cmos image sensor with direct output.

hi. i’m looking for a cmos image sensor with analog output for a project. i need one where the output continuously varies to reflect the light level on the currently selected pixel.

I’ve never heard of such a sensor. All of the CMOS image sensors have the pixel value clocked out by rows and columns.

It may be possible to build a custom FPGA to read out only the one pixel value you are interested then send the value to a DAC to get an analog Voltage out.

If you only need a single pixel value why not just use a very small photodiode?

i need to be able to read a varying level before moving on to the next pixel

How many pixels are need to cover what FOV ? I don’t know of any image sensor that doesn’t output it’s data in some time sliced way. The exception might be some line sensors (an array of 1xN sensor diodes), that might output 1 selectable sensel at at time. N might range from 4 to 100. And the array was IR … IIRC.

deepthought:
i need to be able to read a varying level before moving on to the next pixel

The easiest way is to read out all the sensor pixels (an image) and measure the pixel of interest.

This can be done with just a Web Cam and some software.

i intend to make a depth camera by having a phase shift measurement circuit hooked up to the output, then feeding that into a microcontroller.

after some frustrating googling

http://www.datasheetarchive.com/dl/Data … 476850.pdf

and

http://www.datasheetarchive.com/dl/Data … 013769.pdf

look like they would work, but i have no idea where to get one.

Let me guess a bit here. You’re going to have an illuminator, AM it’s output and detect some relative range by comparing received to illuminator phasing ? Not unlike a CW radar. Does any imaging chip give you realtime power readings ? Don’t they all just accumulate photon knocked electrons, effectively acting as a rectifier and cap, stripping out any higher frequency AM ? An if the AM is low enough frequency, can’t you just use a normal movie imager ? What kind of shutter speeds will be required if you wish to range gate your image ? Or are just going to compare the relative amplitudes of illuminator in all the pixels to get relative depths (after accounting for the varying reflectivity of the scene) ?

http://xxgcxy.gdut.edu.cn/YPJ/151projec … ga1000.pdf

on page 5.

wouldn’t the pixel stay connected to the output amplifier until INCR_X or INCR_y is pulsed?

Yes a pixel can stay connected.

Think about the nonlinear response, and what that means for your calculations.

Note also the time constant (datasheet 5.1), which may severely limit the acquisition speed. You may have no more than 100 pixel readings per second or so. But apparently, you can reach higher speeds if you have more light.

If you really want a Fuga1000, you can try to contact Cypress Belgium. They took over Fillfactory some years ago.

Have fun!

Sorry, I forgot one buy-out. Now On-Semi took over the former Cypress image sensors who took over former Fillfactory.

Anyway that’s not much help since they do not offer Fuga1000 anymore: http://www.onsemi.com/PowerSolutions/pa … ?id=101682 .

If you really need to stick with pixel-by-pixel processing, I’ld suggest to move to the conventional integrating types of sensors. Especially black/white, with a built in 10bit-or-more ADC and not too many automatic functions. The phase detection moves into the microcontroller code. The integrating pixel needs a reset after each readout which should be easy to combine in the sequencing of the ADC control signals. You now only need a cmos sensor that allows the RoI (region of interest) to be 1x1 (or maybe 2x2).

Suddenly you should have plenty of choice on digikey, rs components or maybe even here at sparkfun.