I am about to start designing a crossover (my first design- I've built a few before). I need a little helpto understand if my approach is correct.
This is a system for my in-laws. They are setting up a home theater. These don't have to be perfect, but I'd like to get them as close as possible, without getting too complex (read, expensive) with the crossover. The drivers are a Morel MDT-20 tweeter, and a jbl/vifa Parts express buyout 4-1/2" midwoofer (http://www.partsexpress.com/pdf/299-432.pdf)
The driver measurements I have from parts express show a pretty flat frequency response for the tweeter and 4-1/2" midwoofer. The sensitivity of both are about equal, also. The woofer extends out flat to 5.5KHz, where it has a small resonance peak, followed by a bigger dip.
My plan was to use a second order LR at 3K. The phase of the woofer is 180 degrees out at that frequency, and the tweeter (mdt-20) is at about 0 degrees at that frequency. My theory was that with a second order network, I should be able to run normal polarity on both drivers and have them sum well with the woofer's phase being at 180. Of course, these are free air responses, and I don't know if the enclosure will affect the phase response of the driver. I assume the frequency response will change somewhat.
Also, I'm not sure how the different radiating points of the drivers will affect phase and the summation at the crossover point. I see now why software is needed for doing this.
My plan, for now, is to use speaker workshop to take some driver response measurements with the drivers in the enclosure. Any ideas of where to go from there, or what might work for a crossover? I'm quickly running out of time, so I'd appreciate any suggestions. Thanks,
Steve
This is a system for my in-laws. They are setting up a home theater. These don't have to be perfect, but I'd like to get them as close as possible, without getting too complex (read, expensive) with the crossover. The drivers are a Morel MDT-20 tweeter, and a jbl/vifa Parts express buyout 4-1/2" midwoofer (http://www.partsexpress.com/pdf/299-432.pdf)
The driver measurements I have from parts express show a pretty flat frequency response for the tweeter and 4-1/2" midwoofer. The sensitivity of both are about equal, also. The woofer extends out flat to 5.5KHz, where it has a small resonance peak, followed by a bigger dip.
My plan was to use a second order LR at 3K. The phase of the woofer is 180 degrees out at that frequency, and the tweeter (mdt-20) is at about 0 degrees at that frequency. My theory was that with a second order network, I should be able to run normal polarity on both drivers and have them sum well with the woofer's phase being at 180. Of course, these are free air responses, and I don't know if the enclosure will affect the phase response of the driver. I assume the frequency response will change somewhat.
Also, I'm not sure how the different radiating points of the drivers will affect phase and the summation at the crossover point. I see now why software is needed for doing this.
My plan, for now, is to use speaker workshop to take some driver response measurements with the drivers in the enclosure. Any ideas of where to go from there, or what might work for a crossover? I'm quickly running out of time, so I'd appreciate any suggestions. Thanks,
Steve
You can get pretty darned close using tools available to you at the FRD Consortium and PE's published data files.
Baffle Simulator will give you a baffle curve.
Frequency Response combiner will allow you to combine the results with PE's data (PE uses 2Pi measurements). But most importantly, you can extract the minimum phase data with this (whether you combine with baffle step response or not) which is important for accurate crossover design.
They also have a nifty and very good crossover design worksheet.
To combine frequency response and baffle step, import driver data into the first column. import baffle step response into Box Target. Switch to the "Normalize" view, set your range on the left side, and normalize both. I usually do 20-20khz and 540 data points. Switch to Combine. Set Baffle to to 2Pi and combine. If it fails, try setting to IEC, combine, then back to 2Pi and combine. Seems to be a bug. The result should put the response at 20hz about 6dB down from default. Use Accurate phase data (not Quick Phase), extract. Export from column seven (may have to click "Save" once to get the right Export dialog box).
Consider an MTM for efficiency sake. Also consider exactly how much baffle step to include. For HT stuff I usually do 3-4dB (which has a gentle roll-off from 200hz down with fully compensated data files, but measures flat in near-wall placement which is common). I also usually design ported box response to have a gentle roll-off - again picked up by room gain usually.
Good luck, and have fun!
C
Baffle Simulator will give you a baffle curve.
Frequency Response combiner will allow you to combine the results with PE's data (PE uses 2Pi measurements). But most importantly, you can extract the minimum phase data with this (whether you combine with baffle step response or not) which is important for accurate crossover design.
They also have a nifty and very good crossover design worksheet.
To combine frequency response and baffle step, import driver data into the first column. import baffle step response into Box Target. Switch to the "Normalize" view, set your range on the left side, and normalize both. I usually do 20-20khz and 540 data points. Switch to Combine. Set Baffle to to 2Pi and combine. If it fails, try setting to IEC, combine, then back to 2Pi and combine. Seems to be a bug. The result should put the response at 20hz about 6dB down from default. Use Accurate phase data (not Quick Phase), extract. Export from column seven (may have to click "Save" once to get the right Export dialog box).
Consider an MTM for efficiency sake. Also consider exactly how much baffle step to include. For HT stuff I usually do 3-4dB (which has a gentle roll-off from 200hz down with fully compensated data files, but measures flat in near-wall placement which is common). I also usually design ported box response to have a gentle roll-off - again picked up by room gain usually.
Good luck, and have fun!
C
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.