Let's Make Robots!

PhotoTransistors (Top or Bottom?)

I am following in Mr. Oddbot's footsteps and once again, walking the road of cheap, IR-LED based obstical sensors. Put simply, I can not seem to replicate the results of the work of others. Let's do some bullet points:

  • Want to build simple, 2 or 3 phototransisor object sensor
  • Have been folowing This Post both to the letter, and with many variations
  • I am basically (no, actually) remaking the Dagu Compound Eye, but smaller and "on edge"
  • For bench testing, I am using only one "set" of parts (1-4 IR LED's, 1 or 2 Phototransistors, one ADC)

Problems:

  • The Compound Eye works incredibly well, indoors it can catch my hand 10" (25cm) away. 
  • I have replicated the Compound eye's circuit exactly (each part matching the specs of the parts used on the C.E.) but I cannot get the ADC numbers like the Compound eye. My set-up has maybe 5" (13cm) range, with a total output fluctuation of maybe 1v. I.e. 2v -3v total range from nothing in front to white paper 3cm away.

Tried/Questions:

My first question is why the NPN phototransistor is on the "top" of the circuit. Is there an advantage to this location? In every schematic I see, we have collector to 5v+ and the emitter to a pull-down and ADC. Every other NPN-thing I have ever done had the collector going to the ADC/Pull-up and the emitter going to ground. Why is this different?

What's up with the transistors? I used countless "general purpose" transistors in my tests (the tests above were done with no transistors) and I could see no effect in the output. I went through 2222's, 3904's, 4401's and a darlington chip. All worked (and I know we are talking a volt/current issue here as well as impedence issues) but I could see no change in output on the O-scope or ADC numbers by adding a transistor. The total range of voltage outputted was basically the same as a straight output of the phototransisor when using say, a 2n2222. It seemed to be performing the task equally well with or without this "amplifier" circuit.

To finish this off, I basically tried every configuration of this kind of sensor. Pull-ups, pull-downs, transistors, transistor-to-transisitor-to-transistor set-ups, pull-up/down resistors from 1k to 10M, etc. etc, etc.  --All in all, I could not seem to do better than this 5" (13cm) range nor did I ever see the kind of numbers I can see from the compound eye. What the hell am I doing wrong, and how do I make a simple R/L or R/C/L IR obstical sensor? Sheesh! I used to know how to do this kind of stuff.

 

--You guys using magic parts or something, Oddbot?

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

To begin with, do not follow that old post! I could not duplicate the results.

Follow this post instead: http://letsmakerobots.com/node/10822

I put the phototransistors to the top so that as the light got brighter, the value would increase. You could put the transistors to the bottom but all that will do is invert the signal.

 

I have re-built the compound eye as well and it works even better than the original one, because of the arrangment of the LED's/photo transistors.

The photo transistors are connected in this way only for logical reasons: No light - 0 V. Full light - VCC. You can use the inverted configuration (emitter connected to ground and the resistor connected to collector) and would get: No light - VCC. Full light - 0V.

I have revisited this recently myself and it was just as frustrating for me. I've tried to build a distance sensor a few times using the Ir receivers with digital output but failed miserably. So I decided to buy some phototransistors instead.
I couldn't get the same phototransistors as oddbot uses in his compund eye. They give a much better voltage response than the ones I bought, being only in the microamps they needed an amp to work with.
So until I'm feeling inspired again and placing an order for other parts I have suspended my experiments with this.
I had thought of buying a compound eye. At the price I could steal the parts but I think they come assembled now don't they?

Chris, I have a Dagu compound eye and have played with it. My results varied depending on the light in the room. I have some fancy halogen lights above my desk. When I turn these on the compound eye is useless. When I test in a room with normal lightbulbs I get reasonable results, but I have never been able to manage to track a white card at 25cm...

Yeah, I know...  "Obsticle"...

Obstacle  heh.