Tang Band W5-1138SMF 6th (or 4th) order bandpass
Hi all,
I've never built a 6th or 4th order bandpass before so I'm hoping for advice from anyone with experience of building one.
I've seen that port diameters are roughly double a regular reflex but otherwise, it seems to behave like two reflex loaded speakers - one rear and one front - in terms of calculations. Is that about right? Did your calculations produce the predicted result?
If you built one, did it meet your expectations or was a regular reflex a better idea?
Did you use ports or passive radiators?
Below is the WinISD plot for a single Tang Band W5-1138SMF driver. Is this prediction at all realistic?
The front enclosure is 21 litres, 27Hz, which needs a 10cm diameter port 146.5cm long. The rear chamber is even longer I think, so I will have to use 6.5" passive radiators for this project. I would prefer 8" but they would require too much extra weight, so I'd be concerned about sag.
I'm also curious about doing this as an isobaric pair - roughly double the weight on the passives but almost half the cabinet size.
Any building tips would be greatly appreciated
Thanks.
Hi all,
I've never built a 6th or 4th order bandpass before so I'm hoping for advice from anyone with experience of building one.
I've seen that port diameters are roughly double a regular reflex but otherwise, it seems to behave like two reflex loaded speakers - one rear and one front - in terms of calculations. Is that about right? Did your calculations produce the predicted result?
If you built one, did it meet your expectations or was a regular reflex a better idea?
Did you use ports or passive radiators?
Below is the WinISD plot for a single Tang Band W5-1138SMF driver. Is this prediction at all realistic?
An externally hosted image should be here but it was not working when we last tested it.
The front enclosure is 21 litres, 27Hz, which needs a 10cm diameter port 146.5cm long. The rear chamber is even longer I think, so I will have to use 6.5" passive radiators for this project. I would prefer 8" but they would require too much extra weight, so I'd be concerned about sag.
I'm also curious about doing this as an isobaric pair - roughly double the weight on the passives but almost half the cabinet size.
Any building tips would be greatly appreciated
Thanks.
Last edited:
Hi Karl,
Yes, a narrow passband - just 1.5 octaves it seems - and no ripple or gain/loss (0dB with -3dB at 25Hz and 73Hz, -6dB at 23Hz and 85Hz) so it's flat in the passband. It took a while to get that right in WinISD That range is exactly what I need to pair with 3" 15W satellites with AMT tweeters - they're 87dB/W. As a result, I don't need lots of SPL and I love the idea of not needing a crossover on this speaker - I can use an active 12dB/octave adjustable low pass before the amp and that will be fine. The lack of big inductors/caps should lift the sound quality if the bandpass design/build is done right I hope.
Yes, a narrow passband - just 1.5 octaves it seems - and no ripple or gain/loss (0dB with -3dB at 25Hz and 73Hz, -6dB at 23Hz and 85Hz) so it's flat in the passband. It took a while to get that right in WinISD
That range is exactly what I need to pair with 3" 15W satellites with AMT tweeters - they're 87dB/W. As a result, I don't need lots of SPL and I love the idea of not needing a crossover on this speaker - I can use an active 12dB/octave adjustable low pass before the amp and that will be fine. The lack of big inductors/caps should lift the sound quality if the bandpass design/build is done right I hope.
Last edited:
Add up the cost of that driver and the two PRs you plan to use with it. I'll bet that you can find a better driver at that cost which can be used in a simple sealed or vented box that's the same net volume of that 6th order BP design.
To answer your question about resonances, you shouldn't get vent resonances of you use PRs. Looking at your proposed build from a cost and complexity perspectives though, it might be better to go with a better driver in a simple vented box.
To answer your question about resonances, you shouldn't get vent resonances of you use PRs. Looking at your proposed build from a cost and complexity perspectives though, it might be better to go with a better driver in a simple vented box.
Thanks - that's what I thought based on previous experience of sub building.
The driver costs what most "inexpensive" sub drivers cost. The PRs cost ~US$3 each on Taobao. I'll need one to cut up so I can weigh the cone/surround/spider separated from its basket so I can accurately add weight but "wasting" US$3 won't hurt. How much is a 5mH inductor?
?? 6.5?????????? ????? ???-???
I understand the compromises of a 6th BP, which is why I wanted 0dB gain. As I understand it, narrower bandwidth = more gain = more group delay = worse. Is it more complicated than that? "Don't do it" is not that helpful tbh. But thanks anyway ;-)
BTW - the site in your link has some useful maths - does this prove reliable when actually building one? I guess variations in T/S etc means it isn't wholly reliable?
Last edited:
No need to do that. Just use it as-is and add weight until you achieve the target resonance frequency as specified in the sim.
The maths will get you into the ballpark, but using a good simulation program will be better. Even then, prepare to have to tweak it even further to get the best results.
FWIW, my first build was a 6th order BP design, a replica of the old Bose AM5 series subwoofer. That was way before the Internet, so no info about it on my website. These days my preference lies with offset TLs.
GD will be worse with a 6th order BP build. Depending on the frequency range you're trying to cover with it, that might not be all that noticeable.
I've looked at group delay (GD) on WinISD and a reflex loading has the lowest group delay until it is about 1 octave from the tuned resonance when it starts to spike (in this case tuned to 35Hz, so the rise starts at 70Hz). That spike is almost double the group delay of a 4th order bandpass at the same frequency. The 4th order has more delay above that frequency but it isn't terrible, so reflex vs 4th order is really a trade off between a large spike in delay vs a moderate but broader spread of delay. It's also possible to tune a 4th order to have the same group delay curve as a reflex. Bear in mind this is a sub for 90Hz and below, and the advantage of the reflex's lower group delay only at >70Hz is lost on me. The models I've done for 6th order have the same group delay as the 4th over most of the range until they hit their lower resonant frequency. This is when they spike and it's worse than a reflex. In this case, that's 25Hz but there is so little musical content at such low frequencies that I wonder if that's really a concern. Also worth considering that only the 6th order design produces output so low - the other designs do not - so that's a "muddy or inaudible" choice. So to my mind, the main advantage of a 4th order is really the easier build. I am happy with the challenge of a 6 but a 4 might make more sense overall.
Last edited:
Here's a graph of group delays for the W5 1138 in different designs. They can be changed substantially so it's just one example of many possible. Perhaps the 4th order is the best choice overall though...
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Thx, aim to please
It's beyond the software I have to simulate the group delay and other effects of a second order crossover, which for 100Hz 4 ohm is 12.8mH and 200uF. However, I've seen others measure it and say it is around 4ms for 100Hz. Maths suggests it would be 5ms (180degrees of phase shift = 1/2 wavelength = 5ms @100Hz I think). Plus such a huge coil will have to have an iron core, which saturates/distorts, and 200uF would need to be an electro cap and they sound bad too. They both work as dampers, storing and releasing energy, giving the amp a hard time, etc. Overall, subwoofer crossovers are bad news with almost nothing in their favour over other options. So I'm curious about a bandpass and would really like to hear from people who have built one - so far that's Brian and it seems he doesn't think it's worth the effort. Perhaps an electronic 24dB/octave filter is the only satisfactory, simple and cheap way to do this.
It's beyond the software I have to simulate the group delay and other effects of a second order crossover, which for 100Hz 4 ohm is 12.8mH and 200uF. However, I've seen others measure it and say it is around 4ms for 100Hz. Maths suggests it would be 5ms (180degrees of phase shift = 1/2 wavelength = 5ms @100Hz I think). Plus such a huge coil will have to have an iron core, which saturates/distorts, and 200uF would need to be an electro cap and they sound bad too. They both work as dampers, storing and releasing energy, giving the amp a hard time, etc. Overall, subwoofer crossovers are bad news with almost nothing in their favour over other options. So I'm curious about a bandpass and would really like to hear from people who have built one - so far that's Brian and it seems he doesn't think it's worth the effort. Perhaps an electronic 24dB/octave filter is the only satisfactory, simple and cheap way to do this.
Last edited:
I've done a lot more reading and modelling and I'm going for a 4th order. If I manage to build it correctly, I should be able to get better or matching performance than a reflex in terms of SPL over the bandwidth I need. Plus group delay is less than a reflex+crossover, and more consistent because it can avoid the rapid rises and huge delays near the tuning frequency. It also has less dramatic low end roll off, while the top end roll is tailored to my needs.
So I've ordered some parts... no going back now ;-)
So I've ordered some parts... no going back now ;-)
Most of the parts have arrived, just waiting on a pre-built 12 litre enclosure to use as a test mule.
I weighed all four of the PRs and they vary by +/-2g around 120g. I cut one apart and the full cone, spider and surround together weigh 18g. I cut down the spider and surround to 1/3 so 1/3 spider + 1/3 surround + cone weighs 9g - a realistic cone weight to add to - target cone weight for the design is 88g - so adding 79g is fine. The PR diameter is 12.6cm including 1/3 of the rubber surround. To measure displacement, I have a constant diameter bucket of 19.4cm so I fixed on a ruler to the inside and measured the height change of the water in the bucket. The cone displaces 0.3cm and the basket 0.1cm. That gives displacement as 0.4*3.14*9.7*9.7=118 removed from the front volume. The hole it mounts in will add volume but I'm still waiting on the box to measure that.
The Tang Band driver will be mounted flush to a panel a few cm behind the PR - there'll be just enough space so they won't touch at xmax. The driver's fat rolled surround will take up a tiny amount of front volume but the slightly concave recessed cone will add a little so overall, the driver's net effect on enclosure volume is too small to be concerned about.
In an effort to minimise group delay to get the cleanest sound, I'm going to aim for a small front enclosure, built to be 2.1 litres. To correct any error, I guess I can add some wadding to boost the effective volume, or add solid blocks to lower it. While the small volume lowers group delay, it also lowers efficiency and means more weight on the PR to tune the front enclosure to match the rear enclosure's resonant frequency. Matching the two resonant frequencies gives the flattest response curve. Efficiency is down 5dB from "standard" and the driver is only 82dB/W with a 40W continuous rating. It won't be loud at max power but should be similar SPL as the little satellites in the low 90s.
So, waiting on the box.
I weighed all four of the PRs and they vary by +/-2g around 120g. I cut one apart and the full cone, spider and surround together weigh 18g. I cut down the spider and surround to 1/3 so 1/3 spider + 1/3 surround + cone weighs 9g - a realistic cone weight to add to - target cone weight for the design is 88g - so adding 79g is fine. The PR diameter is 12.6cm including 1/3 of the rubber surround. To measure displacement, I have a constant diameter bucket of 19.4cm so I fixed on a ruler to the inside and measured the height change of the water in the bucket. The cone displaces 0.3cm and the basket 0.1cm. That gives displacement as 0.4*3.14*9.7*9.7=118 removed from the front volume. The hole it mounts in will add volume but I'm still waiting on the box to measure that.
The Tang Band driver will be mounted flush to a panel a few cm behind the PR - there'll be just enough space so they won't touch at xmax. The driver's fat rolled surround will take up a tiny amount of front volume but the slightly concave recessed cone will add a little so overall, the driver's net effect on enclosure volume is too small to be concerned about.
In an effort to minimise group delay to get the cleanest sound, I'm going to aim for a small front enclosure, built to be 2.1 litres. To correct any error, I guess I can add some wadding to boost the effective volume, or add solid blocks to lower it. While the small volume lowers group delay, it also lowers efficiency and means more weight on the PR to tune the front enclosure to match the rear enclosure's resonant frequency. Matching the two resonant frequencies gives the flattest response curve. Efficiency is down 5dB from "standard" and the driver is only 82dB/W with a 40W continuous rating. It won't be loud at max power but should be similar SPL as the little satellites in the low 90s.
So, waiting on the box.
Last edited:
All the stuff arrived so I built it - didn't take that long - the box was already made and the internal partition fitted very well and was easy to fix in place with a few screws and EVA sealing tape. A first test was disappointing though. The sound quality was fine, and the range was good too only a little bit of the "one note bass" effect of a narrow bandwidth. However, the output was much lower than expected. So I moved the internal partition to increase front enclosure volume and that increased the output but the sound wasn't good - noticeably a "one note bass". It seems to me the software predictions understated the effects of the various compromises.
So then I removed the partition and made it into a 11 litre reflex with a 6.5" PR that has a cone mass of 55grams. That works really well. It needs a filter - either a 4th order, which would be 5mH, 150uF, 1.2mH, 24uF, or an electronic filter before the amp.
Experiment over. And what can I say except Brian you were right.
So then I removed the partition and made it into a 11 litre reflex with a 6.5" PR that has a cone mass of 55grams. That works really well. It needs a filter - either a 4th order, which would be 5mH, 150uF, 1.2mH, 24uF, or an electronic filter before the amp.
Experiment over. And what can I say except Brian you were right.
- Status
- Not open for further replies.