Hello All

I am working on a PMDC H Bridge motor drive (24V/16A). My PWM frequency is 20KHz and duty cycle in the application varies from less than 10% to 100%. I am using a Fluke TRMS 179 DMM for current measurement (in series up to 10A).

I am seeing that at 100% PWM (no switching) the current measurement is accurate and as expected. However, as the duty cycle is lowered these is an error in the meter current and the actual current (measured at the power source - battery).

This error increases inversely with duty cycle. (Lower duty cycles higher error). Also the error is higher at higher loads. In some cases the error is 200% - 300% of actual current. Clearly the Fluke 179 is a wrong instrument for this type of PWM current measurements.

So the question is are there other meters that can accurately measure switching PWM currents at frequencies like 20KHz. I need handheld solutions that can be used by installation personnel on the field.

Thanks
Richa

 

Richa, indeed you are correct in your suspicion the Fluke 179 is not appropriate for this job - it has a bandwidth of 1kHz.

In this particular case, I would try to borrow a DMM with higher bandwidth (Keysight U1272/1273/U1281/1282, Brymen BM859/869, Fluke 287/289) but even still the crest factor (the overall shape of the wave) might affect the accuracy of the measurement.

Another alternative is to get a current probe for an oscilloscope, which brings additional benefits such as a closer visualization of the waveform output of the PWM.

 

The problem is caused by the sampling rate of the ADC in the DMM. You would need a meter with extremely high sampling rate (multi-Mhz) to get good resolution when measuring a 20KHz PWM signal.
An well damped analog ammeter (Avo model 8) would give accurate readings at that frequency. If the current to be measured is very high you could use an analog voltmeter with a shunt.

 

You could use a small shunt resistor in series with the motor and measure the average DC voltage across that with an RC low-pass filter.
The AC frequency response of the meter is then not a factor in the measurement accuracy.
A filter RC time-constant of about 0.1s (e.g. 10kΩ and 10µF) will give -80dB of filtering.
The shunt and filter could be put in a small plastic box with banana jacks.

 

What is your expectation for the RMS value of a rectangular wave of a given duty cycle D?