Hello all,
I have three long routes that cover "droplifters," machines for raising or lowering units to another level. These are driven by a motor mated directly to a cyclo drive(no gears, just two cyclo discs and many other small moving parts.) This turns a 16-foot long shaft that has two large sprockets with chains near either end. the shaft is held in two large SKF flange-mounted bearings. The motor turns in one direction to raise or lower and then reverses to return to the starting position.
I have not given many specifics because there are many droplifters of many different sizes-they are only different in function. All are only loaded in one direction, so I can only take readings at the same portion of each cycle. Most of the droplifters travel through their cycles at a constant rate-the motors turn about 1760 rpm(but only for about five seconds) and the shaft turns about 20 rpm. Calculating what going on inside that cyclo is like smacking yourself in the forehead with a ball pein hammer-it only feels good when you stop

All of that to get to this: I have a lot of equipment and not-so-much time. The current setup for measurement I am using is five points on the motor(2hor, 2vert, 1ax), and the cyclo the same. I take a horiz and a vert on each of the large bearings and an axial just on the end bearing. The fmax for all the points is 120,000 CPM at 1600 LOR. The duration between cycles is dependent on sales volume...and gasoline prices , so it has been sloooowww as of late. Looking at my setup, what do I need to change? What do I need to leave? Or do I need to give more info? Should I just take peakvue on the slow speed bearings? On the droplifters that have two bearings at each sprocket that the chains ride on, do I need to monitor every bearing? (Some have as many as 14). Any help speeding up my time will be greatly appreciated-and I don't mind criticism. Anybody who can make reference to a mechanical engineering class they took twelve years before I was born may just know a thing or two more than me.

By the way, what's with the post time? It is 12:19 pm CST here. I thought that this may have been in another part of the world, but it would have to be on an island west of Hawaii.

Richard,
For some reason the correct time will be displayed after you Login!

quote:

the motors turn about 1760 rpm(but only for about five seconds) and the shaft turns about 20 rpm.

Hmmmmmmm!

Does that mean the motor only makes 146 revs each cycle and the shaft 1.6667 revs? How much time is there between each cycle?

quote:

By the way, what's with the post time? It is 12:19 pm CST here. I thought that this may have been in another part of the world, but it would have to be on an island west of Hawaii.

Richard,

Have to chosen the time to be displayed from your "my space" and "preferences". You can choose the time you want to be displayed here. Maybe this is what you need. Then maybe not.

Richard,

There is surprisingly little going on in a Cyclo reducer. There is the high speed bearing which is a standard ball bearing unit, the eccentric bearing which is a cylindrical roller bearing with no outer race (it uses the inner diameter of the cycloidal disc or discs) and the low speed bearings. The pins and rollers in the ring gear are not a likely source of vibration. I have attached a PeakVue spectrum of one with a broken pin on the low speed shaft assembly.

Unless these are allowed to operate with water contaminated lubricant, the eccentric bearing is generally the first thing to fail. If water is allowed into the reducer, you can have pins, rollers and discs with lots of corrosion and the unit will continue to run for possibly years. I have had these things come into the shop with gallons of water for lubricant and once in an emergency, put one back into service with an un-nerving amount of corrosion and had it last at least another 5 years.

Sumitomo has always been fairly forthcoming about giving out the commercial numbers for the bearings, but the eccentric bearing may be a problem. I think SKF makes it but it is proprietary for Sumitomo. There is an SKF number on the inner race, but I'm not sure if SKF publishes fault frequency data for them or not.

I usually take a PeakVue and a velocity reading on the high speed end and another on the low speed with fmax set as you would for anything with rolling element bearings since there is not a gearmesh frequency to worry about. The spectrum I am attaching shows harmonics of output shaft speed and 2 times the number of humps on the cycloidal disc which is the nominal ratio plus 1 ((25+1)x2).

It sounds like you have an 87:1 ratio which should be a single reduction unit, so you won't need to add anything for a second reduction. If you really only get 1 2/3 revs on the low speed shaft, I would leave out the velocity reading on the low speed end and just use PeakVue with one average and fmax and lor to to match the duration of operation.

Good Luck,

Danny Harvey

cyclo.doc (60 Kb, 28 downloads)

Thanks for the time heads up, guys. I went into "My Space" and fixed it. Now Ralph, to answer your question, yes,on some droplifters the motor only turns about 146 times per direction and the shaft 1.6667. This is the short-travel droplifter and it just raises or lowers a "skillet" a little bit so that it can pass under the floor of the conveyor. Some run for as long as 30 seconds per direction, as those that have to lift or lower from a third floor level.(Yes, the ratio is much different for these.) But the quick ones only allow me to take 2 averages per reading. On one production line in particular, the rate of production is at present 10 units/hr.! If I don't miss one, it will take an hour and a half to take readings on this one. Other production lines move as many as 160/hr..or even more! This is impressive when you consider what we make here.

Thanks for the cyclo information, Danny. Very informative! As far as I have been able to research, the eccentric bearings are indeed proprietary to Sumitomo and because of this, they are not forthcoming with the fault frequencies.

By the way, Ralph, how do you do the quote thing? I was experimenting with this and haven't quite gotten it right yet.

Richard,

Were youo able to get the SKF number for the eccentric bearing from Sumitomo? If so you may be able to get the at least the dimensions from SKF.

I don't think Sumitomo is being secretive. After all, you can't buy the eccentric bearings anywhere else so you are already a captive market. As far as I know the only person they have with vibration training worked solely on the parallel shaft reducers and knew nothing about the cyclos. Also that guy was transferred to another position a few years back and as far as I know his position was eliminated.

Good Luck

quote:

By the way, Ralph, how do you do the quote thing? I was experimenting with this and haven't quite gotten it right yet.

To reply to a post starting with a quote,

highlight what you want to quote, right click and copy

Then click on the reply tab and the quote will be placed there automatically.

to place a quote in an already started reply or new thread,

highlight what you want to quote, right click and copy

choose where you want to place the quote and then type a [ then type the word quote and then type a ] then right click and paste

then type a [ and type a / and type the word quote and type a ]

Richard,

It looks like you are trying to use a single channel (1-accelerometer)vibration analyzer (collector) for a task that it is poorly suited. If you must continue with this tool, then reduce number of measurement points (2 to 3 on motor and 2 on gearbox) and restrict data to overall velocity, HFD acceleration level, and velocity spectrum (1 or 2 averages). For small motors and gears, use only 2-measurement points (1-motor, 1-gear). Demod/PeakVue waveform/spectrum is nice if measurement period is long enough.

Another survey option might be to "buy" time on the machine for a proper survey or conduct survey during schedule process/production down-time.

A better tool might be several accelerometers (4 or more with magnet bases) connected to a IOtech ZonicBook (or MAARS portable system) multi-channel data acquisition system, so that a full process cycle can be measured with simultaneous measurements on each component. Measurements during the "so-called" no-load operation can be also very usefull for detecting many types of faults.

Another "better tool" is ultrasound, since the measurements can be made quickly and with fewer measurement points. I simply suggest thinking outside the box (vibration data collector) for this task.

quote:

As far as I know the only person they have with vibration training worked solely on the parallel shaft reducers and knew nothing about the cyclos. Also that guy was transferred to another position a few years back and as far as I know his position was eliminated.

Hey Danny, we have gotten a lot of good information concerning cyclos and other Sumitomo equipment from Virgil Cochran,an engineer in Chesapeake. I don't know how much expertise he has in vibration, but he seems to know what we are talking about and has been very helpful. (Thanks Ralph for the help with the quote thing.)
Walt, indeed you are right-but I am stuck with using the same data collector. But you got me to thinking...too often I often focus on my frustrations and limitations as a contractor and do not often think "outside the box." I will experiment with some of your suggestions and see if I can come up with some more ideas of my own. Thanks all of you veterans for the advice.

Looking outside the square for these types of drives is the key.
Has any RCM been carried out on these items. You might find that VA is not the best technique for the reults you need.
Ultrasonic's can provide some very good and easy measurements that can be trended. Is Oil analysis a better tool for the drive, would it be better to trend temperatures as well? What amount of lead time to failure are you trying to obtain? What type of failures have occurred in the past? And how often do the failures happen?

Answer these questions and you will know how to be succesfull in monitoring this plant.

This is just my opinion

Good luck with it all and keep asking the questions, the more answers you get the better you become.

Hooch

Hooch,
May be just your opinion, but it's a good one...
We already do oil analysis on each of these drives and our conveyors/lube crew has adhered the little plastic temp gauges to the motors and gearboxes(not as accurate as a temp gun but still a pretty good idea, I think.)We check this also each route. This plant has only been running for two years, and none of this type of equipment has failed as of yet. Here we have no redundant equipment and few spare drives or motors-I'm sure it is this way in most new plants now. We presently collect data monthly (or try to) and look for a change. My biggest worry was that we would not see what we needed to see because of improper methods or setup and the first failure would blindside us. I may just be a little fuzzy this morning-what it RCM?

Any lead time to failure would be better than stumbling around in the dark. Oh, well. Back to data collecting.

RCM = Reliability- Centered Maintenance to answer your question. You usually carry out an analysis of the plant or particular machine and you identify key areas that are prone or could be prone to failure. Then you rank the probibilty and work out the mean time to failure and you design your program around the results of this study. This usually involves everyone in the work place being involved and giving the input. Like I said before the more questions you ask the better,do not limit your questions to the plant supervisor, go and ask the workers and cleaners what they know about the operation. You may be supprised what info you find out.
With your oil analysis are you just getting a particle count or are you doing filtergram analysis on the samples as well?
These drives that you explained have a lot of moving parts and to get arise in particles is one thing but if you can identify what has caused the particle ie, a cutting, slide, or wear type of action this may help in determining the overall life of the drive. People often do the oil analysis to get a particle count but if they carried out the next step it could help so much more. Does your lab that processes the samples have the expierence and equipment to give these results?
Hooch

Richard--

Have you considered "grouping" some of your points (especially PeakVue and standard vibe) using a Split Signal Adapter (CSI's Model 624A)? I did this at my mill and have shaved a chunk of time (I estimate between 10-15%) off my route data collection time--while still getting plenty of data to chew on back at the office...

Tony