DSP - how, why, tro...
 
Notifications
Clear all

DSP - how, why, troubleshooting, different hardware, etc.

Page 5 / 6

Nixem
(@nixem)
Eminent Member Registered
Joined: 3 months ago
Posts: 23
 

@ajc9988

Melamine (magic eraser sponge) foam would be interesting too, not sure if it's easier to simulate.


ReplyQuote
ajc9988
(@ajc9988)
Estimable Member Registered
Joined: 9 months ago
Posts: 213
Topic starter  

@nixem - If you can give me the information for the material to be modeled, such as the behavior (isotropic, etc.), Young's Modulus, Poisson's Ratio, Shear Modulus, density, I should be able to do something. Nice to haves are Yield strength and tensile strength, but no necessary, so to speak.

So it is a question of tracking down the material's properties. If you have those where it can be inputted, then it can be modeled.


ReplyQuote

Nixem
(@nixem)
Eminent Member Registered
Joined: 3 months ago
Posts: 23
 

 

Melamine foam macroscopic parameters

@ajc9988 

*Hard for me to paste on my phone, I'll try to edit it later.

Edit: trying to fix links

 

 

This is what I have found, I'll try to find more parameters when possible:

See page 3/6: 

https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.704.8096&rep=rep1&type=pdf

 

 

Not sure if this is helpful, I am not an engineer.

https://core.ac.uk/download/pdf/145050898.pdf

https://www.matelys.com/publications/JRD08.pdf

 

It's hard for me to interpret, but are people having a hard time getting Young's modulus of melamine foam?


ReplyQuote
ajc9988
(@ajc9988)
Estimable Member Registered
Joined: 9 months ago
Posts: 213
Topic starter  

@nixem - You'll have to give me some time to run the model. That first source seems to have what I need, from a cursory glance.

I have been using primarily a mesh size of 0.5", which is good to about 1kHz. To model for 20kHz, the mesh size needs to be 1/10th of that, and that will push the edge of my system on ram and storage needed to model (depending how high the modal analysis is pushed due to the insulating nature of the material). Now, if you want to buy me more ram or another larger SSD...

But, in all seriousness, I'll see what I can do and how many TB that model will take. So far, the most I've seen a model take is a couple TB on my 8TB HDD (so, yes, a little slower due to spinning rust instead of SSD). And this workstation only has 32GB of ram, so guaranteed to take up more than I have in ram. CPU is fine (8 core Intel 10700KF overclocked to 5.2GHz all core, 5.1 for AVX workloads), especially for the serial portions of some of these calculations. I just need more ram and more storage (and I'll take a newer GPU, if anyone feels like granting a wish these days, lol).

I'll make more of an effort to look around in the next little bit.

Question: How thick are you wanting this material to be?


ReplyQuote

Nixem
(@nixem)
Eminent Member Registered
Joined: 3 months ago
Posts: 23
 

@ajc9988 

 

You don't have to model it for me, but I really appreciate it!

 

I was just throwing something interesting into the mix. Melamine foam I'm talking about are those cheap cleaning pads, you can get a 50 pack for $13 on Amazon, dimensions around 10cm x 6cm x 2cm each. 

 

Just throwing out ideas of cheap things that might work well that I've seen other people use.

 


ajc9988 liked
ReplyQuote
ajc9988
(@ajc9988)
Estimable Member Registered
Joined: 9 months ago
Posts: 213
Topic starter  

@nixem - so about 2cm. I may give it a go sometime in the near future.


Nixem liked
ReplyQuote

ElliottDesigns
(@elliottdesigns)
Reputable Member Moderator
Joined: 2 years ago
Posts: 254
 

@ajc9988 Just got 17920 taps processing on my teensy! It works! Also, it's a ram limitation, it's only taking 6% CPU usage! Insane! And to think people spend £220 on a miniDSP 2x4 hd when it can only handle 4096 taps. What a rip off 😂

Elliott Dyson - Mechanical Engineering Student and 3D printing & Design Freelancer


ajc9988 and Nixem liked
ReplyQuote
ElliottDesigns
(@elliottdesigns)
Reputable Member Moderator
Joined: 2 years ago
Posts: 254
 

Also if running minimum phase rather than linear. It's less than 20ms of processing time, if I remember correctly it might even be less than 10!

Elliott Dyson - Mechanical Engineering Student and 3D printing & Design Freelancer


ReplyQuote

ajc9988
(@ajc9988)
Estimable Member Registered
Joined: 9 months ago
Posts: 213
Topic starter  

@elliottdesigns - Nice! looks like you just need to work on the memory now. Did you buy with the extra chip to solder on? IIRC, the teensy has where you have that blank memory spot where you can solder one of two types of memory. Just trying to see if you can increase the memory slightly, thereby potentially going even farther.


ReplyQuote
ElliottDesigns
(@elliottdesigns)
Reputable Member Moderator
Joined: 2 years ago
Posts: 254
 

@ajc9988 nah, I won't risk it, not only are the surface mounted components too miniscule for me to attempt to desolder them, I have no reflow plate or oven

Elliott Dyson - Mechanical Engineering Student and 3D printing & Design Freelancer


ajc9988 liked
ReplyQuote

ElliottDesigns
(@elliottdesigns)
Reputable Member Moderator
Joined: 2 years ago
Posts: 254
 

@ajc9988 Anyways 17920 taps is plenty enough for me. You can get very detailed EQ in the bass region which is really quite uncommon in the commercial market (as far as I'm aware). I'll probably then use Biquads for the high pass and low pass so I can use it to make my two way, and then correct for the crossover phase with the FIR convolution, that way I don't have to reprogram it to work with 2 different lists of filters, one for left, the other for the right. Not only would that be a real pain to code, but I'd be left with 8960 taps for room correction and a crossover slope, which isn't quite enough for the accuracy I want, especially with the bass region

Elliott Dyson - Mechanical Engineering Student and 3D printing & Design Freelancer


ajc9988 liked
ReplyQuote
ElliottDesigns
(@elliottdesigns)
Reputable Member Moderator
Joined: 2 years ago
Posts: 254
 

@ajc9988 as for your simulation, I imagine 1khz is more than enough for a sealed box because I very much doubt anything above that not being damped by the MDF. But for things like transmission lines, a smaller mesh size could be very useful!

Elliott Dyson - Mechanical Engineering Student and 3D printing & Design Freelancer


ajc9988 liked
ReplyQuote

ajc9988
(@ajc9988)
Estimable Member Registered
Joined: 9 months ago
Posts: 213
Topic starter  

@elliottdesigns - oddly enough, for the setup I did for linear (sorry, the software being used doesn't have nonlinear analyses components and I don't currently have access to some that do, although I maybe can convince someone at some point down the road to allow me to use their resources), it was using insulation that pushed the first mode above 1,500Hz with 1" rockwool, and 2500 for 2". Note, I probably needed to adjust the critical and structural damping on those runs, which I did not, meaning that the accuracy is in question, instead I had done the same to see apples to apples for the other runs, whereas looking up melamine put the damping coefficient at 0.1, instead of what I have been using, so just want to premise that the numbers are wrong, even though the trend is correct.

As such, insulation just is so far above plain MDF or using constrained layer damping (As said, I've been using the same value, which for structural MDF, should be approximately the same for MDF in CLD, although it didn't account for the damping coefficient of the rubber mat put between the layers, even though it does factor in the physical properties otherwise of the rubber mat in elasticity, etc., showing a large reduction in displacement under pressure and driving higher the frequency to mode).


ReplyQuote
ajc9988
(@ajc9988)
Estimable Member Registered
Joined: 9 months ago
Posts: 213
Topic starter  

@nixem - The results are bad. They are about the same as one vertical brace (which is less than horizontal). I haven't tried it in conjunction, which one vert and one hor brace is the same as roughly 5-6 horizontal braces meeting certain criteria.

Personally, I would not use the melamine. It just isn't as promising as others out there. I am trying PET next.


Nixem liked
ReplyQuote

Nixem
(@nixem)
Eminent Member Registered
Joined: 3 months ago
Posts: 23
 

@ajc9988 

Interesting! I wonder if it or other open foams do better for higher frequencies.

Is the PET you are using like the Dacron polyester fabric fiber for upholstery?


ReplyQuote
Page 5 / 6

Newegg

Share:
%d bloggers like this: