I was just wondering how many of you guys use alarms to choose which machines to closely examine and how many of you look at all the data that you take?

Neither one really. I only look at points with notes attached or points that have had problems reported previously. I have a ton of points that I haven't looked at on a computer in years.

I am VERY slow at data collection (I do look at all data as it's collected) but quick at reporting.

I'm like Winston.... when I see a point in the field that has an overall that is significantly different from the previous survey, I enter a "notecode" that alerts me to look at it back at my desk. If I see (spectrum or waveform) or hear, or feel something that seems odd, I also enter a note code. Usually I will also collect "extra data" to aid in analysis.

I pretty much ignore alarms. I use the parameter sets mostly to trend things as opposed to alarming. 80% of all points are going to be OK so I don't waste time looking at things that are not a problem.

I use Odyssey software and have magnitude (trend) alarms and also spectral alarms. The spectral alarms are based upon historical data for an individual machine. I only look at machines that break an alarm. I'm very confident that I won't miss anything with the set-up that I have. I'm not prepared to look at all data that I collect.

Big Al,
This is an interesting thread in one respect. The same issue came up in a major difference in the way Odyssey users work vs. CSI/RBM users. I was a CSI user and looked at ALL of the data, not only in the field but at my computer. As an Odyssey user I have become very confident of the alarms and reports and use them exclusively. Of course, that doesn't mean I don't still use my senses while taking data. I use NOTES to mark suspect machines. Also, the dataPAC alerts to the percentage increase over the previous reading as it's taking the data.
Just for grins, how many of the responders to this thread look at all of the data AND are CSI users?

I'm using CSI and look at all the data everytime and feel I have to - it's very busy and drives me nuts when you have to look at the date on every spectrum as it may jump to a different survey from point to point.

We live and die by the alarm settings (which must be adjusted as you get to know a machine better). Our usual sin is calling something too soon, when it still has life in it, this is largely because we are still getting used to Peakvue.

Having good alarm sets can save a lot of time, so you can observe as you go on the stuff that does go into alarm.

Alan

I look at every point's spectrum, waveform and trend using CSI's "Autoplot". Attach labels to points of concern from survey to survey and have never been able to force myself to trust alarms, no matter how they have been "adjusted" over time, trust only trends from AP Sets. Have some "special" AP Parameters that trigger certain areas, as well as a "live" display on the analyzer of the spectrum and waveform, that aid in whether or not to acquire "extra" data on a point while in the field and or add a "note" of a potential problem.

Feeling, seeing, smelling, hearing and etc are all good tools for pre-analysis also.

Autoplot allows for rapid analysis of every point, in my opinion, but I must say from past training classes I have taught, this procedure does take some getting used to. Old habits are hard to break.

Quote from Alan:

quote:

Our usual sin is calling something too soon, when it still has life in it, this is largely because we are still getting used to Peakvue.

I have found this to a very common problem in many of my training classes,......especially using Peakvue, call too soon.

Quote from Sam:

quote:

it's very busy and drives me nuts when you have to look at the date on every spectrum as it may jump to a different survey from point to point.

Sam, have you tried "Autoplot" in CSI? This does not do what you describe here. Seems like going from point to point via the "tab" key press or the "Shift Tab" key press or using the mouse click from point to point is "bad" about doing what you described.

I guess in my opinion, "time or speed" is not what a customer is wanting, but dependable calls from the analyst, whether "in house" or "contracted".

Only my opinion.

In the late 1980's my own software had features like alarm set points; when a threshold was violated you'd get a printout with frequency/amplitude reading and a message to say what it is and the problem - add voice and it'd talk to you. Producing quality PdM route data from a quality system and I think you'll find acceleration does it right the first time. The processing, re-processing and processing some more doesn't enhance IMHO.

Ralph,

I use the plot function and open the box which has all the buttions to push: I don't use a tab or shift tab. I have the v. 4.93 I think so contract time on software is up - I guess CSI will invent a new version to entice people to renew their contract. I hear bad press on the new version already --- any thoughts? I'm looking for justification penny wise to move in a new direction. I'd automatic diagnostis that works.

quote:

I use the plot function and open the box which has all the buttions to push:

I only have 4.90. This "box" sounds interesting! Where is it located, how do I open it? What buttons "all the buttons to push" are in the box?

Your analysis software you mentioned, from the "80's", what happened to it? Sounds like it was a great tool.

quote:

Originally posted by Sam Pickens:
- it's very busy and drives me nuts when you have to look at the date on every spectrum as it may jump to a different survey from point to point.

The above is very true and I wonder whether or not CSI is aware of it.

As you move through several measurements using the tab button it advances to the most recent data at first. When it hits a reading that was missed or is no longer taken, it moves to the most recent reading but then stays on that date. Most annoying and CSI is very, very aware of how annoying it it I assure you

Ralph,

Like most stuff that was on an ISO standard and GPIB/IEEE standard - it made sense so it became obselete. If you remember the old Entek package; mine was close to that or the old B&K package. Mine was open and would print the spectrum with a listing of all frequencies with a display of all amplitudes in g's, IPS and Mils with the trend. It also had cesptrum at the push of a button if you wanted it.

I really had trouble believing people were shifting to blackboxes in 1983. I had a TEC unit which was a really good blackbox 4:1 over CSI IMHO, but it bit the dirt too.

The white elephants are now pink. But the older I get the better I like the pink one's. Getting older now, but like my grandmother - still don't need glasses, drink right out of the bottle.

David and Danny,
You're right about the date skipping. Its a pain to have to watch. My eyes automatically go to the date when I'm looking over data, but every once in a while, you miss one, then go the rest of the route on an earlier date. THEN, you have to try and figure out how far back you have to go to get back to current readings.

I am wondering how you guys are looking at data in order to have so much of a problem with the wrong date jumping in on you.
I must be doing something wrong.

Help me out here! What am I doing wrong?

Ralph,
He is probably referring to the "Plot Button Box". You can turn it on in PLOTDATA by selecting the View menu and then selecting Plot Button Box. So, how do you access Autoplot?

Dennis-

I'm using CSI and watch the data while collecting it and if I feel, smell, or hear something out of the normal I make note to check it back at the office. Otherwise I use the alarms to look at the data back at the office.

Ralph,
If I take data on the back side of the papermachine where the accelerometers are wired out to a box, and out of, say 100 points on the machine, 3 or 4 accelerometers are bad or missing or the wires have been cut or pulled out by a papermaker for some reason, then when I go through the data it will jump dates when i come to that point. Usually I scroll through each point and take a quick look at it. When i take a look at the bad one, it will default to the last good reading on that point. When I move to the next point, it won't jump back to the current reading, but will stay on the date of the last bad reading. It will stay on that date the rest of the reading unless I catch it and forward the date up to the present reading.

IchiBan,
I access Autoplot with the F10 key in version 490.
OK! Plot Button Box!!! That shows me I have been doing this CSI stuff too long!!!!!!!
I am so comfortable with the way I have been doing it for years that these "new" features don't even get checked out by me!!!!! Maybe I had better check out and see what other things are "new" huh? I started CSI with DOS and sort of graduated upward as they (CSI) changed their game plan and also have never really created a craving for this thing in my right hand called a mouse, I am a keyboard man!!!

JimB,
How do you go from point to point?