To be deleted...

[nul-points]

I had some time, while my planned Friday overnight test was running into yesterday, to think about your original LED intensity question, Conrad

We know that there's no point trying to compare the circuit LED drive level witb a DC test setup because of course by the very nature of a pulse circuit, we have a duty cycle - and since this is a flashlight, and not a piece of test equipment, say, our user requirement is just that the flashlight should be 'bright enough' to illuminate what we need to see in the dark - our brain doesn't care if our eyes happen to think that two different Lumen levels appear to be the same, as long as we can avoid stepping on our pet armadillo in our back yard, at night

So - our comparison should certainly test LED intensity from two drive signals 'chopped' at the same periods; and then we want to compare the intensity with, and without accompanying storage

The previous tests were close to what we wanted, but not quite there - due to the initial voltage of 4.5V on the supply cap, the LED was initially bypassing the pulse operation for both test setups (hence the odd 'double-peak' start to the intensity profile);  only after this initial current surge for both test conditions did the supply voltage discharge below the effective forward voltage for the LED at around 2.5V

Using an initial supply voltage just below 2.5V, i've been able to repeat the tests and still get sufficient intensity to provide useful photometric data - the peak level and the duration of illumination will both be reduced of course, so the final values for average intensity will be significantly less, too

The graphs below show the valid results for what we wanted to compare earlier - Conrad, i'm sure you'll be pleased to find that the you were correct in thinking that the average LED intensity of this 'storage' circuit is less than that for the grounded LED which just allows the output energy to deplete the supply directly

The peak intensity is still the same for both conditions, as i explained in the first attempt at this comparison

However, you may be surprised to find that the difference in average intensity between the two conditions is only 20% (average photometric intensity levels were 0.04 & 0.05, respectively, for the 'storage' and 'grounded' setups)

So the doubling of illumination runtime in this storage circuit cannot be anywhere near explained by a reduced illumination level

An interesting test, thank you Conrad

np

[nul-points]

... my bat42 just arrived...

I made a pcb with the 2x bat42's, BC547, 10x 10F supercaps, 100K pot, 10mm led, 100uF buffercap.
The 10x 10F supercaps a divided 2x 5 in series making 2 stacks of 2F caps, backed/loaded by 2x 750mAH 1.2V rechargeables for now.

...

Where does the power come from powering the led?  Is it because its kind of floating compared to ground, so a measuring error?

Regards Itsu

...a few comments on points you raised (here or on the vid):

- the BAT42s will only make an improvement in the efficiency if the 1N58xx Schottkys were leaky
- the LED is clamped to the common line by the BAT42 at its anode, so there won't be a floating measurement type issue with the LED power reading (more likely, i guess is that you're not getting a good value for the supply power (may need a CSR inline with the input & get the scope to do math on that?)

timed measurements of a set voltage discharge from a (measured/calculated) i/p supercap should confirm i/p power nicely

Conrad, re. some of your suggestions above, i developed this circuit initially with a pulse motor towards the end of last year and have some results from that which i may be able to add here - also used the circuit just to feed back to the supply;

the question is not, i believe, what happened to the extra energy, in the case of the grounded LED (that circuit arrangement just feeds the current straight back into the supply where it depletes the source 'charge' (aka 'gorge') on the i/p capacitor, as in conventional supply-circuit-load arrangements - no, the question should be 'how has the use of the input energy been extended?', and i believe that the answer is that the energy is still in 'transit', as it were, on the way back to the supply, but has been diverted into temporary storage before being re-used for more work on the return journey


hope this helps address some points raised - thanks all
np

[conradelektro]

Today I saw three different LED-flash-lights in a local shop.

The prices are in Euro and all had the three batteries (3 times 1.5 V AA, or 1.5 V AAA = 4.5 Volt supply, the two round lights in a package had 6 x 1,5 V AAA = 2 times 4.5 V) included.

That seems to be the market to compete with. And without batteries one gets these LED flash-lights even cheaper. It seems that "brightness" is the desired goal, not so much "run time". Of course one would like both, but to increase "run time" one has chosen LEDs with higher light output (per Watt) and not a more complicated circuit.

Greetings, Conrad

[nul-points]

...
That seems to be the market to compete with. And without batteries one gets these LED flash-lights even cheaper. It seems that "brightness" is the desired goal, not so much "run time". Of course one would like both, but to increase "run time" one has chosen LEDs with higher light output (per Watt) and not a more complicated circuit.
...

That's certainly a good representation of the current market - both for the flashlight itself, and the cells (usually primary, eg alkaline)

I see a niche market for something a bit smarter than just a plain flashlight, though - i think people would be willing to pay more for a device which is rechargeable (say solar + micro USB?), has a control to optimise either intensity or duration, is efficient ...and can be used to recharge a variety of portable devices  eg. phones/tablets etc,  via a standard USB outlet

C'mon, Conrad, even you would pay 30 Euros for a tool like that!  ;-)

All the best
np

PS  i'm currently using one of those circular 24 LED units above as my LED head - works well with 2 or 3 AAA cells - today i used it with 3 cells i/p to recharge the 9V NiMH for my DVM (obviously the flashlight had energy left afterwards, i was just using its capability to charge an output battery which is 2-3x the i/p battery voltage)

[TinselKoala]

The simplest way of driving a LED pretty efficiently: use three AA batteries in series (they will start out at 4.5 Volt and run down slowly to 3 Volt in 90% of their run time), choose a resistor in series with the LED, so that the LED has its nominal current at 4.5 Volt. It is hard to beat that very simple circuit in terms of run time and light output per energy input.

Greetings, Conrad

Actually... the series resistor has two faults: First, it dissipates power itself, and second, as the supply voltage changes, the current through the LED and resistor changes, hence the brightness of the LED changes.

It is actually "simpler" (for some values of simpler) and more efficient to use a linear current sink instead of the resistor. The AP2502 that I have been playing with for some time is such a device. This 4-channel linear device is more efficient than a series resistor and keeps a constant current of 20 mA (or 40 or 60 or 80 mA) through an LED as long as the supply voltage is over the LED's forward voltage. It even works well with pulsed JT type circuits. It has an "Enable" pin that can be used with the usual PWM scheme to vary LED brightness consistently. The advantages are that it does not waste power the way a resistor does, and it keeps the current (and hence the LED brightness) constant as the power supply voltage varies. The disadvantage is that it comes in a SOT-23-6 package!

http://www.youtube.com/watch?v=X9wxuRZV-Ro

In terms of LED efficiency (lumens per watt of electrical power) it is hard to beat the Philips LumiLEDs. See the data sheet below: