Fast High Dynamic Range Light Sensor

[mribble]

I'm looking for a large dynamic range highly sensitive light sensor.  I've looked through a lot of data sheets and often these types of sensors say they have a dynamic range similar to the human eye which is about what I'm looking for.  Here's an example of one that meets these requirements: http://www.mouser.com/ds/2/38/V02-2315EN-106111.pdf

The problem with that sensor and others I've looked at is their integration time, which is basically just their ADC speed, is too slow.  The sensor I pointed out above has it's fastest integration time being 13.7 ms.  I am looking for something under 1ms.

Is anyone familiar with such a part or know how I could do a paramerterized search for one?  It seems like something must exist, but after looking through a bunch of datasheets I'm starting to wonder.  It would be nice if it's i2c, but that isn't required.

The fallback is to make my own using a photodiode and some opamps, but I'd much rather have a pre-assembled chip.

[Dago]

I'm looking for a large dynamic range highly sensitive light sensor.  I've looked through a lot of data sheets and often these types of sensors say they have a dynamic range similar to the human eye which is about what I'm looking for.  Here's an example of one that meets these requirements: http://www.mouser.com/ds/2/38/V02-2315EN-106111.pdf

The problem with that sensor and others I've looked at is their integration time, which is basically just their ADC speed, is too slow.  The sensor I pointed out above has it's fastest integration time being 13.7 ms.  I am looking for something under 1ms.

Is anyone familiar with such a part or know how I could do a paramerterized search for one?  It seems like something must exist, but after looking through a bunch of datasheets I'm starting to wonder.  It would be nice if it's i2c, but that isn't required.

The fallback is to make my own using a photodiode and some opamps, but I'd much rather have a pre-assembled chip.

What do you specifically want to measure? Do you want to measure intensity as seen by the human eye or just generally intensity or?

It's fairly simple to make your own with a photodiode + opamp. Just a transconductance amplifier + the photodiode and then you can sample that over whatever period of time. I currently have a setup like that on my work project with a 24bit ADC measuring some NIR-channels over a very wide dynamic range.

[mribble]

I want to detect small light changes (0.1%) in a large variety of outdoor situations from nighttime to daytime.

I'm very cost sensitive because I will be making a lot of these.  That's the big reason these inexpensive prepackaged soutions were interesting to me, but maybe the reason they are inexpensive is they don't have any bandwidth in their ADC.

I'll also admit I'm not experienced in this area of analog sensors.  I'm sure I could make something work, but then optimizing for cost (hoping for just a few dollars) is hard without experience.  I've actually thought not including an ADC and instead doing it entirely in the analog space would save cost, but that would also probably limit my software options later (I'm a software guy so I like software options and this project does have a micro to control these with extra 10 bit ADCs).

[katzohki]

If the data speed is too slow, try finding something with an analog output and do your own ADC with a fast microcontroller.

If you need help optimizing for cost, don't be afraid to ask here in the forums. That can be something that is more complicated than it seems and it comes with experience.

[mikeselectricstuff]

Take a look at Avago ADPS9007 logarithmic sensor (analogue output) - wide range, don't recall the speed offhand but I'm using to read ambient in the gaps in a 250Hz LED PWM cycle.

[Precipice]

I want to detect small light changes (0.1%) in a large variety of outdoor situations from nighttime to daytime.

Doesn't that sound more like something AC coupled than a wide dynamic range sensor?
Edit: Well, something with actual analogue HPF & LPF stages. Maybe notching, too, to remove mains and at least a couple of harmonics. Anyway, a bit of appropriate analogue filtering might make things easier, without needing lots of dynamic range, fast conversions, DSP-like maths, all sorts of hard and expensive things.

[eliocor]

If in your environment there is artificial light (incandescent/CFL/LED) you'll have to remove the "hum" made by those emitters.
The sensor integration time is here for a reason...
Otherwise you'll get a slight pulsing output (if using incandescent lamps) or even worse if using CFLs or LEDs.
Please select a rather beefy microcontroller to remove those unwanted fluctuations.

[con-f-use]

You might want to take a look at the chips recommended here: https://www.eevblog.com/forum/projects/good-light-sensing-chip/

[mribble]

Mike, there was a typo in your part (letters transposed).  In case others have trouble finding it, I'll include the correct part number: APDS-9007.  It looks like a decent start if I want to create my own circuitry to sense it.  The standard micro controller's ADC wouldn't be sensitive enough to make that work without additional circuitry.

Precipice, I think I understood 20% of what you said   Sorry.  I either need to go study for a few hours to figure it out or you need to remember I'm not an analog dude.  You need to take it easy when a guy says he's a software person.  From my perspective just tossing an ADC to get the full analog resolution into digital is simple.  The problem is I know that's also the most expensive option.

Confuse, that was a helpful link.  Nothing fit my needs exactly, but this one seems pretty good except it has a 3 ms integration time (best I've found in that class of sensor, but still slower than I want).  http://media.digikey.com/PDF/Data%20Sheets/Austriamicrosystems%20PDFs/TSL2571.pdf

Thanks for all the help!

[mikeselectricstuff]

Mike, there was a typo in your part (letters transposed).  In case others have trouble finding it, I'll include the correct part number: APDS-9007.  It looks like a decent start if I want to create my own circuitry to sense it.  The standard micro controller's ADC wouldn't be sensitive enough to make that work without additional circuitry.

Depends on what resolution you want. I just fed it into a PIC ADC (10 bit) with ISTR a 47K pulldown resistor.

[tggzzz]

A reverse-biassed PIN photodiode driving a FET-input op-amp in transimpedance mode will give you high linearity, high dynamic range, and high speed. Very simple and easy to make.

Then use an ADC of your choice to digitise the transimpedance amplfier's output. Your application might require a switched-gain amplifier in order to balance dynamic range and quantisation noise.

If you can relax the linearity and speed constraints, you could use the un-baissed PIN diode in logarithmic voltage mode, followed by a voltage amplifier.

Naturally there will be the tradeoff that higher speed requires higher bandwidth => higher noise => reduced dynamic range. Either your application can tolerate those fundamental physical limits or it can't.