Will this dynamo blinky circuit work?

I'm trying to get a dynamo-run headlight/taillight pair up and running. I've got a test circuit working on a little bottle dynamo, but I'd like to get feedback on it (i.e. will this blow up and spray battery acid on me?) before committing it to a circuit board.

The basic idea is this: I have a rectifier that that goes straight to some power LEDs (nothing fancy) on the headlight, and then, in series with that circuit, I run the following schematic as a blinky light for the rear. The little X1 set of terminal is just run in series with the front LEDs. That is, the positive terminal of the rectifier leads to X1:1, and the X1:2 leads to the positive terminal of the next white LED. Assume that the transistor, capacitor and resistor values are appropriate (I didn't include the front light in the schematic because I'm really slow at making schematics).

I found initially when I ran this without the battery that it would cause the front to pulse in time with the blinking in the back. Then I hooked up the battery (3xAAA Ni-MH, because that's what I had) in parallel with this and it made the pulsing in the front go away and also made a nice blinky stand lamp (bonus!). **Obviously, this guess and check method of testing should not be tried by anyone**

I ran it for a few seconds just on a bottle dynamo in my truing stand and it worked beautifully - is there anything dangerous or stupid about this setup that will fail if I run it on a bike?

LEDs are Luxeon K2 red emitters (Vf typical = 2.95).




EDIT: I should mention, without the dynamo aspect of this, this is a pretty standard blinky circuit. I was inspired by this: http://www.instructables.com/id/Two-transistor-LED-Flasher-Circuit/step3/Circuit-Diagram/

The idea is that, rather than having one light blink and disrupt the entire circuit by drawing, and then not drawing, power, it has two lights that light in sequence (when one is off, the other is on), so that the power draw is more or less constant for the circuit, as a whole.

On the plus side, I don't think the circuit will explode.

On the down side.... I get the feeling that this is an excuse to put together some circuits that you found on the web, as opposed to starting out with some function that you needed to achieve, and then finding/designing a circuit that would perform the function.

There are intrinsic problems with wiring a flashing circuit in series with a LED that you want to operate as a steady light. Any change in the oscillator's current can affect the white LED(s). The battery does provide a parallel path for current to flow when the oscillator circuit isn't conducting, but does so through R5. Not ideal...
You also have the issue of properly charging the battery. Speaking from theory as well as practical experience, it can be easy to damage a battery from overcharging in a circuit without charge regulation. I'd have to know the specific details of the battery and the charging current to know if it is an actual concern. If you are using AA cells or smaller, it probably is. If you are using anything other than nicads, it probably is.

I have to ask why you want two LEDs that alternately are powered. Assuming that they aren't mounted very far apart (6 inches?), they will simply appear as a single steady red light from a distance. If you want a flashing light, just rig up a LMC555 oscillator and use it to drive a power mosfet to switch power to a single red LED. If you want a steady red light, just wire a single red LED in series with the white LEDs in front.
Honestly, the circuit shown was an antique when it was taught to me in EE classes in the early 80's. It's not a great oscillator, but it does provide an opportunity to teach people about transistors, RC networks, etc. There's no reason to use a circuit like this since the 555 was invented.
Besides... isn't Q1 supposed to be wired up with the emitter connected to X1-1?

Steve K said:

On the plus side, I don't think the circuit will explode.

Click to expand...

I have a low bar for success in life - I will count this in the "success" column.

Steve K said:

On the down side.... I get the feeling that this is an excuse to put together some circuits that you found on the web, as opposed to starting out with some function that you needed to achieve, and then finding/designing a circuit that would perform the function.


There are intrinsic problems with wiring a flashing circuit in series with a LED that you want to operate as a steady light. Any change in the oscillator's current can affect the white LED(s). The battery does provide a parallel path for current to flow when the oscillator circuit isn't conducting, but does so through R5. Not ideal...

Click to expand...

Not quite - I was trying to figure out a way around the problem you describe. By having lights that alternate, rather than a single flasher, I hypothesized this circuit would solve that problem. It seemed to, once I added the batteries. I'm going about this in much more of a trial-and-error way than a deductive process. I have a degree in biology and had no interest in circuits until I received a dynamo hub for a recent present - consequently, I'm groping around for a solution somewhat blindly until my understanding can catch up.

Steve K said:

You also have the issue of properly charging the battery. Speaking from theory as well as practical experience, it can be easy to damage a battery from overcharging in a circuit without charge regulation. I'd have to know the specific details of the battery and the charging current to know if it is an actual concern. If you are using AA cells or smaller, it probably is. If you are using anything other than nicads, it probably is.

Click to expand...

Good to know. I'll look into nicads as a stand-light solution. I'll look at your circuits to see how you handle the charging, if not through a resistor.

Steve K said:

I have to ask why you want two LEDs that alternately are powered. Assuming that they aren't mounted very far apart (6 inches?), they will simply appear as a single steady red light from a distance. If you want a flashing light, just rig up a LMC555 oscillator and use it to drive a power mosfet to switch power to a single red LED. If you want a steady red light, just wire a single red LED in series with the white LEDs in front.
Honestly, the circuit shown was an antique when it was taught to me in EE classes in the early 80's. It's not a great oscillator, but it does provide an opportunity to teach people about transistors, RC networks, etc. There's no reason to use a circuit like this since the 555 was invented.

Click to expand...

They will appear to blink from the side, and I'm more concerned about how they look from 5-15 meters than anything further than that. Just something to make my light stand out from background noise.

I'm a little afraid of the 555 in that circuits like that are above my pay-grade at this time - I appreciate analog circuits, too. It sounds like I need to suck it up and learn the thing, though. Is there any way to eliminate the issue of one flasher making the rest of the lights flash?

Steve K said:

Besides... isn't Q1 supposed to be wired up with the emitter connected to X1-1?

Click to expand...

Yes... the diagram was incorrect. It's fixed.

Issue 1: addressing the problem of hooking up an oscillator in series with the headlight LEDs.... the simplest solution is to just replace the battery and associated diodes & resistor with a power zener diode. Pick a voltage for the zener that is adequate to power the oscillator portion of the taillight. Pick a package size adequate to dissipate the power.
Alternate solution: replace the zener with some combination of LEDs. Might as well get some light from that wasted power. This is essentially what I do in the circuit mentioned next...

Issue 2: standlight. I use a nicad because it works for me. More accurately, I use a AA cell because it works with the boost converter that I use, and I use a Nicad cell because nicads are more forgiving of continued trickle charging and sloppy charge regulation. It might be the case that your best solution is a supercap instead. You'll still need some sort of basic charge regulation, though.
For charge regulation, I just charge to a fixed voltage. I'm conservative with the voltage in order to be kind to the cell. This is viable because the cell's capacity is much larger than I need. In one design, I just use the voltage drop across some yellow LEDs, minus the drop across a silicon diode, to charge a nicad cell....
http://www.flickr.com/photos/kurtsj00/4803579536/in/set-72157617009273346

In a later, more complicated design, I set up a voltage reference for charging a nicad cell...
http://www.flickr.com/photos/kurtsj00/7483782976/in/set-72157621965148305

Issue 3: blinky lights. The 555 is nearly the cornerstone of modern electronics (supported by the 2N2222 transistor and 1N914 diode). It is soooo handy that it's worth going through the datasheet and application notes. However, if the cross-coupled transistor oscillator works for your needs, then don't let me talk you out of it. I would suggest reviewing the info on the 555 though, just to see what the options are:
http://www.ti.com/product/lm555

There's also a cmos version, which works well at low voltages.

Awesome. I'm going to have to take this in baby steps, so I'll figure out the series oscillator first, since my top priority is getting blinking that only happens on one end. I was thinking about doing something similar with the series LEDs - this clears up something I was confused on.

The data sheet for the 555 is much more helpful than the 555 example projects I've found - those haven't really cleared up anything about the functioning. Thanks for the helpful response, as always.

Alternating blinky LEDs do work; I've followed enough of them over the past winter. However, you don't have to have them. I have 2 solid rear lights, a single LED in a mudguard housing and a Lucas motorbike rear light stuffed with ½ watts-worth of red LEDs on the back of the rack.

Switching to solids would simplify the ccts, however you run them, but I agree with your philosophy generally. I appreciate a suck-it-and-see attitude. Worst that can happen is you'll need a few new components. If you have one, any old model train controller at quarter power ought to provide you with a bench supply for the purposes of soak testing.

And I wouldn't worry about the 555 chip, it's still not really the digital age. There's a chap roundabout here who was reprogramming PICs. Good luck to him I say.