Ski Slope FFT Vibration Data

One of the most frustrating issues to troubleshoot, “ski slope” is the common term for when vibration viewed in the frequency domain appears high at very low speeds then ramps down to almost nothing at higher frequencies. When ski slope occurs, overall vibration measured in the velocity scale is too high to be “real.” Ski slope is not good data and should be discarded. Many analysts make the mistake of trying to analyze this data, using it to schedule machinery repair or as valid contribution data to a machine’s overall vibration trend.  

Bias voltage shift is what causes ski slope. Accelerometer output is AC voltage vibration signal “riding” along a DC voltage called the “bias voltage.” Typical bias voltage is 8-12 VDC. Bias voltage must remain steady otherwise the analyzer will mistake it for vibration. This occurs when ski slope appears. The shifting DC bias voltage appears to be low frequency vibration in the eyes of the vibration analyzer. 

Overshock followed by lack of settling time or electrical interference are the primary causes for bias voltage shift and results in ski slope. Magnet mount accelerometers are most susceptible because of the strong pull force. Even if the technician is careful, the sensor can be over-shocked when the magnet grabs to the machine surface. This will not usually damage the accelerometer, but the sensor must be allowed to settle before a reading is taken. Vibration analyzers have the ability to measure the DC bias voltage. If the technician observes that the DC voltage value is moving they must wait until it settles before acquiring data. In the case of electrical interference it is recommended that the sensor’s cable shield be terminated to a clean earth ground.

Depending on the analyzer, DC bias voltage and vibration data may be difficult to measure at the same time while acquiring data. If Fast Fourier Transform (FFT) plots have a ski slope shape or overall vibration measurement in velocity is unrealistically high, technicians can use a “T adaptor” and digital volt meter (DVM) to monitor the bias voltage while acquiring data. Using the DVM in DC voltage mode while acquiring data gives technicians the ability to watch the bias voltage as the analyzer is acquiring data. If the DC voltage shifts by more than 0.1 VDC, there could be grounding or settling issue with the accelerometer. 

I applied this technique personally while troubleshooting permanent mount accelerometers in a HVAC motor and fan monitoring application on the roof of a pharmaceutical company. The sensors had integral cables and were routed to a junction box outside the enclosed motor and fan. These are notoriously noisy applications. Accelerometers with the best shielding and cabling are recommended. The installer did not ground the sensor cables, allowing both the ground wire and shield to float. With HVAC unit turned off, sensor bias voltage was steady at close to 10.0 VDC. But when HVAC units were turned on the bias voltage was constantly moving from 8.0 VDC to 12.0 VDC. Tying ground and shield to clean earth ground stopped the bias voltage shift, allowing us to collect meaningful vibration data.  

Finally, industrial sensor manufacturers are quick to advise customers to buy accelerometers with greater range (80 g’s or 500 g’s instead of the typical 50 g’s) when ski slope occurs. Certainly this can be valid advice especially if a sensitive accelerometer is being used with only 10 g’s range. But most industrial accelerometers are 100 mV/g with 50 g’s range. If vibration is exceeding the sensor’s range, signal clipping will occur which can also lead to ski slope. But these applications are rare within the industrial realm. To get a sense of how much physical vibration 50 g’s really is, take a portable shaker table and crank it up to 10 g’s, then 15, at 1800 CPM. Your desk will be shaking and if your door is open you will most certainly grab the attention of your coworkers. Remember that is just 1/5th of the vibration required to over-range a 50 g accelerometer. 

Before buying new high-range accelerometers, check the bias voltage of your old sensors. If it is unsteady, check cable grounding and sensor mounting then test the bias voltage again. There may be over-shock due to loose mounting or possibly electrical noise due to cable grounding.