Voltage Measurement in Data Acquisition Applications

What is voltage?

Voltage is the difference in electrical potential between two points. It is sometimes also called tension or pressure, because of the analogy between water and electricity. 

Think about a closed system of water pipes that includes a pump. The pump drives the water around and around the pipes, creating a difference in pressure that drives the water. This pressure pushing the water allows it to do work, such as turning a turbine, for example.

In this analogy, water represents electricity. The pump represents a power supply. The pressure that the power supply creates in the circuit is the voltage potential, and the rate at which the water moves represents the current.

AC vs. DC voltage

Voltage can be either DC or AC, depending on the current that is carrying it. In DC systems, the current never changes direction. It is unidirectional, i.e., it does not change polarity.

But in AC systems, the current alternates directions, crossing 0V in a positive direction, then turning around and crossing 0V again in a negative direction. You can see both DC and AC voltage (and current) represented in the graphs below:

The most well-known sources of DC voltage are the common AAA battery or the much bigger one that starts (or even powers!) your car. The most well-known source of AC voltage is the 120 VAC or 230 VAC that powers our homes and businesses.

The terminology of voltage measurement

You may have heard terms like “single-ended referenced,” "single-ended, non-referenced” and “differential” and wondered what they really meant in terms of making voltage measurements. Well, it’s all about the reference point. Remember that right at the beginning of this article we established that voltage is the electrical potential difference between two points.

Single-ended measurements

Single-ended measurements are those that are made with respect to the ground. The signal is carried only on the positive wire, and the other wire is referenced to the ground. But who provides this ground? 

If the measuring instrument provides the ground reference, then the measurement is classified as “referenced, single-ended.” This is often abbreviated as RSE.

On the other hand, if the signal itself provides the ground reference, then the measurement is classified as “non-referenced, single-ended”, abbreviated as NRSE.

Differential measurements

If we ignore the ground and measure between two points in a circuit, we are making a differential measurement. It’s called differential because we measure the difference between these two points. We’re really measuring two signals instead of one in a single-ended scenario. Each differential channel really has two preamps inside it, and these inputs are floating with respect to the ground.

Floating signal sources

You may also hear about a signal source being a “floating” one. This simply means that it is not directly referenced to the ground. A common example of such a source would be a battery. In the case of measuring from the outputs of floating signal sources, the measurement system needs to provide the ground reference. 

What do average, RMS, and peak-to-peak voltage mean?

Voltage can be quantified in several ways. Let’s take a look at each of these common terms and what it means.

The average voltage (\(V_{AV}\))

As the name already states, the average value during one period. For pure sinusoidal signals, the average will be zero, because the amount of current in the first positive half of the waveform is equal to the current in the negative half cycle. These currents cancel each other out resulting in zero. So taking just one half of a cycle we can define the average by taking the peak-to-peak (max) value and multiplying it by 0.637.

The RMS voltage (root-mean-square aka \(V_{RMS}\)) 

The RMS voltage is the square root of the arithmetic mean of the squared function values that define the continuous waveform. A common way to calculate RMS is to multiply the peak value by 0.707. RMS is the most common way to express AC voltages.

The Peak voltage (\(V_{PK}\) or \(V_{MAX}\))

The Peak voltage describes the highest voltage within one period.

The peak-to-peak voltage (\(V_{PK}\))

The peak-to-peak voltage defines the entire amplitude of the positive and negative peaks within one period. 

The crest factor

The crest factor is a ratio of the maximum peak values divided by the RMS value of an AC waveform. Since DC voltage levels and square waves don’t have peaks, they have a crest factor of 1, whereas a pure sine wave has a crest factor of 1.414.

Please note that in most measuring systems, the average, RMS, peak, and crest factor values are typically calculated over a period of time, usually a subset of the data acquisition system’s selected sample rate. This is a very useful way of presenting these values.

For example, in DewesoftX data acquisition software, the user can select any of these values and have them calculated at a divisor of the selected sample rate. Here is an example of the setup screen, where you can select which statistical values you’d like to display/record:

The picture below shows how it looks on the display screen. Channels can be displayed in a variety of graphical widgets, from simple numerical displays to strip charts, bar graphs, and more.

What is common-mode voltage (CMV)?

Common mode voltages are signals that are present on both leads of a signal source. In reality, there should not be identical signals on both leads, so the common mode is usually noise that has crept into the signal chain.

The best way to eliminate or reduce common-mode voltages is to make a differential measurement. 

To explain, let’s step back a little. In the single-ended measurements mentioned above, we are using one preamplifier to measure the positive signal line. If the noise gets into the signal, how can we tell? How can we know what is the real signal and what is the noise? 

Perhaps from experience, we can see 60 Hz riding on top of the signal, but it’s a challenge.

The most basic approach to eliminating common-mode signals is to use a differential amplifier. This amplifier has two inputs: a positive one and a negative one. The amplifier measures only the difference between the two inputs.

What is a common mode rejection (CMR)?

Electrical noise riding along on our sensor cable should be present on both lines - the signal positive line and the ground (or signal-negative) line. Signals common to both lines will be rejected by the differential amplifier, and only the signal will be passed through, as shown in the graphic below:

This works great, but there are limits to how much common-mode voltage (CMV) the amplifier can reject. When the CMV present on the signal lines exceeds the differential amplifier’s maximum CMV input range, it will “clip.” The result is a distorted, unusable output signal, as shown below:

So in these cases, we need an additional layer of protection against CMV and electrical noise in general (as well as ground looping, which will be discussed in the next section) - isolation.

An isolated amplifier’s inputs “float” above the common-mode voltage. They are designed with an isolation barrier with a breakdown voltage of 1000 volts or more. This allows it to reject very high CMV noise and eliminate ground loops.

Isolated amplifiers create this isolation barrier by using tiny transformers to decouple (“float”) the input from the output, by small optocouplers, or by capacitive coupling. The last two methods typically provide the best bandwidth performance.

What is a ground loop?

Unless they are prevented, ground loops can be a serious problem for measuring systems. Sometimes called “noise”, a ground loop is caused by inadvertently referencing electrical equipment to more than one path to the ground - any difference of potential at these grounding points causes current to flow between them, inducing a current loop. This causes distortions in the signal which, if high enough, can ruin the measurement. 

In the picture below, the measurement amplifier is connected to the ground (GND 1) on one side. An asymmetrical shielded cable is used to connect the sensor, whose metallic housing is placed on a conductive surface at GND 2. Due to the length of the cable, there is a difference in potential between GND1 and GND 2. This potential difference acts like a voltage source, coupling with the electromagnetic noise from the environment.

If the sensor could be decoupled from GND2, it might solve the problem. But sometimes this is not possible. Furthermore, sometimes the referencing of a cable shield is required by safety rules, and therefore should not be removed.

The best solution is to use a differential amplifier within the signal conditioner which is isolated. With this one change, the problem is solved.

Ground loops can also come from the instrument itself, via its own power supply. Keeping in mind that our measuring system is plugged into power, which has a ground reference. It is critical therefore to decouple this reference from the signal-handling components of the instrument to ensure that ground loops cannot be created within the instrument.

This scenario can become dangerous if there is a wiring fault. Looking at the high current path from the power supply, what will happen if the return line is broken? All of the energy will be routed through the signal conditioning part of the DAQ hardware. This could result in the entire system being damaged or destroyed, and even dangerous potentials for the human operator of the instrument.

By isolating the signal path completely from the power supply, it is not possible for the scenario above to occur.

Galvanic isolation

Given all of the information above, it seems clear that our measuring systems should have at least differential and preferably isolated analog signal inputs. 

But when you’re reviewing the isolation specifications of various measurement systems and signal conditioners, you may find it specified with terms like: 

• channel-to-ground,

• channel-to-channel and

• bank isolated. 

What do these terms mean, and how do they relate to each other?

Channel-to-ground isolation

Channel-to-ground isolation defines the maximum voltage there can be between a channel’s input and instrument ground. Normally an instrument’s ground is referenced to the ground of the power supply. By isolating signal ground from chassis ground we can eliminate most ground loop problems.

Sometimes this is also referred to as input-to-output isolation. All of the channels share a common ground, which is isolated from the ground or earth potential of the instrument. This would not be a limitation if only one signal source were connected to the system. But when additional signals are connected, each with ground potential differences, it can lead to noise across all signals and common mode problems. 

If two or more channels share a common ground, then they are not galvanically isolated. Take care when an instrument mentions only input-to-output or channel-to-ground isolation. 

Channel-to-channel isolation

Channel-to-channel isolation defines the maximum voltage there can be between a channel and any other channel. Channels cannot share a ground bus, for example. Each channel must also be isolated from the rest of the system, e.g. the system’s power supply voltage, chassis ground, and so on. If all channels are isolated from each other, then they are necessarily also isolated from the ground, so channel-to-ground isolation is included within channel-to-channel isolation.

So, if a system has channel-to-ground isolation, it does not necessarily mean that it has channel-to-channel isolation. But, if a system has channel-to-channel isolation, then it must also have channel-to-ground isolation.

Several data acquisition systems from Dewesoft provide both channel-to-channel and channel-to-ground isolation, as shown in this short video:

Bank isolation

Bank isolation is typically employed by high-channel-count systems. In this scenario, channels are isolated in groups that share common isolation. This can be done either to save money or out of necessity. 

Isolation components take up space and consume power. In very high-density measuring systems sometimes there is literally not enough room for each channel to be separately isolated.

When you look at the Dewesoft compatible signal conditioners for voltage later in this article, you will see that the SIRIUS HD (High Density) signal conditioners are bank isolated in pairs, where every two channels share an isolation path. All other SIRIUS signal conditioners are isolated per channel.

Learn more about Isolation:

The Importance of Isolation in Data Acquisition SystemsElectrical isolation is a separation of a circuit from other sources of electrical potential. Learn about importance of galvanic isolation in DAQ systems.

What are “CAT” safety standards?

Looking at instrumentation specifications for high voltage inputs, you will notice terms like “CAT II” and “CAT III”, with some voltage levels next to them. What do these mean?

“CAT” refers to the IEC (International Electrotechnical Commission) measurement category of how and where instruments are used with respect to living circuits. There are four categories, listed by the Roman numerals I, II, III, and IV. 

These CAT values essentially refer to the location of the instrument. The lowest value of this means that the instrument will be located farthest from the high voltage and transient potentials, while CAT IV refers to locations very close to the source of high voltages and therefore transients.

Then within each of these CAT values, you will often see voltage values that refer to the instrument’s ability to withstand transients up to that value. An instrument that is rated to CAT II-1000V is of course able to withstand many higher-level transients than an instrument that is only rated only to CAT II-600V.

CAT value example locations

Basically, the closer the equipment is connected to the primary source of voltage power, the higher the possibility of dangerous transient voltages getting into the instrument, and therefore the higher the CAT number is.

Transient overvoltage protection by working voltage and CAT

In the table above, “working voltage” refers to DC or $AC_{RMS}$ to the ground. The peak transient refers to 20 impulses at the given voltage level. The test source impedance comes from V/A.

Notice also that the source impedance for CAT III and above is 2 Ω versus 12 Ω for CAT II! 

According to A=V/R (Ohm’s law - a basic tenet of electrical engineering), a 2 Ω source has six times the current flowing as a 12 Ω source. This is why the CAT number itself is more important than the voltage value that follows it. For example, a CAT II 1000V-rated instrument is not superior to one rated to CAT III-600V because of this source impedance difference.

So how do you know what CAT level and voltage to require for your high-voltage application? Because high voltages involve not just the safety of your equipment but the safety of human operators, always consider the worst case in terms of dangerous high-voltage transients and choose an instrument that will protect you and your colleagues.

Dewesoft signal conditioners with CAT ratings

What is signal overload (overmodulation)?

When signal levels are higher than expected, they are clipped by the ADC converter, resulting in wrong measurements, which means you have to do the test all over again. This is called signal overload, clipping, and overmodulation, among other terms.

Engineers have been struggling with high dynamic range signals for decades. Dynamic range refers to the difference between the smallest and largest amplitude excursions of a signal. Imagine that you are making a measurement and most of the time the signal is in the millivolt range. Still, occasionally it jumps up to 80 V. If you set the input range to 100V in order to avoid clipping when the signal rises, the resolution of the signal when it’s in the millivolt range will not be optimal.

Engineers have addressed this by inputting the same signal into two channels of their measuring system and setting the gains differently. This more or less solves the problem but it creates two more problems: 

1. You need twice as many measurement channels in your system.

2. Analyzing the data is far more difficult because you must manually combine the data sets from the two channels after the test to create a composite dataset for each channel. This is an enormous workload and additional analysis burden, especially when multiplied by many channels.

A better solution would be a DAQ system where each input channel really had two ADCs inside, each set to a different gain, and a fast processor that would automatically choose the one that best represented the signal and combine them into a single data stream. 

Does that sound too good to be true? It has already been done by Dewesoft with their SIRIUS DualCore® ADC technology. Each channel amplifier has two ADCs that always measure the high and low gain of the input signal. This results in the full possible measuring range of the sensor and prevents the signal from being clipped.

Dewesoft’s DualCoreADC® technology achieves more than 130 dB signal-to-noise ratio and more than 160 dB in dynamic range. This is 20 times better than 24-bit systems and 20 times less noise.

Voltage transducers

Every instrument that can measure an analog voltage signal can do so directly, right? So why would we ever need a voltage transducer?

Nearly every DAQ system and data logger in the world can directly accept low and medium voltages in the ranges of 0-10V or 0-50V, so we do not need a sensor or transducer to reduce or convert this voltage in any way. From 50V to approximately 1000V there are signal conditioners available for DAQ systems such as the Dewesoft SIRIUS HV module, which can directly and safely accept these voltages and internally step them down so that they can be digitized, displayed, and stored.

But at higher voltages, or in any case, when life-threatening currents and voltages are present, it is essential to use a high voltage transformer to step down the high voltage and isolate the human test operator from dangerous voltage and current. Such a device is called either a Voltage Transformer (VT) or a Potential Transformer (PT).

The typical PT sensor includes a transformer to step down a very high potential - even higher than 10kV - down to a safe level. It can be placed in series with or across the circuit being monitored. The transformer’s primary winding has a large number of turns compared to the secondary. 

Because the data acquisition device connected typically has a very high impedance, very little current will flow, therefore the PT’s secondary winding experiences almost no load at all. Most PTs output between 50 and 200V, which nearly every DAQ system can accept.

PTs are available for outdoor usage and those designed for indoor usage. There are also those designed for electrical metering applications. There is also an alternative to the pure transformer type which uses a bank of capacitors after an intermediate transformer to further step down the voltage. These can be less costly because the relatively low step-down ratio intermediate transformer is less expensive than the conventional wound transformer with a high step-down ratio. 

A third variant is optical VT. Optical VTs are usually found in power substations, and not often in DAQ applications. Since they operate on the principle of the Faraday effect, whereby the polarization of light is affected directly by a magnetic field, they are inherently isolated. They are also extremely accurate.

Voltage transducer applications

• Energy Production and Distribution high voltage power line testing, synchronizing generators with the main power grid, 

• Aerospace - engine and power system testing

• Automotive - electrical circuit system test, hybrid, and electric motor tests

• Transportation - electric subway cars, third rail and pantograph tests, electrical energy distribution centers

Voltage transducer pros

• They provide essential safety to the test engineer and technician

• Easy to use

• Most models do not require external power

• Long-life operation

Voltage transducer cons

• Can be expensive

What is aliasing?

Let’s say your AC voltage is a 10 kHz sine wave, but you only take one sample from it every second. Obviously, the resulting recording is going to be completely wrong. 

Between each sample that you take, 10,000 sine waves will have gone by. The resulting “signal” will look like a waveshape, but it will be completely wrong. It will be an “alias” of the actual signal. It is dangerous because you get something that looks like a signal, but of course, it’s wrong.

Let’s use a practical example. The graphic below represents the real signal that we are trying to measure:

Now imagine that we are not sampling it fast enough. In fact, we are sampling at the rate shown by the dots in the graph below:

You can probably already see what our “signal” is going to look like after this process:

This is a completely wrong result: it does not resemble the actual signal at all.

The most obvious solution to aliasing is to simply sample faster, to ensure that the frequencies of your signals will never be higher than the sample rate can handle.

But in a practical sense, this is not always possible. Sometimes unexpected transients happen, for example. 

If we filter in the analog domain before the ADC, we can prevent the aliasing problem from ever occurring. Note that it is still important to set a high enough sample rate to capture the frequency range of interest, but at least with Anti-Aliasing Filters (AAF), we will avoid false (“alias”) signals from destroying the integrity of our measurements. 

The ideal AAF would have a very flat passband AND a very sharp cutoff at the Nyquist frequency (essentially half of the sample rate). 

Typical AAF configuration: a steep low-pass analog filter before the ADC prevents signals more than half of the maximum bandwidth of the ADC from passing. This is what Dewesoft does with its 16-bit SAR ADCs as found in SIRIUS-HS modules.

However, with their 24-bit Delta-sigma ADCs, as found in SIRIUS, KRYPTON, and IOLITE data acquisition system for nearly the entire product line, Dewesoft DAQ systems have an additional DSP filter on the ADC itself that automatically adjusts based on the sample rate that the user has selected. This multi-stage approach provides the most robust anti-aliasing filtering available in DAQ systems today. 

Dewesoft’s fast sigma-delta ADC technology is the best possible approach to preventing aliasing.

Devices for voltage measurement

Voltage is one of the most often recorded signals, and virtually every DAQ hardware in the world can measure it to one degree or another. Other instruments can also measure voltage, such as laboratory oscilloscopes and voltmeters, just to name a few.

Voltmeters are very accurate but have low bandwidth, while oscilloscopes have very high bandwidth but are not as accurate, for example. The table below breaks them down by low and high ranges, accuracy, and bandwidth:

The numbers are just generalizations: there are many instruments on the market, and their specifications can vary greatly.

In addition, you sometimes need to measure very tiny voltages, i.e., in the microvolt range ... and all the way into the thousands of volts. A voltage preamplifier with multiple input ranges is needed to convert these widely different signal levels to a normalized output that can be digitized. Sometimes a tiny voltage is riding on top of a large DC offset, which is yet another challenge that many measuring systems have trouble with.

Dewesoft DAQ devices for voltage measurement

PT = Potentiometric Transformer is absolutely required for safety, isolation, and voltage division

Learn more:

The Importance of Isolation in Data Acquisition SystemsElectrical isolation is a separation of a circuit from other sources of electrical potential. Learn about importance of galvanic isolation in DAQ systems.