Not all analog-input smoothing happens in software. Many analog input modules clean up their readings in the hardware itself, sampling far faster than they report and averaging a whole block of those fast samples into each value the PLC sees. Chosen well, the size of that block can cancel out the steady electrical hum that mains power induces on field wiring before the noise ever reaches the application. This guide explains what block averaging is, how averaging over a line-frequency cycle rejects 50 or 60 Hz interference, how it contrasts with software filtering, and how it interacts with the module's update rate and effective resolution.
Block averaging in one line: Block averaging is a hardware technique in which an analog input module samples its converter much faster than it reports, then averages a block of those samples into each value passed to the PLC. Averaging over a whole mains cycle cancels the 50 or 60 Hz hum induced on field wiring, so it is commonly used for mains-frequency rejection. Being done in hardware, it cleans the signal before any software filter sees it, at the cost of a slower module update rate.
An analog input module does not have to report every reading its converter takes. Many modules oversample - running the analog-to-digital converter far faster than the rate at which they hand values to the PLC - and then combine a block of those rapid samples into a single reported value by averaging them. The PLC receives one clean number per scan, but that number is the average of many underlying conversions rather than a single instantaneous reading. This averaging happens entirely inside the module, in its hardware and firmware, before the value is presented to the application.
The reason this works as noise reduction is the ordinary statistics of averaging: random noise that varies from sample to sample partly cancels when many samples are averaged, so the averaged value is steadier than any single conversion. The more samples that go into each block, the more the random component is smoothed, at the cost of taking longer to collect them. The module is doing, in silicon, a job similar to what a software average would do, but it is doing it on the raw high-speed conversions the PLC never even sees.
Because this all occurs before the value reaches the control application, block averaging is effectively invisible from the PLC program's point of view - the module simply delivers cleaner readings. That is a strength and a subtlety: the smoothing is real and useful, but an engineer looking only at the PLC logic will not see it there, so understanding a signal's true responsiveness means knowing how the input module is configured as well as what the program does with the value.
The most valuable trick block averaging enables is the rejection of power-line interference. Field wiring runs alongside and is coupled to equipment energized at the mains frequency - 50 hertz in many regions, 60 hertz in others - and this induces a small, steady oscillation on analog signals at exactly that frequency. Because the interference is periodic and always at the same frequency, it can be cancelled rather than merely smoothed: if the module averages its samples over a time span equal to one full mains cycle, the positive and negative halves of the interfering wave sum to nearly zero, and the hum is largely removed from the reported value.
This is why the integration time of an analog input is often chosen to match, or be a whole multiple of, the mains period. Averaging over one 60 hertz cycle, for example, spans exactly one full oscillation of 60 hertz interference, so that interference averages out; averaging over one 50 hertz cycle does the same for 50 hertz. Some modules offer a configuration for the local line frequency precisely so their averaging window can be aligned to it, and getting that setting right for the region is what makes the rejection effective. A window that does not match the mains period leaves part of the hum uncancelled.
This kind of synchronous rejection is something a generic software filter does less cleanly, because it targets a specific known frequency rather than smoothing all fast movement indiscriminately. By tuning the averaging window to the mains period, block averaging removes the one dominant, predictable source of electrical noise while disturbing the real signal as little as possible - a more surgical approach than simply damping everything fast. It is one of the main reasons averaging is done in the module hardware where the timing can be controlled precisely against the converter.
Block averaging is not free; it trades speed for quality. Collecting a whole block of samples takes time, so a module that averages more samples per reported value updates more slowly. There is a direct tension between how much averaging is applied and how often the module can deliver a fresh value, and choosing the block size means deciding how much update rate to give up for how much noise reduction. For a fast signal that must be seen quickly this argues for a small block; for a slow, noisy signal it argues for a larger one.
There is also a relationship with effective resolution. Averaging many samples can improve the usable resolution of a reading, because the averaging suppresses the small random fluctuations that would otherwise obscure the least significant bits, letting genuine fine changes show through more clearly. So a longer averaging window often buys both cleaner noise rejection and a steadier, finer reading - again at the price of a slower update. The right balance depends on whether the signal's value lies in its speed or in its precision and quietness.
It helps to see block averaging as one stage in a chain of smoothing that a field signal can pass through, each at a different place. The transmitter may damp the signal at the source; the input module may block-average it in hardware; and the PLC or SCADA may apply a software filter on top. These stack, so an engineer who wants a signal to be both clean and responsive has to account for all three rather than tuning one in isolation. On a cloud SCADA platform such as Merobix, where values arrive from many remote sites, knowing that a signal was already averaged in the input module - and configured for the correct local mains frequency - explains why it looks clean and how quickly it can really respond, and it guards against piling on further software smoothing that would make an already-averaged signal needlessly sluggish. The clearest results come from deciding, across the whole chain, where each part of the smoothing should live.
Mains-powered equipment induces a steady oscillation on field wiring at the line frequency. If the input module averages its samples over a time span equal to one full mains cycle, the positive and negative halves of that interfering wave sum to nearly zero, cancelling the hum. This is why the averaging window is often matched to the local 50 or 60 hertz period, and why some modules have a line-frequency setting.
Block averaging happens in the input module hardware, oversampling the converter and averaging a block of samples before the PLC ever sees the value, and it can be tuned to cancel a specific frequency like mains hum. A software filter runs in the PLC or SCADA on the value the module already reported. The two stack, so a signal can be averaged in hardware and then filtered again in software.
Averaging more samples per reported value takes longer to collect, so a module that averages heavily updates more slowly. There is a direct trade-off between how much noise reduction the averaging provides and how quickly the module can deliver a fresh reading. Longer averaging can also improve effective resolution, but it is chosen against the need for a fast update on quickly changing signals.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.