Sine-on-Random - Clarification 1

[Krunklock]

Greetings,

I am having some difficulty with some colleagues overseas in convincing them, or myself, that we are running a sine-on-random specification correctly.

When the sine tone is sweeping through the profile...will the tone be visible on the random profile? In simple terms...should I be able to see a spike in the random profile where the sine tone is sweeping at?

I'm new to vibration testing, and unsure if the two signals should be completely seperate of each other...or does the sine tone ultimately affect the random profile at that specific frequency? Should my Grms level change on my random profile as the sine tone sweeps through and increases in amplitude?

Also, when using average control with 3 ref point during S-o-R testing, is it normal to just average the random signal for all 3 points and use 1 point for the sine tone, or is it common to average both the sine and random portions?

 

[IRstuff]

I think you should see the sine tones. Otherwise, it's ALL random, just not uniformly random. A sine on random usually means that there are specific tones that always present, like the vibration tones from an engine or helicopter rotors. You should not expect those to be randomly present (unless the helo is serious trouble ;-)), they should always be present.

TTFN

FAQ731-376

 

[btrueblood]

1. yes
2. Both input accelerations will be sensed by the control accelerometer, so the feedback to the controller will effectively be the sum of the two inputs. Yes, at higher frequencies, the sine tone may affect the Grms of the random profile, depending on how the control is specified. Ideally, the random would be controlled independently of the sine tone, i.e. the sine input would be ignored in computing the random profile (or a notch filter tuned to the sine tone would be applied to remove the sine tone), and similarly, the peak g's only, at the frequency of interest, would be used for control of the sine tone. In reality, most spec's. for SOR (well, ok, I only really know about MILSTD810, and a few proprietary aero corporate spec's)) only spec the random g^2/hz upper limit...so it can be a little tricky to set up. A lot of the peak recorded will be due to how good your equipment is at resolving the FFT. Most good vib controllers have a section in the manual about setup for this test, or at least somebody in their shop shoud know. See:

http://www.bksv.com/doc/bo0226.pdf

Good luck.

 

[floattuber]

It depends. I've run one SOR where the levels of the tones were so small in comparison to the random level, that the tones either didn't show up or were barely noticable. Also, with our spectral dynamics software, we can choose for the tones to be additive or not.

But yes, in general, the tones are visible.

Our software doesn't allow for one accel to control sine and another to control random. The control accels average all vibration regardless of whether it's random or sine.

 

[IRstuff]

Depends... We usually get specific amplitudes for the random and specific amplitudes for the sines, and the sines are usually quite noticeable above the din of the random.

TTFN

FAQ731-376

 

[Krunklock]

I am definitely appreciating the feedback...you guys are totally BA.

Just so people know what I'm working with:

Sine:

Freq [Hz] Amp of Accel [g]
100 3
200 10
1000 10

rate: 0.5 oct/min

Random:

Freq [Hz] PSD [g^2/Hz]
10 0.1
100 0.1
100 0.00001
1000 0.00001
1000 0.05
2000 0.05

RMS: 7.7 g^2
Crest factor clipping: 3
DOF: 190

Average Control using 3 inputs: 2 on fixture, and 1 on DUT


1] I don't really see how it's physically possible to try and shake something and also NOT shake something (the psd between 100 - 1000 Hz seems incredibly low).

2] I believe my controller is only capable of averaging the random signal on all three inputs, as well as Ch 1 on the sine profile. Ch1 is on the fixture. I don't feel that this is correct, and I have been pressuring the controller supplier for a software update (which is pending but it isnt going to be available soon enough).

3] How would I apply a notch filter to the random profile? How much does that affect the 'harshness' of the random profile and the sine profile? Should my RMS stay at 7.7 g^2 throughout the entirity of the test, or should it creep upwards to the 10-12 g^2 level and back down as the sine sweeps upwards and back downwards?

4] I am being challenged by our HQ in the mother country that our testing is completely incorrect. I want to be able to defend our actions and explain that we are indeed testing correctly, but it just may be open to interpretation.

Also...I really wish I knew about this place earlier...I've been scrambling all over these inter webs asking questions...you guys are full of awesome.

 

[btrueblood]

"1] I don't really see how it's physically possible to try and shake something and also NOT shake something (the psd between 100 - 1000 Hz seems incredibly low)."

Agree, the step changes at 100 Hz and 1000 Hz are unrealistic. The steepest rolloff I can recall ever testing was 24 dB/octave, and it is hardware dependent (both the shaker and the fixture/DUT).

"2] I believe my controller is only capable of averaging the random signal on all three inputs, as well as Ch 1 on the sine profile. Ch1 is on the fixture. I don't feel that this is correct, and I have been pressuring the controller supplier for a software update (which is pending but it isnt going to be available soon enough)."

If Ch1 is an accelerometer attached to the test fixture or DUT, then it sees the combination of the random and sine inputs, correct?

"3] How would I apply a notch filter to the random profile? How much does that affect the 'harshness' of the random profile and the sine profile? Should my RMS stay at 7.7 g^2 throughout the entirity of the test, or should it creep upwards to the 10-12 g^2 level and back down as the sine sweeps upwards and back downwards?"

Somehow the controller needs to know how much voltage to crank out to the shaker head at the sine frequency. One way is to do it "open loop" - i.e. add a certain amount of amplitude at the freq. of interest to the random profile. Other control strategies attempt to measure the output acceleration directly, by using a notch filter to remove the random profile. It is also possible that the same FFT used for the random control is being used (on a line-by-line basis) to compute the g levels at the freq. of interest...but that is a coarser level of control.

The reality is, no filter is perfect, and no FFT has infinite resolution...there is some slop-over from both the random to the sine and from the sine to the random.

If your HQ says it's all wrong, ask them for the details of the infinite-rolloff filter they have requested in their spectrum definition.

What controller are you using? What does the manual say regarding SOR?

 

[IRstuff]

"random" is often used in place of what is actually happening, which is that there are sympathetic vibrations or resonances at or near the sine inputs. What this means that often, the random vibe has no physically meaningful correlation to the real environment, it's just a way to create a test that's "good enough" for qualifying a system.

While I agree that the steps are unrealistic, the fact that there's a region that has low vibration is not necessarily surprising.

TTFN

FAQ731-376

 

[btrueblood]

"While I agree that the steps are unrealistic, the fact that there's a region that has low vibration is not necessarily surprising. "

Agree. Though the 4 orders-of-magnitude change does make me curious...then again, with active damping and suchlike systems becoming more common, it could be real. The problem is in defining +/-3 dB tolerances at 100 Hz, or at 1000 Hz. Normally, you'd expect to see ramps at some specified rate (xx dB/octave) between the level changes.

 

[Krunklock]

Alright, I apprieciate all of the previous help...it's worked out great.

I just have one question (more like one question, with more than one part to said question).

In the attached image is the configuration setup from an Unholtz-Dickie controller (VwinII I believe). They have Random Only (Filtered Tones) selected.

If anyone uses this controller, maybe you can help out the most, but to people that don't use VwinII, can you help me understand what they mean by that?

Are they just seperating the signal so they display only the random and the sine tones on a seperate chart? Or does that actually affect the sine tone levels and thus alters the test?