Tweet
Unnecessary Backstory: About two decades ago, my parents were building a new house and gave a budget to build a theater room. I was still in high school and loved listening to music. Initially we were just going to do a Bose all in one speaker system with a big screen, since that is what was popular at the time. Thankfully another friend was more obsessed with AV at the time and pointed me towards some forums. I got inspired by seeing theaters posted in online forums. They had a meet-up demo of some guys theater, and I went with some people and my dad, who promptly increased our budget... but everything is so still dang expensive. So I got into the DIY section of the forums for subwoofers and inevitably ran into the guys building DIY speakers and this forum, then got to listen to some RS TMWW speakers that the user chasw98 built. Eventually we built three CJD Khanspires for a matching L/C/R speakers behind an acoustically transparent screen, four of CJD's 2-way speakers for the surrounds with enclosures built into the false ceiling, and 4 x 18" infinite baffle subwoofer using a false ceiling as the airspace. I tried to build my own short 3-way speaker and it went well but the final product didn't measure how I predicted and it left a sour taste in my mouth that I left some performance on the table. A few years later I got an itch to build a big horn loaded TMWW speaker with AMT tweeter and mid using Marcel Batik's design tool: https://at-horns.eu/
The Meaningful Part: I previously built a test baffle for the speakers to test very simple horns. Decided to test building a crossover there. Below is the rural testing set-up with foam to suppress tile first reflection:
Using a calibrated mic and DATS, I measured the RS-225 woofer and the AMTPRO-4 individually to generate FRD and ZMA files, then measure together to estimate the time alignment in XSim. Then I built a simple crossover for the mid and woofer. The notch filter was to flatten an ugly 10kHZ peak when using this driver without horn loading, and is the reason for the impedance dip at the very high frequencies. The tweeter is not modeled here, but will be for a future project; though with 2 woofers can be built the same way with 2nd and 3rd order crossovers and L-pads and not dip below 4 ohms.
This is not a speaker you want to listen to. This study's only concern was the crossover. My test setup limits me to measuring down to 300hz from impulse gating the measurement for room reflection effects, and is irrelevant with an open baffle like this anyways.
Here is the modeled crossover response and the schematic made in XSim:
Then built the crossover, which is the mess below the speaker in the first picture. Then measured the actual response (red line) of the crossover against the predicted response (blue line) using XSim:
Here is out of phase actual (red) vs predicted (blue):
Disclaimer: Apologies for not being explicitly clear on the measurement process and time alignment estimation. I'm still learning and can HIGHLY recommend this guy's courses on crossovers and the one on measuring speakers which taught me how to build a simple crossover that measures in real life how it was modeled: https://www.udemy.com/user/marius-tanasescu-2/ No I haven't been paid, I just deeply appreciate his content and how it helped me finally get better results.
The Meaningful Part: I previously built a test baffle for the speakers to test very simple horns. Decided to test building a crossover there. Below is the rural testing set-up with foam to suppress tile first reflection:
Using a calibrated mic and DATS, I measured the RS-225 woofer and the AMTPRO-4 individually to generate FRD and ZMA files, then measure together to estimate the time alignment in XSim. Then I built a simple crossover for the mid and woofer. The notch filter was to flatten an ugly 10kHZ peak when using this driver without horn loading, and is the reason for the impedance dip at the very high frequencies. The tweeter is not modeled here, but will be for a future project; though with 2 woofers can be built the same way with 2nd and 3rd order crossovers and L-pads and not dip below 4 ohms.
This is not a speaker you want to listen to. This study's only concern was the crossover. My test setup limits me to measuring down to 300hz from impulse gating the measurement for room reflection effects, and is irrelevant with an open baffle like this anyways.
Here is the modeled crossover response and the schematic made in XSim:
Then built the crossover, which is the mess below the speaker in the first picture. Then measured the actual response (red line) of the crossover against the predicted response (blue line) using XSim:
Here is out of phase actual (red) vs predicted (blue):
Disclaimer: Apologies for not being explicitly clear on the measurement process and time alignment estimation. I'm still learning and can HIGHLY recommend this guy's courses on crossovers and the one on measuring speakers which taught me how to build a simple crossover that measures in real life how it was modeled: https://www.udemy.com/user/marius-tanasescu-2/ No I haven't been paid, I just deeply appreciate his content and how it helped me finally get better results.
. Avoid USB mic not just for the modern "VCAD ways", but also because measurement with timing reference can provide better understanding of phase, delay, time of flight if you can see the real change in excess phase and impulse location relative to t=0 as audio sources are moved around. Avoid Xsim as well for the "outdated" category, it is end of life software from 2015, VCAD is the continuation with all the same features + 100x more.
Comment