Filtering

Ask questions relating specifically to the PicoDiagnostics NVH equipment and software here.

Filtering

Postby Mark Dalton » Fri Mar 24, 2017 8:21 am

Just wondering if anyone has had a go at using the filtering option in the new NVH software.
I've got a car with a Diff whine, the whine might actually be characteristic of the Diff, but as much as anything else I'm trying to get a handle on what is "normal" for this DIff so that I can then do a car to car comparrison.
When I do a band pass I get no sound at all,
when I do a low or a high pass I'm not sure it's making any difference.
I'm not even sure I'm even looking at the correct frquency rage. I was looking at frquencies below 300 Hz, but in Steve Smith's post post49441.html?hilit=filter#p49441 his range goes all the way up to 1800Hz and where I have either the Mic recording or the accelerometer, Steve has the Mic and 3 Accelerometers in his example. What hardware would I need to acieve this?
Is the recording I have just to noisey to even be able to use it meaningfully?
Attachments
GTS W'Accel.pddata
(744.01 KiB) Downloaded 33 times
GTS W'Mic.pddata
(11.9 MiB) Downloaded 44 times
User avatar
Mark Dalton
TwoWaves
TwoWaves
 
Posts: 81
Joined: Mon Dec 01, 2014 11:35 pm
Location: Melbourne

Re: Filtering

Postby PicoKev » Wed Mar 29, 2017 10:33 pm

Hi Mark,

Please do not think that there "Is no one out-there" or you are being ignored. 8)

Unfortunately Steve really is the main man when it comes to NVH and he is currently in orbit around Europe training. I did see him briefly before he jumped on an aeroplane to Italy, and I did mention this post and he said he will reply as soon as he can when he gets back.

You asked about hardware? if you can post the details of the NVH kit you currently have and which scope you use it with, that will help identify what you would need to add to your current tooling.

Regards,
Kev.
PicoKev
TwoWaves
TwoWaves
 
Posts: 91
Joined: Thu Apr 05, 2012 11:16 am

Re: Filtering

Postby Mark Dalton » Wed Mar 29, 2017 11:41 pm

Thanks Kev,
I'll have a look for my Scope Model and my NVH Kit just has the single axis interface.
User avatar
Mark Dalton
TwoWaves
TwoWaves
 
Posts: 81
Joined: Mon Dec 01, 2014 11:35 pm
Location: Melbourne

Re: Filtering

Postby Steve Smith » Mon Apr 03, 2017 3:15 pm

Hello Mark and thank you for the post.
Sorry for the mega late reply too

Can I ask Mark, the car in the pddata files, is this the vehicle with the diff whine and if so can it be heard in the cabin? (Customer complaint)
Listening to the audio in the GTS W’Mic pddata file I cannot hear a diff whine.

I have chosen a stable section of your capture at 88 km/h (1453 rpm) in which to view the frequency spectrum as this helps to remove transient engine and transmission noises and keeps “variables” to a minimum.

I will go through a couple of techniques that will help with audio analysis here:
The first tip is to ensure we have the latest software as this helps with new features and bug fixes
The next is to ensure you have the maximum frequency of interest set to cover the noise of concern.
Here Mark select Options > Advanced Options, select Graph and increase the Frequency of Interest to 3000 Hz

The maximum frequency of interest is limited to 350 Hz when using an accelerometer and so to enable you to use your accelerometer for frequencies above 350 Hz, inform the software you have a microphone connected to channel D (Even if you don’t)

It’s a cheat I know but it will enable the use of the accelerometer to identify frequencies of interest above 350 Hz without the intrusion of back ground noise (detected when using a mic)

I have used this technique in the following post
post49431.html?hilit=WHINE%20IN%20REVERSE%20GEAR#p49431

ACCELEROMETER CAPTURING AT 1200 HZ.jpg


Using a Mic and Accelerometer simultaneously will help you confirm the frequencies identified by the accelerometer as the offending frequency of interest by listening to the recorded data played back into headphones

In answer to you question, “What hardware would I need to achieve this?”
Here you could go for the following upgrade kit:
https://www.picoauto.com/products/nvh-a ... pgrade-kit
This would provide you with an additional NVH interface and therefore the option to use a Mic and Accelerometer simultaneously

If you then required 2 x accelerometers and a single Mic you could add https://www.picoauto.com/products/nvh-a ... pgrade-kit

Moving on:
Frequency analysis of audio recordings is not easy! (I am certainly no expert and really need to work on this practice)
You mentioned your recording being “to noisy to even be able to use it meaningfully” What you have captured Mark is exactly what was going on around the bottom of the differential housing at the time of test and this is the problem. (There is so much noise that is not relevant that is also captured)

The filtering works well if we have identified an offending frequency (E.g. 1000 Hz) but this also introduces another challenge. Often when filtering you remove the fundamental frequency with a “Band Stop Filter” (to see if the noise has been removed) but you can still her the noise during playback

This is because the fundamental frequency is accompanied with a 2nd and 3rd harmonic that give the original noise its characteristic sound (warmth/depth) and so it remains audible. To Band Stop these frequencies (remove the noise completely) you would have to remove a large section of the spectrum, 1000 to 3000 Hz whereby you also remove components that may have been of interest.

An example could be when trying to separate pulley noises that reside in close proximity to one another.
E.g.
Pulley 1 rotates at 1000 Hz
Pulley 2 rotates at 1100 Hz

If Pulley 1 is the offender and we use a Band Stop Filter 1000 to 3000 Hz we have also prevented the majority of Pulley 2 frequencies from being audible for diagnosis.

Contact Microphones are one possible solution but they also introduce noise that may convince you that you have a transmission issue when in fact you are simply listening to characteristic transmission operation.

Think about listening to a normally functioning generator via a stethoscope, you can hear way too much “activity” that is simply not audible in operation by the naked ear.

The technique therefore would be to identify the offending noise, frequency and amplitude in the cabin via a cabin Mic (with the customer). Then add accelerometers, contact mics and standard mics in close proximity to the suspect component or zone of the vehicle

Between all these devices the originally identified frequency in the cabin will also be present in the other Mics and accelerometer, but their amplitudes will change so confirming you are very close to the source. (It’s not easy)

Development is ongoing here at Pico with Mics and software including noise cancelling. (Said too much already)

Coming back to your captures Mark (Sorry for waffling on)

When chasing noises in the spectrum, set your spectrum to cover the expected frequency of components on your vehicle (Differential Whines at speed are normally 800 to 1500 Hz)

Review a steady area of interest in the signal history and locate the highest peaks in your spectrum (Set here from 0 to 3000 Hz).

SPECTRUM RANGE SELECTION.jpg


Choose the highest frequencies (131 and 778 Hz), the rest of the spectrum can now be temporarily ignored by dragging the frequency view to the right. (This will improve resolution)

FREQUENCIES NOT RELEVANT.jpg


Now compare these peaks with those measured by the accelerometer and you have 2 peaks that have both been detected by both sensors (accelerometer and Mic)

ACCELEROMETER TEST.jpg


MIC TEST.jpg


We now have 110 Hz and 136 Hz “suspect frequencies” for investigation that cannot be linked to a known fundamental frequency such as Tyre or Engine.
However 110 Hz is close to a 3 order Propshaft vibration.

With all the above mentioned Differential whines are a lot higher frequency and so the 778 Hz peak would need to be considered.

I hope this helps, take care…….Steve
Steve Smith
Pico Staff Member
Pico Staff Member
 
Posts: 480
Joined: Sun Aug 25, 2013 9:22 am

Re: Filtering

Postby Mark Dalton » Fri Apr 07, 2017 9:16 am

Hi Steve,
Thanks for the reply, it's just the kind of information I was looking for, it's all good learning material.
The recording does have the whine the customer is complaining of in it, it's not particularly loud and it may be that it's more noticeable in the cabin than outside.
I was thinking.....It kind of makes sense that you want to get as close as you can to a source of noise to identify it, but maybe in cases like this it's not the ideal solution. There's so much white noise outside the car that interferes with your noise of interest. But presumably the car manufacturer designs the car to keep those frequencies out the car, but not the ones that aren't supposed to be there, so maybe inside the car could be the best place to isolate a noise or it's frequency at least.
Back to the issue...I did say in my original post that we believed the whine to be characteristic of this diff. One of the things I like about the NVH kit is the ability to do a quantitative analysis rather than a subjective analysis. So in cases like this I can compare with one or more other cars, then go back to the customer or the manufacturer with evidence and clearly demonstrate the issue is normal or abnormal.
In this case we were able to demonstrate to this customer what he heard was totally normal.
Thanks again for all your help on this one, every little bit of knowledge helps for when the next one comes along.
User avatar
Mark Dalton
TwoWaves
TwoWaves
 
Posts: 81
Joined: Mon Dec 01, 2014 11:35 pm
Location: Melbourne


Return to PicoDiagnostics NVH

Who is online

Users browsing this forum: No registered users and 0 guests