Baud rate with LED & Phototransistor (or diode)

It's really a non problem (as you have discovered) - my son was in my lab the other day working on a UNI project to use the indicator LED from a device as a receiver (the idea came from a paper from Mitsubishi) and no RX hardware other than processor pins. Using 10mA to drive the led on the TX side it could be made to work at 100kbaud, with a few extra parts I'm sure he could have got 1M baud. This scheme used the current from the RX led to charge the sample cap in the uP's ADC, running the ADC at 300k s/s the 0 reading was about 20 counts peak and the 1 reading about 512 counts. (Not suggesting you do it like this - but it illustrates how easy it is to get an optical link. Using a proper RX device and an op amp I think you can easily drop the TX current by a factor of 10.)

Michael Kellett

You said that the device needs to be powered externally, so it might make sense to solve the two problems simultaneously.

In high voltage power transmission system, current is measured with a current transformer (CT), however, taking down the secondary current would require complicated isolation e.g. from a 400 kV line. Sometimes fibre connection is used, a laser on the ground sends light (and transmit data if needed) along the fibre to the CT and a photocell generates the power required by the electronics. The electronics then sends down the measurements on an other fibre. It might even be possible to directly reflect (via a controlled LCD etc.) light from the uplink to the downlink fibre.

Or look at the contactless smart cards, in which the base station powers the card electronics and the card then disturbs the RF field around it, which the master station can detect.

I only mentioned FSK and remote power because it has been done before in a custom chip set. It wouldn't surprise me if somebody has a write up on this.

Then again, there is always the question of scale. You make a couple hundred of thousand of something, it pays to minimize the interface,

It's more like a couple of hundred of, without the "of thousand" in there.

Well, IRDA is a well-studied and supported protocol, with data rates in the megahertz range. I've always wondered why it lost out to Bluetooth and WiFi.

Cheap sensors are phototransistors; cascode 'em into a fast NPN switch for best speed.

line of sight without obstructions is a problem in many applications.

Which still isn't very good. A photodiode running into the base, with a base-emitter resistor, will speed up the turn-off more than a cascode will. It's mostly a recombination issue.

Cheers

Phil Hobbs

Which is why 6N136 is so much better than 4N35. Give or take geometry or something, I'm sure.

You can convince a 4N35 to run as fast as a 6N136 in the default circuit, but you can get a 6N136 going much faster still by similar methods.

Tim

and

Bluetooth

Add much more limited range, and not usably sunlight resistant.

?-)

Ha! Here in Seattle, we scoff at this mythical 'sunlight'!

Yep. In one of my seminars was a fellow from Seattle... had a nice tan which I commented on. He replied, "Tan? No. It's rust" ;-) ...Jim Thompson