Most multimeters are bandwidth-limited to below 1khz while the current ripple of most SMPS is well above that. Even benchtop multimeters usually have bandwidth up to ~300kHz, which is not enough to accurately measure the output current of a high-frequency SMPS (800kHz or 1MHz). Yet measuring the output current of a SMPS with a multimeter is very common practice. So what am I missing here? How can measuring the current of a decently large ripple at a frequency higher than the bandwidth of your measuring device be accurate?
\$\begingroup\$
\$\endgroup\$
2
asked Jan 31, 2020 at 6:40
-
5\$\begingroup\$ 1k ohm frequencies? More to the point, what's wrong with measuring just the average DC voltage, ignoring the ripple? \$\endgroup\$– Dmitry GrigoryevCommented Jan 31, 2020 at 6:43
-
\$\begingroup\$ If you want to measure and see variations in current at that frequency, you cannot use a multimeter. You need to use an oscilliscope. You may find some really expensive current probes for oscilloscope or some hall effect chips that convert current to a linear voltage that can be read by an oscilloscope. \$\endgroup\$– NoumanQCommented Jan 31, 2020 at 6:53
Add a comment
|
3 Answers
\$\begingroup\$
\$\endgroup\$
2
Usually when we measure the output current to evaluate a DC power supply we are primarily interested in the DC component, and some mV or tens of mV of ripple on top (either from the power supply or because the load current has an AC component) on top of volts are not important at all.
If you want to turn things around and actually measure the ripple voltage or current and ignore the 99% of DC component, you need something with sufficient bandwidth and sensitivity. Usually that would be an oscilloscope with some kind of a probe such as a current probe. Low-value shunt resistors are problematic at high frequencies since even a bit of parasitic inductance can greatly affect the voltage drop. For example a 5nH inductance will affect the reading across a 50m\$\Omega\$ resistor by 40% at 1.6MHz.
answered Jan 31, 2020 at 7:32
-
\$\begingroup\$ Thanks for the response! Can you elaborate a bit on how you calculated 40% error at 1.6MHz? I'd like to be able to replicate this, it seems extremely helpful. \$\endgroup\$ Commented Feb 1, 2020 at 3:42
-
\$\begingroup\$ The resistance and the reactance add in quadrature so when the reactance is 50m ohms the total signal is \$\sqrt{R^2 + X_L^2}\$ or \$\sqrt{2}\$ times the correct number. \$\endgroup\$ Commented Feb 1, 2020 at 4:03
\$\begingroup\$
\$\endgroup\$
As an add-on to @Spehro's answer, the problem with current measurement will typically be with a load drawing high current pulses. The switching pulses of the SMPS will be internal and for the most part filtered out by the output smoothing circuit to present a steady DC output voltage.
answered Jan 31, 2020 at 8:08
\$\begingroup\$
\$\endgroup\$
DC measurements if done correctly will be filtered to allow only the Nyquist spectrum for each sample. AC measurements need to be done by using AC coupled to coax then terminated by 50 Ohms to avoid spurious noise. Casual measurements may be taken with the 20MHz DSO filter but DMM's are at risk to errors from crosstalk to long unbalanced leads.
