loading

Measure Muscle Sounds! Part 2: Silicone Embedding

Picture of Measure Muscle Sounds! Part 2: Silicone Embedding
This is the long-awaited follow up to the super awesome Measure Muscle Sounds! instructable.

To see what you can do with these sensors, check out these videos.

Silicone embedding is the trickiest part of the process of building your own muscle sounds sensor. This is why I chose to make this a completely separate instructable. This instructable will definitely test your patience and manual ability, so be prepared to screw up a couple boards before you get the hang of it.

So without further delay... let's get started!

 
Remove these adsRemove these ads by Signing Up

Step 1: Complete part one of the instructable

Picture of Complete part one of the instructable
This instructable is a follow-up for the Measure Muscle Sounds! instructable published a while ago. Make sure you complete that instructable first unless you just want to try embedding something else (like your younger brother) in silicone and you just want to know how to do it.

Step 2: Bill of materials

Picture of Bill of materials
spacer2.png
33077-1009-1-3ww-l.jpg
This instructable requires a few custom materials and tools. The lab where I made the sensors already had most of the materials/tools available or they were custom-made there, so I apologize for not being able to direct you to specific suppliers/part #. If you try the instructable, please post where you got your materials from!

Part: Nylamid spacer (Diam: 13mm, Thick: 2mm)
Supplier: Custom-machined
Qty: 1 per sensor (reusable)

Part: 20A Shore RTV Silicone
Qty: ~2g per sensor

Part: 65A Shore RTV Silicone
Qty: <1g per sensor

Part: Acrylic (transparent plastic) board with circular indentations
Specs: 25mm minimum width and length
Centered indentation depth: 0.4 - 0.6mm
Centered indentation diameter: 4 - 6mm
Additional indentations at a minimum 30mm distance in any direction
Supplier: Custom-machined
Qty: 1

Other: Assorted modeling hand tools, precision scale, metal roller, silicone solvent.

NOTE: You can get the silicone and solvent from Nusil Technology. The modeling tools, you can probably find in a good art supplies store.

Step 3: Glue the spacer on the microphone

Picture of Glue the spacer on the microphone
spacer2.png
spacer1.png
As seen in the picture below, the microphone must be coupled to a sealed air chamber directly underneath it. In order to achieve this, we need to embed the whole thing in silicone. This is done in a two step process: first, we build a case for the board making sure we leave the space for the air chamber, and second, we cover the whole thing from underneath with a silicone membrane sealing the chamber.

The air chamber is quite small (equivalent to a cylinder of 13mm in diameter and 2mm in height). This is where the nylamid spacer comes in. You need to glue the nylamid spacer directly to the microphone (figures 2 and 3). You also have to make sure the spacer is centered with the microphone. Note that the microphone is not centered on the board so the plastic spacer will be slightly offset and the board cannot be used as a reliable reference. Be careful not to let glue into the microphone because this will definitely make it useless and you will have to start all over!!. You can carefully glue a small piece of paper to the microphone first to protect it.

In terms of the glue, this is still an unresolved problem (suggestions welcomed). I have only used glue sticks because you have to make sure you will be able to remove the spacer afterwards (i.e. you need a weak glue), but if the glue is too weak, you won't be able to hold the spacer in place during embedding and the whole thing will become really difficult and frustrating. I encourage you to try a few different glues and post your results here, a glue stick is definitely NOT the best solution.

Step 4: Mix the 20A shore silicone

The RTV silicone comes in two parts conveniently marked "part A" and "part B". You need to mix thoroughly both parts in a 1:1 mass ratio. Try to do this in as low a temperature and as dry environment as you can. Once mixed, the silicone will start to vulcanize at room temperature (RTV stands for room temperature vulcanized) and it will become increasingly difficult to work with.

You will need about 2g total of the 20A shore silicone (i.e. 1g of part A + 1g of part B) for each sensor. If you are making more than one sensor at a time, it is ok to mix all the silicone required at once. However, once mixed, the silicone must be used ASAP. You can store the unused mixed silicone in a plastic bag in your freezer, but I don't recommend it (I would rather use the extra silicone to model something else, like a fake finger to mix with candies for halloween).

Spend a few minutes mixing the silicone with a clean metal roller in the same way you would mix pasta or cookie dough. In fact, a pasta roller may work pretty well to prepare the mix. Make sure there is no grease/oil on any of the surfaces the silicone touches. Grease will prevent the silicone from vulcanizing so wash your hands or wear surgery gloves. Also, make sure no air bubbles are left in the mix.

Step 5: Create the silicone case

Picture of Create the silicone case
Use some more glue to fix the nylamid spacer (with the sensor attached) on a flat surface. You can use the back of the indented acrylic platform as a flat surface. Of course, by "back" I mean the side that is not indented.

You will get a "sandwiched" spacer where the "top slice" is the sensor and the "bottom slice" is the flat surface. At this point, you are ready to start placing some of the mixed silicone all around the sensor and spacer. You want to seal every single space and make sure there will be no air bubbles trapped. Don't worry about the electronic components, they work with very low currents so they won't overheat or short-circuit even when embedded.

Start by carefully shoving small quantities of silicone in tight spaces using the different modeling tools, then pack some more silicone around those spaces. Pack tightly, especially in the space between the spacer and the PCB around the microphone, and ensure the microphone is always centered on the spacer before continuing. Repeat until the silicone shows up around the board and use a flat spatula to pack the silicone some more from all 4 sides. My cases usually end up more square than the one shown (I didn't make this one).

Also make sure to cover the whole top of the board. It is not necessary to cover the accelerometer completely, so you can use it as a reference to even out the silicone at the top. Be prepared to spend up to 2 hours in this step, lousy embedding will make your sensor useless.

Step 6: Cure (vulcanize) the silicone case

Picture of Cure (vulcanize) the silicone case
bottom_emb.png
Place the silicone-embedded sensor into an oven at 60 deg C and allow curing for approximately eight hours. you can probably speed up the curing process with higher temperatures. However, some of the tiny air bubbles could pop and create fissures which may cause air leaks, making your sensor useless. You can actually see the air bubbles in the examples below (the silicone is not smooth and has too many reflections). This sensor was cures at 60 deg C for 8 hours, so the bubbles are probably due to lousy embedding.

After curing, carefully remove the sensor from the plastic board and then remove the plastic spacer. The case should look like the second figure below but with no bubbles!!!

Step 7: Mix the 65A shore silicone

Now mix some of the 65A shore silicone just as explained in step 4. This silicone will be used to create the bottom membrane that seals the case from below. You will need about 1g per sensor. In fact, you need less than this, but if you can't measure less than 1g with good precision, it will be difficult to maintain the 1:1 mix ratio, so you should go for the safer 1g measure (i.e. 0.5g of part A + 0.5g of part B) and use the remainder for the nail of your fake halloween finger.

Step 8: Roll membrane on indented board

Picture of Roll membrane on indented board
After cleaning the plastic board thoroughly, use the 65A shore silicone you just mixed and a small metal roller to produce a thin membrane centered on one of the indentations of the plastic board. Be sure to pack silicone into the indentation before flattening the membrane out. The indentation is in fact optional, but it will help secure good coupling between the skin and the sensor.

The thickness of membrane should be approximately 0.5mm, although I have no idea how to ensure this to happen (more suggestions welcome).

Step 9: Apply solvent to case edges

Picture of Apply solvent to case edges
Using a small brush, apply silicone solvent to the edge of the silicone case, on the microphone side, where it will contact the membrane. This allows the surface of the silicone to dissolve a bit to facilitate proper curing and sealing with the membrane. Also make sure the area of the membrane in the board is larger than the entire sensor case before continuing.

Step 10: Apply membrane and cure

Picture of Apply membrane and cure
Mount the sensor onto the flattened membrane making sure that the microphone is centered on the indentation in the plastic board. Use the excess silicone edge to seal the membrane onto the sensor by pulling it upwards and around the case.

Allow the sensor to cure at 60 deg C for eight hours.

Step 11: You are done!

Picture of You are done!
After curing, your sensor will be ready for testing. In order to do that, you can tap slowly on the membrane and check the microphone's output signal. You should be able to see the wave in a scope very clearly (for details on the electronic circuit check the first part of the instructable.). The membrane should quickly restore itself when deformed, otherwise, there may be an air leak. There is no reliable solution to air leaks... your best bet is to trash that sensor and start again.

By embedding in silicone, the sensor sensitivity has been passively increased. You have also enabled the microphone to measure really low frequency vibrations such as those produced by contracting muscles. Try placing the sensor in your chest and watch/listen your heart beats!

Check out the videos:

http://www.komodoopenlab.com/index.php/Portfolio/CMASP#vids

And above all, have fun!

NOTE: The techniques presented in this instructable are not optimal and still pretty rudimentary. Some of the disadvantages include a lack of consistency in the performance and sensitivity of the sensors built (due to the lack of precision in manufacturing). You are encouraged to suggest different, easier and/or more appropriate ways to solve any of the steps of this instructable. Your input will be extremely valuable in making these sensors affordable and useful in prosthetics.

Orngrimm1 year ago

It may sound silly and weird, but have you tested the latex of condoms to be used as the membrane instead a self-made-one? The latex is very thin yet durable and for sure biocompatible (in THIS applicational field).

drhoff3 years ago
to get the right thickness, put a spacer the same thickness that you want the membrane to be under each end of the roller and roll across.
Operon (author)  drhoff3 years ago
That is a very good idea, thanks!
hi, is there a reason you used silicon to seal the chamber as opposed using a tape or film to seal it? this seems a bit more complicated than it needs to be, to me at least.
Operon (author)  nave.notnilc5 years ago
No particular reason. We thought we would eventually manufacture the whole thing automatically, so it wouldn't really matter if it was a bit more complicated doing it manually, but that hasn't happened so I think it is worth to try simpler methods. I think tape or film would work ok but I would still suggest to ensure there is some kind of bump so the skin is in contact with the membrane at all times. If you make it with tape and get decent results, it might be worth adding you as a collaborator so you can add your alternative to the instructable.
hi i resolved the problem of the membrane. use two pieces of plexiglass separated by something (i used a 0.5mm pcb) if you give the pcb the shape you want you will get perfect membranes. i also did the same for the case of the microfone . if somebody is interested i can send some pictures. i have some question: 1) why do we use on oven considering that rtv silicone solidificates at room temperature? 2) why is the membrane made of sil. shore 60 ?'can we use the shore 20? 3) how are the dimension of the air chamber calculated? Thank's and ciao
Operon (author)  marcomilazzo6 years ago
That's great marco... I have added you as an author so please do add your info, steps, edits and/or pictures. My responses: 1) You just have to wait more... the oven is to speed up the curing process 2) I made some tests where shore 60 came out working better than 20, but I don't think the difference is too significant. I think I attached the paper with the data to this or the first instructable. 3) mmm.. I am not sure what you mean... again, this was done empirically, I just tested a whole bunch of sizes and the one reported here worked best.
moya.edge6 years ago
Hmmm...to deal with the consistency in the performance and sensitivity, it might be possible to compensate using a neural network. Maybe, just maybe, some of my friends can help deal with this type of problem. I can provide more information in the next year or so, no promises, this is not an easy problem to solve or maybe you solved it already since post. :)
Alliraph7 years ago
Hello Operon, I read about the selection of silicone hardness from your thesis. Is there any alternative if I can't get the silicone with hardness of 20A and 65A? As in one of your experiments, you concluded that "The softest or the hardest silicone types (shores 20A and 65A) are recommended for MMG measurement, as they exhibited the least variability during the tests." I wonder if a difference of ±5A would affect the measurement much. What do you think? Did you use any mold while embedding the components? Packing the silicone to the sensors like you described gives me an idea that the viscosity of the silicone is very high like modeling putty. I assume the silicone you used is for potting and encapsulation purpose, which generally has low viscosity. Is it not? As I know most of the silicone encapsulation use pouring method which requires mold. By the way, thank you so much for sharing this instructable. That's really a cool stuff. I wish I'll be able to build one.
Biotele7 years ago
I build something similar to your microphone enclosure and amplified the signal. I noticed something puzzling. When then the enclosure is air tight and the membrane is pressed there is low frequency high voltage rumble generated by the microphone. When amplified it sounds like rain on a large drum. If i make a small vent that rumble disappears. Is there something about electret microphone which make them act like that at high pressure? Are you getting these results.
Operon (author)  Biotele7 years ago
That is a bit weird... I remember getting high frequency noise when touching the microphone's case but it disappeared after embedding. I don't remember any low frequency rumble. Just out of curiosity: how did you enclose your microphone? Mine was not enclosed at high pressure (I imagine that could damage the mike)
Biotele Operon7 years ago
I used a bottle cap to make the air chamber and a stiff plastic membrane to close the cap. I noticed from your pictures that air chamber is actually just a bubble in front of the mic.
endolith8 years ago
The paper linked from the other instructable doesn't really explain the advantages of this type of system. In which situations is it better/cheaper than surface EMG?

I am specifically interested in detecting independent movement (preferably just the beginnings or intent to move) of the muscles in the forearm that control fingers. Is this realistic?
Operon (author)  endolith8 years ago
Is this better than EMG? This is still an open question. Some say it is (in theory), but it hasn't really been sufficiently proven. The main assumption is that muscle sound sensors are less sensitive to placement (because sound travels farther than electricity through tissue). They are also supposed to be cheaper, but these are the only two reasonable advantages I have heard about. They have disadvantages as well, like being slower than EMG and VERY sensitive to movement. The idea was abandoned by NASA early after they proposed (and patented) it, but they may still have some good uses, and if the limitations are resolved, they could definitely become a great alternative to EMG. For now, we just need to perfect their construction.

Is this cheaper than EMG? Depends, EMG sensors for prosthetics run at about US$700. So, if you make your own muscle sounds sensor it will definitely be cheaper, although I bet that is also true if you make your own EMG sensor.

Can it detect finger movement? This is actually something for which muscle sound sensors could be more useful than EMG sensors because the EMG sensors can't really detect deep muscle activity, but the tendon movement of finger flexors and extensors can be easily detected by the muscle sounds sensors. The only big problem is always limb movement (or any movement other than what you want to measure). You would have to figure out how to filter it out, perhaps by using the embedded accelerometer. You can check my thesis to get some ideas. Also, the PRISM lab has recently done some research into this.
Biotele Operon8 years ago
Electrode maintenance in EMG is a pain. Electrodes are good for a short while, but for long duration you start looking at implantable electrodes. You get a lot of motion artifact in emg which in your case you deal with by using an acc. But in emg the microphonics (change in impedance of the electrodes due to motion) are random and hard to filter. You get a lot of line noise with EMG which in your system is definitely not a problem. check out my dual power supply to power your system.
Operon (author)  Biotele8 years ago
I added you as a collaborator... maybe you want to add a link to your instructable and some instructions on how your power supply may be used.
Biotele Operon8 years ago
Thanks, I amt thinking of build your device... but a bit simpler.
endolith Operon8 years ago
"like being slower than EMG"

What does "slower" mean in this context?

"although I bet that is also true if you make your own EMG sensor."

Yeah. I was thinking of building something based on the OpenEEG design.

"but the tendon movement"

Oh, excellent. I hadn't thought of that. EMG only measures muscles, while sound is created by the tendons moving around, too?

I'm not really sure how movement and muscle sound would be measured differently by the accelerometer and the microphone. They'd both show up as the same kind of vibration in my mind. I'll read through your thesis and see if I can get a better grasp of this.
Biotele8 years ago
Can you use hot glue in place of silicon? Silicon is nasty to work with.
Operon (author)  Biotele8 years ago
You sure can try... the problem is always the precision though (getting good enough measurements in a consistent manner)... I agree, silicone is horrible to work with, but it is the best I have been able to do, although I am sure there must be a whole bunch of materials worth to try.
Biotele8 years ago
You need to use new amputees to model your system. New amputees muscles are not atrophied and changed and they still controlled in a natural fashion. New amputees can learn intrinsically to use a prosthesis if it is fitted IMMEDIATELY after surgery. This is because the subtle myofeedback loops that go to the spine are still alive. You take these for granted but they can make a difference of 6 month in phys-hab. This was observed by a polish doctor in the Napoleonic war. The retreating French army could not stop and so new amputee had to be immediately fitted with a stump and made to walk. They walked quite well.
oscampo8 years ago
Thank you so much Jorge!!! I can't wait for this instructable and, I used a similar procedure and I have ready some specimes for doing characterization. In my sensor, the 13x2mm air chamber only reports a max SNR of 19.0 dB. But, as you write, this procedure with the silicon is not precise. I have had thinking about another geometry in the air chamber, maybe as a cone, but now I don't have the time for trying. But in a couple of months I will try to test with conic or another geometries. Thanks again!!
Operon (author)  oscampo8 years ago
No problem!! I have added you as a collaborator in the instructable. It would be great if you could add steps on how to measure the SNR. You can post some pictures of your sensors as well. Don't hesitate to make any recommendations you consider useful!