Infrared LED / detector design

Wow, I had no idea I'd receive this much help here. Looks like I found the right forum!

I really don't want to seem like the typical forum dullard and have y'all hold my hand all the way through this. Anyways:

I built an emitter driver circuit from an 555 I had in the parts drawer. It works very well and seems to happily source upwards of

150mA.... so I'm planning to skip the transistor and just use that to drive the emitter.

The detector circuit is being a little annoying however...... or maybe I'm just being perfectionist again. I'm using a phototransistor, collector to ground, emitter to +5VDC through a 1k resistor. I pick the output off the emitter to my scope.

So the leading edge has a ramp to it, the trailing edge is just what I expect. Larger values of R-sub-L increase the leading edge angle, while an R-sub-L or 0ohms will make a nice square wave while pulling

30mA....... unsuitable for a battery powered application especially considering this is only a sensor. ARG!

Anyways, I believe the advice is correct. The emitter and detector are right beside each other so I don't need to make this any more complicated than I need to.... I've got the timing circuit right there for use on the detector circuit.... I'll probably do that, run a HI/ LOW pass filters and be done with it..... Unfortunately I've been fighting with the detector square wave thing so I haven't gotten to that point...... I considered cleaning up the square wave with a one- shot, but that will trim the leading edge.... I dunno.

I'll bookmark all the Ir modules you listed. Thank you. I haven't given up on components just yet. I'm still having fun :)

Oh, I pointed a TV remote at my circuit and could see the data packet. COOL!

Try using it as a photodiode. Leave the emitter or collector open circuit and use two remaining pins as a photodiode. I forget which junction works better; it seems to me B-E is supposed to absorb most of the photons, but B-C has better electrical characteristics (i.e. will stand more than 5V reverse bias).

The downside to photodiodes is the low current, which makes voltages move slowly. To solve this, you use a transimpedance amplifier, so the diode drags a few microamperes out of a "virtual ground", not really needing to produce any voltage gain, while on the other side of the op-amp, real volts appear, along with as much current as you'd like. By running the photocurrent into a base-emitter junction (as on a phototransistor), you increase the current by hFE times, but hFE varies considerably, so you get poor results.

If you can figure out how to process your signal primarily in the current domain (current sources, sinks, mirrors, etc.), you might get acceptable performance without drawing too much total. It's also potentially easier to handle wide range signals, since you only have to worry about leakage currents (which can be arbitrarily small, given a good transistor choice) instead of voltage offsets. Of course, this is something chip designers are better at, not the least reason for which is because discrete transistors don't match as well.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

Don't know if it is of any use, but MT8870 clones are still available.

You could make 40 kHz primary modulation and DTMF on that as secondary.

..

uit

rks

e

ts

t

er to

s are

The Texas Instruments website has lots of tips on TIA for photodiodes, mostly from their Burr Brown acquisition. Even if it takes actual money (ack!), I suggest using parts with specifications. Digikey or Mouser will do the job.

Usually for a project that you release into the wild, you never depend on one safety mechanism. For instance, if the thing is going to run into a wall, you would have your photo sensor scheme and maybe a simple bump detector, even if it is just a plain old switch. Put a big red button on the top of the robot for a master power kill. Nuances like this separate the skilled engineers from the Phil Allison's of the world. ;-)

On a sunny day (Tue, 23 Feb 2010 02:04:05 -0800 (PST)) it happened " snipped-for-privacy@sushi.com" wrote in :

Some insects have these feeling 'hairs', something like this:

/ OOO \

Som pieces of spring wire (not pointy, maybe with a rubber tip), and a microswitch could simplify things.

Jan Panteltje a écrit :

microswitch

Nah, just leverage the so criticized ROHS and let your Pb free prototype grow whiskers.

--
Thanks,
Fred.

.

What's the time base on that 'scope pic? Or, at what frequency are you chopping the LED? As Tim says a photodiode is faster than a photo- transistor. You've also got a lot more choice in area size with photodiodes. The other thing to be aware of is that IR LED's tend to be slower than their visible cousins. So that delay and turn-on time you are seeing might be from the LED. You can test this by blinking a red LED at your circuit and seeing if the respose time changes.

George H.

croswitch

I didn't want to get into it, but I would try bumpers made out of coax. Stiff enough that they are self supporting, weak enough that they will flex. You can get a small voltage generated by flexing coax since the cable has a piezoelectric property. This requires a high impedance amp. I had a sample of some coax especially designed for this "feature", though I don't know if the product made it to market. Here it is from a different manufacturer:

On a sunny day (Tue, 23 Feb 2010 13:49:48 -0800 (PST)) it happened " snipped-for-privacy@sushi.com" wrote in :

Nice stuff! Of course it depends a bit on the speed of the robot, I think for low speed you can use this cable, for high speed you need to start breaking - or steering away earlier, and maybe use IR or the feelers on micro switches.

[...]

The simplest collision detectors I have seen are based on spring wire ("piano wire") like that sold at hobby shops for a few dollars. You put a loop on one end and screw that down securely, then place the other end between two screws or a hoop of wire to form the other side of the switch; depending on your needs, you can bend the wire to make it into a NO (normally open) or NC (normally closed) switch. You will need to "debounce" the switch, just as you would with any mechanical switch.

Here's one of the odder "antennae" I've seen, where the wire is attached to a microswitch (which also needs debouncing).

How to Hack a Rumble Robot (with video)

"Antennae" like this have one problem, though: they tend to get caught in things and/or get bent. If your robot shape allows it, you can bend the wire so that only the "middle" is exposed:

------------------ (very sloppy "bow") / \ Body ====== / ================= \ ==== (=) Mounting (=)\(=) "Contact" screw \ screws

You still need to debounce the contact, but there is no free end to get tangled up or tripped over.

Then there are "flex" sensors that change resistance based on how much they are bent. And hundreds of other approaches.

However, for anyone starting out with robot building I'd strongly recommend choosing a simple bumper switch design and only adding one or two before spending a lot of time thinking about the fancier stuff. It's all too easy to get so sidetracked by dreams of sophisticated hardware that you wind up losing interest in the project because it takes so long to finish. On the other hand, if you start out "simple and stupid", the emotional feedback from seeing one's creation mobile and making noises -- even if it is occasionally bouncing off walls and running into the cat -- helps keep one involved, and provides an incentive to keep improving things bit by bit.

Good luck!

Frank McKenney

--
  The tendency to achieve our foreign policy objectives by inducing
  other governments to sign up to professions of high moral and
  legal principle appears to have a great and enduring vitality in
  our diplomatic practice.
                -- George F. Kennan / American Diplomacy 1900-1950

The time base is 10ms/DIV, 2V/DIV (both channels). The top waveform is the emitter driven by a 555 at roughly 30kHz. The bottom waveform is the detector circuit output.

I'll have to see if I have any photodiodes.... Or if I can get any locally. I've got an order put together for DigiKey.... kinda waiting on money though.

..........sorry for the delay y'all. I didn't realize this conversation had gone to page 2!

d

Good point. I'll do that, especially to keep me from burning up the drive motors.

mi=3D

I actually did this (Revision #1) with two lengths of straightened coat hanger. I works fine. I just wanted something a little more nerd- like :) Plus the Ir is something I've always wanted to play with... This is just an excuse to get started.

Are you serious!?!? Wow, that's incredible..... But now that you mention it, that's how many strain gauges work.

Ya, it's no making any skid marks on the floor, but it ain't slack either. At it's speed a collision would shake things up a bit. The limit switches from the original design switched BEFORE it hit the wall. The Basic Stamp reacted in uS since all it was doing was waiting for the switch. It'd be nice to have the 2-3" of Ir range.

Thank you Frank.... very cool.

I'll see about posting a video of the bot.... It's nothing fancy..... I don't have a Youtube account, but that's easy enough to remedy I guess.

o-

d

If money is an issue Phil H. says you can use solar cells as photodiodes. I've never tried this and I don't know about the speed issue. You could just run things at a lower frequency.

Photodiodes use to be cheaper, the price seems to have doubled in the last few years. George H.

What!? No kidding? Interesting, I've got several small 0.3V cells.... I'll see if it works, and at what frequency

cool :)

BPV22NF is good medicine for 900 nm-ish, and almost free.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net

Just an update on building this system from components if anyone is interested:

I have the emitter driven by a 555 timer at 30kHz

The phototransistor is outputing a respectable square wave, inverse of the emitter.... The Ir is reflecting off a surface about 2" away (for now) and it was apparent I needed an op amp. No problem though.

The High and Low pass RC filters are a little broad in their descrimination (SP?) but I'm still going to play with them some more.. Right now it looks like the RC filter will work if I separate my two sensors by a modulation frequency of at least 20kHz.... I was hoping for a little more like a 1kHz or 5kHz at the most difference between left and right.... but we'll see. It would be nice to use the 555 I already have setup to modulate on the detector, but I haven't figured out how to do that yet.

I need to look back through all the advice given here again to pick info...... and I found all the components advised on DigiKey.... All of the tuned receivers we out of stock when I checked though. I need to check it again and look at Mouser for the same components.

According the Phil H's book the solar cells are slow.. He quotes a 3dB corner at 20kHz. But this is using his transistor cascode trick on the front end. The nice thing is you've got a lot of area! Mcuh more than your phototransistor.

George H.

oto-

end

t -

Ahh, speaking of the master. Thanks Phil I'll order a few. I was getting PIN diodes from photonic detectors, But they sold out to someone/ slashed the product line an raised the price. So now I'm using diodes from OSI optoelectronics.. (which use to have a differernt name.) I should have tried these years ago because they seem much better than the previous diodes. (better in the there is less C with the same active area.)

George H.