More on electronic compound eye....

Working from this circuit, I replaced the pot and the resistor with an LDR (dark resistance 1M or so and filtered daylight resistance 2k or so).

So far with testing during day and night in front of a window, it seems to work fairly well.

Except that it is very sensitive to the direction that the LDR is pointing. If it is pointing away from the window then it totally stuffs up the circuit and you don't get a voltage drop change when you pass your finger over the phototransistor.

Is there a question?

Why do you need two light-sensitive elements in the circuit?


Rob

replaced the pot and the resistor with an LDR (dark resistance 1M or so and filtered daylight resistance 2k or so).

Actually, you did not use an LDR, but rather a silicon photo transistor! The photo transistor is like a silicon photocell, with a spectra response of:

File:Response silicon photodiode.svg - Wikipedia

Looks like your sensor is picking up infrared.

Ray

I replaced the pot and the resistor with an LDR

I think he has both.


Rob

Quote
I replaced the pot and the resistor with an LDR
I think he has both.

I'm thinking not... Would not make any sense. I think the pot he replaced is now the photo transistor... Write-ups and schematic do not asgree, however.

Ray

Yeah, I'm just going by what (I interpreted) he said, which means that the resultant circuit would make no sense. We'll just have to wait for clarification I guess.


Rob

Graynomad:
Yeah, I'm just going by what (I interpreted) he said, which means that the resultant circuit would make no sense. We'll just have to wait for clarification I guess.


Rob

You are correrct Rob - I have both an LDR and an IR phototransistor.

It means that the compound eye, up to a point, has the ability of self adjutment
beteen night time/indoor operation to day time/filtered sunlight operation.

It has been working fairly well in front of my lounge room window during day and
night.

It is obviously not perfect because the LDR responds to a broad band of visible
light while the phototransistor responds to a narrow band of IR.

But clearly there is enough of a correlation between ambient IR and ambient visible
light to make the circuit workable. Am consistently getting an ambient voltage at the LDR - MPSA13 collector junction of 2-3V......which drops when you stick you finger over the phototransistor/IR LEDs.

The circuit is critically dependent on which way the LDR is facing. i.e. towards
or away from the window, and also whether it is being inadvertently shadowed so
I will need to be conscious of this when applying the circuit.

Graynomad:
Is there a question?

Why do you need two light-sensitive elements in the circuit?


Rob

To avoid having to manually adjust the electronic eye between night time and day time, as with the original pot and resistor, it either has to be an LDR or a digital pot controlled from the arduino.

I had an LDR laying around and so I tried it and it seems to work fairly well.

And it would seem far easier to mount a number of small LDRs on a matrix board than it would be to mount a series digital pots.

It is obviously not perfect because the LDR responds to a broad band of visible
light while the phototransistor responds to a narrow band of IR.

Be aware tht window glass attenuates uv and may also stop ir.

Im not exactly sure what you are doing but daylight ldr sensors always use plastic lenses never glass.

But clearly there is enough of a correlation between ambient IR and ambient visible
light to make the circuit workable. Am consistently getting an ambient voltage at the LDR - MPSA13 collector junction of 2-3V......which drops when you stick you finger over the phototransistor/IR LEDs.

Iv re read that and think i understand.

If i point a camera out a window and into the room , i would expect a difference o 2 to 5 f stops exposure depending on size of window and direction of sun.

So i think the ldr pointing in the direction of the eye is the correct orientation for your application.

Would not a sensor that measured in LUX with the spectral response of the eye be more desirable?

I think it would.
What he seems to be attempting though is a simple analogue compensation circuit with components to hand.

I think thought that the circuit he is using is not sufficiently sensitive at low light levels.

Boardburner2:
But clearly there is enough of a correlation between ambient IR and ambient visible
light to make the circuit workable. Am consistently getting an ambient voltage at the LDR - MPSA13 collector junction of 2-3V…which drops when you stick you finger over the phototransistor/IR LEDs.

Iv re read that and think i understand.

If i point a camera out a window and into the room , i would expect a difference o 2 to 5 f stops exposure depending on size of window and direction of sun.

So i think the ldr pointing in the direction of the eye is the correct orientation for your application.

I will have to think about how to put the components together very carefully.

If the LDRs were pointing straight up then clearly the skylight in our kitchen area would work better than my lounge room that has north facing windows and gets very directional filtered sunlight.

It will still be a difficult prospect to get the eye to work in all but a narrow range of lighting conditions.

I may yet have to resort to interchangeable ‘heads’ for my hexapod - one head for even ambient lighting and one for directional lighting for example.

interchangeable 'heads' for my hexapod

I think that's the first we've heard it was on a robot.

I would look at this as two entirely different problems, detecting ambient light and "other" light, whatever that is. Then you have two separate problems to solve that don't interfere with each other.

Then have an algorithm in software that makes a decision based on the inputs.


Rob

Graynomad:

interchangeable 'heads' for my hexapod

I think that's the first we've heard it was on a robot.

I would look at this as two entirely different problems, detecting ambient light and "other" light, whatever that is. Then you have two separate problems to solve that don't interfere with each other.

Then have an algorithm in software that makes a decision based on the inputs.


Rob

I was looking at the digipots but they all communicate with the arduino via I2C or what ever and, if I was to say have 5 'facets'/phototransistors in 2 'eyes' either side of the head, then that is going to get rather complicated fitting the ICs in a limited space plus the fact that I will have multiple I2C connections with the arduino. That I was I have been thinking that the LDRs would be far simpler in that respect.

Are there any other ways of achieving self adjustment that I have not come across or otherwise considered?

The only other way I have considered doing it is by using a jfet as a voltage controlled resistor.

There are 15 PWM pins on the arduino mega, which is plenty, and the interface is simple with respect to code.

Interchangable heads.

You might consider the weston light meter.

They all used selenium cells except the ranger 9 which used a cd ldr and was the most sensitive ever made.

Ranging was done by using an optical attenuator for the high range which was basically a perforated plastic plate , like perf board or veroboard.
Brilliant bit of kit.
Circuit was nothing more than battery and galvo with series resistor and ldr.
Very sensitive to low light levels.

Light has a very large range compared to voltage.

So detection for practical use is going to take some sort of ranging arrangement, from photo experience 2 ranges , high and low should be sufficient.
As previously described this is done mechanicaly by the eye or camera iris for photography.

Perhaps a differential amp with digiot controlled gain could do this.

Boardburner2:
Light has a very large range compared to voltage.

So detection for practical use is going to take some sort of ranging arrangement, from photo experience 2 ranges , high and low should be sufficient.
As previously described this is done mechanicaly by the eye or camera iris for photography.

Perhaps a differential amp with digiot controlled gain could do this.

The purpose of my electronic eye was not going to be for accurate ranging, merely detection of a hand or object in close proximity to the hexapod head which would initiate a 'threat' response similar to that of a bull ant.

I was also going to have a couple of simple electronic antennae that would do the detecting of obstacles.