I have been doing some readings with a SKF Microlog and the overall measurement is always lower than other equipments. Does anybody know how overall vibration measurement is done with the Microlog? A technical data sheet would be appreciated.

Any chance that the device is measuring using a RMS detector and scaling it to read peak by applying a gain of 1.414? That method is valid only for an undistorted sine wave, something that rarely exists in a real machine. I'm not that familiar with the Microlog but see if you can change to a peak detection method.

John

It may be according to Hoyle? If for example you're comparing to CSI!

Normally the OA is the square root of the sum of the squares in a given baseband frequency. Some use 20 kHz and other 40 kHz regardless of the baseband frequency. You loose some analytical capabilities that way in my opinion.

I am meassuring 0-Peak values overall velocity.

My understanding of how the SKF box determines overall values is similar to Sam's. I've heard smart people call this "derived peak". I believe vibration purists prefer "true peak"...which determines the peak amplitude directly from the time trace, without first performing an FFT on the data. True peak reflects non-periodic events, such as impacts. I don't believe derived peak reflects non-periodic events as well.

So...if your "other" equipment employs true peak, and you are comparing that data to the SKF box's derived peak, the difference may be in how the data is massaged.

When I see differences between my SKF data and other data, I consider three possibilities:
1. The filtered range of my SKF data is different than that of the other data.
2. Non-periodic events, such as impacts, may be reflected in the true-peak data, but not evidenced as a periodic event in the derived peak calculation.
3. The SKF box employs digital intergration. The other data may employ analog integration. I suspect differences may occur as a result of this (ref previous posting: http://maintenanceforums.com/eve/forums/a/tpc/f/3751089011/m/8101044562 )

I'm not saying that any of these means of data collection are wrong. It does, however, point out pitfalls in comparing data.

George

Go to these links, the articles will probably tell you far more than you would like to know about instrument discrepancies.

http://www.bently.com/articles/articlepdf/2Q05_Understa...crepancyReadings.pdf

http://www.bently.com/articles/articlepdf/1Q06_Readings_Differ_pt2.pdf

This can be a serious matter so know what you are "measuring."

John

There may be two simple reasons, one of which was already covered. If you use the SKF collection in the normal mode where OA, FFT, and TW are stored together, the OA is calculated for the spectral bandwidth of the selected point. If there are high (or possible low )frequency components of vibration that fall out of the bandwidth, they will not be included in the OA amplitude.

The second possibility is you may be using an accelerometer with a very high bandwidth and a low resolution setting, such as 600,000 CPM and 400 lines. This results in a spectral bin width of 1500 cpm. When you integrate the system throws out the first bin and any vibration present below 1500 cpm goes with it.

To get around this and simulate what may occur in other systems two measurements are setup per point, one for OA and one for spectral analysis. This way you can get the overall value accross the valid range of the sensor and setup the spectral bandwidth to typical specifications for sleeve or anti-friction bearings. Just an opinion and not necessarily a fact.