IR remote circuit running out of steam?!

[Gorgon]

Hi Mike,
I've just done some testing on your circuit. I built it this evening and I connected it to my old basement computer. I don't have your software on it, but used a terminal program to send normal serial data. I used 9600Bd, a speed I think will be a good simulation of your IR codes.<p>I connected the +V (GND) line to pin 5(GND) on the 9pin D-sub connector, and the -V(TXD) to pin 3 on the connector.<p>It took some time to charge the capacitor, after 1 min the voltage over it reached 10v. The capacitor consist of 2x 2200uF or a total of 4400uF. I use the LD271 IR-LEDs, but can't see the IR output, only scope the voltage over them. <p>I measured the voltage over the 22 ohm to the correct level around 0.6V. This will give a current through the IR-LEDs of 27 mA. This is not very high so I added one 1N4148 diode to the 2 in series, making a total of 3. This increased the IR current to 54mA, a more comfortable level.<p>I then shorted one IR-LED, running on one, with no difference in current, as expected.<p>I then tried to empty the capacitor by transmitting continously (Holding a key down for autorepeat). I scoped on the voltage over the capacitor and it took 30 sec to lower it 2V, with 8V left. The voltage over the 2 IR-leds is only about 2.6V when transmitting and the voltage drop over the 22 ohm resistor is 1,2V. A total of about 4V. Giving a headroom of 6V from idle state. This give at least 1 min of continouse transmission before it gets iffy.
I also saw that the voltage did drop fast in the beginning and then slowing down. This is natural because of averaging of the transmitted character pattern.<p>The conclusion is that the circuit construction is sound. With the boosting of the driver current to 50+ mA it should reach longer as a remote control.<p>I found that the drive current for the indicator LED (an old red led I found) was on the low side. If you want any usable indication you must use a high efficient type of led. The current is between 1.5-2 mA.<p>I did not see any sort of light from the LD271 diodes, and I did not expect it at 950 nm.<p>All together this does not help you too much with your construction. <p>TOK

[Bernius1]

I admittedly didn't read the whole thread, though most suggestions seem viable. But here's my take;<p>The first sentence in the Ckt descrip.says 'during idle time' the cap is charging. So I think the ckt is designed as intermittent (TV remote) and not data x-fer.
That said, the 4700uf's charge curve is lower/longer, so your data stream is being 'smoothed' by the cap, hence less range/signal.
Your critical components are; D2, the cap, & the 2.2K resistor. Dropping the resistor raises all the currents, not just to the cap. So Avg current draw on the port becomes an issue. LOWERING the cap value shortens charge time, but will also lower burst brightness at first x-mit. Changing D2 to a Schottkey will gain precious millivolts, which will be realized at the cap (which spans the rails). If you gain .4V, it'll be avaiable across the output Q's & IR diodes.<p>The way I see it, if the LED output (arbitrarily) in lumens was, 100,60,48,40,36,33,31,29,29,29,29 with each successive burst, then a Schottkey and 3300 or 2200uf cap would be lower, with a higher average ; 80,66,58,52,47,43,40,38,38,38,etc. Either way, the drop in lumens will mirror the cap's charge time, inversely.
Moving the 2.2K resistor to the other side of the node may help also, but I can't predict side effects from here. And it's your $$ and equipment !!

[scag]

Hi all
Ok many things to try out, and some more confusing infomation that i need to sit down. I will get back to all this but I am away for 2 weeks from the 8th June. My partner and I are cycling to Cardiff then cycling the London to Brighton Bike Ride on the 19th. I will check emails on Tuesday night.<p>Many thanks for all your help. I will be back at this ASAP<p>Many thanks
Mike