When using a light-to-frequency device, what can be done to eliminate jitter on the rising and falling edges of the output?
The jitter is most likely due to ambient light incident on the detector, specifically 60Hz or 120Hz flicker from AC powered lights. The best way to remove this other than by shielding and filtering is to integrate, or accumulate pulses from the device over an integral number of AC cycles. For 120Hz, 50ms would be a good integration time. This timing must be fairly exact, since we are relying on getting an integral number of AC cycles during this time. Experience has shown this to be quite effective. For example, it can be done with a microcontroller such as the BASIC Stamp, as used on our TCS230EVM, by specifying the argument to the COUNT instruction to be 50.
How can I eliminate infrared (IR) and view only the visible light spectrum?
Place a Hoya CM500 or Schott BG18 (or BG39) filter on top of the device to remove the IR. You can purchase a
Hoya filter from Newport Glass (www.newportglass.com).
What is the dynamic range of TSL230R?
The TSL230R has an input dynamic range of 160dB; that is, it can measure light over a range of 100,000,000-to-1.
The TSL230R datasheet states that the output can be either a pulse train or a square wave. Is there a method of choosing the output type of the sensor or are there two versions of the device with different output types?
There is only one version of the device. The output is a pulse train when the “divide by 1” frequency scaling mode is selected (S3=L, S2=L). In all other modes, the output will be a square wave with 50% duty cycle.
From comparing the datasheets, it appears that light-to-frequency converters can sense a wider range of irradiance responsivity than the light-to-voltage sensors. Is this correct?
Yes. In practice, the light-to-frequency (LTF) converters have a much higher dynamic range (DNR) than the light-to-voltage (LTV) devices. The TSL235R has a typical dynamic range of 1.25 million (saturation frequency / typical dark frequency) * Sensitivity Range or ((1.1MHz / 0.4 Hz) * 100). The LTV devices are limited by their output swing and their typical output noise voltage. We will assume here that any serious LTV application in which we are concerned about dynamic range will take into account the dark voltage, and thus it will be subtracted out from the output voltage.
Now we need to step back for a minute. Although we have generated some interesting information here, we have made several assumptions resulting in a comparison that is not quite apples-to-apples. We have correctly used noise as the limiting factor of the DNR for the LTV devices, while overlooking the bandwidth in the LTF discussion. What is the bandwidth for the frequency output devices, anyway? The answer is that the bandwidth will be determined by the time spent counting pulses, or the gate time. An implicit, although not necessarily true, assumption when stating frequency is that we have counted n pulses over a period of 1 second. In this case, our bandwidth is one Hz. If we compare the LTV device with a bandwidth of one Hz, we get significantly better DNR. But is it practical to build a 1Hz filter, and wait one second for a measurement? Maybe in some applications, but that may not be a practical solution for most applications. Additionally, to take advantage of that dynamic range would require a 24-bit ADC, which is not very cost-effective. Here is where we begin to differentiate the strengths and weaknesses of each device.
Typically LTF devices are used in applications where measurement time is not critical, or if it is, there are assumptions made about the light intensity being measured, particularly that the intensity is high enough so that the period of the frequency is much greater than the required measurement time. Dynamic range is a big advantage of the LTF devices, and the benefits of that dynamic range can be realized at very low cost in the form of a 16-bit counter in a microcontroller.
Light-to-voltage devices, on the other hand, may be used in applications where light is rapidly changing, or where an AC light signal or pulse is being measured. In such applications measurement time may be much more important than dynamic range, and an 8- or 10-bit ADC might be used. Even if we wanted to take advantage of the dynamic range of the TSL250R at its full usable bandwidth, we would need at least a 16-bit ADC. So we are limited, in fact, by the practical limits of building a cost-effective system rather than by the device itself.
With a light-to-frequency converter, what is the relationship between the light intensity and the output frequency? Will the output frequency be 0 Hz if the sensor is in total darkness?
With absolutely no light at all, theoretically the output frequency should be 0 Hz. However, there is an offset or "Dark Frequency" which is specified in the datasheet. The frequency at the output (fO) is given by the equation:
fO = fD + (Re)(Ee)
where fO is the output frequency, fD is the dark frequency, Re is the device responsivity for a given wavelength of light in KHz/(uW/cm^2), and Ee is the incident light irradiance in uW/cm^2
What is the simplest way to decrease the integration time on an analog output linear array?
Decrease the time between successive SI pulses.
Is there a way to reduce the integration time on an analog linear array without increasing the frequency of SI?
Yes. The integration time can be controlled using the SI and HOLD pins. HOLD is clocked in on the rising edge of CLK. This opens the switches terminating the integration. The data can then be clocked out when convenient. However there is a limit to how long the data can be held since there is always some leakage, or droop, on a hold capacitor. The practical limit is around 100 msec, and this limit will decrease, as the droop increases, with rising ambient temperature.
Do you have any linear arrays with a response that approximates the human eye?
All of our linear arrays respond to light in the 300-1100 nm range. The best option would be to use one of our linear arrays along with a photopic filter such as the Hoya CM500 or Schott BG18 filter.
Is it possible to cascade multiple packaged TSL3301 devices in a row?
The TSL3301 device cannot be cascaded. Physically, due to the packaging, the devices cannot be butted together without a large gap. Electrically, there is no carry, or chain, output to link the devices together. However, as suggested in the datasheet, the TSL3301 die itself, in its unpackaged form, can be cascaded both physically and electrically through the use of its serial chain in (SCIN) input. Pleasecontact TAOS if you have a requirement for a TSL3301-type device of greater length.
When using a linear array for edge detection, how far can the linear array be positioned from the object and still be able to detect the edge?
There is not really a clear-cut numerical answer to this question because it depends on the type of lighting that you use, the intensity of that lighting and how far the lighting is from the detector. And all of these things are driven by how much space you have to work with. Two suggestions that I would make are: 1. Locate the linear array sensor as close to the object as possible. The closer the object is to the detector, the better. 2. Use a collimated light source, not a diffuse source. If the light source is more like a spotlight, it will produce a better shadow and improve your resolution. Also, the farther the light source is from the object the sharper the image will be due to increased collimation.
Why does the TSL2550 have two channels?
Conventional silicon detectors respond strongly to infrared (IR) light and have maximum sensitivity in the infrared
range close to 880 nm which the human eye does not see. Accordingly, if conventional detectors are used as ambient
light sensors, inaccurate results will occur due to differences in the IR spectrum of various light sources.
This can lead to significant error when the infrared content of the ambient light is high, such as with incandescent
lighting, due to the difference between the silicon detector response and the brightness perceived by the human eye.
This problem is overcome in the TSL2550 through the use of two photodiodes without the use of an IR blocking filter.
One of the photodiodes (Channel 0) is sensitive to both visible and infrared light, while the second photodiode
(Channel 1) is sensitive primarily to infrared light. An integrating ADC converts the photodiode currents to a
digital output which in turn is input to a microprocessor where illuminance (i.e. ambien
