Originally Posted by Longeze
A few misc. questions & observations:
Why is it that it appears people are mainly interested in waveform amplitude when displaying graphs?
Has or can anyone overlay the original signal plot over their measured response plot? (I don't have a phone with an accelerometer).
Has or can anyone do an FFT or spectrum analysis of the system response to a white noise signal source and without EQ(flat) as well as with EQ applied?
Am I missing something, or is the amplitude gradient on the graphs shown, fairly course? If so, is it possible to increase resolution by limiting the sample range and taking multiple samples?
I would think that doing the above analysis with various isolators & configurations would be instructive.
Longeze....great questions and basically, yes, there's been some discussions in the VibSensor thread about the original EOT signal and what it looks like after an FFT analysis. Below is the FFT of the signal for that time stamp straight from the disc.
Basically, if the ULF system is responding in a linear fashion, its response should look very similar.....i.e.; a linear increase in power as the square wave reduces in frequency in discrete 5Hz increments.
For our hobby, member
has created an rpi based DIY vibrometer. Details about it are in the link below. There's also the VibSensor app (link below) which has given us hobbyist an accurate and easy method to measure the ULF TR response above 10Hz. Below 10Hz, the actual weight of the phone begins to introduce measurement error. But since the EOT demo scene ends at 10Hz, it's a pretty good signal to use when trying to characterize our ULF reproduction systems and comparing to the original EOT signal below.
The DIY vibrometer using rpi shouldn't have the 10Hz limitation since it's lighter in weight. However, I don't think anyone has actually tried it yet to my knowledge.
Regarding white noise, I believe there's similar FFT traces in the VibSensor app also for a 0-50Hz content signal and this signal can also be used to characterize and supplement the EOT signal when evaluating our ULF TR systems. Using both together gives a picture of how accurately our ULF TR systems are replicating the original signal, both with discrete content signals (EOT sample) and with mixed signals content which is more real-world (the 0-50Hz White Noise sample).
DIY vibrometer using rpi: https://www.avsforum.com/forum/155-d...n-meter-6.html
VibSensor Thread: https://www.avsforum.com/forum/113-s...st-thread.html
EOT FFT Analysis
Originally Posted by Longeze
My thinking when using the innertubes is that there will be a significant reduction in point load stress, creating a more uniform displacement of the baffle with minimal distortion.
The tubes are dirt cheap and offer adjustable response as I've said.
The tubes offer potentially greater range of motion and less coupling between planes than "conventional" isolators.
They're independently tunable for load. Since my GF & I are mainly the ones using the hovercouch, I was able to tune each side separately.
Their response in both compression & rebound is likely to be more linear than current Boss tech, which typically uses conical section (progressive rate) damping. IE. the more you displace the cones, the stiffer the response becomes. An argument could be made that this is desirable, however in performance vehicle suspension systems, technology has moved away from this thinking back towards linear spring technology to keep the wheels (couch) under control. Progressive rate springs are used as a shotgun approach to accommodate a wide variety of loads. My goal is to approach "critical damping" for the hovercouch to maximize its ability to reproduce it intended operational frequencies as accurately as possible while maximizing amplitude for a given energetic input.
FWIW, my approach to tuning it was similar to tuning the suspension on a motorcycle. I'd bounce up & down on it & vary the PSI until it felt a sufficient range of motion and experienced no more than 1 cycle of rebound. I had my GF do the same on her end.
FWIW, With my amp set to 1 O'clock, the Boss-hovercouch will shake the crap out of my house-3 rooms away, if I set LFE to 0db. I typically listen at 0db with the LFE set to -12db. Max on the MV knobs is ~5 O'clock with 119W RMS Max available per driver.
To keep the tubes in place around the JBL's, I drilled holes for zip ties to keep the tubes captive to the extent that they can't work their way under the speakers. The zip ties were kept fairly loose. I think thin soft rope/cord would be preferred, but I didn't have any of the proper size on hand.
If you want to go really deluxe, you could tie all the tubes together on a common air line into a single Schraeder valve, for ease of adjustment/refill. I found that it was so easy to remove & replace individual tubes, that I'm not going to bother with it as I planned.
I apologize for the less than scientific "data" provided. I realize that it's pretty much entirely anecdotal, but given the super low cost of the tubes & the ease of implementing them, I strongly encourage others to try them. I'm pretty damn happy with the results. YMMV
Your thought process is spot on. By controlling the inner tube pressure, the effective visco-elastic properties of the suspension system can be changed rather dramatically. That inner tube pressure combined with control of the speaker cavity pressure inside each inner tube (different tire diameters surrounding each JBL), can be used to customize the BOSS experience. One could probably write a thesis project just from the use of inner tubes and how the pressure and diameter of the inner tubes could be tuned to optimize the BOSS experience.
Just as a point of reference, during the initial BOSS development, I kept the suspension constant (20-30 durometer visco-elastic rubber). This was done to control the experiment and minimize the suspension variable. This allowed focus on other factors I was worried about most when looking at all the data being collected. there were 8 different accelerometers placed at different locations around the 4x8 sheet of plywood and all 8 were measuring simultaneously. This was done primarily to see what factors affected platform bending which I wanted to minimize to keep the BOSS response linear and natural.
There could easily be another set of experiments conducted looking at just the suspension factors outlined if one really wanted to chase further optimization of BOSS performance. I concluded going above 30-40 durometer has a largely negative impact on performance. But, I think there's a lot of runway left with experiments going below 20 durometer and the incremental benefits could be staggering in terms of TR sensitivity and power optimization. Then there's the hover-craft effect which could be yet another thesis by itself
BTW....it sounds like we may be cut from the same cloth so to speak. I have a Mechanical Engineering degree from General Motors Institute with specialties in Machine Design and Vehicle Dynamics.....pretty cool stuff! Far from what I'm doing today as a career, but it's still one of my passions and bubbles to the surface once in awhile with hobbies such as this