dumb led trick for lfo's

[duck_arse]

here's something I been playing with, just to show I don't know how to work transistors properly.

if you have a triangle lfo producing about 3V p-p centred about 4V5, this circuit will blink opposing leds on just the tips of the triangle wave. so it goes: positive peak, one led blinks on - falling from positive, led blinks off - negative peak, other led blinks on - rising from negative, led blinks off. 2 led blinks per cycle.



the chain of resistors determines the on-off ratio, and the brightness. values as shown, there is an off dwell between blinks, but R2-R5=100k produces an on overlap instead of dwell. try a 200k pot to find the resistor values to match your lfo and personal blinking preferences. I'm using superbright leds and they are not blinding with the 10k's, but it turns out neither the transistor base resistor nor the CLR has much (any?) influence on the brightness. you'll have to wait for an engineering type to wander by and explain what's doing.

I tested a shirtload of transistors last night before I realised today that it's a pair of emitter followers, so any transistors will should work. the only problems I had were with some bc557A and bc558C, which turned on and wouldn't blink. well, they would blink quite well when I turned them arse about, which was how I had the circuit drawn and working originally, so just about anything, as I said. (I don't have any 2n50xx to test with.)

I have been hacking at other configs, but they detune to a point that about equals the emitter follower, so I won't embarrasses myself with backwards drawn transistors. yet. I don't have any P-channel mosfets to try, so I couldn't convert the top half of the circuit.

I'm using a comparator and integrator dual opamp osc, tuned to match the phase45 osc output, so it should work with that style oscillator circuit, but don't quote me. whip it up on your breadboard, lash it onto yr fave osc and juggle the resistor values. and let me know how you go with it.

[midwayfair]

Nice! I always forget to mess with PNP transistors, but I've seen something like the NPN side of this. Lots of good uses for this, not just with LFOs -- I bet you could use it to trigger opposed envelope responses, too.

[Kipper4]

intresting mate. mmmhhh

[PRR]

> neither the transistor base resistor nor the CLR ....

"CLR"?

> ... has much (any?) influence

There's two limits on the current. The emitter resistor, and the base resistor divided-by the transistor Beta. You have 10K in one spot and about 120K/100+ 1.2K in the other spot. The 10K has by-far the most effect.

(The base resistance is really a mess of 33K, 120K, 82K, and 33K; I won't work it out because for any reasonable Beta I suspect the 10K still dominates. Idiot-proof: even if all those resistors were series, 268K, it still reflects like 3K to the emitter side so the 10K is the big choke. And clearly there's some parallel action, not pure series.)

Some application of Thevenin could reduce R1 R3 R4 R6 R6 to two resistors. Input impedance would change, perhaps higher, which is usually fine.

I bet it is *very* sensitive to triangle amplitude and power supply voltage. (Also temperature, but probably not-much in the range of temps you'd want to play in.)

There's other ways to do it. Any "precision" you want. But this isn't a precision chore.

Neat example of Complementary Symmetry.

[tubegeek]

CLR = "current limiting resistor" usually....

[duck_arse]

@prr - the sum total of my knowledge of mr thevenin's work equals zero.

does Beta differ from hFE in any appreciable/applicable manner? brightness changed with supply voltage, but the symmetry seemed constant, even if the on to off ratio increased with increased supply.

resistors R1-R8 can be reduced to a 180k from lfo to each base, and a 180k from base to positive for Q1, base to ground for Q2.