Accurate Measurements on pulsed system's harder than you think.

[poynt99]

PW/TK,

These familiar "Gotchas" raise their ugly heads from time to time. Last time I involved myself with this one in particular was with Luc on his Recirculating BEMF circuit.

This was back in 2009, and I did up an analysis which explained his observations and showed, as often seems the case, that erroneous assumptions were being made.

Here is that analysis: Luc_flyback01.pdf

A lot of time and effort went into this document (as did all my documents uploaded here), and aside from a few too many references to "RMS", it is still relevant, and should I believe shed some light on this scenario as well.

A couple of excerpts in summary:

The Rbulb intensity (or heat) can not be used reliably to indicate the total amount of power used by the entire circuit!

When there are other components in the circuit (such as resistors and coils) and the bulb is in series with them and the power supply, the bulb's intensity is only indicative of the power being dissipated in the bulb itself. It does not indicate how much total power is being taken from the power supply and being used in the whole circuit.

Peace Out.

[EMJunkie]

Perhaps this is a good opportunity to ask one of the most important questions of all: "How is it that a change in Temperature changes the resistance so much? What is going on to make this happen?"

We know this is Nonlinear Conduction - From this we can see that the "Conductivity" of the Filament has changed. Temperature is the cause. Why?

What else can change the Conductivity of an Element?

   Chris Sykes
       hyiq.org

[verpies]

Incandescent bulbs have a very large positive temperature coefficient.  As such, they present very non-linear I versus V behavior. 

Yes, incandescent bulbs are not linear but that does not mean that they are not monotonic.
Light bulb's brightness depends on the current flowing through it and Tinman is correct that more current always means more brightness....and in his experiment this is all that matters.

I am more concerned how his scopes calculates the average of these pulses, e.g. from the screen data, from memory or from the waveform period...

[TinselKoala]

Hmmm.... I think some people are still missing some points and being distracted by Red Herrings.

1. The current measured by monitoring the voltage drop across the CVR is also the current flowing through the bulb. The elements are in strict series so the same current is flowing through them both. Right?

2. The brightness of the bulb is an indication of the power being dissipated in the bulb. A dimmer bulb means less power dissipated in the bulb, a brighter bulb means more power dissipated in the bulb. This is true regardless of factors like the temperature coefficient of resistance of the bulb, and the duty cycle of pulsation.  Right?

3. The instantaneous power being dissipated in the bulb is Watts=I²R, where I is the current at the instant of measurement and R is the bulb's resistance at that instant. Right? And this is also equivalent to Watts=V²/R, so solving for R we have R=V/I by Ohm's Law. Right?


4. When the capacitor is connected, the current through the bulb is (relatively) constant, so there is no difficulty with the "mean" value of the current. So the power dissipated in the bulb is also constant. The resistance of the bulb can be calculated by R=V/I. We know I from the CVR measurement, but what is V?
Question: Does the voltage drop across the bulb as measured by the scope give us the "V" value for this equation when the voltage is constant, duty cycle 100 percent?

5. When the capacitor is _not_ connected, the current through the bulb is pulsed. So the power being supplied to the bulb is no longer constant. Depending on the thermal lag of the filament, this power is "smeared out" or averaged over some time interval, so the bulb is actually dissipating some power even when the filament is not receiving any current. Hence it can appear to be glowing steadily even though its current supply is pulsed. During the current peaks, as measured by the scope, the power dissipated is related to the _square_ of the current, adjusted somehow for the thermal lag and the temperature coefficient of resistance of the filament. Right?
Question: Is it legitimate to use the simple "mean" value of the current to calculate the average power dissipated by the bulb in this case, since the power supplied during the peaks is related to the square of the current?



Are the Red Herrings starting to jump out of the bucket yet?

[TinselKoala]

Yes, incandescent bulbs are not linear but that does not mean that they are not monotonic.
Light bulb's brightness depends on the current flowing through it and Tinman is correct that more current always means more brightness....and in his experiment this is all that matters.

I am more concerned how his scopes calculates the average of these pulses, e.g. from the screen data, from memory or from the waveform period...

I believe the scope calculates the measurements from the data displayed on the screen, or, in the case of my Rigol, it can also calculate between cursors. Yesterday I did a rough visual analysis of TinMan's scope traces and I think it is calculating the "mean" value correctly. A better question might be, as I've tried to clarify above, "Is the mean value of the current (and/or voltage drop) the appropriate measurement to use here when estimating the power dissipated in the bulb?"
Looking at the screenshot below, is the "A" area the same as the "B" area? I think it is, pretty close anyway, which tells me that the scope is calculating the mean value correctly. I think.