Colin's LED torch circuit

So I've built one of Colin's (colin55) circuits that I think is pretty remarkable, his LED torch (B). It works really well.

But when I run a simulation of it in LTspice, there's a big discrepancy between his description of its behavior and what Spice reports. (See attached screenshot and Spice sim file.)

He claims it runs at 500kHz, but LTspice shows it as only about 150Hz. (Thanks to Eric Gibbs for showing me how to measure frequency with LTspice.)

Now, the inductances I used for L1 and L2 are only a guess (see my other threads here about trying to figure inductances). My homemade transformer is wound on a little bitty piece of ferrite broken off a larger piece (a cable loop). I suppose this could affect the frequency of the flyback operation, but not that grossly.

Anyhow, as always, very curious to hear explanations of what's going on here. Again, it's an amazing little circuit. Lights a "super-bright" LED really brilliantly on just one AA cell.

I believe you have mistakenly used mH (milli-Henry) instead of µH (micro-Henry) for the transformer inductance values in your simulation.

Well, OK, so do you have any idea what the (approximate) actual values of these coils might be? I sure as hell don't.

Again, let me describe them. I broke up a larger ferrite loop (the kind used to put ribbon cables fhrough to reduce RFI) into pieces. I used a piece about 1/8" square by about 3/8" long as my core, and wound the transformer coils on it (60 turns/40 turns).

Another circuit simulation I ran similar to this one would not work at all with µH values. I haven't tried this one with smaller values (will do so right away).

This is fairly frustrating; I had hoped that some of the "experts" here could have at least hazarded a guess as to what inductance values I'm dealing with here. All I need at this point are ballpark figures. Nobody's health or safety is riding on this.

========================================

OK, further experimentation has revealed that, yes, frequency does increase with reduced inductance! Who woulda thunk it?

At 100µH/220µH, F goes up to about 15kHz.
At 10µH/22µH, F goes up to 180kHz.

I guess by the power of deductive reasoning, assuming Colin's value of 500kHz to be accurate, I could figure what the Ls really are here.

At 3µH/7µH, F is 440kHz. So I guess these are close to the inductances I have, correct?

I will say that as frequency increases, the waveforms get uglier and uglier. Lots of spurious spikes and gaps in the pulses. Not that it really matters in this application. I do wonder, though, how much RFI this little device could potentially be emitting ...

hi cz,
Look here for inspiration, these are 'Joule Thief' circuits.

joule thief circuit - Google Search

carbonzit said:

-...
At 3µH/7µH, F is 440kHz. So I guess these are close to the inductances I have, correct?

I will say that as frequency increases, the waveforms get uglier and uglier. Lots of spurious spikes and gaps in the pulses. Not that it really matters in this application. I do wonder, though, how much RFI this little device could potentially be emitting ...

Click to expand...

Yes, that's likely close to the real inductances.

It will put out some RFI certainly. The amount depends somewhat upon the layout of the circuit (long leads are antennas for the RF signal), the frequency, and any shielding there may be. If you bring an AM radio near by you can probably hear some of it.