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

[EMJunkie]

Now you are oversimplifying.  Sometimes the use of RMS values for measurements will be more appropriate, depending on what you are doing.

But as I've said, and as Verpies and Poynt99 have also confirmed, the proper way to do _power_ calculations, if you have the data as you do from a DSO, is to do the instantaneous multiplication of the actual per-sample values of voltage and current, which generates an instantaneous power waveform, then take the _average_  (not RMS) value of that IP waveform. This procedure works for all waveforms, whether complex or not, all load types, and all power factors. All the phase shift and other complications are taken care of by the sample-by-sample calculations performed in the scope, typically at hundreds of thousands of samples per second or more, so the errors caused by this "numerical methods integration" are very small.

Yes, I realise. Possibly a good idea to check with both equations anyway. Its not hard. Just a case of gathering the data which one will be doing anyway.

As one equation partially verifies the other.

   Chris Sykes
       hyiq.org

[EMJunkie]

So turn up your power supply until the scope/DMM reading is 2.5 V and then re-do your power calculations using the scope's new RMS value and let's see how it comes out.

It looks like your Duty Cycle is accurate across the scope screen. I have found that horizontal measurements are usually more accurate than the vertical ones, since there is less noise and you aren't limited to the 8-bit ADC precision levels.

Yes, this what I did in the first place, some drift slipped in there.

I should have rechecked it again before posting. Was only five minutes between experiment and taking Pics. Still shows others something else to watch for also.

   Chris Sykes
       hyiq.org

[EMJunkie]

For Giggles:

The 1KHz Signal is from a High Side Mosfet Driver. The 1K Resistor is measured to be 1K.

Mean Calculation: Red Channel

V / R = I = 5 / 1000 = 0.005 Amperes

I² * R = P = (0.005² * 1000) * 0.5 (Duty Cycle) = (0.000025 * 1000) * 0.5 (Duty Cycle) = 0.0125 Watts

RMS Calculation: Yellow Channel

V² / R = P = 3.458² / 1000 = 11.957764 / 1000 = 0.011957764 Watts

Now I must have something wrong here? We see: 0.000542236 Watts difference!

   Chris Sykes
       hyiq.org

I am still seeing a possible problem if the resistance changes in time, or am I seeing something that is not there?

In all of the Equations we use, Ohm's Law, if R changes during the operation of the circuit, then the Power through the Circuit Element in question will change.

Eliminating the problem, like TK has pointed out, can only be done with a Current Sensing Resistor with absolute minimal Inductance and using this Voltage Drop, Across the Non Inductive Current Sensing Resistor, CSR.

   Chris Sykes
       hyiq.org

[TinselKoala]

Actually the Ohm's Law relationships are definitions of each term in terms of the others. So, since the current depends on the resistance (if voltage is constant) then even though the resistance isn't explicitly included in the particular formula you may be using, it is still there hidden in the other variable values. So P=VxI does have the resistance in there, because I = V/R.

Here's where it becomes important to understand how the scope calculates "Average" or mean values. It is adding up values from very short timeslices and then dividing by the number of slices. So even if the resistance (or voltage, or current) may vary during the "on" time of a pulse, the sampling system will catch it, as long as it isn't varying too fast. But at 1 gigasamples per second..... well, you can see that even very small, very fast changes will be caught by the system and will give the true average. This is actually a problem sometimes, since the scope may be including noise or random glitches in the average. Hence, the scopes will have "Bandwidth limiting" that can be selected which will cut down on the presence of this kind of noise in the input samples to the averaging function.

On my Rigol, the 20 MHz bandwidth limiting is indicated by a "B" symbol in the Channel V/Div display at the bottom of the screen.

[EMJunkie]

Actually the Ohm's Law relationships are definitions of each term in terms of the others. So, since the current depends on the resistance (if voltage is constant) then even though the resistance isn't explicitly included in the particular formula you may be using, it is still there hidden in the other variable values. So P=VxI does have the resistance in there, because I = V/R.

Here's where it becomes important to understand how the scope calculates "Average" or mean values. It is adding up values from very short timeslices and then dividing by the number of slices. So even if the resistance (or voltage, or current) may vary during the "on" time of a pulse, the sampling system will catch it, as long as it isn't varying too fast. But at 1 gigasamples per second..... well, you can see that even very small, very fast changes will be caught by the system and will give the true average. This is actually a problem sometimes, since the scope may be including noise or random glitches in the average. Hence, the scopes will have "Bandwidth limiting" that can be selected which will cut down on the presence of this kind of noise in the input samples to the averaging function.

On my Rigol, the 20 MHz bandwidth limiting is indicated by a "B" symbol in the Channel V/Div display at the bottom of the screen.

Yes, this is the way I see it also, one can be interchanged to work out the other. Each are complimentary.

Ok I see what you mean, the over all Mean Sampling is fast and accurate enough to see any circuit changes most of the time. I must admit, I was not think clearly when posting the last post and really did not need to post it, so I am sorry for the Interlude there.

I think still, the oscilloscope is the most awesome machine we Humans have ever built. We would be lost without them!

   Chris Sykes
       hyiq.org