NP,

thanks for the info, need to digest most of it still though.

We had an internet outage all morning, so need to catch up now.

I changed the feedback led now back to green, so we should almost be there :-)

Measured again the voltage across the now green feedback led with the scope to be 1.47V rms @ 2.54mA rms and 800uW
The DMM (Fluke 179) shows 0.792V, but it is operating out of its specs, so unreliable.

Removed the 2200uF and 0.1uF caps (i allways include them when working on pulsing systems).

The sharp negative green current pulse visible in my last screenshot above is the same as the current pulse through the (now again) green led.
So it seems that the feedback signal (current) through the green led is fed back into the battery pack now.

I have combined the supply input current (in white) with the current (in green) through the green led in the below screenshot.
There you see that the both sharp pulses (green is reversed for better comparing) are the same.

Rest of the post will be analyzed later today.

Itsu

Hi nul-points. Going by Itsu's scope power measurement values posted on his diagram in reply # 37 above,
below is what the efficiency works out to. If the scope math function that Itsu is using
calculates all the instantaneous power points by multiplying all the corresponding current and
voltage waveform points and then takes the average (mean) of all those calculated instantaneous
power points, then those scope power calculations should be pretty close to actual. This method
would take into account any negative going parts of voltage and current waveforms as part of the
calculations.

Just taking an RMS measurement of a complex asymmetrical spikey waveform which goes both
negative and positive may not give an accurate reading, since the meter or scope may only
look at just the positive or negative half of the waveform to calculate the RMS value.
It would depend on how the meter or scope calculates the RMS values, so I wouldn't trust
such RMS readings on asymmetrical waveforms which go both positive and negative without knowing
exactly how a particular meter or scope determines the RMS values.

Based on Itsu's posted power measurements in his reply # 37:
-------------------------------------------------
Input power from battery: 28.4 mW

Total circuit power consumption: 1.4 mW + 2.3 mW + 21 mW = 24.7 mW
(power consumed by base resistor and base circuit of transistor should be fairly small)

Efficiency = (24.7 mW / 28.4 mW) x 100 = 86.97 = 87%

nul-points, I will have to take a closer look at your comments later this evening when I have time,
but the calculated efficiency above here looks about in the right vicinity for that type of circuit.
The power consumption of circuit components like transistors is not normally included in
efficiency calculations, but, anyway, I understand why you are doing that, and it doesn't change
things by that much anyway since the power consumption of the transistor is relatively low.

Itsu, with your white LED measured as consuming around 21 mW, was the white LED glowing very very bright,
like almost flashlight bright? At 21 mW, it seems to the me the white LED would be glowing super bright.

Void,

The way you describe the working of the scope i use is correct.
It takes 250M(ega) samples / sec across the whole display screen, then average (mean) them.

I also have to agree with your effeciency calculations (±87%).


The white load led is/was indeed very bright as to unable to look into.

Itsu

I received my BC327-25 transistor, so build a new setup using a PCB instead of the breadboard.
The breadboard is kind of unstable and i want repeatable results.

The setup is shown below.

Using:

the new BC327 transistor,
the same transformer,
the same 50K pot,
a green feedback led,
a white (10mm) load led,
a ceramic 47pF cap,
a 100uF electrolytic capacitor and
the 3.7V 3.1KWh (840mAh) battery pack

It has an on/off switch and elevated leads so to be able to put the current probe almost anywhere.
The transistor and potmeter are removeable.

Will do again the measurements this weekend.

Regards Itsu

The new circuit layout looks good Itsu. Nice to have those raised wires
for easy access by your current probe. Ok on the white LED being very bright
in the previous test. That passes the sanity check.

For the scope math calculation, if you mean that the math function does
multiply each corresponding point for the current and voltage waveforms
to get the instantaneous power for each point set, and then takes the mean of
all those multiplied values, then from my understanding that is a quite accurate way to
determine the average power. I have used that same method in the past to analyze
joule thief circuits and similar, but I had to do the multiplication of the waveform points
and take the average using an Excel spreadsheet to do the calculations on all the points.