User:Pranav Rathi/Notebook/OT/2011/10/11/Device for studying acoustic and mechanical noise in optical setups

From OpenWetWare
< User:Pranav Rathi‎ | Notebook‎ | OT‎ | 2011‎ | 10‎ | 11


All optical setups are subjected to the mechanical and airborne acoustic noise generated by various electronic and mechanical devices such as computers and air-vents in a lab. Optical setups such as optical tweezers are very sensitive to these noises. Optical tweezers like ours is so sensitive that it almost worke like a microphone. Currently it is reading all kinds of mechanical and airborne noises which are very troublesome for DNA unzipping and overstretching experiments. So I had to do a full scale investigation of these noises[[1]]. To do it I had to create this device: I named it Acoustic & Mechanical Noise Reader (AMNR). This device is nothing but a condenser microphone inside the ear-plug of a stethoscope. But it is so sensitive that it can read the mechanical and acoustic noise up to -95DB (believe it or not), and it just uses a regular computer sound card with a Lab View V9 program. This whole setup has three parts; AMNR, noise investigator and helper program and, tone generator and speaker.

The whole purpose of this device is following:

  • Identify the airborne noise frequencies around the tweezers and their source.
  • Identify the mechanical noise (vibrations) frequencies in the optical table and their source.
  • Help me chose and design the right setup (stages and sample holders) on the microscope.
  • Help me in designing and searching ways to isolate the optical trap form acoustic and mechanical noise.

Most of this I discussed in the “Noise issue with the optical tweezers’’ page of my note book[[2]]. In this page I will just discuss the construction and demonstration briefly.


The hardware is pretty simple:

  1. Classic ll S.E. stethoscope by 3M Littmann. This is one of the best in the market used by cardiologists, with nominal price.
  2. Condenser microphone. Omnidirectional mic with frequency response 50Hz to 1.5kHz.
  3. 3.5mm headphone lead.
  4. Investigator and helper Lab View V9. Link to download: [3].
  5. Tone generator Lab View V9. Link to download:[4].
  6. 5.25" speaker by Virtual Reality Sounds Labs. The speaker has frequency response from 100Hz to 20kHz, which is more than enough for me.

I do not know any specifications of the sound-card and external amplifier i am using to generate the tone of different frequencies.


Construction of AMNR is easy just solder the condenser mic on the one end and lead on the other end of a co-axial wire. Snug the mic into the ear-plug of the stethoscope and tape it as shown in the slide 2. To activate the mic just put the lead into the mic jack of the sound card.

Noise Investigator and Helper helps with reading and identification of the noise frequencies; slide 3&4. This program takes the signal from mic and plot it on a waveform graph than it does FFT to distinguish the different frequencies in the signal with their relative power in DB. This program can also write and read the data for future use. Helper part of the program can help in many ways; it can help with deciding which material is better for certain frequency range, it can help with deciding which setup is better and it can also help with choosing which part on optical table has less mechanical noise. It does it by feeding the power values of particular frequency from FFT power spectrum graph into an array and then taking a mean and feeding it into an another array (you can choose your desired name for that value and feed it). Than the whole process is repeated again for different frequencies and compared the values in the last. The most amazing feature i just added to this program is that i can listen to a specific band of frequencies through a band pass filter. So now i can listen and see a specific sound and filter-out others slide 17.

Tone generator is another program of this setup; slide 5. It can generate a tone of any frequency, but it does not matter because every electronics and output device (speaker) have its own frequency response range, so only the frequencies in that range will workout. This program is helpful with testing of frequency response of different materials and setups (geometry). Now for example I can generate a range of frequencies to test the mechanical frequency response of the optical table at certain distance or place on the table. Same can be done with microscope or other stages or breadboards or different isolation materials/setups by reading the generated signal through AMNR.


<html> <div style="width:700px" id="__ss_9816973"><strong style="display:block;margin:12px 0 4px"><a href="" title="Device for studying acoustic &amp; mechanical noise">Device for studying acoustic &amp; mechanical noise</a></strong><object id="__sse9816973" width="700" height="500"><param name="movie" value="" /><param name="allowFullScreen" value="true"/><param name="allowScriptAccess" value="always"/><embed name="__sse9816973" src="" type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" width="700" height="500"></embed></object><div style="padding:5px 0 12px">View more <a href="">presentations</a> from <a href="">pranavrathi</a>.</div></div></html>

I have tested this device in many ways to check if it can serve the purpose. One of the tasks is to scan the electromechanical devices for mechanical noise (vibrations; mechanical noise can propagate through wires and contact-surfaces). So in search of the source, these devices should be scanned first.

Search for mechanical noise source

I have two CPUs in the lab; CPUs have moving parts like fans and hard-drives which produce mechanical and acoustic noise. I have scanned one of the CPUs by touching the stethoscope sensor-head to the skin; slide6. Slide 6 shows the signal on the waveform graph on left and FFT on the right. In FFT all different frequencies can be seen with their relative DB. Same is done with AC vent-wall in the lab; slide 7. I tested that 150Hz was one of the most prominent frequency generated by the AC. The wall suggests the same. In slide 8 I scanned the optical table which suggests somewhat the same power spectrum but it is not correct since the airborne noise in the lab overwhelms the mechanical noise in the table (since the table is on active vibration isolators and not in contact with any electromechanical devices, there is no mechanical noise there). The test results suggest that this device can successfully search for a mechanical noise source.

Search for airborne noise source

In slide 9 I did a test for airborne noise by hanging the AMNR close to the vent in the hood without touching any surfaces. In slide on left the waveform shows the interference pattern, on right it shows the spectrum with 150Hz dominating. Through this process a source/sources of airborne noise can be identified.

Material Vs material by mechanical noise (vibration) transmissibility test

This is a transmissibility test. In the test I generate a signal at 6 different frequencies (100, 200, 300 400 500, 600Hz) with same power by a woofer speaker. I had three (aluminum, acrylic and acrylic plate with fewer surfaces) sample-holder plates to test. I tested the plates one by one by putting the plates on rubber feet (so no noise is transmitted to optical table because rubber has really high attenuation above 50Hz) on optical table, speaker on one end and sensor-head 2.5 inches away. I cutout a gape on diaphragm to detach the diaphragm form the frame so at low power the speaker does not generate the sound but it generates the vibration of that frequency. This is to avoid any airborne sound contamination in the mechanical noise signal, because the sensor-head is only 2.5 inches away from the speaker. To make sure that there is no contamination I kept the speaker at lowest possible power and scan it to check for airborne sound. Once I am sure that no airborne sound is generated I put the sensor-head directly on the back of speaker and register the DB. I do it for all six frequencies (this is the reference DB to calculate the transmissibility=output(at 2.5 inches)/input). Since the speaker is woofer it can only generate the frequencies efficiently from 100 to 700Hz, if I want to push more I can push it to 1000Hz, but I do not need that, because the most problematic range is 100 to 600Hz.

I begin the test with speaker off to register the airborne noise spectrum to determine the lowest DB received at all 6 frequencies, because during the test the results will not be better than this value, so it is good to know this value at all 6 frequencies. I used the tone generator to generate the signal and give it to the speaker, slide 10. Slide 11 shows the setup; the signal is generated at the speaker propagated through the plate (material) and received through the sensor-head for all three plates. The result is compared in slide 12. Blue curve which belongs to aluminum plate, suggests lowest attenuation. Acrylic and acrylic plate with fewer surfaces are better. I designed these plates especially the one with large circle and triangle (fewer surface) such that they have less transmissibility (high attenuation) for mechanical noise of 100 to 600Hz by choosing the plastic over metal and fewer surface so they are affected less by airborne noise and less material is available to propagate the vibrations. Geometry is very important factor in acoustic and mechanical noise absorption and propagation, so i tried to design it such that it strongly favors the attenuation at lower frequencies <600Hz and drift the resonance to very high frequencies > 1.5kHz.

The sample holder plates hold the sample in the trap, so any vibration in the piezo-stage can propagate to the sample and make the data noisy. So by choosing the right material and design this problem can be minimized for some frequencies. Plastic is good for propagating lower frequency noise (<100Hz), but for higher frequencies it is a good attenuator. The goal behind carving a circle and triangle in the stage is to have a fewer surfaces to expose to airborne noise and less material available to propagate the vibrations. The result suggests accordingly.

Piezo-stage holder material test

Above tested sample holder plate screwed over piezo-stage which is screwed over holder stage. This holder stage comes between xy-motorstage and piezo stage. So by choosing it correctly the mechanical noise coming from xy-motorstage can be over-damped /attenuated so it does not reach to the sample.

I have three stages made out of two materials to test. I have one aluminum stage and two polypropylene stages. I have two designs in polypropylene stages; one is regular with two large rectangle boxes carved out so it has fewer surfaces (less material) in contact (slide 14) and second stage with rubber feet (slide 15). This design I tried that the holding-screws do not touch the stage main-body and the main-body does not touch the xy-motorstage. Four rubber feet keep the stage off the ground and in-holding rubber-cups hold the screws away from the stage. So any mechanical noise is efficiently attenuated by the rubber before it reaches to the stage main-body, since the rubber used is a good attenuator of frequencies above 50Hz.

The test setup is shown in the slide 13; the speaker is kept upside down on the table with an aluminum plate on the top. The drawback of this setup is that the results are affected by the characteristics (material and geometry) of this plate so I start with measuring the signal directly off the plate at each frequency. This would be reference DB. Before starting the test I made sure that the speaker was at lowest possible power at each frequency, so there was no airborne sound contamination.

Now in slide 14 I stack up the aluminum holder stage and piezo stage with sensor-head on the top and record the DB at each frequency. I repeat the procedure of each polypropylene stage in slide 14 and 15.

The results are presented in the slide 16. I recorded each data point many time and the mean is plotted (I did not bother to calculate the error bars). The top blue curve is on the floor data. There is not much difference between the aluminum stage and polypropylene, but the stage with rubber cushion is definitely better. Here the frequency response of all three is controlled by their material and geometry, maybe that’s why we see a peak at 300Hz in all three. {{#widget:YouTube|id=p-MA-xA8XV8}}


It looks like that this device is definitely helpful in fulfilling the purposes I outlined and hence useful in dealing the noise problems due to acoustics and mechanical in optical setups.