Lights with very low lows...how low is to low ?

Well i need to ammend that thought and say budget and really good low low with no pwm and a decent high as well .. we can sit around and talk about very expensive lights that do it and cheap garbage that does a low low low but it has to be a flashlight too not a joule thief .. it needs to turn around and do at least 100 lumens too .

I'm with you Boaz, hope our prayers are answered soon. I am surprised there aren't more options based on a potentiometer(sp). I know these are fairly cheap, they must be hard to implement.

Its impossible to use a pot. in a light-they only take a very small amount of power. All the "infinite" output lights use 8 byte microprocessor controllers, so 256 modes, or something less.

Your probably right, I thought I had heard a T1A used that approach, and knowing the part in general is fairly cheap...

The T1A puts out like 70 measly lumens of blue-green, donut holed light....

If we can put the pot into the torch, I Know you can use the $3 single mode drivers based on the AX2002 or the PT4105 chip and have a low LOWER than the V10R picture above, and the PERFECT ramping rate (TRUE visually linear ramp) to 100% (700 or 1000mA). I currently run that setup in a battery pack and its seriously awesome. I have been thinking of how to implement this into a torch though, and it always ends up not being possible, or just plain ugly.

(btw i consider sticking a knob out of the torch (like they do for mags) ugly. I want something where you twist the head and it ramps brightness. Ive considered the "hotpots" - membrane based pots from sparkfun for example, but they cost as much as the torch :/

How about two buttons like a vb 16 ?

The latest iTP A3 does 1000 Hz PWM and the A2 does around 2400 Hz PWM. It's not noticable by ordinary humans.

2 buttons require a MCU to handle the interface, that is either not available, or is beyond the ability of most DIYers by going into programming territory.. It is possible though, but by that stage you are already beyond most budget lights? maybe.

Is there a really small POT that works like a regular interior light dimmer, click on/off, turn to dim?

Physically there are potentiometers with a switch at the low end. The switch can handle up to a couple hundred milliamps, maybe up to an amp tops. You could then connect the pot to the circuitboard described earlier.

But you still have the problem of physically installing the potentiometer into the torch.

If your thinking of using the wall type dimmers, I don't think its as simple as putting one of those in, and its likely it wont work as well. Those have a potentiometer for the knob, but thats usually connected to a circuit to do the dimming of Mains power.

And those almost certainly have significant parasitic drain in "off".

Please explain your aversion to Pulse Width Modulation. Do you have compact fluorescents in your house? They flash at 120 Hz. Do you watch TV? Go to the movies? None of them generate continous light.

A major advantage to PWM, that has been rarely, if ever, mentioned here, is the lack of colour shift. All leds suffer from it at low currents that some people find it objectional. If they are fed a high current in pulses there is zero colour shift.

Cheers,
Angus

I think you meant 8 BIT. Don't forget that they can also operate on 16 bit words, so the limit would be not 256 levels but more like 2^16 or 65536 levels if needed. Don't knock 256 discrete light levels anyway. How many grey levels does your digital camera generate per colour channel? 256? Way more than our eyes can even distinguish.

I just breadboarded a pic 12f683 driving a small led to test its pwm module. It easily generates 1 microsecond pulses at 1000 per second, in increments of 1 microsecond. In other words at least 1000 discrete light levels can be produced.

Then I wrote a program that alternated the brightness of the led between two different levels until I could finally see a difference. It turned out that starting at 500 microsec, it took a value of 520 before I could determine the second was barely brighter, that is a 4% difference.

If that holds across all brightnesses then if 1.04^n = 100, then only n = 118 discrete levels should be needed, at least for my old eyes. So 256 levels is way more than enough.

Cheers,
Angus

If you want perfect modes why not follow Tido's thread http://budgetlightforum.cz.cc/node/799 and do it yourself?

For example the pwm module in the pic12f683 has a range of at least 1000 to one. If used in a 100 lumen flashlight then the lowest possible level would be 0.1 lumen. Is that low enough?

Cheers,
Angus

Whoa dude-calm yourself. I didn't "knock" 256 light levels. Oh, and BTW, your eyes perception of light is roughly the cube root of the actual output, not logarithmic.

My apologies. No offense meant.

However I said nothing about anything being logarithmic. All I said was it took a 4% increase in the stimulus for my eyes to perceive that there had been any change at all.

There are a number of mathematical models that describe response, the cube root model being but one. Look up the Weber-Fechner law or Stevens' formula for example. They all produce reasonable fits to the data.

And don't forget how the magnitude of the brightness of stars is measured.

Cheers,
Angus

If that holds across all brightnesses then if 1.04^n = 100

Thats logarithmic.... f(x)=ab^x.

There have been "studies" on CPF, where people found that cube roots most accurately modeled what they were seeing, namely in the V10A.

I may have unintentionally misdirected this thread. If that is the case, I apologize.

Thinking of how the expensive lights handle the situation, I wondered if something as simple as a POT could be implemented into a cheap lights tail to get a similiar result. I know it's too good to be true, but imagine a RC-G2 with this ability? Anyhow, I'll keep dreaming.

The pot needs to be connected to the driver. Need to bring a wire to the driver if its in the tail. Novatac/HDS does the electronic switching with a spring for the conductor. Liteflux uses a brass tube. Nitecore uses a piston to mechanically transfer the switching action to the switch on the PCB. If we were to include such an electrical pathway, we can simply put a pot in the tail to connect to the circuit. That could work...