Electronics/Acoustics project ideas for Thesis

[Exprymer]

Hello!
I'm finishing my Electrical Engineering in about 1.5 year and i intend to make a Acoustics/Electronics based Final Project/Thesis.
I'd ask you opinions of what would be a useful project and maybe offer some ideas.
Some Ideas that i already have in mind:

1- Subwoofer-stand like the Genelec W371A, dsp based, featuring auto eq and the Anti-SBR feature. This project would have a dock that would receive the signal from the audio interface and apply the crossovers and configure the DSP. Further Thinking is necessary on this one.

2 - USB DAC with auto-eq integrated, using the auto-eq Repository.

3 - Active Absorption by Shunt loudspeakers (like the PSI AVAAs).

Any other Ideas are welcomed,
John.

 

[alex-z]

#1 is already really easy. Use a miniDSP 2x4HD, set a crossover at 80Hz, then measure with Room EQ Wizard and tell it to auto-generate filters. You supply the house curve, and amount of headroom available, it calculates filters for you.

#2 seems like a similar project, just applied to headphones instead.

#3 seems the most promising. Active room mode treatment is an expensive niche, so any work done on making it more affordable is valuable.

You could also do something like design your own DAC + amp combo, which directly falls under electrical engineering.

 

[Exprymer]

Hey, @alex-z . Yeah, i could use the miniDSP but i wanted to build my own using embedded systems like STM32 together with and DSP like ADAU1701. Also, i don't know if the midiDSP does anything other than FR correction. I wanted to be a little more in depth with maybe Hilbert Transforms or using All Pass filters tuned by Machine learning to decide the best Impulse response for the subs. Knowing that it would be 2 subwoofers, the second one would output an convoluted signal that fills the gaps on the FR not by PEQ or FIR alone but with phase correction.

#2 is actually similar indeed. I want to design the power stage and the DSP stage.

#3 is a Topic that i've been reading extensively. It's indeed a new and yet expensive topic that could be developed in a diy situation.

 

[bigjacko]

How about distortion cancel? Ha anyone did anything like this? I am thinking there might be two ways to do it.

First way is to measure distortion in real time, either measure frequency response or driver itself. This way the processing time needs to be short and will never be on time.

Second way is to measure distortion before playing music, either frequency or driver data like klippel data. This way the corection will be on time but needs to calibrate after a certain time.

 

[DonH56]

Since you said "project/thesis" I assume this is not for a PhD (and thus need be original work)? Undergrad or masters level? How complex does it need to be, what resources do you have available?

I built a servo-sub circuit when I was in college. Wasn't exactly a senior project, just something for an acoustics grad class, but mainly because I needed a sub. The circuit was similar to the Rythmik design (I should have patented mine back then) and included an active crossover, all-pass filter for delay matching, and LDI servo circuit, all analog op-amp based (no cheap DSP chips back then).

There are so many good room-correction schemes already that I'd be hesitant to do that. Not that it would not be fun and useful, but I usually look for something I need (selfish). But it sounds like a neat project.

Active absorption sounds interesting. There are also some new'ish membrane absorber/diffusion schemes that would be fun to implement. The latter essentially uses specially-shaped cavities to get broadband absorption from a very thin panel. Would require machining or woodworking skills, however (I have a wood shop, just no time to play in it).

HTH - Don

 

[Exprymer]

How about distortion cancel? Ha anyone did anything like this? I am thinking there might be two ways to do it.

First way is to measure distortion in real time, either measure frequency response or driver itself. This way the processing time needs to be short and will never be on time.

Second way is to measure distortion before playing music, either frequency or driver data like klippel data. This way the corection will be on time but needs to calibrate after a certain time.

Hello, Actually i think there are distortion correction technologies already available. The distortion comes from the non linearities of the woofer so some technologies apply a sort of feedfoward control on the signal to compensate for these non linearities.

 

[Exprymer]

I built a servo-sub circuit when I was in college. Wasn't exactly a senior project, just something for an acoustics grad class, but mainly because I needed a sub. The circuit was similar to the Rythmik design (I should have patented mine back then) and included an active crossover, all-pass filter for delay matching, and LDI servo circuit, all analog op-amp based (no cheap DSP chips back then).

That's quite interesting actually. I didn't know of the use of servos for sub design. In my mind they were not as fast.
This could be useful for the Active Absorption unit, which i can use the servo to excite a membrane in order to mitigate que pressure. Thanks for you contribution!

It's my final Graduation Project.

 

[DonH56]

That's quite interesting actually. I didn't know of the use of servos for sub design. In my mind they were not as fast.
This could be useful for the Active Absorption unit, which i can use the servo to excite a membrane in order to mitigate que pressure. Thanks for you contribution!

It's my final Graduation Project.

Not as fast as far as bandwidth, but distortion is audible, and since they tend to be driven to the highest power levels thermal effects are more prominent. There are a number of ways to implement the servo and sensor, probably worth doing an overview leading to the one you choose. I used a dual voice coil woofer and used the second coil for feedback; that mitigates some of the effects like thermal that an accelerometer based approach does not. My control was entirely analog; now, you could investigate using a DSP to do it (probably how I would do it today, had I not grown old, fat, and lazy and just bought one ).

 

[SIY]

Hello!
I'm finishing my Electrical Engineering in about 1.5 year and i intend to make a Acoustics/Electronics based Final Project/Thesis.
I'd ask you opinions of what would be a useful project and maybe offer some ideas.
Some Ideas that i already have in mind:

1- Subwoofer-stand like the Genelec W371A, dsp based, featuring auto eq and the Anti-SBR feature. This project would have a dock that would receive the signal from the audio interface and apply the crossovers and configure the DSP. Further Thinking is necessary on this one.

2 - USB DAC with auto-eq integrated, using the auto-eq Repository.

3 - Active Absorption by Shunt loudspeakers (like the PSI AVAAs).

Any other Ideas are welcomed,
John.

Where are you? Depending on your answer, I might have something interesting for you.

 

[Exprymer]

Where are you? Depending on your answer, I might have something interesting for you.

I Reside in Brazil, at the moment. Makes things quite difficult haha

 

[Exprymer]

Not as fast as far as bandwidth, but distortion is audible, and since they tend to be driven to the highest power levels thermal effects are more prominent. There are a number of ways to implement the servo and sensor, probably worth doing an overview leading to the one you choose. I used a dual voice coil woofer and used the second coil for feedback; that mitigates some of the effects like thermal that an accelerometer based approach does not. My control was entirely analog; now, you could investigate using a DSP to do it (probably how I would do it today, had I not grown old, fat, and lazy and just bought one ).

That was my thoughts exactly, the servos i've seen were not reliable enough for >60hz. The heat issues is another problem indeed. But on active absorption i don't really think Xmax will be a major factor.

 

[DonH56]

That was my thoughts exactly, the servos i've seen were not reliable enough for >60hz. The heat issues is another problem indeed. But on active absorption i don't really think Xmax will be a major factor.

My original design targeted 300 Hz and was stable to about there. Again, that was all-analog... But I would PM @SIY because if he has something you can do I bet that would be a killer project.