My watsonseblog dated 2007 Jul 14
Back in the early days of electronics, they would assemble several neon lamps with 1 meg resistors and 1 uf caps in a box, along with a 90V B battery. The lamps would flash in random patterns. These were called a “Do Nothing Box”. I guess this is my version of a Do Nothing Box.
I came across a project at W. D. Sherman’s site that is similar to a project I built a few years ago. I used a PSO (Phase Shift Oscillator) to drive LEDs so that they turn on and off gradually, not abruptly as the usual flasher does. The PSO uses three RC networks, each delaying the signal 60 degrees, for a total of 180 degrees delay between the collector and base of a transistor. Along with the 180 degrees in the transistor itself, this gives 360 degrees total and this positive feedback causes the circuit to start oscillating, putting out a sine wave.
I had long thought about putting an additional transistor between each of the RC networks, and driving a LED with that transistor. I had seen similar circuits in circuit handbooks and other texts, so the idea wasn’t new. But I had never used an enhancement mode MOSFET for the transistor. Since the 2N7000 enhancement mode MOSFET acts like a vacuum tube and doesn’t draw any current on the input, the biasing is much easier than with a regular BJT transistor. And with the very high input impedance, small capacitors below 1 uF, and megohm plus resistors can be used. But it does require more than 2VDC at the gate to turn on the 2N7000.
I built the circuit at Sherman’s site with the following changes. He didn’t specify the LED color, but I used red LEDs. I used 5 sections instead of 7. I used 470k resistors instead of 499.9k. I used 100 ohm resistors instead of 120 ohms. And I changed the most important part, the timing capacitor, from 4.7 uF to 0.33 uF or 330 nF. I first used 1 uF, but it was still too slow, so I changed them to 0.33 uF. I think they could be even smaller, 0.22 uF would be better. Or use the more common value of 0.1 uF and change the 499k to 1 meg.
I used three AA cells for 4.5VDC power. The circuit works well, each LED fading out and the next one gradually brightening up. The red LEDs drop about 2V, and it requires at least 2V to turn on the 2N7000, so the circuit is limited to operation above 4VDC supply voltage – below that it just quits. The total supply current varies with the LED brightness, so the supply current is constantly changing in the 50 to 100 mA range. I would say that three 1.5V cells is too low a voltage, it should be a minimum of 5V and preferably 4 cells for 6VDC.
And for LED colors such as green and blue, 6VDC may not be high enough. Or 2 or 3 LEDs can be connected in series, and a 9VDC supply can be used. If three sections are used, then multiple LEDs per section could be used in series to give the effect of the flashing lights around the edge of a marquee. Right now all of the LEDs are in a line (see picture), but I may try putting them in a circle. I think 9 LEDs around a circle at 40 degrees apart would give a nice effect. I may try two concentric circles, with 9 LEDs on the outer one and maybe 5 LEDs of a different color in the inner circle. And connect them up so they counter rotate. It’s soothing to watch the LEDs go through their patterns of oscillation. One could sit and watch them in a darkened area to relieve stress. Or just put them on a shelf and let people ask what they are for!
I was staring at the blinky LEDs today, Saturday, and it seemed that they were off longer than they were on. I’m still running it from three AA cells at about 4.5VDC. The 2V drop across the LEDs limits the voltage at the gates to 2.5V or less and that seems to be the reason why it’s unable to operate below 4V and why it’s operating marginally now at 4.5V. All it needs is to have a bit higher voltage across the rest of the circuit when the LEDs are not conducting. Well, I had an idea that if I put a resistor across each LED, the voltage at the LED cathode could be pulled up to nearly the supply voltage when the FET is not conducting. So I put a 22k across one LED and it really did make a difference. I then warmed up the soldering iron and put a 22k resistor across all of the LEDs, and it made an obvious difference – the LEDs stay lit longer. Now it will still function down to 3VDC. The 22k is a very small fraction of the 499k with which it’s in series when the LED is off, but it draws negligible current when the LED is on. If you think the effects it causes are too strong, you can increase the value to over 100k and as it goes higher, the effect is less noticeable until there’s no difference with or without the resistor. I’m guessing that this may be when the resistors are nearing a half meg or more. But by adding these resistors, the circuit will operate below 4V, and the LEDs will also speed up a bit and stay on longer. They will allow you to leave the 1.5V cells running until they are down to 1V, which gives many more hours of operation on a set of 3 cells.
Back to experimenting…