Photodiode Question

Lots -- I did something like this a few years ago and saw several milliamps.

But as you really should know, the real answer is "it depends". It depends on the diode area, whether it's shaded, what the customer thinks "bright" is, ect.

Order a few from Digikey, cobble up something with a 9V battery and an ammeter, and go measure it! Then use the sensitivity on that data sheet to extrapolate to what you'd expect from other similar diodes.

One thing that you will learn is that your eyeballs have an amazing capability to accommodate changes in lighting, and that "bright" for a room doesn't hold a candle to "bright" for a sunny day.

As Tim said, "It depends". Tim mentioned the "obvious suspects" - how big is the photodiode? (1 mm x 1 mm or 1 cm x 1 cm - it will make a difference...), how bright is "bright"? (IIRC, the eye's photoresponse is logarithmic, but a photodiode is linear over several orders of magnitude in illumination), etc.

One thing Tim didn't mention is the frequency response of the photodiode. Do you and your client only care about DC illumination, or does the photodiode have to be able to deal with the illumination switching on and off at several tens or a few hundred kHz. In general, if only DC is of concern, the photodiode can be designed to have better responsivity - the current out per unit optical power in.

Also, with the photodiode be reverse-biased or zero-biased? (Reverse bias will improve the responsivity) but not applying an external bias voltage across the PD will make things simpler.

Depending on what the client's needs are, you might want to consider using a photoconductor instead of a photodiode.

Bob Pownall

Photodiode could be not a very good idea, as they depend on everything. A good part for luminance detection is TPS851 from Toshiba. See the datasheet for details. If you want to isolate the light sensor from being illuminated by your own backlight, you can time multiplex the backlight and the sensor.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

For simple detection of ambient light level with no requirement of high linearity, I find LDRs more sensitive and easier to work with. They correspond fairly well with human spectral response too.

Those guys are all nuts.

For backlight control, you can use one of the ambient light sensors such as the Intersil ISL29000 or TAOS TSL2560, for instance. They look after all of that nonsense, and produce an analogue or digital output proportional to just the perceived brightness (i.e. luminous intensity), without being fooled by all the IR from incandescents, for instance.

Full zenith sunlight is about 800W/m**2, so figuring 0.3 A/W average responsivity, a typical 2.3 mm square photodiode such as a BPW34 will never produce more than ~1.3 mA without optical concentration of some sort. Indoors it's a factor of 10**3 to 10**4 dimmer than that, so you're looking at a few microamps in bright room lights.

Cheers

Phil Hobbs

Then I'm sure glad I didn't make any suggestions.

John

Nah, it's just my rhetorical way of trying to get people to do a bit of calculation (or even experiment) before giving advice. I know, I know, I'm on Usenet, but we must live in hope. ;)

Cheers

Phil Hobbs

=A0 =A0 =A0...Jim Thompson

Hello Phil,

Maybe they aren't nuts. A good ambient light sensor (with human eye sensitivity and lambertian radiation pattern) is more expensive than just a photo diode or photo transistor.

Searching for "ambient light sensor" or "daylight sensor" will probably give some afffordable components. I agree that just a photo diode is not a good idea as their maximum sensitivity is mostly in the IR range and radiation pattern will be too far from lambertian.

Best regards,

Wim PA3DJS

There are certainly cheaper ways to measure light, at some level, but it depends what you want to do.

The ALS parts are way under a buck in volume, and come in chip scale packages if you like. That gets you the right spectral response, built-in PGA and ADC, with an I2C interface. They're slow as molasses, of course, but that doesn't matter in that application.

Night lights use CdS or CdSSe photoconductors, but they don't have to have any kind of stability or repeatability, and they don't. (CdSSe's resistance can vary 5X due entirely to previous illumination history.) Phototransistors have somewhat similar issues, due to the poorly specified beta.

Cheers

Phil Hobbs

That could be made to work as well. ...Jim Thompson

My task is to produce a PWM output proportional to a photodiode current.

I guess I need to find a typical one and measure it myself. ...Jim Thompson

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

Well, you didn't specify what kind of proportionality, so I guess I win firsties.

Tim

Geez, I haven't futzed with PUTs in ages. Does anybody still make them?

... for certain definitions of "PWM."

But yeah, PUTs were fun.

John

PUTs aren't that hard to find. Maybe you're thinking of UJTs?

Anyway, if they're on the expensive side ($0.25 in singles?), you can Jeorgify your own from a 2N3904 and 2N3906 ($0.06/ea in singles)... the equivalent circuit is as indicated (i.e., like an SCR, but take the second junction).

Tim

=A0 =A0 =A0 =A0...Jim Thompson

er

=A0 =A0 ...Jim Thompson

=A0 =A0| =A0 =A0mens =A0 =A0 |

=A0 | =A0 =A0 et =A0 =A0 =A0|

=A0|

=A0 =A0 =A0 |

Hello Jim,

For the experiments, take one (Photo diode/transistor, LDR, ambient light sensor) with a wide viewing angle (somewhat over 90 degrees). Measure the response also for a filament lamp, fluorescent tube/lamp, white LED lamp and sun under low elevation angle. You may also experiment with white plastic as a diffuser. It reduces sensitivity, but increases viewing angle when using narrow viewing angle photo diodes/transistors.

Compare the values with you own eye perception.

Good luck with the experiments and best regards,

Wim PA3DJS

Except that's current-to-frequency :-) ...Jim Thompson

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

On-time is constant, so it's PWM. You didn't say if frequency had to be constant. No specs, remember? ;-)

Tim

I recall a recent lawsuit (maybe the Hynix/TI thing) over a design for an ambient light sensor chip that used semi processing to get human-similar wavelength response.

Scads of people offer ambient light sensor chips.

John