Accurate Current Measurements with Oscilloscopes
Most engineers consider the oscilloscope their first tool of choice for most hardware development work. Yet very few
engineers ever consider how accurate their scope is. Even fewer actually look into it but would be very surprised by what
they learned. Before taking the time to make current measurements with a scope, it pays to take the time to understand how
accurate your scope is at the settings you need to use for your low current measurements.
|The Basics of Scope Accuracy|
Most of the major oscilloscope manufacturers place great importance on the timing aspects of their products. While timing
accuracy is often spec’d in double-digit ppm, voltage measurement error on the same scope can be as much as the signal level
you need to measure in a scope’s lowest volts/division setting. Multi-gigahertz sample rates are fairly common today in
mid-range digital scopes today yet most of those scopes only utilize 8-bit A/D converters.
Below is the formula for the voltage measurement accuracy from the datasheet for a popular line of mixed signal scopes from
one of the leading scope manufacturers:
Analyzing this formula, the accuracy is presumably determined by two major factors, (1) the analog front-end and amplifiers
and (2) the resolution of the A/D converter (given as 2mV in this case but on some scopes this value increases at higher
volts/division settings). When trying to determine a measurement level visually, you must also consider conversion/scaling
error in converting the measurement value into pixel coordinates on the display and the potentially huge human factor involved
in translating the information on the display into numeric values. The table below shows the voltage measurement error based
on this formula for several volts/divisions settings, with and without a display vertical offset (all error values and
percentages are +/-):
|DC vertical accuracy = ± (DC vertical gain accuracy + Display vertical offset accuracy + 0.25% full scale)|
- Full scale = volts/division x # of vertical divisions displayed
- DC vertical gain accuracy = ± 2% of full scale
- Display vertical offset accuracy = ± (0.1 x volts/division + 2mV + 1% of voltage offset)
|Full scale range
|Error in volts with 0 offset
|Error in volts with 4 division offset
|Error when voltage = 1 division ( 4 division offset)
|Error when voltage = 1 division ( 0 offset)
|Error when voltage = 50% of full scale ( 0 offset)
|Error when voltage = full scale ( 0 offset)
Several things should be obvious from this:
What isn’t as obvious is the impact of the display voltage offset. If you change from 100mV/division to 50mV/division to
zoom in on the waveform the error only degrades from 30% to 32%. If you also drop the 0V level from the center of the display
to the bottom of the display when making this change, the measurement error goes from 30% to 36%. Another thing that you can’t
see here is in this case the manufacturer requires a 30 minute warm-up period for the scope to meet the datasheet
specifications. If you turn on your scope and start taking measurements immediately or if your scope hasn’t been calibrated in
years then your measurement accuracy likely won’t even be this “good”.
- The accuracy degrades considerably as the volts/division setting is reduced but in the case of this particular family of
scopes even at the 5 volts/division setting the measurement of a 5V level can be wrong by almost 30%.
- The accuracy at full scale isn’t really important unless you have voltages that require the full scale to be displayed.
More important is since the result of the accuracy formula is a voltage, that voltage relative to the voltage you are trying
to measure is really the accuracy that matters. For example, at the 10mV per division setting, the full scale accuracy is 6%
but if you are trying to measure a 10mV signal the accuracy is only 52%.
- Even the best accuracy for a voltage equal to the volts/division setting is not very good at ±28% (with 0 offset).
|So How Bad is it Really?|
All of this begs the question “how bad can it really be in real life?”. The images on the next page show a current waveform
captured on one of mid-range scopes belonging to the family of products covered by the datasheet that the formula above was
taken from. The waveform on the left was taken using the "standard" probe for the scope and the one on the right with the
CMicrotek µCP100 current probe. The waveforms show the current draw from a lithium-polymer battery for an 8-bit micro as it
wakes up from a sleep state, performs a timer based background task and then returns to sleep.|
|Vertical scale: 10mV / division||Vertical scale: 5V / division|
|Voltage/Current conversion: 1mV = 1mA||Voltage/Current conversion: 1V = 1mA|
|Peak current: 25mA||Peak current: 16mA|
|"Average" current during task: 14.25mA||"Average" current during task: 8.3mA|
|Sleep current: 5mA||Sleep current: < 1mA|
|Scope measurement error: +/- 5mA||Scope measurement error: +/- 1.5mA|
These examples show measurement accuracy in line with the calculations above. Given the measurements with the "standard"
probe in the example above (operating current 200% too high and sleep current over 10X too high compared to the micro’s
current specs), an engineer could spend several very frustrating weeks trying to solve a power consumption problem that
doesn’t exist. This example also illustrates the importance of amplifying the current waveform so that it can be displayed
and analyzed more accurately as with the µCP100 example. The wave form captured with the µCP100 current probe are also impacted by
the scope’s measurement inaccuracy but are much more closely aligned with reality. This wave form is good enough to show the
increased current draw when the micro’s internal A/D convert is turned on and several A/D conversions are performed (the
slight “bump” at the fourth vertical reticule from the right edge of the display).
|Why Not Just Use a Larger Sense Resistor?|
It can be tempting to increase the current waveform amplitude by simply using a higher value sense resistor in order to be able to
use a scope on a more accurate volts/division setting. There are a couple of reasons this may not help:|
- Doubling the value and switching to 20mV/division only improves the error by 10%. Increasing the value by 5X and switching
to 50mV/division improves the error from 52% to 36% but for this test setup would also increase the voltage drop at the peak
to 125mV. Depending on the source voltage and circuit design this could be enough to interfere with the circuit’s operation.
- A decreased voltage level can have an unexpected impact on the current draw of a circuit. For most digital circuits, the
current draw will actually decrease slightly as the power supply voltage decreases (due to reduced resistive losses). Other
circuits require a certain power level so if the voltage level is decreased the current draw must be increased by a proportional
amount to maintain the power level. This will be the case for a battery voltage feeding into a voltage regulator or for the
power supply for a wireless radio transmitter.
In general, a scope may not be the best tool for highly accurate current measurements but it can still play a huge role in
developing firmware-based, power-managed products. Even when used with a low-end 8-bit scope, the µCP100 current probe can
provide accurate enough measurements to assess what the firmware is doing. This discussion also points out the importance of
understanding the accuracy of the scope you are working with. Even “high dollar” scopes can have surprisingly poor voltage
measurement accuracy and the best scopes on the market may provide poor results if not properly calibrated or not allowed to
warm-up as called for by the manufacturer.
µPower AnalyzerTM and µCurrent ProbeTM are trademarks of CMicrotek, LLC|
Copyright 2017 CMicrotek, LLC