 mm (T-1) leaded package and narrow 17° half-angle allow the device to be bent over and pointed at the sample, reducing sensitivity to light from other directions (although the inside of the box was painted matte black, there will inevitably be some small amount of stray light inside).
Selecting the Illuminators
The goal in selecting LEDs for illuminating the sample was to throw a lot of light towards the sample and not much elsewhere. As with the photodiode, the T-1 leaded package allows the device to be directed towards the hole over which the sample is placed and the lens-like dome reduces the angle over which the light is spread. Because of the desire to distinguish specular and diffuse reflectivity (and, again, the tight budget), a single LED was used for each color.
A secondary goal was to avoid too much variation in the illuminance provided by each LED color so as to simplify the photo-detector design. Now, naively, this would mean picking LEDs with similar candela ratings. However, the candela is a photometric unit, meaning that it measures light as perceived by the human eye, not a radiometric unit, which measures the power of electromagnetic radiation, including visible light, without regard to the vagaries of the eye. The sensitivity of photodiodes is given in the radiometric unit of amps per watt (0.62 for the SFH 229). In fact, as the human eye cannot see ultra-violet and infra-red light, it is not possible to give a candela rating to LEDs of this type.
At 555 nm, a luminous intensity of one candela (1000 mcd) corresponds to a radiant intensity of 1/683 watts (1.464 mW) per steradian. At all other visible wavelengths, the eye is less sensitive so that to achieve one candela the radiated power has to be greater. The sensitivity of the eye is characterized by the CIE 1931 standard (see Resources, below). For example, at 650 nm, the standard gives the sensitivity of the eye as 0.107 (relative to 1.000 at 555 nm). So, to produce the same radiant intensity as a 555 nm (green) LED of 1000 mcd, a 650 nm (red) LED should be rated at only 107 mcd. As mentioned above, candela ratings are not expected for LEDs with wavelengths not visible to the eye. Instead, the data sheet should directly specify the radiant intensity.
The critical choice was therefore a green LED with a very high candela rating. Within the time and cost constraints, a part from Kingbright was located that has a typical luminous intensity of 8,000 mcd (8 candelas) at its recommended drive current of 20 mA but was somewhat blue/green in color with a wavelength of 525 nm. The radiant intensity was calculated as 15 mW per steradian. This figure formed the basis for the selection of the infra-red, red, blue and ultra-violet illuminators.
Designing the Photo-Detector Circuit
As mentioned above, the customer insisted that the sample be placed at least six inches (152 mm) above the PCB carrying the LEDs, photodiodes and other circuitry. The LEDs, sample and specular photodiode were arranged to be at the corners of an equilateral triangle of side length 176 mm. The 50 mm diameter hole over which the sample is placed subtends, at this distance, a solid angle of 0.0634 steradians at the LED. So, the total power incident at the hole will be about one milliwatt (using the 15 mW/sr figure of the blue/green LED). (Several second-order factors are being ignored here, such as the angle between the hole and the axis of illumination and the falloff in radiant intensity away from the LED axis, but we're looking for some ball-park data to help with circuit design.)
What happens to the light when it hits the sample was difficult to assess with the information given. To get some idea of what will happen, the (unjustified) assumption was made that all the incident light would be dispersed evenly into the half-sphere (or 2.π = 6.283 steradians) inside the box from the sample. The reflected radiant intensity would then be 0.16 mW per steradian. The photodiode datasheet gives the radiant sensitive area as 0.3 mm2. This tiny area subtends a solid angle of 0.0000097 steradians at the center of the sample area. So, the power incident on the sensitive area of the photodiode will be 1.55 nW. With a (peak) sensitivity of 0.62 A/W, the photodiode will respond with a current of 0.96 nA. This is scarily small, especially as the dark (leakage) current of the photodiode is given as 5 nA maximum (50 pA typical).
Two things were apparent at this point. First, the signal conditioner for the photodiode will have to be really, really sensitive. Second, the customer has made things quite difficult by insisting on the large distance to the sample. Incident power falls as the square of distance to the source. This square law applies both to the illumination of the sample by the LED and the illumination of the photodiode by light reflected from the sample. Therefore, moving the sample down from six inches to three inches above the PCB would result in sixteen times as much photo-current. Peering closely at the photodiode, the suspicion was generated that the small sensitive area given in the data sheet did not account for the focusing effect of the lens. As things did not turn out as badly as the above calculations suggest, this was probably the case.
The signal conditioner was designed as a current-to-voltage converter followed by an inverting variable gain stage. A very low input current operational amplifier was used with a 10 MΩ feedback resistor to get a gain of -10 millivolts per nanoamp. The cathode of the photodiode is connected to the positive supply and the anode is at a virtual ground point formed by the operational amplifier action. This circuit point on the PCB was surrounded with grounded copper to block any current leakage across the surface. The second stage turns the signal "the right way up" and provides voltage gain defined by an I2C potentiometer set by the microcontroller. The output goes to the A-to-D converter.
Testing by ECROS Technology verified that the optical reflectivity measurement system operated according to the design and met the stated requirements of the customer. In spite of the short time allowed for the project, the customer has not yet used the delivered system and so no final report can be made about the project's success.