Amphion does this purposefully in some of their designs in order to create a hybrid cardioid bass response with many small vents in a specific pattern on the sides of the enclosure nearest the bass drivers. So in theory, you could create a minor cancellation If you co located the wavefronts of the driver and vent. Placing the vent on the rear of the enclosure eliminates that possibility. Using 4 small vents placed in each rear corner of the enclosure might effectively reduce some resonance and modes though if you wanted to experiment.
I’ve done a lot of research about passive cardioid speakers in the past six months or so, but I didn’t really make the connection between that and aperiodic. Some also add vents near the driver on the baffle if the goal is to also add mid/low-mid directivity control. Sigberg Audio does this with their Manta and documents the design on ASR. I think that cardioid frequency response is a bit closer to high tuned ports with moderately high port losses compared to aperiodic frequency response which use very high port losses to damp a high Q knee. So basically the difference appears to be location and and amount of damping.
Four in the corner: Are you talking about a novel way to deal with standing waves?
Attachments
OK, I won’t derail this tread any longer; I hate how that always happens. I was hoping for feedback on the use of these two drivers, which I got, but it quickly took a turn into a discussion of modeling software and enclosure types. Thanks for your help guys! I’ll follow up when I build something.
That was my thinking if the goal was purely the aperiodic function….why not make it count on as many design fronts as possible. Take the surface area required my your model, divide it by four and proceed.
I ran a couple sims in PCD using traced manufacture data if anyone’s interested in seeing how the SB 8 inchers can work together. I also show how just the SB20FRPC looks with a baffle step compensation circuit. Sims, graphs and measurements are about as rare as a Sasquatch sighting in these full range parts, so I thought these might help someone understand what’s going on with these drivers in a real baffle and with various circuits in place. These include minimum phase, baffle diffraction and enclosure effects. The baffle I used is 14.5” wide by 23” tall.
As far as the frequency response graphs go, the top black line is the sum of the SB20FRPC and the SB20PFCR with the SB20PFCR rolled off 1st order. The middle grey line is the SB20FRPC as a full range by itself with a baffle step circuit in line. The lower grey line is system phase for the 1.5-way which has minimal error and exactly matches that of the corrected single driver alone.
The second picture is just to show the filter I used to hit a true first order target for the woofer. Ignore the .01mH value. That’s just a placeholder so that I can implement the XO/elliptical notch shown directly below. I also included a zobel network. It’s not absolutely necessary, but without it, a bit more of the upper end hash of the woofer contributes to the treble.
So, it looks like you can end up with 3-4db higher sensitivity and exactly the same frequency response and phase as a the SB20FRPC with a baffle step circuit. You also get double the cone area for bass, higher maximum output and reduced distortion for both drivers. The minimum impedance is about 4.1 ohm at 165Hz, max is 8 ohm or so and the electrical phase is basically a straight line. This should be a very easy speaker for any amp to drive. The cost of this is about $70 per side, a larger enclosure and minimal additional effort.
As far as the frequency response graphs go, the top black line is the sum of the SB20FRPC and the SB20PFCR with the SB20PFCR rolled off 1st order. The middle grey line is the SB20FRPC as a full range by itself with a baffle step circuit in line. The lower grey line is system phase for the 1.5-way which has minimal error and exactly matches that of the corrected single driver alone.
The second picture is just to show the filter I used to hit a true first order target for the woofer. Ignore the .01mH value. That’s just a placeholder so that I can implement the XO/elliptical notch shown directly below. I also included a zobel network. It’s not absolutely necessary, but without it, a bit more of the upper end hash of the woofer contributes to the treble.
So, it looks like you can end up with 3-4db higher sensitivity and exactly the same frequency response and phase as a the SB20FRPC with a baffle step circuit. You also get double the cone area for bass, higher maximum output and reduced distortion for both drivers. The minimum impedance is about 4.1 ohm at 165Hz, max is 8 ohm or so and the electrical phase is basically a straight line. This should be a very easy speaker for any amp to drive. The cost of this is about $70 per side, a larger enclosure and minimal additional effort.
Attachments
Hello!
I’d like to build a tqwt/transmission line box. If I understood well you are using this design:
In the past I used this simulator and for this speaker I got this design:
Is there anybody that can suggest me if your design or the one got from simulator is better?
Thanks for your support!
Carlo
I’d like to build a tqwt/transmission line box. If I understood well you are using this design:
In the past I used this simulator and for this speaker I got this design:
Is there anybody that can suggest me if your design or the one got from simulator is better?
Thanks for your support!
Carlo
Attachments
If you want to confirm a given design with alternate dimensions then I would suggest to use Hornresp and do the simulation on your own. I'm saying this not blame you but if you want to tweak some more you can simply do it directly on your own and ask more specifically if your changes are noteworthy or not as expected.
FYI
from HiFi-Forum.de
https://www-hifi--forum-de.translat...&_x_tr_hl=en-US&_x_tr_pto=wapp&_x_tr_sch=http
from HiFi-Forum.de
https://www-hifi--forum-de.translat...&_x_tr_hl=en-US&_x_tr_pto=wapp&_x_tr_sch=http
I finally dredged up the screenshot I took of the shaping circuit I used. BSC with a notch to tame the 3-7k shout. Easy to adjust by swapping out resistors. R2 is probably best between 4-7ohm depending how much you like that boosted 3-7k range. So I may have ended up changing the resistor values before finally locking it in. For 13.25" (~336mm) wide baffle I found the BSC coil would be fine between 2.5-3.5mH. Up to 3.5mH gets rid of the drooped upper bass a bit better, just depends how much db/watt overall you are willing to sacrifice and what sounds good to you. I stuck with the 2.7mH for now.
Last edited:
Looks good to me…..if you can reach your desired SPL and keep the fullrange excursion under control I consider that a win. I’ve always been a fan of the .5 alignment and it’s had some added bonuses when used in a typical MTM where the .5 inductor is on the bottom woofer instead of a straight up parallel wiring……..the system retains some point source character that otherwise gets swallowed up by lobing in the midrange.
I used a 3.2 mH and 6.8 Ohm the values I have at home (calculated values for a 17" wide baffle is 3.3 mH and 5.6 Ohm). I used Dayton Omnimic to measure the response at 1 m in a rather probematic room.
Kudos to SB! The two drivers measure the same and the results are consistent with the published curves!
The balance gets better with the baffle step
I will also try the other part of the filter to tame 3-7 kHz peaks
Kudos to SB! The two drivers measure the same and the results are consistent with the published curves!
The balance gets better with the baffle step
I will also try the other part of the filter to tame 3-7 kHz peaks
I tried the suggested LCR trap and it does reduce the two peaks but it also deepens the dip between them. I might load the whole thing in Xsim and have two high Q filters one aiming at 3.5 kHz and one at 5 kHz.
The LCR trap did improve the sound! When I used a LCR trap on a Goodman Axiette Alnico the frequency response improved but not the sound, rather the opposite in fact.
Very nice drivers those SB. I might build Hobby HiFI Jericho 2023 horns for them...
The LCR trap did improve the sound! When I used a LCR trap on a Goodman Axiette Alnico the frequency response improved but not the sound, rather the opposite in fact.
Very nice drivers those SB. I might build Hobby HiFI Jericho 2023 horns for them...
I run SB20FR in OB with a 15" woofer, crossed at around 300Hz 1st order.
No sign of any of these peaks and dips when taking sweeps with Spectroid, an Android app.
In my experience LCR traps are only good in theory, if one wants the SPL graph to look flat, not so much when the sound is concerned.
No sign of any of these peaks and dips when taking sweeps with Spectroid, an Android app.
In my experience LCR traps are only good in theory, if one wants the SPL graph to look flat, not so much when the sound is concerned.
The best way to decide if it needs attention is probably to measure multiple spots in room and average them. Then see if it’s even still there. Probably not once beaming of an 8” driver is considered.
Measurements of distortion and energy storage(ringing) might tell you if it’s benign or needs attention regardless.
Lastly, I’d say that if you’re experiencing brightness or thinness then it’s probably a lack of baffle step comp. Not that single band over 5k
Measurements of distortion and energy storage(ringing) might tell you if it’s benign or needs attention regardless.
Lastly, I’d say that if you’re experiencing brightness or thinness then it’s probably a lack of baffle step comp. Not that single band over 5k
Here are some measurement, for some reason the scale autocorrected to 4 dB instead of 3dB so it flattanened the curves a bit.
The graph is bitmapped not vector graphics, and thanks to an old PC laptop with a low res screen this really shows.
I agree with Satx a peak that is only present at really straight on axis but not 15 degrees offaxis onwards, then reducing that peak on axis will cause a dip in the overall response and thus not an improvment but quite the opposite.
With music the LCR filer is a definite improvement but a double trap at 4 and 6 k might be eveb better
The graph is bitmapped not vector graphics, and thanks to an old PC laptop with a low res screen this really shows.
I agree with Satx a peak that is only present at really straight on axis but not 15 degrees offaxis onwards, then reducing that peak on axis will cause a dip in the overall response and thus not an improvment but quite the opposite.
With music the LCR filer is a definite improvement but a double trap at 4 and 6 k might be eveb better
