Skip to main content

Replies sorted oldest to newest

I am using a LM2907 frequency to voltage converter.
The input is the pulse train from a magnetic pick-up probing the tooth gear of a diesel engine flywheel, or gearbox or from a 1024 .. 4096 pulse/rev incremental encoder mounted on NDE of a electric motor drive, or any frequency modulated ac signal. This circuit functions as a frequency demodulator.
The output can be fed into any frequency spectrum analyser (vibration datacollector).
The frequency spectrum reveals torsional vibration content, excited torsional resonance frequencies, pump blade pass pulsation, speed governor (in-)stability, etc.
Simple low cost home-made test equipment.
A cmos 4046 phase lock loop circuit can also serve very well to detect what goes undetected using conventional accels!

Attachments

Interesting thread...
Question: Do these converters (Trig Tek & Copp Tek)just convert the pulses in time domain to a frequency peak? or do they also do the demodulation to actually measure the torsional resonance?

Some years ago, I was asked to provide startup rate info on a DG. I used a prox probe to collect data from the toothed flywheel. Wrote data to file and played back later.
TEDIOUSLY, I measured average gear passing rate to get instantaneous speed.
A few years later, I discussed this with Tony DeMatteo who mentioned that the Copp Tek would do this on the fly.

I was never asked but thought that if I could demodulate the data, I could measure torsional resonances during startup or coastdown.

I have to believe someone has done this. Yes? No?

Jim Powers
quote:
Arie,
You mentioned putting the tape on a coupling. Isn't it better to measure at a free end since the coupling may be close to a node? Strain gauge measurements are best made near a node, conversely.


Rick, what you state is generally the case if you are investigating the 1-node torsional response. But like a study of lateral vibration the system should be modeled in advance to get some feel of potential response and mode shape. It is entirely possible that a 2-node response might exhibit highest amplitude at a coupling and there would be the loaction for a demodulation measurement device but not necessarily ideal for strain gages.
Eventhough for a coupling mode the node happens to be at the coupling element, the modal amplitude of the inertias (including that of the coupling hub)are often higher depening upon the inertias on either side of the drive train. Probably, zebra tapes on the coupling hub should be fine for capturing the torsional levels at this mode, which is often dominant and of interest.

Kindly suggest me some suppliers who can supply good reflective zebra tapes (precise spacings)at pitches as desired by the application.

Thanks
Jeyaselvan
Vendor,

Oh I was waiting for that pitch. I had to cut and print stripes from 1x1m sheet for a specific use once as I could not find a supplier so I think I have for the rest of my interest in torsion... It has been our major export success to Florida about a stripe per year for a couple of years.
By that it is 1m long, 20 mm wide 2mm black stripes with 2mm reflective, base is 3M reflective material for road signs. I have used it with Monarch tachos. Olov
quote:
Eventhough for a coupling mode the node happens to be at the coupling element, the modal amplitude of the inertias (including that of the coupling hub)are often higher depening upon the inertias on either side of the drive train. Probably, zebra tapes on the coupling hub should be fine for capturing the torsional levels at this mode, which is often dominant and of interest.


I would not count on that approach if the desire is to investigate the 1-node torsional. The slope of the mode shape in the area of the node is such that inferring amplitudes in the rest of the system is highly inaccurate based on an amplitude acquired at the node. I also very highly recommend that a model of the system be created in advance of field measurements. All too often people get the cart before the horse and do a field study, then come away with questionable data, and end up repeating the field test. If for instance you are investigating the 2-node torsional, then a spring-mass model will indicate the optimium places to acquire both FM incremental data and strain gage data (if needed). It is also desireable to measure at multiple points because this can add credibility to the normalized mode shape curve and thus the spring mass model. A third opportunity that can often be accomplished when a spring-mass model is initially established is to see what if any change might be accompished by a simple system change that can yield a measurable change in the response. I once had a situation where a gear manufacturer was being sued for what was believed to be mesh frequency at the 3-node torsional. The addition of a simple split disk was predicted to shift the 3-node frequency by about 10 Hz. The theoretical sprin-mass model had predicted the "original" 3-node critical and when the disk was added the system shifted the anticipated 10 Hz. In this instance, the gearbox manufacturer did not believe the 3-node torsional to be a problem, in large part due to the fact that a response around 200 Hz would require, from a practical standpoint, huge amounts of energy to develop destructive torsional amplitudes. In the field study, multiple measurement points matched the calculated mode shape curves for the first three modes; in addition the system responded as anticipated by the model when the disk was added. The result was the end user dropped a lawsuit against the gearbox manufacturer in the face of theoretical data that very closely matched the measured data for three modes, with and without the added disk. The credibility of the spring mass model was thus without question and clearly indicated that it was not the gearbox.

As a side note, I have had success with "homemade" zebra tape made by running normal reflective optical tape through a high quality printer capable of banner or continuous printing. I had layed out tape intervals based on the circumference at the measurement location taking into account the ends butting together properly. Initially I was quite surprised how well it worked and later became aware of someone using this technique to investigate torsional issues in large steam turbine generators.
quote:
quote:Arie,You mentioned putting the tape on a coupling. Isn't it better to measure at a free end since the coupling may be close to a node? Strain gauge measurements are best made near a node, conversely.


John, yes you are right. The coupling is a tricky location for an optical 'zebra' measurement.

Often I have analysed the diesel engine flywheel high frequency magnetic speed pick-up ac signal with good result. Because quiet often the first nodal point is somewhere in the middle of the torsional flexible coupling adjacent to flywheel, not at the flywheel.
Also, an incremental encoder pulse train derived from an encoder at NDE will seldom be a nodal point.
The zebra method is a second choice for me and should be used when magnetic pick-ups or encoders are not available.

I used custom printed "zebra tape" with infrared sensors and FM at multiple power plants (Nuclear and coal-fired).

It is really difficult to install the tape and usually it comes off at full speed and when the shaft heats up. To address these problems we developed a new contactless probe: Analog torsional Vibration Sensor: http://www.horvathresearch.com/atvs/
No more encoder tape and expensive FM or TIMS hardware needed!

In a Torsional vibration analysis system, the angular vibration of the machine shaft along with its axis of rotation is measured. The power transmission of the machines that use couplings or rotating shafts is also considered. If Torsional vibration is not controlled or correctly measured, it can cause unplanned machine breakdowns. For instance, the gear rotor of oil pumps can cause a failure if the torsional vibration is not controlled and monitored.

A Torsional vibration analysis system can effectively measure and analyze the angular vibrations at the main shaft of the machine. The earliest manifestation of any emerging fault can be detected in this region of the machine

How does this torsional system work? Does it require multiple pulses per revolution? Can measurements be made at engine flywheel where starter motor is typically located? If two speed sensors are required, then are they located adjacent to each other or at some other index angle? Does one of the two speed sensors detect TDC or simply detect the same pulse(s) as the other speed sensor? How is the system setup or tuned to a specific engine? Can system be used on a reciprocating compressor? Can system be used on electric motor driven drive train?

Walt

Add Reply

Post
×
×
×
×
Link copied to your clipboard.
×