Dynamo switching help please

[Edocaster]

Thanks for your reply. My circuit for the 2x XM-L light I have is basically the same as the one I linked to by Syc, but with a smoothing cap and a 68 ohm resistor for the standlight.

I think my calcs on efficiency might have been off. Apparently, a 68 ohm current limiting resistor, when the supercap is at 5.5v (i.e. fully, fully charged) and the LED about 2.7v (Cree XM-L Vf seems quite low at low current) results in about 41mA (probably less due to supercap internal resistance?). Which is only 0.115W lost in the resistor. I think my initial guesstimate took into account capacitor resistance - although I don't actually know what it would be (how does the ESR rating for a capacitor play into the internal DC resistance?).


I can live with 0.115W.


The PNP transistor route... I was thinking of having the transistor to keep the standlight off until the base goes low (which is what I think jdp298 is hinting at). But I'm a bit confused where to connect the base - would it be like Q1 in your circuit, connected to the top voltage rail? Or does it need to be connected to ground?


A smoothing capacitor is pretty much something I will need. Your zener suggestion makes sense. However, the growing complexity of any transistor circuit is putting me off though, especially as the basic circuit seems to last a good few minutes...

 

[Steve K]

the ESR of some supercaps is pretty significant. The associated losses are probably not too big in this application. The current limit resistor can be modified to compensate, usually.

I'd have to see your schematic in order to say if it would work or not.

 

[Edocaster]

Hi - sorry for the late reply. Downloaded LTSpice to draw the diagram (no idea how to use the rest of the software). So, would this work (bridge rectifier and smoothing cap omitted for simplicity)?

 

[Steve K]

I think your general idea of how to use the pnp transistor is valid. Some tweaking of values and other stuff might be needed to actually make it work. LT Spice would make it easier to get the bugs out of the circuit, but maybe it's easier to just breadboard it.

Some thoughts/concerns:

1. I think the supercap could charge up to twice the LED Vf. The schottky diode will only drop 0.4v or so while there is some current flowing. Once the current drops down to a few mA or so, the voltage drop across the schottky will decrease. I'm still partial to my circuit for controlling the charging voltage of the supercap, or go back to the circuit where there is always some current flowing through the schottky, through R1 and to D4.

2. I'd either get rid of D6 or R2. I'm inclined to just get rid of D6. Try using 2.2k ohm resistors for R2 and R3 to get about the right voltage and current for the PNP base. I'd also put a small cap across R3 to keep the transistor turned off during the low voltage part of the dynamo's rectified AC output. 10uF perhaps?

3. did we already discuss the pro's and con's of wiring 4 LEDs in series? More light at high speeds, but less light at low speeds. If you spend time below 8mph or so, you might not be happy with 4 LEDs. I'd suggest mocking the light up with non-permanent materials to try out both. My headlight switches between 2 and 4 LEDs, depending on the bike's speed, and there's really not that much of a difference. Play around with it and see what you think.

 

[Edocaster]

Thanks for looking at this. This is one of the only places on the internet where these ideas even get looked at!


I think putting a medium-sized resistor between C1 and R1 (150+ ohms) could be done to ensure there is always some current going from the supercap to the standlight (to stop C1 being overcharged). For a small loss across that resistor, it will also help ensure the standlight never totally goes out.


As for D6, that's there as you suggested with a smoothing cap the voltage wouldn't drop fast enough to switch the transistor before the main lights went out. I pretty much have to design around the smoothing cap. I'm guessing the smoothing cap also makes the suggested 10uF cap across R3 unnecessary? I actually wouldn't mind the standlight turning on prematurely when the dynamo goes low - I'm more worried about the opposite (gaps when there is no main light and no standlight).


The D6-R2-R3 route could waste some current when the wheel is moving, depending on what base current I need to achieve. Is there any sense in replacing the PNP with a P channel MOSFET (as I understand they don't need a gate current to run)?



Finally, on the 4 LEDs point, I'm planning on building the Pilom curcuit 9 or 10, so with a switch between voltage doubler and full-wave rectifier mode. I did toy with just shorting out LEDs, but that's a no-go for the smoothing cap, I think (yep - I told you I had to design around the smoothing cap). The voltage doubler should ensure I can always get the 12V or so required.

 

[Steve K]

I still have concerns about a smoothing cap messing up the detection of a stopped dynamo. You'll have to build it and see how it works out. :)

 

[Edocaster]

Your help has been invaluable, in saving me time and money. Which is to say, I think I will probably just use the ordinary passive standlight circuit. It seems that controlling the standlight via transistors detecting voltage is almost mutually incompatible with a smoothing cap.

Nevertheless, in case it helps anyone looking at this thread in future, one last resort which springs to mind is to place the D6 R2 R3 string before the bridge rectifier (omitted from my schematic), replacing D6 with an ordinary diode facing the right way to prevent reverse current. It would only work on half of the dynamo waveform, so your suggestion of a small cap across R3 would be necessary. I think that might work.

 

[Steve K]

Depending on your interests, you might want to build the basic circuit and play around with the more complicated one in your spare time. Or maybe play around with microcontrollers and really make something fancy! I have some ideas that I'd like to play with, but I just need more spare time.

The commercial LED bike lights have really improved a lot in the last 5 years or so, making it less practical to make your own. Perhaps the real value is making a good standlight to go with the commercial LED lights??