Purpose:
I started a previous thread with an experiment showing that the orientation of the woofer under test had a large effect on its T/S parameters, hence its predicted behavior in an enclosure. Several forum members pointed out the possibility that it was the proximity of a reflective surface, rather than driver orientation, which was responsible for much of the effects. One poster recommended hanging a driver from chains far from any reflective surfaces.
Here, that experiment is largely repeated. The drivers are kept further from reflective surfaces. The hanging method is also used and compared to the clamping methods. Separate testing of the individual voice coils of the dual-voice-coil woofers was eliminated.
Methods:
The Audio Concepts DV-12 was used. This is a 12" paper cone, cast frame, dual voice coil (6ohm + 6ohm) woofer designed for sealed cabinets. SpeakerWorkshop by Audata was used for data collection, T/S parameter estimation, and frequency-response simulation. A ChipAmp.com LM3886 kit was used for power at ~0.32VAC. A Wallin Jig and Behringer UCA-222 was used to interface with the software.
Two identical drivers were tested using the free-air and added-mass methods using three orientations:
In addition, the two drivers were firmly bolted face-to-face and modeled as a single driver. For single-driver testing, voice coils were wired in parallel for a DC resistance of 3 ohms. For face-to-face testing, each coil from one driver was wired in parallel with a coil from the other driver. Then the two pairs of parallel-wired coils were wired in series, for a DC resistance of 6 ohms.
104g of mass was added, producing ~30% shift in Fs for single drivers, and ~15% shift for drivers tested face-to-face. A maximum length signal (MLS) was used with a voltage of about 0.32VAC. The Behringer UCA222 does now allow adjustment of recording level, and 0.32VAC was the largest signal which would not result in clipping of the inputs.
Results:
Data from Woofer A in the horizontal position is suspect! The impedance curves clearly show the minimum impedance of driver A in the horizontal position, both with and without added mass, is ~6ohms. This is likely due to a bad connection. All other impedance curves are as expected. Some example impedance curves can be seen in figure #5. Another note: SpeakerWorkshop gives the user the choice of entering DC resistance manually, or allowing SW to calculate DC resistance from the impedance curves. SW’s calculation of DC resistance differed from that measured with a multimeter, and I found that using manually-entered DC resistance resulted in more consistent T/S parameters, so I opted to input DC resistance manually. This was not done in the first experiment.
The sixth figure is a chart showing all T/S parameters calculated by SpeakerWorkshop for all conditions tested. The seventh figure is simply a graphical representation of select data in picture #6.
The top graph in figure #8 shows simulated frequency-response curves of both drivers in a box volume (undamped) calculated to achieve a Qtc of 0.707. The chart in figure #9 is the box volume used to plot curves in said graph. The bottom graph in figure #8 shows simulated frequency-response curves of the same drivers keeping Vb=100 liters.
Conclusion:
The data are difficult in interpret and a bit perplexing. Testing the driver vertically clearly results in lower Fs, higher Vas and slightly higher Qms. Note that when tested face-to-face, modeled as one driver, these orientation-dependent differences disappeared. This suggests that whatever is causing the differences is cancelled by the cones moving simultaneously in opposite directions relative to their frames. It is curious that the hanging method produced results more closely matching the horizontal orientation than the vertical orientation, even though the driver can be seen hanging a little closer to vertical than horizontal. The results leave me wishing I had tested the back-to-back drivers hanging from the ceiling.
The big question:
So, how big should I build my box? The differences in Vb needed to obtain my desired Qtc of 0.707 are indeed huge for the different measurement methods. The differences also demonstrate that different parameters are, indeed not cancelling each other out. I plan to use the woofers horizontally in the face-to-face compound configuration, close to the floor. I’ve been warned by members of this forum that modeling the two woofers as one woofer and building according to those parameters is the wrong way to do it. I should just get the T/S parameters for “the” driver and cut the Vb in half. But what are the T/S parameters for my woofers?
I started a previous thread with an experiment showing that the orientation of the woofer under test had a large effect on its T/S parameters, hence its predicted behavior in an enclosure. Several forum members pointed out the possibility that it was the proximity of a reflective surface, rather than driver orientation, which was responsible for much of the effects. One poster recommended hanging a driver from chains far from any reflective surfaces.
Here, that experiment is largely repeated. The drivers are kept further from reflective surfaces. The hanging method is also used and compared to the clamping methods. Separate testing of the individual voice coils of the dual-voice-coil woofers was eliminated.
Methods:
The Audio Concepts DV-12 was used. This is a 12" paper cone, cast frame, dual voice coil (6ohm + 6ohm) woofer designed for sealed cabinets. SpeakerWorkshop by Audata was used for data collection, T/S parameter estimation, and frequency-response simulation. A ChipAmp.com LM3886 kit was used for power at ~0.32VAC. A Wallin Jig and Behringer UCA-222 was used to interface with the software.
Two identical drivers were tested using the free-air and added-mass methods using three orientations:
- solidly clamped to a test jig vertically (see Figures #1 & #2)
- solidly clamped to a test jig facing up about 4ft (1.2m) from both the floor and ceiling (see Figure #3)
- hanging from chains roughly 3ft (0.91m) from the ceiling and 5 feet (1.5m) from the floor (see Figure #4)
In addition, the two drivers were firmly bolted face-to-face and modeled as a single driver. For single-driver testing, voice coils were wired in parallel for a DC resistance of 3 ohms. For face-to-face testing, each coil from one driver was wired in parallel with a coil from the other driver. Then the two pairs of parallel-wired coils were wired in series, for a DC resistance of 6 ohms.
104g of mass was added, producing ~30% shift in Fs for single drivers, and ~15% shift for drivers tested face-to-face. A maximum length signal (MLS) was used with a voltage of about 0.32VAC. The Behringer UCA222 does now allow adjustment of recording level, and 0.32VAC was the largest signal which would not result in clipping of the inputs.
Results:
Data from Woofer A in the horizontal position is suspect! The impedance curves clearly show the minimum impedance of driver A in the horizontal position, both with and without added mass, is ~6ohms. This is likely due to a bad connection. All other impedance curves are as expected. Some example impedance curves can be seen in figure #5. Another note: SpeakerWorkshop gives the user the choice of entering DC resistance manually, or allowing SW to calculate DC resistance from the impedance curves. SW’s calculation of DC resistance differed from that measured with a multimeter, and I found that using manually-entered DC resistance resulted in more consistent T/S parameters, so I opted to input DC resistance manually. This was not done in the first experiment.
The sixth figure is a chart showing all T/S parameters calculated by SpeakerWorkshop for all conditions tested. The seventh figure is simply a graphical representation of select data in picture #6.
The top graph in figure #8 shows simulated frequency-response curves of both drivers in a box volume (undamped) calculated to achieve a Qtc of 0.707. The chart in figure #9 is the box volume used to plot curves in said graph. The bottom graph in figure #8 shows simulated frequency-response curves of the same drivers keeping Vb=100 liters.
Conclusion:
The data are difficult in interpret and a bit perplexing. Testing the driver vertically clearly results in lower Fs, higher Vas and slightly higher Qms. Note that when tested face-to-face, modeled as one driver, these orientation-dependent differences disappeared. This suggests that whatever is causing the differences is cancelled by the cones moving simultaneously in opposite directions relative to their frames. It is curious that the hanging method produced results more closely matching the horizontal orientation than the vertical orientation, even though the driver can be seen hanging a little closer to vertical than horizontal. The results leave me wishing I had tested the back-to-back drivers hanging from the ceiling.
The big question:
So, how big should I build my box? The differences in Vb needed to obtain my desired Qtc of 0.707 are indeed huge for the different measurement methods. The differences also demonstrate that different parameters are, indeed not cancelling each other out. I plan to use the woofers horizontally in the face-to-face compound configuration, close to the floor. I’ve been warned by members of this forum that modeling the two woofers as one woofer and building according to those parameters is the wrong way to do it. I should just get the T/S parameters for “the” driver and cut the Vb in half. But what are the T/S parameters for my woofers?
Attachments
The isobaric push pull (face to face) arrangement that you plan is a different animal than either driver alone. The enclosed area between cones will also affect the new single driver’s compliance.
That said, half the size of a normal sealed box for that arrangement sounds correct, but you pay a big price in output level, hope you are partial to “girl with a guitar” type of music.
Looks like your vertical results average fairly close to the speaker’s usual TS parameters.
That said, half the size of a normal sealed box for that arrangement sounds correct, but you pay a big price in output level, hope you are partial to “girl with a guitar” type of music.
Looks like your vertical results average fairly close to the speaker’s usual TS parameters.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.