To be deleted...

[conradelektro]

So i still make some measurement error there (allthough i also used a csr in the battery lead which makes it even more confusing) or the signals (pulses) interfere somehow.
It annoys me that i do not get any logical data out of this setup.

@itsu:

I watched your video https://www.youtube.com/watch?v=NyOTWG0F914 and extracted the attached picture.

Thinking about your measurement problem (illogical result) it comes to my mind, that the only cause could be the board with the caps (see arrow on the picture).

There seems to be an unintended partial short or a cross-connection between input and output on this board. If it is not too much work, a separation of the input caps and the output caps onto two separated boards would be a prudent measure? Just to be sure and to eliminate all possible causes.

I like the black box and the calibration of the LED light output very much. Such careful measurement methods are needed to be sure about the "true happenings" in a circuit.

Measuring batteries over days and weeks is probably not very useful. The chemistry of batteries is very hard to track accurately. I am sure that nobody suspects that the "miracle" is in the batteries, therefore one should use caps, may be 1 F or 10 F super caps (one for input and one for output) to bring down the measurement cycle to a few minutes. Once one has a pretty good idea with the caps one could return to batteries and a test cycle of days or weeks.

Greetings, Conrad

[nul-points]

Inspired by Itsu's use of a solar cell for photometric readings, i've setup a similar cell in a sealed tube with the multi-LED head i'm using at tbe moment

Current draw by the circuit is 60mA, so you were right, Itsu, using a more heavy-duty load certainly moves the readings into a more accessible zone  - fortunately, it also reduces tbe run time, which in my case should be around 12+ hours for me, using 750mAh NiMHs

I can see by a glitch on the supply line trace, causing a similar glitch on the intensity trace, that there will likely be a closer correspondence between supply voltage and sensor o/p than i anticipated, compared to my results posted above, using a photodiode

I've had to apply some filtering to the sensor because the light pulse peaks were overloading my datalogging system, and i think that the filtering is an order of magnitude more than i used with the photodiode, so this will produce a certain difference in behaviour

However, since this test is only to determine the intensity profile of the LED drive within its own setup, from full input charge to end of run discharge, nothing needs to be quantified to refer to any other test results

More later
np

[nul-points]

Here's an example of the sort of application i anticipated for this approach:  at the centre is the basic flyback converter; arranged around the periphery are a variety of inputs, outputs and funky bits (no, i've no idea what that means, either!)

The general idea is to provide a compact 'workhorse' which can accept energy input via a variety of methods (storage, active, and renewable) and which can then convert or transfer that energy into light or a different storage type, respectively

The simple controls allow selection of function (charge/light) and power-level

The 'funky bits' are related to efficiency

I've called this device the 'flyback flash-lite' for reasons which escape me

More later
np

[itsu]

Conrad and NP,

thanks for your suggestions, but i have sorted it out, as expected it was a measurment error, my error.
It has to do with calibrating/degausing the current probe which was done the wrong way causing all current measurments to be around zero (kind of when you use AC coupling on your scope).
I never used this wrong calibration way, and i don't know why i did it here, probably my head was more at the PC problems.

Now the data makes sense (allthough somewhat dissappointing), see screenshots.

1st is the supplying battery/supercaps, it shows it delivers 103mW
2th is the receiving battery/supercaps, it shows it receives 29mW (thus charging the battery/supercaps)
3th is the led under test in the blackbox, it shows it consumes 32mW (backed up by the solarcell which shows 1126mV = 36mW)
(current controller was set in all 3 screenshots to 10mA/Div., so similar as Ch4 shows)

We are missing 103-(29+32)= 42mW, which probably can be accounted for in the 100K resistor, bc547 transistor, 2x bat42 diodes and transformer.

I did notice in this high power setup that lowering the 100K pot, there will be a point where the pot starts to smell and gets really hot.
Lowering the ohmage decreases the oscillator frequency and increases the leds output (higher pulse duty cycle)

Anyway, i am glad thats out of the way.



Nice going NP, i like that: 'flyback flash-lite'.

I was also playing with the idea to throw in some small solar cells, so this flashlight would probably never run low when used/stored outside (car/camper).

Need to run over your new circuit while doing some rundown tests on my high power setup, i do see L1 still is in, its still needed?


Thanks all,  regards Itsu

[nul-points]

Hi Itsu

Glad to hear that you sorted the power measurement - and that you didn't have to carve up your suoercap board to find it!

The circuit in these initial tests is based on a conventional boost converter type arrangement - the switching device draws some current through the primary inductance to ground, then when it switches off, the stored charge in the inductance is released through the LED into the output storage.  The majority of the remaining energy will be returning via Q1 emitter to deplete the input supply

The intention of the whole setup us to re-use that part of the energy which has been diverted into the outout storage to    extend the runtime - hence the swapping of the output & input stores - a proportion of the input energy can be used again  - its already lit the LED once, then it can be used to light it again - and then maybe a third time - as in the 2nd graph i posted above (see my energy estimations there)

So - don't be disappointed!   We are making progress!

...probably a good idea though not to adjust VR1 too low in resistance - the increased base current of Q1 could overheat the resistance track inside the Varpot - they're only intended to dissipate low mW of power!

My 2nd circuit variation uses a different arrangement - no output swapping

Yes, the inductor L1 is necessary - the sink for the LED flyback energy is back at the input and the inductance assists this action.  On my test circuit C2 is not populated at the moment

Sounds like a good application for this device - travelling off-grid, etc

Hope the high current tests go well - don't forget to factor the storage swap runtime energies into your totals

All the best
np