5
\$\begingroup\$

I have looked at the TSL235 light to frequency converter and for measuring ambient light it seems expensive (~$1.50). What application would you have for using a light to frequency converter vs a standard light sensor (CDS/LDR/diode) to measure the light?

\$\endgroup\$
1
  • \$\begingroup\$ from myself I can only add: entire light-to-frequency family, inluding TSL235 and it's partners: TSL220, TSL230, TSL240 etc, although ubiquitous, has been obsoleted by TI some time ago... vendors still have their lot of them, but I'd stick with newer solutions. AFAIK the current line handling similar tasks is called "Optical Sensors", and consists mainly of OPT101 ti.com/product/OPT101 - available in PDIP for hobbyist purposes. \$\endgroup\$
    – user20088
    Aug 3, 2015 at 11:22

3 Answers 3

10
\$\begingroup\$

The data sheet for the TSL235 points out 2 important advantages of the device: 1. The output is in digital form without the use of external components, and 2. the device can be interfaced directly to a microcontroller. It is temperature compensated, has excellent linearity, and the package includes an integral lens for the light input. Combining all these features with others given in the data sheet makes the price very competitive with trying to achieve the same results with individual components.

\$\endgroup\$
8
\$\begingroup\$

The TSL235 is a subsystem that is self-contained, calibrated, and very linear with guaranteed limits on behavior. See this:

enter image description here

Compare to an inexpensive (10-15 cent) photodiode, the OP980.

enter image description here

Only a minimum is stated, no maximum.

Similarly, an even cheaper phototransistor such as the PT19-21C, the minimum is half the typical and no maximum is stated.

enter image description here

Calibration in manufacturing is something that most companies will try to avoid. It adds cost directly and indirectly- the adjustment labor itself, the procedures and test rigs, and dealing with field re-calibration. A part you can simply drop in and it works is much better, if possible.

So, if you have a limited amount of space, don't want to have to make calibration adjustments, and the application is not especially price sensitive then it might be a good choice. For example, adjusting the backlight etc. on an LCD panel to respond to ambient light. You don't need any extra circuitry to amplify the weak signal, nor any calibration.

For something sloppy and cheap like inhibiting motion detector security floodlamps when it is light out, it would probably be sub-optimal to use such an expensive device. LDRs are traditionally used.

There's another reason to use silicon photodiodes, phototransistors or a sensor such as the TSL235- they don't contain RoHS proscribed substances (cadmium) like CdS LDRs do.

Note that there are much cheaper devices available that provide similar functionality- especially if calibration does not have to be that good- 48 cents with I2C interface (fully digital and does not consume a timer resource in the micro).

\$\endgroup\$
3
\$\begingroup\$

While the sensor itself is a bit expensive, it can interface directly to a low-voltage (2.7 v) digital device, and you don't need a transimpedance amplifier to condition a photodiode, and you don't need an A/D converter.

This device would be particularly useful when used with an FPGA, which can easily provide a frequency counter. With an MCU, not so much. It is also much superior to a CdS/LDR in terms of linearity of response and wide dynamic range.

\$\endgroup\$
3
  • \$\begingroup\$ Although most do, some MCUs don't have an A/D converter: for those using a light/frequency converter can also be very useful. \$\endgroup\$
    – Ale
    Aug 3, 2015 at 7:07
  • \$\begingroup\$ @Ale - Implementing a high-resolution frequency counter in an MCU is not easy. Among other things, you need to be able to disable all interrupts during the count period. \$\endgroup\$ Aug 3, 2015 at 15:11
  • \$\begingroup\$ I was assuming to have an hardware counter (e.g. the TIMER1 on PICs), but indeed not all MCUs have one... \$\endgroup\$
    – Ale
    Aug 3, 2015 at 16:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.