I made one cabinet with the mid driver hard mounted and with 12 g/l wool stuffing in the midrange box. The other speaker was made with a gasket of 2 mm thick closed-cell-foam and with 16 g/l wool stuffing. They both measured the same... which annoyed me. I was hoping to see a difference.
Yes it does, I sort of forgot about the factory data curve. The prototype box was made of plywood, and the foam board fit around the plywood box to form a simulated baffle. So there is not much difference in the physical construction of the two boxes.
No they are not the same, unfortunately. When I made the prototype I had a preliminary cabinet design, but the final design changed quite a bit.
In the prototype box, the two side walls are very close to the driver, while the top and bottom are about 2 inches away from the magnet. In the real cabinet midrange enclosure, the driver is located very close to three walls. I am suspicious that this could be a factor.
Thanks for all the comments ! .......... j.
The visualization helps a lot.
My problem solving approach goes like this.
Fold a sheet of paper lengthwise down the center. On the left side of the paper I jot down what I know. On the right side of the sheet I jot down what I want to know, do or fix. I also list the tools available to use, easy ones first
I want to flatten and or smooth the FR / resonance plus reduce distortion.
One thought is that all the SB Acoustics SB15 family of drivers use the same cast frame and will fit in the same enclosure front baffle cut out.
Another thought is the speed of sound is ~1130fps and the wave length of 2200Hz is about 6 in. Seems quite possible there are peeks and nulls bouncing around inside the mid enclosure and leaking out through the soft compliant rubber surround.
Thanks DT
Easy cheep stuff first:
I might try small white cotton stockings stuffed with Dacron pillow stuff placed in the mid-enclosure, kind of a Meta Material Helmholtz solution. Sort of like pillows in your living room.
My problem solving approach goes like this.
Fold a sheet of paper lengthwise down the center. On the left side of the paper I jot down what I know. On the right side of the sheet I jot down what I want to know, do or fix. I also list the tools available to use, easy ones first
I want to flatten and or smooth the FR / resonance plus reduce distortion.
One thought is that all the SB Acoustics SB15 family of drivers use the same cast frame and will fit in the same enclosure front baffle cut out.
Another thought is the speed of sound is ~1130fps and the wave length of 2200Hz is about 6 in. Seems quite possible there are peeks and nulls bouncing around inside the mid enclosure and leaking out through the soft compliant rubber surround.
Thanks DT
Easy cheep stuff first:
I might try small white cotton stockings stuffed with Dacron pillow stuff placed in the mid-enclosure, kind of a Meta Material Helmholtz solution. Sort of like pillows in your living room.
Last edited:
That isn't what I meant by compliant mounting. Foam tape is about sealing, the energy transfer from driver to box happens through the bolts or screws. If you screw straight into the wood the energy will transfer, like a thermal break is needed to avoid heat transfer. It might be difficult now to accomodate well nuts as the cabinet is already built. The wooden box being attached to the foam is a much less rigid structure than a box glued and fixed to another well braced and glued box.
If you want to test the inner wall theory I would try and get some automotive felt, also the same sort of stuff used in the cheap removalists blankets. Looks a purpley sort of colour. That is the best material I know of to have on the walls of a cabinet to damp the backwave. That in combination with foam, wool or fibreglass is very effective.
There is resonant behaviour from the driver in both measurements, it is seen in most off axis angles but changes frequency. Something about the final cabinet is anplifying it or not damping it as well. Prime candidates for that are going to be mounting, box construction and distances between structures that can create or amplify a frequency.
My understanding is the 45 degree facets is mostly for aesthetics in your Project. And when relative Small as in you Project, I would guess diffraction is more or less similar to flat baffle.
Maybe open cut side of foam core have a bit of absorbtion ? (Wild shoot and just a thought).
@hifijim: Im myself in the middle of preparing Project with same Dayton woofer. While testing I added bucking magnet.
I have no final conclusion. I just tested free Air with/with out magnet. So far I would say with extra magnet Sound was definitely Improved. Clearly more punch. I will do more testing also with measurements.
Maybe open cut side of foam core have a bit of absorbtion ? (Wild shoot and just a thought).
@hifijim: Im myself in the middle of preparing Project with same Dayton woofer. While testing I added bucking magnet.
I have no final conclusion. I just tested free Air with/with out magnet. So far I would say with extra magnet Sound was definitely Improved. Clearly more punch. I will do more testing also with measurements.
Yes that is true. The impedance curve for the driver shows a wrinkle at 2.2k on every curve for both drivers, including the free air impedance sweep with no cabinet. This means the source of the resonance is within the driver, and the most likely source is the rubber surround. As you said, somehow the two cabinets are interacting with this resonance in different ways, creating a peak that is 1 dB higher in the final cabinet.
At this point, there is no way to incorporate well nuts. The best I could do is isolate the driver with o-rings as @augerpro describes in his thread https://www.diyaudio.com/community/...n-methods-shootout-thread.356130/post-7111405
I don't have any o-rings or rubber washers of the right size, but next time I am at the hardware store I will pick some up.
Thanks for the suggestions.
Actually, the bevels have a noticeable effect in the tweeter range.
Here is a filter design that is getting close. There is still more refinement to do, but it is approximately on target.
The in room response (PIR) looks quite nice, as does the power response. The on-axis is nice.
The 2k bobble is mitigated somewhat by lowering the upper crossover from 3k down to 2.6k. The tweeter and mid responses complement each other this way, and smooth out the bobble a bit. Still, it's presence is a major factor in voicing the midrange. Visually, the graphs look ok, but I will have to do some careful listening to know for sure. I was very focussed on keeping the 2k peak lower than the 100-200 Hz fundamental region. By lowering the crossover to 2.6k, I sacrifice some of the DI performance, and now I have a small dip from 2.5k - 3.5k.
As always, I welcome all comments.
The in room response (PIR) looks quite nice, as does the power response. The on-axis is nice.
The 2k bobble is mitigated somewhat by lowering the upper crossover from 3k down to 2.6k. The tweeter and mid responses complement each other this way, and smooth out the bobble a bit. Still, it's presence is a major factor in voicing the midrange. Visually, the graphs look ok, but I will have to do some careful listening to know for sure. I was very focussed on keeping the 2k peak lower than the 100-200 Hz fundamental region. By lowering the crossover to 2.6k, I sacrifice some of the DI performance, and now I have a small dip from 2.5k - 3.5k.
As always, I welcome all comments.
Here is an interesting earlier version of the filter. The woofer and tweeter circuits are the same, but the midrange is different. It uses an amp side resistor, and this changes many things. The high pass section now has much larger series capacitor, and much smaller parallel inductor. The response is identical, as in perfectly identical, to the above filter (rev 2.07).
Here are the two filter responses... solid is this filter (rev 2.06), dashed is the current design (2.07)... as I said, identical.
The biggest drawback to the 2.06 filter is that the amp side resistor on the midrange wastes a lot of power, and causes the impedance to be unnecessarily low from 50 Hz to 1 kHz. I don't mind low impedance if it comes with higher sensitivity... but low impedance due to current flowing through resistors is not helpful.
So 2.06 filter is a dead end... but I thought it was interesting.
j.
Here are the two filter responses... solid is this filter (rev 2.06), dashed is the current design (2.07)... as I said, identical.
The biggest drawback to the 2.06 filter is that the amp side resistor on the midrange wastes a lot of power, and causes the impedance to be unnecessarily low from 50 Hz to 1 kHz. I don't mind low impedance if it comes with higher sensitivity... but low impedance due to current flowing through resistors is not helpful.
So 2.06 filter is a dead end... but I thought it was interesting.
j.
@hifijim ,
I hope that a little conversation about the QTc of the mid-range driver and enclosure is not to far off topic.
I am thinking that if the volume of the mid-enclosure is doubled from 6L to 12 L the QTc would go from 0.71 to 0.58, the pressure inside the enclosure due to Voice Coil excursion would decrease by half. The mid-range lower crossover frequency is well over two octaves above Fs. There will be near 0 impact to the mass loading of the driver.
The shape of the surround will have much less push pull effects caused by air pressure modulation inside the enclosure. There will be reduced interaction between the resonances of the cone and surround materials.
Thanks DT
I am thinking of placing a near field microphone right at the half round roll of the surround to measure distortions..
I hope that a little conversation about the QTc of the mid-range driver and enclosure is not to far off topic.
I am thinking that if the volume of the mid-enclosure is doubled from 6L to 12 L the QTc would go from 0.71 to 0.58, the pressure inside the enclosure due to Voice Coil excursion would decrease by half. The mid-range lower crossover frequency is well over two octaves above Fs. There will be near 0 impact to the mass loading of the driver.
The shape of the surround will have much less push pull effects caused by air pressure modulation inside the enclosure. There will be reduced interaction between the resonances of the cone and surround materials.
Thanks DT
I am thinking of placing a near field microphone right at the half round roll of the surround to measure distortions..
Last edited:
The amp side resistor won't waste any more power than a resistor after. It's highpassed, so it won't dissipate the full bandwidth.
Have you thought about the other midrange layout? Series cap and coil, then shunt cap and coil? Some people prefer this. I have a feeling cascading the tweeter and its filter after the midrange highpass could also be beneficial.
I also find that breakup peaks are easier to kill off with a combo of tank-cap across the lowpass coil, and then the LC across the driver. Same parts count as you have currently, and you still add resistance at breakup.
Have you thought about the other midrange layout? Series cap and coil, then shunt cap and coil? Some people prefer this. I have a feeling cascading the tweeter and its filter after the midrange highpass could also be beneficial.
I also find that breakup peaks are easier to kill off with a combo of tank-cap across the lowpass coil, and then the LC across the driver. Same parts count as you have currently, and you still add resistance at breakup.
Nice build thread! Have you considered impedance seen by the driver? IME mostly midrange could benefit, possibly woofer. It could lower distortion from drivers if impedance seen by driver is higher, but it varies between drivers. Not current drive, but same principle.
If I had to guess, your test cabinet wasn’t completely sealed so it behaved like an aperoidic vent. FWIW, I always add a vent to my mid enclosures. It should take that bump down a decibel or so. Worst case it doesn’t do anything….just seal up the vent and explore other options……but my bet is I’m correct.
Hello,
I often drill a 5/64 inch hole to place a 1/4 inch microphone inside a enclosure for in box pressure measurements.
At least a couple of times I have not remembered to seal the small round hole when the in box measurements were complete.
Sealed or not the 5/64 inch hole is too small to make a difference in the near field measurements.
No need to worry about a little enclosure leak, too much concern over nothing.
Thanks DT
Hello Jim,
SB’s measurement is on a baffle either IB, quasi IB or IEC.
You measurements is on 2 different enclosures.
So my hypothesis is that this resonance is related to the mounting of the driver.
One way to check if related to the enclosure/mounting is to do measurements at different drive levels. May I suggest Re-measurement at -20dB, -10dB, +10dB, max SPL
ie. 0.283V, 0.895V, 8.95V, max voltage (according to V = sqrt (P x R))
to see how closely the region of interest follows the expected dB loss or gain.
Another way is to do STEPS at different levels eg. 1dB steps from -20dB to + 16dB at the frequency of interest.
Only once one diagnoses the problem, then one can move onto management ie. various methods to try to mitigate the problem.
SB’s measurement is on a baffle either IB, quasi IB or IEC.
You measurements is on 2 different enclosures.
So my hypothesis is that this resonance is related to the mounting of the driver.
One way to check if related to the enclosure/mounting is to do measurements at different drive levels. May I suggest Re-measurement at -20dB, -10dB, +10dB, max SPL
ie. 0.283V, 0.895V, 8.95V, max voltage (according to V = sqrt (P x R))
to see how closely the region of interest follows the expected dB loss or gain.
Another way is to do STEPS at different levels eg. 1dB steps from -20dB to + 16dB at the frequency of interest.
Only once one diagnoses the problem, then one can move onto management ie. various methods to try to mitigate the problem.
Thanks for the thoughts, @tktran303 ... As I said in post #366, the impedance curve for the driver shows a wrinkle at 2.2k on every curve for both drivers, including the free air impedance sweep with no cabinet. This means the source of the resonance is within the driver, and the most likely source is the rubber surround. Somehow the two cabinets are interacting with this resonance in different ways, creating a peak that is 1 dB higher in the final cabinet. Your hypothesis is a mounting difference. The cabinets are not causing the resonance, but the real cabinet is exacerbating the resonance, or perhaps both cabinets are damping the resonance, but the prototype box is damping it a little better.
At this point, it is a rather academic concern. I don't think I can do anything about it without an unacceptable level of surgery on the cabinet. I will try out Fluid's idea of soft rubber o-rings / washers to isolate the driver from the screws, but beyond that, the cabinet is now a fixed variable.
However, I am skeptical that the structural mounting is a factor. I don't think this is a structural resonance issue. I think I would have detected a 2k prominence when I probed the cabinet with the stethoscope. Another bit of circumstantial evidence is the fact that the 2k resonance is very well damped. In the burst decay plot, it dies down below -30 dB in just a few cycles. Structural resonances are typically higher Q.
Instead, I think the acoustic energy in the cabinet is the cause of the difference. The prototype cabinet has more distance from the cone to the side wall than does the real cabinet. In essence, the driver has more "breathing room" in the prototype cabinet. This difference in acoustic energy behind the cone is affecting the resonance of the rubber surround.
In both the prototype and the real cabinet, I have stuffed the enclose space with loose fiber wool to 16 g/liter. That completely fills the space with the wool being slightly compressed. It is a very well damped acoustic. And yet I believe the close walls constrain the energy around the cone and surround. If the rubber o-rings make a difference, then we will know that my theory was wrong...
I consider the 2k resonance to be a manageable concern. My current passive filters reduce it to a 2 dB bobble peak to valley... which is not much of a concern. Of course in an active speaker, I would apply a DSP PEQ to reduce it to nothing, but in a passive speaker, there is a practical limit to how many notch filters we can apply.
====== ======
I am still refining the passive filter. I would like to thank @wolf_teeth and @A4eaudio for some interesting ideas to try.
At this point, it is a rather academic concern. I don't think I can do anything about it without an unacceptable level of surgery on the cabinet. I will try out Fluid's idea of soft rubber o-rings / washers to isolate the driver from the screws, but beyond that, the cabinet is now a fixed variable.
However, I am skeptical that the structural mounting is a factor. I don't think this is a structural resonance issue. I think I would have detected a 2k prominence when I probed the cabinet with the stethoscope. Another bit of circumstantial evidence is the fact that the 2k resonance is very well damped. In the burst decay plot, it dies down below -30 dB in just a few cycles. Structural resonances are typically higher Q.
Instead, I think the acoustic energy in the cabinet is the cause of the difference. The prototype cabinet has more distance from the cone to the side wall than does the real cabinet. In essence, the driver has more "breathing room" in the prototype cabinet. This difference in acoustic energy behind the cone is affecting the resonance of the rubber surround.
In both the prototype and the real cabinet, I have stuffed the enclose space with loose fiber wool to 16 g/liter. That completely fills the space with the wool being slightly compressed. It is a very well damped acoustic. And yet I believe the close walls constrain the energy around the cone and surround. If the rubber o-rings make a difference, then we will know that my theory was wrong...
I consider the 2k resonance to be a manageable concern. My current passive filters reduce it to a 2 dB bobble peak to valley... which is not much of a concern. Of course in an active speaker, I would apply a DSP PEQ to reduce it to nothing, but in a passive speaker, there is a practical limit to how many notch filters we can apply.
====== ======
I am still refining the passive filter. I would like to thank @wolf_teeth and @A4eaudio for some interesting ideas to try.
I have coated the cone to surround transition on several drivers with various glues etc, with various results, but there has alway been an improvement to the FR wiggle coming from the surround. Maybe something to try?
Can 6L cause any significant issues with a 400Hz HP? >110dB SPL at 400Hz has a 5" moving about +/-1mm for ~10cc each way. That's under 0.2% Vb delta.
Jim,
I believe this is the most practical too. Please consider double decoupling - a rubber gasket between the frame and cabinet (if you haven’t already) and rubber o-rings between the screws and the frame. Be careful not to over-tighten