2 Lux3, BB 400/500 on 4AA

[Flatscan]

I'm modding a bicycle headlight to run 2 Lux 3W emitters on 4AA NiMHs. My current plan is to run the emitters in series off either a BB400 or 500. I'm assuming that 2 emitters on 4AA will have comparable runtime to a sandwich running on 2AA, giving the BB400 3 hours on 2300mAh batteries. I would like to move up to the BB500, but I want around 2 hours of runtime on the same batteries. Is this reasonable, or should I stick with the 400?

The headlight features a high/low/off momentary switch and what appears to be regulation. The battery voltage on high drops steadily under load (cheap heavy-duty cells), but the lamp voltage stays pretty level. I want to keep the switch, but running two regulators in a row seems like a very bad idea. Gutting the electronics and using a mechanical switch is a viable, but non-preferred solution.

 

[Doug S]

One issue you will need to address is cell reversal since the BB will operate down to 1.6V. I think that you probably can get 2 hours out of 4X2300 and a BB500.

 

[Flatscan]

Is it sufficient to train 4 cells together and simply not run them all the way down (say, up to 2 out of 3 hours), or do I need a low voltage (<4x0.8V) detection circuit? There is a "low battery" indicator on the stock board, but it triggers far too late. I could ask for guidance on how to hack it over in Electronics if this is a useful/necessary feature.

The jury is still out regarding 400 vs 500.

I'm reconsidering whether the extra output of the BB500 is worth it. I'm not planning to do lots of night riding, mostly "whoops, it's dark," so the extra runtime won't be a factor, except in an emergency.

 

[Doug S]

[ QUOTE ]
Flatscan said:
Is it sufficient to train 4 cells together and simply not run them all the way down (say, up to 2 out of 3 hours

[/ QUOTE ]
Humans are generally too fallible to rely on this approach.

[ QUOTE ]
Flatscan said:
or do I need a low voltage (<4x0.8V) detection circuit?

[/ QUOTE ]
I would say yes. I think that I would set it a bit higher though, minimum 4X0.9V This really doesn't cost you much at all in runtime and gives better protection. 0.9V is probably over 99% discharged. If it is simply detection and not automatic cutoff, I would probably go up to 4X1.0 or 4X1.1V.

 

[Flatscan]

Okay, I've successfully identified and hacked the "low battery" indicator. As far as I can tell, it compares the voltage drop between a resistor/diode pair and a resistor/resistor pair. The resistor/diode value is relatively unaffected by a drop in Vbattery, but the resistor/resistor value falls linearly. I tried using algebra to calculate the proper resistor replacement, but I made an incorrect assumption somewhere, so I threw in a pot, to be dialed in on the finished product.

My earlier observation of regulation was shown to be false, as the Vlight is controlled strictly by a power FET of some sort. However, Vlight on low is obtained by feeding the FET a gate voltage between closed and full-open, so the FET has a voltage drop of up to 2.8V. From my measurements, maximum low Vlight is ~1.2V < 1.6V, so the BadBoy will drop down to direct drive and low current, which should not be a problem.

Should I find a replacement power FET now? It's quite possible that the BB will fry it with excessive current draw on high or burn it out on low if my reasoning is incorrect.

 

[Flatscan]

Quick note: I ended up replacing the transistor, as it was running way out of spec. It was running very hot, so I tried to make a thermal path away from it with some copper wire and AA, which didn't work so well. It eventually heated up to the point that it began flickering the light output.

Radio Shack's transistors in that size package looked to be rated between 200 and 400mA, so I moved up to a TO-220 package. I bought an N-channel power MOSFET and an NPN power transistor with reasonable ratings, I think 4A for each.

For some reason, the MOSFET didn't work. It was dissipating 3.6V, even though its measured Ron was 0.9 Ohms. My guess, in the absence of any explanations that make sense, is that I switched source and drain. Shorting source and drain allowed the BB500 to run the emitters at full.

I then tried the transistor, which worked like a charm. On high, the BB pulls 830mA or so, with no increase in current draw when shorting the transistor pins. The transistor got a little warm. On low, the BB pulls 200-something, and the transistor feels the same as on high. Having a low setting that works makes the light useful as a task light as well.