To be deleted

[itsu]

I used a battery operated SG, but it has only 10Vpp output, see white trace when unloaded (10Khz, 10% duty, AC)


The setup used is the same as your right diagram above, white led, 100 Ohm resistor, probes setup the same.

Yellow is the (sunken) voltage across the led
blue is the current through the 100 Ohm csr (rms value adjusted to show correct value, so no /100 needed).
red is the math trace yellow x blue = power of the led.

So nothing wrong with your scope, its probably the led (diode) doing this.


Itsu 

[Void]

Thanks for that Itsu. Yes, the LED is rectifying the current waveform, so it
shows differently than the AC voltage waveform. Thanks for confirming that!

[Void]

For the current waveform in my AC test, it is a DC pulse waveform with a 10% duty cycle,
so I believe the correct RMS formula for that type of waveform is: Vp x sqrt (duty cycle ratio).
Based on the current waveform in my scope screen shot, this would give:
4 x sqrt (0.1) = 1.26
The scope calculated the current RMS as 1.28, so that is pretty close.
1.28 /100 = 12.8 mA
However, the actual RMS current for the LED should be around 6 mA (estimated), so we can't
use that calculated RMS value by the scope.

For the voltage waveform, we have an AC pulse waveform which the scope is offsetting
with a negative offset. The RMS voltage value calculated by the scope is also not correct for our circuit,
because the LED is only conducting for 10% of the time, but the scope is calculating the RMS value for
the entire waveform cycle. I am still fiddling to see if I can figure out how the scope calculates that 9.54V RMS
value for the voltage waveform. The actual RMS voltage across the LED was probably about 2.75 V (estimated).

At any rate, the RMS values calculated by my scope obviously can't be used to calculate the power consumption
of the LED with a pulse type waveform being applied to the LED.

P.S.
After thinking about it, I guess the way the scope calculates RMS values is to square all the measured sample
values taken over a period of time, then calculate the mean of those squared values, and then take the square root
of the calculated mean. I am feeling too lazy right now to save waveform samples to a CSV file and compute it
all out using Excel, but I might try it at some point to confirm if that is how the scope calculates the RMS values. 

P.P.S.
Since the voltage and current waveforms are pretty close to being a nice rectangular shape in my test setup,
another way to do a sanity check on the measurement of the power consumption of the LED in my AC test circuit
is to use the actual scope measured values (as shown in the scope screen shot I posted above).

If we do it this way we get (I think this is right, but maybe I am overlooking something ):
We are only concerned about the voltage waveform when the LED is conducting,
so, V measured = 0.65 divisions x 5 volts per division = 3.25V
I measured = ( 4  x 1 A per division) / 100 = 40 mA
Duty cycle is 10%, so the duty cycle ratio is 0.1
3.25V x 40 mA x 0.1 = 13 mW
The scope's measured/calculated value for the LED power consumption (using the Math function) was 16.7 mW.
Ballpark, but not exactly really close. Maybe this method of estimating the power consumption of the
LED is not right, or maybe it is not so accurate. Or maybe the power calculated using the Math function was off a bit.

I think this only works out (roughly) because the waveforms are quite clean rectangular waveforms.
However, thinking about it, that should actually be what the LED is doing. According to the scope measurements,
the LED has 3.25V across it and conducts 40 mA during its On time, which is 10% of the time.
That very possibly may not be so good for the LED however, if left running for a long time. 

This is why it is way easier to try to self-loop a circuit that you think might be OU, and see
if it will self-sustain itself.  No measurements or calculations required, and much less chance of
making an error somewhere. 

[itsu]

Hi Void,

on this wiki there is some info on rms values / waveform:
https://en.wikipedia.org/wiki/Crest_factor

So every waveform needs its specific calculation to arrive at the correct rms value.
I can not see how a scope does that, so probably there is some generic way.

But i agree, (power) measurements are no easy task and marginal cop > 1 values should be treated with causion.


Itsu

[nul-points]

Therefor i am planning to take daily measurements and take the average after some time.

Today i did a second measurement and put the results in a spreadsheet, see below.
There you see that the difference between input and the power users is much smaller (about 1mW)

Itsu

interesting observation, thanks Itsu - i'll look forward to seeing how those readings develop


np