1. Improving Signal-to-Noise Ratio (SNR)
Signal-to-noise ratio (SNR) is a term that is used to describe the ratio of the amplitude of the signal to the amplitude of the noise. A larger SNR typically results in a less noisy measurement, which enables you to better overall resolution. Noise in strain readings can be particularly troublesome because of the small signals that are involved in strain measurements. The SNR can be improved by either increasing the overall amplitude of the signal before the noise is introduced into it, or by reducing the amplitude of the noise.
A common source of noise in many applications is the measurement device. Noise introduced by the measurement device adds to the overall error of your measurements, and can obscure the smaller amplitude signals, which will reduce the overall dynamic range of your measurement device.
Noise that is introduced from an external source can often be associated with specific frequencies, and therefore can be filtered out in software if the frequency of the noise is predictable, and does not interfere with the bandwidth of the signal of interest. The most common type of noise is power line interference, which will show up as 50Hz or 60Hz noise in the measurements.
Techniques for rejecting external noise to improve SNR:
- Reduce the length of the strain gage’s lead wire, and keep the wire away from any potential noise sources. However, in many situations this may not be possible.
- Use Proper Shielding Techniques. Make sure that you connect the shield to the reference of the measurement device, which can be COM or EX- (refer to your device documentation), and make sure that it is only connected at one end of the cable. For isolated devices that have a floating ground, the shield needs to float at the same potential as the board’s signals to be effective.
- Twist the leads of the signals together. Twist the two signal leads together, the pair of excitation leads together, and the two remote sense leads together.
- Features of the measurement device that can help improve signal to noise ratio:
- Dynamic Range. Dynamic range defines the noise level relative to the full input range of the measurement device, and is often specified in terms of dB. The NI 9237 has a Spurious Free Dynamic Range (SFDR) of 106dB, which is equivalent to noise levels of about 0.0005 percent of the full input range. This means that very little additional noise will be contributed by the 9237.
- Common Mode Rejection (CMRR). Because noise from external sources is often conducted equally on all wires, a high common mode rejection will reject a large percentage of the conducted noise (for example, the 9237 has 85db of rejection which means that only 0.005 percent of the original common mode signal amplitude will be measured at the input). Using twisted pairs and matched signal wires will help ensure that the majority of the environmental noise gets conducted equally to the leads, which will maximize the effect of the device’s common-mode rejection.
- Remote Sense. When using remote sense, any noise that is conducted to the excitation cables will be canceled out when the data is sampled because the remote sense will compensate for the noise.
- Anti-alias filters. Anti-alias filters prevent high-frequency noise from being aliased at lower frequencies. This feature not only improves the overall noise performance of the device, but also allows software filters to be used very effectively for either filtering out specific frequencies (notch filter) or ranges of frequencies (low-pass/high-pass filter).
- Increase the amplitude of the signal. With strain measurements, this can be accomplished by either choosing a more sensitive strain gage, or by increasing the amplitude of the excitation voltage. Increasing the excitation voltage must be done with care because if the amplitude is increased too much, self-heating errors in the strain gage may outweigh the SNR benefits achieved with the larger excitation.
Maximizing SNR with your measurement device (the importance of gain in strain measurements)
One of the main reasons for applying gain to a signal is to scale a small signal to take advantage of the full input range of the analog to digital converter (ADC) in order to increase the resolution of the input signal. If the ADC of the input device has a variable input range, or a very large resolution, then gain sometimes is not necessary. For example, an input range of +-50 mV on a 16-bit ADC correlates to a resolution of 1.5 uV, which is equivalent to the resolution obtained with an input range of +-12.8 V using a 24-bit ADC.
If the signal path is prone to introducing noise into the measurement, then applying gain early on in the signal path before the noise is introduced can help increase the overall SNR. On some devices, such as SCXI, gain can be beneficial because the signal conditioning circuitry can be located several feet from the ADC. On other devices, like C series modules, the ADCs are built into the measurement device and are only inches away from the connector, so gain won’t improve the SNR.
In general, the two main factors that should be used to determine which measurement device you plan to use are (1) available signal ranges and (2) overall noise-free bits of resolution, often referred to as Effective Number of Bits (ENOB). Built-in filtering can also be important to improving accuracy by reducing aliasing and high-frequency noise from the measurements.
2. Choosing an Optimal Excitation Level
The change in output voltage for a given level of strain increases in direct proportion to the excitation voltage. Increasing excitation voltage thus improves signal-to-noise ratio, but a practical limit is reached when the ill effects of gage self-heating become predominant.
A common request in strain gage work is to obtain the recommended value of bridge excitation voltage for a particular size and type of gage. Unfortunately, a simple answer to this question is not possible because there are many different factors that will affect the optimal excitation level for a particular strain gage configuration.
This section of the document is intended to outline the most significant considerations which apply. However, a more thorough resource about the specific details of determining the optimum excitation voltage for a strain measurement is the Vishay Measurement Group Tech Note TN-502 “Optimizing Strain Gage Excitation Levels.”
Factors Affecting Optimum Excitation
It is important to realize that strain gages are seldom damaged by excitation voltages considerably in excess of proper values. The usual result is performance degradation, rather than gage failure; and the problem therefore becomes one of meeting the total requirements of each particular installation.
The following are factors of primary importance in determining the optimum excitation level for any strain gage application:
- Strain gage grid area.(Active gage length x active grid width.)
- Gage resistance.High resistances permit higher voltages for a given power level.
- Heat-sink properties of the mounting surface. Heavy sections of high-thermal-conductivity metals, such as copper or aluminum, are excellent heat sinks. Thin sections of low-thermal-conductivity metals, such as stainless steel or titanium, are poor heat sinks. Also, the shape of the gage part may create thermal stresses in portions of the structure due to gage self-heating. Long warm-up times and apparent gage instability can result. The situation often arises in low-force transducers, where thin sections and intricate machining are fairly common.
Strain measurement on plastic requires special consideration. Most plastics act as thermal insulators rather than heat sinks. Extremely low values of excitation are required to avoid serious self-heating effects. The modulus of elasticity of the common plastics drops rapidly as temperature rises, increasing visco-elastic effects. This can significantly affect the material properties in the area under the strain gage. Plastics, which are heavily loaded with inorganic fillers in powder or fibrous form, present a lesser problem because such fillers reduce expansion coefficients, increase the elastic modulus, and improve thermal conductivity.
- Environmental operating temperature range of the gage installation. Creep in the gage backing and adhesive will occur at lower ambient temperatures when grid and substrate temperatures are raised by self-heating effects. Thermal output due to temperature will also be altered when grid and substrate temperatures are significantly different.
- Required operational specifications. Gages for normal stress analysis can be excited at a higher level than under transducer conditions, where the utmost in stability, accuracy, and repeatability is needed.
A significant distinction exists between gages used in dynamic strain measurement and those used in static measurement applications. All the various performance losses due to gage self-heating affect static characteristics of the gage much more seriously than the dynamic response. Therefore, it is practical to "drive" the dynamic installations much harder, and thus take advantage of the higher signal-to-noise ratio that results.
- Installation and wiring technique. If the gage is damaged during installation, if solder tabs are partially unbonded due to soldering heat, or if any discontinuities are formed in the glueline, high levels of excitation will create serious problems. Proper technique is essential in obtaining consistent performance in all strain gage work, but particularly under high-excitation conditions.
An optimum excitation voltage is best determined by an experimental procedure. With no load applied, you should examine the zero point of the channel while excitation level is progressively raised. Once instability in the zero reading is observed, you should lower the excitation until stability returns. It is best to perform this experiment at the highest temperature over which you are taking measurements. Also, using larger gages and higher resistance gage, like 350 ohms instead of 120 ohms, decreases the power per unit area dissipated making higher excitation voltage possible.
Reducing noise and increasing resolution are important for making accurate strain measurements due to the very small voltage levels that are involved. Picking the right measurement device can greatly improve the integrity of your strain measurements, but gain and excitation level are not the most important factors in making accurate strain readings. You should pick a measurement device with a large dynamic range/resolution, and an input range that is sufficient for your application. Then if you take steps toward reducing the noise introduced into the system, you can reduce the excitation level to reduce self-heating errors and improve the accuracy of the signal from your strain gage. For static or slow-changing strain measurements you may also consider over-sampling and averaging multiple readings.
4. Other Strain Resources
“Optimizing Strain Gage Excitation Levels”