Measuring Led Driver electrical efficiency.

[HackedFridgeMagnet]

Just wondering what is the best way to measure the steady state electrical efficiency of a low voltage Led Driver.
I haven't got a programmable load.
I am assuming all the waveforms (Vin, Iin, Vout, Iout) are not going to be either pure dc or a low frequency AC sinusoid.
Vin should be solid in this case at 24v.
Assuming Iin, Vout and Iout have a non zero AC component as well as the DC component. The switching frequency on the Driver is around 100khz.

What would be a good way to measure the efficiency.
1% accuracy would be nice, but I doubt I could get that off my cro anyway.

I have 2 meters a Fluke and a BK prescision, and an old HP 1742a scope and another old 20mhz analogue Cro.
I can get access to a couple more  cheap meters to, if that would help.

I was thinking of just capturing all the non pure dc waveforms and multiplying them out to give both Input and Output power as a function of time.
Then I was thinking of averaging these values.

Anyone got a better idea

[Rufus]

Assuming Iin, Vout and Iout have a non zero AC component as well as the DC component.

If the voltage is constant average power is proportional to average current.

The voltage across the LED should be reasonably constant and if it isn't you may be able to parallel some capacitance to make it so (probably giving a small artificial increase in efficiency).

A difficulty is voltage burden measuring the currents. Input is easy just measure input voltage after the meter and if there is voltage ripple stick a capacitor on it. For the output if the driver has a current sense resistor measure the voltage across it and possibly find some way to calibrate it first.

Accurate efficiency measurement is difficult and the higher the efficiency the harder it gets.

[HackedFridgeMagnet]

Thanks Rufus.

If the voltage is constant average power is proportional to average current.

I might switch to a resistive load for the measurements to make this the case.


I was thinking about using a known purely resistive load. I need about 10watts, 1 amp. And accurately measure it's resistance near the operating point of the test.

Then use a kelvin connection( I think that is what it is called) to my multimeter via a low pass filter and measure on DC volts. this should give me the average output power. V2/R.

Then there is only the input current, I haven't looked at this waveform, maybe I can make some other assumptions here.