Scale Response Time: Less Noise, More Speed
June 17, 2009
Tony Boehm
Rice Lake Weighing Systems
Congratulations! Once again, you are being asked to squeeze more speed out of your weight-based process. Where is it going to come from this time, and how are you going to get the scale to settle fast enough to keep up with the rest of the process?
We often turn to digital filtering to provide stable weight data in high-speed applications, but most filtering algorithms use a form of averaging to calculate a stable output. Averaging requires extra sampling, which requires more time, and that is exactly what we don’t have. Instead of filtering out the noise on the back end, maybe we should spend more time getting rid of it on the front end.
For simplification, let’s lump all noise into two categories: component noise and environmental noise. Component noise is inherent in the system given the components you are working with. You can use higher-quality cables or ADCs but the minimum amount of noise you can achieve is limited by the performance of your selections. This is the noise that many installers think they have in their system because that is what the product data sheet from the equipment manufacturer states. Unfortunately, the noise floor for most industrial installations is nowhere near the minimal noise conditions present in the lab where the manufacturers performed their tests.
Environmental noise is all the rest of the noise present in your system.
A basic understanding of the various sources that contribute to environmental noise can go a long way toward minimizing the amount of filtering needed, thus increasing your process throughput.
Before we explore ways to minimize noise, it is important to understand that what we really want is to maximize our signal-to-noise ratio (SNR). SNR is a common term used in electronics lingo that describes the level of signal with respect to the amount of noise present. We can reduce our noise or increase our signal. This may seem obvious, but it is surprising how many system installers overlook this simple idea. For example, if your weight-based process uses bridge sensors, it may be possible to increase the excitation voltage or, at the very least, consider resizing the load cells. Many installations have load cells sized two to four times beyond their needs. If you can cut your cell’s capacity in half, you will double its signal output. This is low-hanging fruit that should not be ignored.
The best way to minimize noise when installing an instrument is to follow the manufacturer’s recommendations. Most of the weight indicators that we deal with have been designed and tested to comply with rigorous emissions and immunity standards. To comply with those standards, some indicators will require shielded cables grounded at the connector, some may ground to the enclosure, and some may require ferrites installed on the cables. Regardless of the method, you can be sure that the manufacturer’s recommendations are actually requirements for their equipment to consistently comply with the tests they so proudly display on the product packaging and labeling.
The power source can also be a source of significant noise, especially in an industrial setting. In the cost-sensitive world of product development, power conditioning can take a back seat to other features. Instrument manufacturers often accept that if an installation has a poor-quality power source it will probably need more conditioning than can be squeezed into the enclosure of the instrument. This is why many seasoned system integrators just include the cost of a good quality uninterruptible power supply (UPS) in the machine’s budget. In some applications, a high-quality UPS can work wonders for cleaning up the power supply and lowering the system’s noise floor.
In summary, increased speed comes from reduced filtering, which can be achieved by increasing SNR. Boosting SNR can be accomplished by increasing the signal or reducing the noise. Following the equipment manufacturer’s recommendations is the best way to ensure your installation techniques will align with the conditions and assumptions made by the team that designed the instrument.
Tony Boehm is a senior electrical engineer at Rice Lake Weighing Systems (Rice Lake, WI) and holds an MS in signal processing from the University of Wisconsin–Madison.
You May Also Like