IR Receiver/Demodulator

Well, that's why I suggested an inductor in the TIA FB path earlier. That is your DC loop. You really need to get rid of DC immediately following the photodiode.

spike

power

gets

LED

capacitor?

The battery delivers about 240mA based on the voltage which develops across the

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

More I think about it, the more I like using an inductor for energy store and discharge into the LED or a chain of LEDs for more bang.

Extra components are the RS latch and comparator for peak current -- 32k clock sets latch and MOSFET on, and peak current resets latch, ready for next edge. But there's already half a quad gate package there for oscillator and gated drive now, so maybe it's doable?

Grant.

It's already been mentioned but using 32khz crystals for filters works very well if you don't care about response time. Typical Q's are around 50,000 so bandwidth is less than 1 Hz. Too narrow for the op. Inverter oscillators are cheap, but suck current so if current consumption is a concern avoid them, they are especially poor at 32khz. If you must use something like a

Another important note on using 32Khz crystals for TX and RX is that they both need to be in the same resonance mode, either serial or parallel. Because the bandwidth is so narrow the frequency difference between operating one in serial resonance and the other in parallel resonance means you take a big hit in sensitivity from being off frequency from each other.

Back more toward the op problem. My own experience agrees with previous suggestions - optical filters help a great deal for sunlight situations. As mentioned optimize your optical path with filtering and directionality as much as possible.

Then toss it all for an RF link when you get frustrated enough.

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

message

filter

hands

watch

Pierce

other

acting

=20

My

capacitance=20

that

you

you

effects.

power=20

but=20

pulse

cr2032=20

capacitor=20

=46rom the datasheets:

Is able to provide about 200 uA continuous. OP's pulses were about 80 ms= at 200 mA. Duty cycle should be less than 0.1% max. Or about 80 seconds between = transmissions. A 5% droop at 3 V is 0.15 V required C follows naturally. About 1.33 = Farad.

OP clearly, is not getting the output level expected, by about 1000 x.

I suggest 3 AAA in series to get the voltage and the current capability = desired.

signal

more

to

across

as

The datasheets i have perused do not support that cell producing 240 mA, = but=20 more like 240 uA. After all the total charge is on the order of 240 = mAhour,=20 and the internal resistance seems to be over 100 ohms.

They'll easily give 10mA or more on load. Product we used them in was UHF AM tx and we used stack of two to get enough power for decent range. I don't recall the detailed numbers as this was back in the '80s.

Grant.

I don't agree at all. Having low gain at DC is fine, to avoid saturation, but if you have zero dc gain you can't tell whether you have no signal or a hardware failure.

But the big problem is nearly always noise--shot noise of sunlight or differentiated amplifier voltage noise.

Cheers

Phil Hobbs

Well, for diagnostics you can let a wee bit of DC sneak past or provide a FET that breaks the inductive path during a test.

Now I don't want to be facetious here but: You mentioned the shot noise pollution. Wouldn't that present enough signal to ascertain that the photodiode is working? I can't imagine it dropping to zilch even at night. If there was a full lunar eclipse maybe but then your DC level would also drop to zero.

Ok, true, but we'd have to quantify that. I have seen IR comms (with much more data rate than here) work clear across a soccer field in Europe. Now I don't have a clue about soccer and field size, but it sure was huge.

Baffled telecopic optics, optically narrow-band filtered, polarized, and operating at a crystal-controlled narrow band modulation rate that was tightly band-passed at the receiver?

Jon

Nope, nope, and nope.

Yes, there was a filter but not super fancy. Somewhat matched to the data rate but the shape factor was nothing to write home about, simple LC circuitry on the cheap.

Yes, but that noise is nor correlated at the carrier frequency, so it can filtered out, no?

Grant.

Are you sure? The detector itself might have been chosen for its responses over wavelength.

Hmm.

Jon

In one of the referenced articles upthread they were bouncing signals off the clouds, but they moved up to red to avoid the water absorption of the IR.

Grant.

...and every time they turned it on Batman showed up.

It would seem that they'd need a *lot* of transmit power to make up for the scattering, which they must be counting on since clouds are quite amorphous.

Down in the kilohertz, it's hard to get enough inductive reactance for this job. A sub-poissonian current feedback loop works OK.

It would if you have something sufficiently smart attached to it. If you have a DC path, all you need is a comparator.

It isn't that hard to do if (a) you have a sunshade, as you've suggested already, and (b) you use lenses to corral some reasonable number of photons onto a reasonably small detector.

RF works better for many of these things because the effective area of an isotropic detector goes like (lambda)**2. Shorter wavelengths mean better resolution but require more precise aiming.

Cheers

Phil Hobbs

I am sure. But I don't remember the photodiode characteristics, it might have been chosen for the purpose.

LC is fairly easy and stable down there. Mechanical filters are sort of rare antiques by now, I wouldn't even know if anyone still manufactures them. Back then Collins did.

Heck, nowadays you could even do active filters. Or take a Cypress PSoC and use the S/C filters in there. The uC part would then almost be a complimentary bonus.

Yeah, it probably wouldn't be so great in a thick London fog. Unless you have enough energy to vaporize the water content :-)

...

Oh yeah... :o)

It was big, side by side fresnal lenses (tx & rx) each about a foot square, I think it might've been a 5W LED they used.

Grant.