HI fellow analyst.
Lately I answer “refurbish gearbox vibration: http://maintenanceforums.com/eve/forums/a/tpc/f/3751089011/m/5111017402“ and I was astonished to learn that for some people ghost frequency in a gearbox is ok. This got me thinking… I would like to expose my view on the subject and learn yours.

First, my opinion is forge in the mining industries where many low speed, heavy loaded gearbox runs with very little maintenance for decade without serious breakdown. I am use to see gearbox runs for 30+ years. I thing that critical gearbox must not produce ghost frequency at any extend unless management is willing to take the risk! In the past 5 years, we changed many gearbox and replace them with drop-in from various shop and all of them except one show ghost frequency. I don’t exactly know how or why ghost frequency (I believe it has to do with machining) but a 40 years old gearbox talk for itself. Original gears and design are free from ghost frequency and the expensive drop-in as well (up to 40000$ difference) so it has to be important. I think that the small machining flaw slightly overloads the adjacent gear tooth and it might even be worst in limit lubrication. I thing that it probably reduce life expectancy of the gearing of 10, 15 or even 20% but I will have to work here for another 20 year to find out! I suppose that it comes down to money; as long as you get what you paid for!

Waiting for your views, best regard, Marcel

I was one individual that stated (in refurbished gearbox vibration) under some circumstances "ghost frequencies" are acceptable. My answer is based on 18 years experience with a major US gear manufacturer, virtually all that time directly involved in vibration and acoustics as related to industrial gearing.

If we look at allowable casing readings over a progression of time we can find many specifications (late 1960's) allowing as much as 20 g's peak with no regard to frequency content on high speed gearing. Those specs also generally didn't make reference to a casing that was made out of fabricated steel or cast, although it is generally accepted that a cast housing offered relative high damping of high frequency vibration and thus for a given pair of gears would present a lower casing level. In this same timeframe Dudley's Gear Handbook cited levels at 10 g's pk as a maximum. I have examined many gears from that time that approach 10 g's peak, many (like the reinjection compressor drives on the Alaskan pipeline) still run today, after 30+ years of service.

Tighter and tighter specifications came from various establishments such as API, AGMA, ISO, etc. so that today the specs seem to be around 5 g's pk or slightly less. This downward trend in allowable levels is an interesting concept. To reach the desired levels many gear manufacturers had to pay great attention to the mounting of accelerometers. In other words to meet the specifications gearing wasn't improved, the accelerometer mount was improved. The refurbished gearbox had a suspected ghost frequency at 2 Khz - 5 g's at that frequency is 20 millionths of an inch. Mounting is very critical as is proper transducer selection. I had pointed out to the originator of that thread his units were in/sec and he shouldn't be using a velocity coil in that range of frequencies.

The energy in the casing is a function of a numerous accuracy factors which can be lumped as transmission error (T.E.). T.E. is defined as the difference between the position that the output shaft of a gear drive would be if perfect, without any errors or deflections and the actual position of the output shaft. This is a measurable quantity.

Looking at API 613 the allowable (broadband) casing level over the range of 2.5 kHz to 10 kHz is 4 g's peak. It has been my experience that the source (and I emphasize source) of the energy, i. e., "true" mesh or "ghost frequency" is not as critical as the amplitude. I may very well have a machine that has a 2 g component at an identifiable ghost frequency and a 1 g component at mesh yielding a broadband slightly over 3 g's peak. In my opinion there isn't anything wrong with that machine. It may even be able to establish that the gear manufacturer had a "bad" index wheel that was the contributor. But fundamentally I've never seen a ghost frequency to be an issue if kept within a reasonable spec level. I have rejected a few in my time but all basically exceeded the applicable specifications by a considerable amount.

Please define Ghost Frequency. Most references I have heard to ghost frequencies were created electronically (aliasing) which could be caused by high GMF frequencies outside the sample. There are also discussions about frequencies being cause by patterns left during the machining process. Not sure which would be relivant in your case? I’m having trouble with the link.

Any gear in the manufacturing process is "generated" on a machine that is comprised of a very complex gear train. If any gear in that train (the manufacturing machine) has an error then that error is imposed on the final product. If for instance the index wheel has 70 teeth but the final product being generated has 100 teeth, then two components may appear in the running spectrum. One, the normal mesh frequency will be 100 x RPM; the other, the "ghost" frequency will occur at 70 x RPM. These components can appear in either the structural borne signal on the casing or the airborne signal.

The correct link that Marcel B is referring to is at http://maintenanceforums.com/eve/forums/a/tpc/f/3751089011/m/5111017402

quote:

Originally posted by Waylon M:
Most references I have heard to ghost frequencies were created electronically (aliasing) which could be caused by high GMF frequencies outside the sample.

At the selected Fmax, a sampling rate of 2.56*Fmax according to Nyquist provides that no alias frequency will be produced in this range.

David,

Sampling at 2.56X Fmax is no guarantee that aliasing won't be a problem unless an analog low pass (anti-alias) filter is used ahead of the A/D converter.

Walt

Walt,

You are right. I actually meant that by selecting Fmax, a matching LP filter is set up in the instrument as well.

Thanks all for the great input. I looked at the couplings on this unit and the motor coupling (KopFlex) has 36 teeth and the pump 34 teeth. Attached is some pictures from last year of the unit that went south in a hurray. This is why I don't want any surprizes!!

CRDPICS.doc (578 Kb, 15 downloads)

Hi fellow,
First, I would like to assure John that I do not challenge his technical integrity or expertise in any way. I just want to create a debate within gearbox user as what is acceptable or not. It is common knowledge that ghost frequencies are manufacturing flaws. It derives from the milling’s gearmesh or dull cutter or … James I Taylor in gear analysis handbook says: “New gear should not be accepted when these fractional frequencies are present.” Page 44 From my understanding of gear, the teeth’s pitch line vibration signal should closely match a perfect sine just like electricity, as it is the perfect energy transmission pattern. I think like M. Taylor as long as management is willing to pay the price difference. All the major gear manufacturer; Falk, Sumitomo, Nord gear, Euro drive, etc provide us with ghost free gearbox! Why they make sure that their drive train do not contain any ghost frequency and why should we settle for anything less? Is ghost frequency catastrophic? No, Is ghost frequency desirable? Certainly not!

Best regard, Marcel

I have nothing to add to this discussion, but would like to ask if anyone knows where I can get a copy of Dudley's Gear Handbook. It is out of print as far as I can find and I have wanted a copy for years.

Thanks,

Danny, I'm not ready to part with my copies (1st & 2nd editions) of Dudley's Gear Handbook but you might check http://www.fetchbook.info/search_Darle_W._Dudley/searchBy_Author.html. They have Dudley's "Handbook of Practical Gear Design" which is almost a good and much later (1984).

John,

Thanks for the tip. I'll try them out.

Regards,

John,

Thanks. That link led me to a 1994 edition which has everything I need and more.