IR Receiver/Demodulator

If the selectivity of the crystal is a problem, it's that it's too selective.

Indeed... I can utilize an 8MHz crystal oscillator. I sacrifice about 20uA supply current by doing so but it is an option.

--------------------------------------- Posted through

I haven't used IR in over a year but I remember getting pretty good range with a TSAL6100 and a PNA4602M detector. Carrier pulsed at 10% and 1amp; modulated at 40%.

If its a remote then its not continuous transmission so battery power shouldn't typically be an issue.

Its odd that the LTE-5228A datasheet doesn't spec radiant intensity for high current pulses. Maybe its not the best emitter for the job.They only provide specs for 20mA that's useless for long range transmission.If an emitter is rated for 2A you would figure they show some specs at higher current pulses. When they don't it makes me wonder.

For example here is the datasheet on the tsal6100.

I believe the AVRs (ATtiny series I'm thinking of) have internal oscillators that can be calibrated via software (you would do that once at the testing stage). There's still a small variation with temp/voltage but the absolute frequency would be closer. Worth looking at them and comparing prices and current comsumption. Also AVRs are single clock per instruction which could mean you can knock the frequency down (programmable via an internal register) and that would save more power.

Mark.

So anything tighter than about a Q=10 could cause you problems. ...Jim Thompson

Yes. The bandpass response of the LC reduces the system noise bandwidth. But optical bandpass filtering will help a lot, as background light will make wideband shot noise, some of which will sneak through the LC with the signal.

Usually, for low Q situations, with a shunt resistor. You could reasonably have an LC with a native Q of 100 maybe, and kill that down to 10 or 20 maybe with a resistor.

How much capacitance does your photodiode have at its operating bias?

John

A crystal filter in the receive path will have too much Q and not let his pulses through.

John

The terminal capacitance is stated as 70pF with the conditions of Vr=0V and f=1MHz. Obviously neither of those conditions apply. There really is not much to the datasheet (PNZ323B)

--------------------------------------- Posted through

Yeah, true, 20-80msec could be a tight squeeze. But LC can also be pretty steep and stable there at 30-some kHz.

if

he

and

at

it

ed

at

#2 was the first thing I thought about too.

Mainly, because earlier today, I purchased a new camera lens (a 500mm mirrror), and I always make it a point to grab a couple Roscolux swatchbooks while I'm at it. Such as this one from BHPhoto:

Hard to beat at $1.95 each. I wonder if there's a gel in there that will do the job?

Probably not - most of them are intended for stage and photo studio lighting.

Kodaks Wratten 87 is a generic IR pass filter. For a quick and dirty trial unexposed and developed slide film is usable. But you could do better with a dedicated long pass IR filter like Schott RG800 or RG850 (or much cheaper cast plastic IR dye filters).

Narrow bandpass interference filters can get expensive. Laser line tuned ones 10nm fwhm are about the cheapest at ~£50.

Regards, Martin Brown

The TSOP module is encased in a black plastic resin (and so is the pin photodiode that I have for that matter). Isn't the purpose of that for creating an optical filter which passes only the band of interest?

--------------------------------------- Posted through

Not having any IR LEDs at the time, I once used a red superbright. Planted directly in front of the photodiode (alignment was fairly critical) and pulsed at 100mA (more than a few Cd worth), I got a usable signal out of the diode. :-)

Tim

You haven't given much detail, but this is beginning to look like a very bad idea.

You're trying to detect a tiny signal in a sea of high intensity noise. What are you using for error detection/correction? Small percentage modulation of the ambient light can wreak havoc with your input amplifiers. Reflection from passing car, shadow from bird flying overhead, bee flying thru the signal path can all saturate your amps long enough to trash the pulse width measurement.

If it's a handheld remote, you've got lots of other issues. If it's fixed position on both ends, have you considered 80 feet of wire? Sometimes low-tech is best.

Not really. It is a crude dye based long pass filter which cuts most but not all of the visible light. Some have a nominal passband fwhm ~50nm at

Easy enough to test it with a white LED torch to see how good the built in filter is at blocking visible light. A layer of black slide film in front might still give you some additional signal to noise.

Or alternatively put the thing out in full sun and measure its operating point with and without a thick black cloth covering the sensor.

Regards, Martin Brown

That isn't a good geometry unless you intend that the remote operator levitates. Your sensor is pointed skywards and the sky is bright. To one side or the other would be better assuming that the thing is in roughly the same plane as the controlling transmitter.

It is only worthwhile if you can increase the current proportionately without going outside the operating envelope. The optimum duty cycle is around a 37% pulse width at correspondingly higher current (puts a shade over 10% more power into the fundamental carrier frequency).

Regards, Martin Brown

Spent a day playing around with the optics which yielded no major improvements. The only thing I was unable to try was the optical filter (as I don't have anything suitable on hand). So, while I wait to get my hands on something I thought I'd try a few of the other suggestions... if nothing else it would be a learning experience.

First I am making a change to my transmitter so that I can use a watch crystal as the time base of the modulation. I've basically added a Pierce Oscillator with its output going to one input of an AND gate. The other input of the AND gate is tied back to the existing microcontroller acting as an enable. Upon enabling the signal is fed to the gate of a FET controlling the LED. Here is a snippet of the schematic...

I am working on the receiver end now and have a few questions. The basic topology would be a reverse biased pin photo-diode loaded with a shunt resistor. The voltage that develops would be run through non-inverting amplifier (perhaps multiple stages) and then through a crystal filter. Again, I have no real analog skills but I can see that loading it with the resistor will lead to saturation from ambient light. What if I was to replace the resistor with the LC tank that someone had suggested. The ambient light is likely to show as DC right? So, I now have a low impedance path for the DC current but my modulated current shows quite nicely. Also, the thought is that I could try the receiver with a crystal filter (extremely high Q) and then without (Q of the LC tank). I am sure there are some flawed thoughts here... any comments?

--------------------------------------- Posted through

As an aside, if you're bandpass filtering 32KHz, could you use a large capacitor from the junction of the resistor and LED to ground, such that when the FET turns on you get a high powered shorter spike? It seems you can't drive more than about 200mA with the LED shown as it's a 50/50 waveform. You're allowed up to 2A though with a 10us pulse. If your photodiode can respond fast enough to that shorter pulse it might mean you can turn the gain of the receiver down and reduce background noise effects. You'd need to do a fourier analysis of the 50/50 waveform at lower power for the 32KHz content and compare to the 32KHz content of a shorter pulse but at higher power. Alternatively use a monostable to create a controlled pulse width and up the current. I'm curious whether that would work or not...

Mark.

Whoops, had a brain freeze, I meant kHz not KHz. I could see there was something wrong but couldn't figure out what. Strange thing is word blindness..