What Is The 3 dB Specification In Industrial Accelerometers?
Introduction
The ±3 dB specification is commonly used by sensor manufacturers, but its meaning is often misunderstood.
Let’s break down what it actually represents.
What Does a 3 dB Change Represent?
An amplitude decrease of 3 dB corresponds to an amplitude ratio of:
0.707 × the original value (or 70.7% of the original value)
So, if the sensor is excited with a 1 g input at its 3 dB cutoff frequency, the output would be roughly 0.707 g.
This is where the common statement that “3 dB corresponds to about 70%” comes from.
This specific value, also known as the half-power point, is commonly used because it is a mathematical transition point in a typical frequency response that marks the start of meaningful roll‑off or transition.
Why Does the Vibration Analysis Industry Use the 3 dB Point?
The ±3 dB boundaries are not arbitrary. They reflect how electrical filters and mechanical systems behave. Every accelerometer has internal circuitry that behaves like a filter. The cutoff frequency of a filter is mathematically defined at the point where:
- Output amplitude has dropped to 0.707 of the input, or equivalently output power has dropped to 50% of the input
- This equals a 3 dB decrease
This point marks the transition between:
- Passband, where the response is mostly flat and predictable
- And stopband, where significant roll-off or rise occurs
Every sensor manufacturer uses ±3 dB to define this transition.
What Happens Beyond the 3 dB Point?
Beyond the cutoff, attenuation increases rapidly. For traditional filters, this occurs at a predictable mathematical rate, e.g.:
- A first‑order filter rolls off at –20 dB per decade
- A second‑order filter rolls off at –40 dB per decade
This steep roll‑off is why using data beyond the 3 dB cutoff becomes unreliable.
Why is ±3 dB Acceptable in Vibration Monitoring?
For most vibration applications, especially trending, machine health, and general condition monitoring, you rarely operate at the exact cutoff frequency. You operate well within the usable bandwidth where the response is essentially flat.
Remember, you control what data you obtain from your sensor, namely through the determination of a proper FMAX setting in your acquisition equipment. The ±3 dB limits simply serve as a general boundary marker that your FMAX should be kept within. Inside this boundary, the sensor behaves linearly and accurately. Outside that boundary, amplitude begins to attenuate, and data reliability falls off.
If a user needs tight accuracy at a particular frequency, they should ensure that the chosen sensor's 3 dB point occurs well beyond this desired frequency, or that the sensor's response is flat/within desired tolerance at this desired frequency. Sensor manufacturers including CTC make this easy to determine, by providing 10% and sometimes even 5% frequency response markers, along with the typical 3 dB point markers. Operating within these limits guarantee respective frequency response tolerance.
Note that we mostly discuss 3 dB of attenuation here as it would relate to a typical low pass filter response. The same can be said for +3 dB of gain, representative of a response that rolls up instead of down. This is typical of accelerometer resonance shapes (if not mitigated my sensor filtering), where the output response rises sharply past the +3 dB point before rapidly tapering off. Operating well within the +3 dB point ensures that sensor resonance is not included in the data acquisition.
Remember too that sensor mounting and usage methods will often affect achievable 5%, 10%, and 3 dB frequency response points, when we talk about sensor resonance. Manufacturer datasheet specifications are almost always for a correctly installed, direct stud mount of the sensor to the equipment. Methods such as magnet mounting can and will reduce achievable 3 dB bandwidth in the field by reducing sensor/magnet system resonance, so always test for your specific application!
