2

I am working on ''Active noise cancellation for air vent'' project. I am using NI myRIO1900 it has FPGA module(Xylinx Z-7010) , Audio Technica ATR 3350 omni directional microphones and Tevion Loudspeakers. I have done the programming using LabVIEW . I am using sinusoidal noise source of (200-500 Hz) as test noise signal to simulate the noise.

I have implemented LMS(least mean squares) algorithm on LabVIEW, as a preliminary results I am getting around 8dB noise reduction but these results are pretty random in a sense only at certain volume levels( if I adjust very carefully), the noise cancellation is happening and the variation in step size is not giving any decent change ( again I have to make volume level changes to see some results).

Over all, I am getting some results but I don't have control over the results.

This is my set up enter image description here
[the pipe length is 1 meter] I know I have to consider secondary path modelling and feedback modelling. I have actually completed the coding of FBFxLMS(Feedback Filtered x LMS Algorithm) also but I don't have any reasonable results with this also.

I have read in 2-3 places that without proper acoustical design it is not possible to reduce the noise effectively. Quoting from this paper page 5

''If the acoustical design of the system is not optimized,the digital controller may not be able to attenuate the undesired noise adequately.''

I am using this paper as my major reference.
The only point I am aware of is using low frequency noises (<= 500Hz) to consider the sound wave propagation as plane acoustic wave. I am completely unaware of any other details on how to better my acoustical design.

I am wondering how to improve the acoustical design of my setup for effective noise cancellation?
What are the problems using just cardboard material to simulate the environment of air vent ?

Please help me if some one has experience with ANC or Acoustical design.

Edit:
I am just using this model as prototype to demonstrate effective and controlled Active Noise cancellation.I don't have any practical model to compare to(right now!).
I am looking for some useful references or previous works dealing with acoustical design of Active Noise Cancellation(ANC).

Edit2:
More details:
1.In the picture the Anti noise speaker has just been inserted without proper sealing.

charansai
  • 129
  • 3
  • 1
    I'm voting to close this as unclear until more detail is added. If someone had to post a question because there isn't enough room in a comment for all of the points, then the post is definitely needing more informaiton. – hazzey Jan 14 '16 at 03:33
  • @hazzey I am sorry but it's unfair. It is impossible to know exactly what to post even though I tried to give as much information as possible in the beginning itself.I provided details for few of Mr Chuck's questions in my original post only but I was not clear enough may be. Rest I could not expect his point of view ! For ex. last 5 questions are on acoustic design which is my actual problem and LMS is kind of obvious most of the times( I am not blaming though, yeah I would have been more clear) – charansai Jan 14 '16 at 04:53
  • 1
    @charansai Closing a question is completely reversible. It allows you time to add the information necessary for good answers to be given. – hazzey Jan 14 '16 at 12:22
  • @charansai - closing a question is perfectly fair within this site, and the closing mechanism was intentionally created for questions like this. As hazzey notes, questions can be re-opened once they have been [edit]ed to narrow the scope of the question and provide details requested by the community. The fact that an entire self-admitted non-answer is required to _begin to elicit_ necessary details tells me this question is much too broad. And with another question guessing at your setup, it tells me that your specific, answerable question is unclear to the community. –  Jan 14 '16 at 15:24
  • Suffice it to say that I have closed your question so you have time to [edit] and narrow the scope of your question to a specific problem. At the moment, it sounds like you need far more information than what the StackExchange Q&A format was made to deliver. –  Jan 14 '16 at 15:25
  • @GlenH7 Hi, I don't have problem on closing the question if it is too broad or unclear but please let Mr Chuck comment on whether question is still not clear enough. Are all of his questions are answered or not he can only tell. I will edit the question to be specific and on point but I can never know till the person who has problem with the clarity of the question will comment whether it is clear or not. Thank you :) – charansai Jan 14 '16 at 16:57
  • @charansai - The question has merely been closed and not deleted. Chuck has sufficient privileges to comment on this post, so there's nothing preventing him from doing so. Skimming his list of questions and your post, I don't believe that you've answered many of the points he's raised. While you have addressed some of them in your comments, you need to [edit] those comments into your question. Other readers won't bother with jumping all over a post trying to piece it together. –  Jan 14 '16 at 17:06
  • I can't understand this ! Is the question still too broad?? and am I not clear enough still !? – charansai Jan 15 '16 at 01:54

2 Answers2

2

I don't think there's enough room in a comment to add all of this, so that's why I'm posting this as an answer.

  1. What do you mean when you say, "I have implemented LMS on Labview"? What is LMS?
  2. What FPGA board are you using? Is this an off-the-shelf part or some home brew hardware?
  3. You're not happy with your test results - what are they? Can you post before and after spectral results?
  4. How are you simulating noise?
  5. How are you measuring audio output? All I see in your image is two speakers and a cardboard duct.
  6. What is the purpose of the duct? Are you trying to simulate a larger duct system? (Note that ducting has its own set of acoustic properties)
  7. How do the dimensions of the duct compare with the simulated duct system? (Assuming this is the reason for the duct - why you're doing this at all to begin with still isn't clear) (Dimensions meaning length, width, height, number of bends, etc.)
  8. How are the speakers inserted into the duct? Is it just a hole they are resting against? Are the holes sealed?
  9. Have you taken into account the fact that butting the speaker face against the duct forms a Helmholtz resonator with the hole and the speaker body? You may be better off using just the speaker cone instead of the entire speaker body.

In general, looking at your setup, I would bet that you don't have the microphone co-located with the noise canceling speaker, so you are not acting on the actual magnitude and phase of noise at the canceling speaker to actually achieve cancellation correctly.

Also, I would point out, as I did above, that your "duct" is very poorly sealed and that fact, combined with the material (cardboard) you've chosen to use, will mean that the chamber doesn't resonate at the frequency you want it to, even if it accurately represented your final system, which I sincerely doubt it does.

I keep mentioning how your test duct compares with the actual duct because longer duct lengths add acoustic impedance and so even if your "scale" test works well it may still not work correctly when installed in the system. Scale testing, and getting scale factors setup correctly, is a science in its own.

What resources have you been using so far in working on this project?

Chuck
  • 3,565
  • 12
  • 28
  • Thanks for the detailed observations. I will edit the question making the question more clear considering your points. To answer your questions here 1. LMS is least means square algorithm 2. As I said I am using ''NI myRIO 1900'' module and it has Xilinx Z-7010 4.I am using sinusoid wave with frequencies varying between 200-500 Hz as noise source (as i have already mentioned) 5.I am using two microphones one infront of the noise source speaker and error microphone is at the end of the box( as I have mentioned in the figure) – charansai Jan 13 '16 at 22:57
  • 6,7. Initially I chose this model just to demonstrate the active noise cancellation principle but now I will consider proper acoustic design as you are pointing out but still I don't have any practical duct to compare the dimensions. 8. They are resting against the holes and holes are not sealed. 9. NO , thanks for the info (y) Can you please explain how this co-locating of microphone with noise cancelling speaker works ? – charansai Jan 13 '16 at 23:04
  • I have till now worked on audio processing details and coding. Can please provide some useful references discussing the acoustical design of ANC. I clearly lag the knowledge,I want to learn but first I want to know the difficulty in managing the acoustic design for a master student with electrical and signal processing background. I will mention the resources (if you mean references) in the edit. Thankyou – charansai Jan 13 '16 at 23:09
2

The first big problem is with your speaker placement; the 'anti-noise' speaker should be as close to the 'noise' speaker as possible. Another big problem is with the speakers you are using; they are way to small for an open-volume system. Even with both of these issues fixed, I doubt you will be able to get significant noise cancellation.

The simple explanation of noise cancellation that you can read on the side of a box of noise cancellation headphones is very oversimplified. What noise cancellation headphones actually do is not to add 'anti-noise'; rather they try to maintain a fixed pressure inside of the closed system of the headphone in the presence of pressure disturbances from outside the system. Even in this small closed system, they aren't very effective once the wavelength of the sound gets close to the size of the closed volume (1 kHz = 34 cm is roughly the upper end for active noise cancellation in headphones).

If you are going to have any success with your system, you are going to need to target frequencies which are even lower than the frequencies you are currently trying; I'd say you should start at 70 Hz. At these frequencies, your speakers don't have the ability to move enough air to create effective sound for either disturbance or cancellation; you need something more like a subwoofer with a driver size of at least 6 inches.

Community
  • 1
Chris Mueller
  • 6,386
  • 1
  • 26
  • 54