IR detector system, biasing of photo diode

It's simple: Just do what I say.

This one cost just less than 2USD, goes to 80kHz:

Adding them on a PCB would could be tricky, since reflections still is a major hassle.

I did look into that, but did not go deeper.

The Lorenz beacon was very ingenious.

The solution with a rectangular loop in the base of the docking station worked well, but the client doesn't want that.

Maybe it's time to convince him to look at other solutions again :-)

The advantage of the photo diode solution, is that it's all analog, so can be build and tested quickly, without ordering a PCB

You can solder some wires onto a MEMS microphone.

Optical: imagine that the docking station has a small vertical wall inside a cavity or equivalent. There's an LED set back on each side, so the robot sees one LED or the other as it moves side to side. The robot has a single photodiode staring ahead.

Drive the two LEDs at different frequencies. Too far left, you will hear BEEP. Too far right, hear BOOP. Sort of like the old airplane landing thing.

Looks promising with two LEDs and a business card.

Digikey has mems mikes for 20 cents. They are pretty flat to 20 KHz.

One would bandpass filter, same as you would with a photodiode.

Shouldn't be too difficult. Even reflections travel at the speed of sound.

Yes, that's how the Luba does it:

On the topic of optics, some LIDAR system uses 940nm, since water absorption removes most of the sunlight interference:

Would 940nm be a good choice for a IR system?

A ferrite rod aerial could be mounted so that it discriminated very clearly betweeen two transmitting loops and it could be located on the board with the electronics. You could wrap it in a cardboard tube to disguise it as a capacitor or something and tell the customer the system works by a newly-discovered magic principle.

He is more of a problem than the electronics.

The credit card experiment is morally equivalent to using two square tubes stuck together. Now mount two LEDS, red and green, to the common wall, at the back of the tubes. Shine that at a wall. It should project two fuzzy squares, red and green, barely overlapping, with a fuzzy orange stripe between them.

That could actually be two IR LEDs modulated at different frequencies, or two photodiodes.

TV remotes work very well, so I wouldn't expect s/n problems with the emitter and detector staring at one another a few feet apart.

Use multiple LEDs stacked vertically along both sides of that back wall (which should actually be a PCB) and apply a lot of square-wave current.

We like the Osram right-angle surface-mount LEDs, but there are lots of parts like that around.

He meant conceptually equivalent. Components don't have morals.

It will look orange to a human being because our optical frequency discrimination system sucks.

TV remotes are doing a rather different job.

Why square wave? It's just the sum of the sine wave fundamental, and all the odd harmonics of that fundamental. A triangular wave is just as easy to generate, and while it has the same harmonics,the amplitudes drop off as the square of the frequency. Generating a sine wave isn't all that difficult.

Scarcely the right part for this particular job.

Perhaps amplitude modulate the signal and look at pulse-width in the receiver? Or maybe you can modulate the receiver supply voltage to mess with the sensitivity? Or look for a side channel to read the AGC setting, perhaps supply current? Or have the robot yaw to do a physical search to find the edges of the digital signal.

If you do decide to go analogue, maybe you can run at 32.768kHz and use a crystal in the filter?

One of my summer jobs in high school was working in the electronics shop of the physics depertment of a local university. They registered me as a fake student so they could pay me 70 cents per hour. I learned a lot.

I designed some stuff for Mössbauer spectroscopy, which is astounding physics. Gamma-ray absorption lines turn out to be pretty high Q.

Yeah, Mössbauer was the basis of the GR time dilation (redshift) measurement at the Harvard clock tower (Pound & Rebka, 1960).

Fun stuff.

Cheers

Phil Hobbs