Well here is an interesting one. I made a ring that extends the driver radius by 30mm increasing the diameter to 210mm. I compared this driver with a pair of drivers spaced 60mm and the results are quite striking. First, pictures of the two setups.
And now the results. Single driver is red, clamshell dual driver is green:
It's difficult to recommend a clamshell arrangement that requires a second driver over (in this case) a 30mm ring of 3mm MDF. Remember the driver with the ring of MDF was drawing 1/2 the power from the amplifier. Had it been a 4R driver and been compared with the pair of 8R drivers it would have had a 3dB advantage over the clamshell arrangement with the same input power.
This result confirms my thoughts about how the clamshell arrangement behaves.
And now the results. Single driver is red, clamshell dual driver is green:
It's difficult to recommend a clamshell arrangement that requires a second driver over (in this case) a 30mm ring of 3mm MDF. Remember the driver with the ring of MDF was drawing 1/2 the power from the amplifier. Had it been a 4R driver and been compared with the pair of 8R drivers it would have had a 3dB advantage over the clamshell arrangement with the same input power.
This result confirms my thoughts about how the clamshell arrangement behaves.
Attachments
Last edited:
Another experiment. This time I measured 1,2,3 then 4 drivers, all with the same 90mm spacing with the front driver facing cone towards the mic, as seen below:
And the results. This time I included the dipole peak from these small drivers. No surprises here.
Note this plot is 5dB/
Ok so, not very scalable it seems.
2 dB more max by doubling from 2 to 4 drivers does not seem to be wurthwhile
Great work guys - and William, thank you so much for taking the time to do all these experiments - so interesting!
I completely agree with you comments above and seems to again align with SL who said ' yeah they sum and two is better than one, but it is still a lot less efficient that just using a baffle.
Thinking about the scaling of your experiment above - presumably the ring would have to be scaled in a similar way on a 15" driver? I.e. the increase in size in baffle would be proportional to the driver diameter?
I completely agree with you comments above and seems to again align with SL who said ' yeah they sum and two is better than one, but it is still a lot less efficient that just using a baffle.
Thinking about the scaling of your experiment above - presumably the ring would have to be scaled in a similar way on a 15" driver? I.e. the increase in size in baffle would be proportional to the driver diameter?
In other words....you have increased the driver diameter by 40% above, so I would need a 21" ring around my 15" sub to do the same thing?
(Hope this makes sense!)
(Hope this makes sense!)
If you currently have a 100mm gap, you will get the same result with one driver and a ring that increases driver diameter by 100mm.
Go the whole hog and put the driver in a 200mm deep H frame, plus the second one above it. You'll gain more than 10dB over your current setup. This is what I would do.
Above the 15"s keep the other drivers nude.
Go the whole hog and put the driver in a 200mm deep H frame, plus the second one above it. You'll gain more than 10dB over your current setup. This is what I would do.
Above the 15"s keep the other drivers nude.
Ok so, not very scalable it seems.
2 dB more max by doubling from 2 to 4 drivers does not seem to be wurthwhile
Umm, it looks interesting?! 🙄
This arrangement doesn't achieve much acoustically. The marketing department might like it though.
Thanks and very interesting!
I would not have room for two stacked in the man cave. And current gap is 15cm, so I would need a similar scaled baffle.
So are we finally concluding here that we can indeed gain 6dB from this configuration when optimized, which is more than we had previously thought (we were only expecting the 3dB from doubling of power when we first did a measuremnt), but this is obviously far less efficient that a decent H baffle? Of more traditional configuration? (I.e. pretty much what SL described.)
I would not have room for two stacked in the man cave. And current gap is 15cm, so I would need a similar scaled baffle.
So are we finally concluding here that we can indeed gain 6dB from this configuration when optimized, which is more than we had previously thought (we were only expecting the 3dB from doubling of power when we first did a measuremnt), but this is obviously far less efficient that a decent H baffle? Of more traditional configuration? (I.e. pretty much what SL described.)
My conclusion is that you can increase spacing up to somewhere around driver diameter. This will then behave in a similar manner to a dipole that has a baffle that is twice driver diameter. This will be true at low frequencies only. The larger drivers and larger spacings will make things fall apart at lower frequencies than my tests showed with those small Scan Speak drivers. I was getting some really nasty peaks and dips as I moved the drivers around. They were very audible. I never gained 6dB with two drivers.
I would put your drivers into an arrangement similar to what SL used in the LX521 and pick up 10dB over what you have now and call it done.
I would put your drivers into an arrangement similar to what SL used in the LX521 and pick up 10dB over what you have now and call it done.
But the saving grace of the clamshell arrangement is that woodworking is not required, any cavity resonance issues are avoided, it's more compact - and bass watts are cheap. Plus, low-order distortion is arguably far less important at low frequencies.
Thanks for your time and analysis!
My drivers were £100 each, so the cost and time to build similar enclosures to SL isn't worth it for me. Also the clamshell allows for the nice isolating sling suspension - which is a fantastic bonus for mechanical coupling isolation, and something that would be more difficult without the 'clamshell'.
I appear to have a good 5-6dB increase in output as measured (you have achieved 4.5-5dB in yours and others have found similar) - which as you and SL both highlight - is not a very efficient use of the extra amp channels and drivers compared to a 'normal OB arrangement' - but is a nice, unexpected bonus, and I have both to spare!
I don't need a further 10dB headroom either with these having such high VD - they go much louder than I require in my man cave and dig deep - as my measurements show!
As I have said all along - this arrangement works for me - my man cave is not very large, and I have only just managed to squeeze naked 15" drivers into the space - I like the clamshell aesthetics too which is nice🙂
It appears then we can draw a line under this?
There is nothing new under the sun!
My drivers were £100 each, so the cost and time to build similar enclosures to SL isn't worth it for me. Also the clamshell allows for the nice isolating sling suspension - which is a fantastic bonus for mechanical coupling isolation, and something that would be more difficult without the 'clamshell'.
I appear to have a good 5-6dB increase in output as measured (you have achieved 4.5-5dB in yours and others have found similar) - which as you and SL both highlight - is not a very efficient use of the extra amp channels and drivers compared to a 'normal OB arrangement' - but is a nice, unexpected bonus, and I have both to spare!
I don't need a further 10dB headroom either with these having such high VD - they go much louder than I require in my man cave and dig deep - as my measurements show!
As I have said all along - this arrangement works for me - my man cave is not very large, and I have only just managed to squeeze naked 15" drivers into the space - I like the clamshell aesthetics too which is nice🙂
It appears then we can draw a line under this?
There is nothing new under the sun!
It's great to see so many people get involved in this and experimenting.
A couple of things have been running in my mind, though.
1- Can we really compare a clamshell of two 15" subs with fs of 23Hz, to what looked like a clamshell of two 4" drivers, with a fs of probably something like 100Hz? I might have missed a post explaining the why...
2- Since with the subs clamshell, the frames are bolted to each other, they act as one solid unit. Placing two drivers in front of each other, not linked together, might have some different interactions?
A couple of things have been running in my mind, though.
1- Can we really compare a clamshell of two 15" subs with fs of 23Hz, to what looked like a clamshell of two 4" drivers, with a fs of probably something like 100Hz? I might have missed a post explaining the why...
2- Since with the subs clamshell, the frames are bolted to each other, they act as one solid unit. Placing two drivers in front of each other, not linked together, might have some different interactions?
Nice
Having seen the latest measurements my temptation to do more has decreased substantially.
If I may attempt a summary:
Bushmeister observed approximately +6dB on-axis near field with the clamshell arrangement.
The Linkwitz model shows that +6dB is expected (and provides a model covering all angles), but makes certain approximations, most significantly: there are two strictly-dipole sources (four point sources). It also implies that the sources are not otherwise coupled: there's no back-action from the sound that affects the source.
The question of whether driver coupling is significant, in the sense of an efficiency change at the driver level was raised. I haven't evidence in support of that at a level that makes it hard to answer the original +3dB/+6dB question.
Some (I for one) thought that, measuring so close to the drivers, there may be some obstruction of the sound from one driver blocking the sound from the other leading to something less than +6dB on-axis, close in - and by conservation of energy, other places where the sound pressure is higher than expected.
Taking all the measurements into account - I hope I am, it's not surprising that the original clamshell shows +6dB, i.e. the Linkwitz model is a good approximation. Perhaps not surprising from the start, given the expertise involved.
One could find situations where the Linkwitz model does not apply, due to sufficiently obstructed cones or tightly-coupled drivers, but that's diverging from the original question.
Is that about right?
Ken
Having seen the latest measurements my temptation to do more has decreased substantially.
If I may attempt a summary:
Bushmeister observed approximately +6dB on-axis near field with the clamshell arrangement.
The Linkwitz model shows that +6dB is expected (and provides a model covering all angles), but makes certain approximations, most significantly: there are two strictly-dipole sources (four point sources). It also implies that the sources are not otherwise coupled: there's no back-action from the sound that affects the source.
The question of whether driver coupling is significant, in the sense of an efficiency change at the driver level was raised. I haven't evidence in support of that at a level that makes it hard to answer the original +3dB/+6dB question.
Some (I for one) thought that, measuring so close to the drivers, there may be some obstruction of the sound from one driver blocking the sound from the other leading to something less than +6dB on-axis, close in - and by conservation of energy, other places where the sound pressure is higher than expected.
Taking all the measurements into account - I hope I am, it's not surprising that the original clamshell shows +6dB, i.e. the Linkwitz model is a good approximation. Perhaps not surprising from the start, given the expertise involved.
One could find situations where the Linkwitz model does not apply, due to sufficiently obstructed cones or tightly-coupled drivers, but that's diverging from the original question.
Is that about right?
Ken
But we still don't have an understanding of the mechanism involved here.
At least I don't...
I explained it a few posts back, but I'll try again.
When an unbaffled driver moves its cone forwards it pressurises the air on that side. If only one driver is used this high pressure air moves around to the other side of the cone to the low pressure region that is there. If two drivers are used in a clamshell, the low pressure region is not on the other side of that cone, it is on the far side of the furthest driver. The volume between the cones is at atmospheric pressure and doesn't change appreciably during operation.
When the path between the high pressure and low pressure sides of a driver in a dipole arrangement is increased the output of that dipole also increases. You'll read more about this on SL's site and elsewhere. What also happens is the dipole peak and first null goes down in frequency. All of these phenomena occured when I increased the baffle diameter and when I clamshelled the drivers. Using these two methods I increased the pathlength by 60mm and ended up with identical results. In my mind this proves my hypothesis correct.
Agree
a nice way to explain why the Linkwitz model applies and, with the usual limits on
"2 d2" etc. there's no change relative to the equivalent single dipole other than the "2" (the way SL set up the math).
I've learned how unimportant obstructions are at the frequencies of concern, in the near field, as well as in the far field.
Ken
a nice way to explain why the Linkwitz model applies and, with the usual limits on
"2 d2" etc. there's no change relative to the equivalent single dipole other than the "2" (the way SL set up the math).
I've learned how unimportant obstructions are at the frequencies of concern, in the near field, as well as in the far field.
Ken
I explained it a few posts back, but I'll try again...
Found it, post 324. Yes, you should have made more of that at the time, given that it's the key to the whole problem.
Ditto Keith - I missed it too - but it has been a very busy thread to keep up with!
I was listening for 3 hours last night with a mate and a bottle of good red.
I stand by my initial assessments of this speaker.
Just outstanding! Goosebumps all over on a few tracks.
The OB bass is amazing, as is the whole set up.
So clean and crisp.
I was such an OB doubting Thomas prior to this build - but to SL and JK and all the other OB pioneers here - thank you so much!
I was listening for 3 hours last night with a mate and a bottle of good red.
I stand by my initial assessments of this speaker.
Just outstanding! Goosebumps all over on a few tracks.
The OB bass is amazing, as is the whole set up.
So clean and crisp.
I was such an OB doubting Thomas prior to this build - but to SL and JK and all the other OB pioneers here - thank you so much!
It's great to see so many people get involved in this and experimenting.
A couple of things have been running in my mind, though.
1- Can we really compare a clamshell of two 15" subs with fs of 23Hz, to what looked like a clamshell of two 4" drivers, with a fs of probably something like 100Hz? I might have missed a post explaining the why...
I used the 15W/8530K01 with an FS of 38Hz and a Qt of 0.5. There are some differences and some similarities. I have some perfect for an OB 15" drivers here but I didn't care to drag them outside in the rain for some full sized testing which would't have taught me any more than these drivers did. The tests were not without merit.
2- Since with the subs clamshell, the frames are bolted to each other, they act as one solid unit. Placing two drivers in front of each other, not linked together, might have some different interactions?
The drivers were not moving in my tests with their 13g mms. Bolting them together would have made little difference. I made little wooden feet for them that kept them quite stable on my MDF test bench.
If my tests have too many limitations for you I encourage you to test a pair of 15" drivers outdoors and post your results. I personally look forward to seeing how your results differ from mine. I believe I've covered expected differences already.
- Home
- Loudspeakers
- Multi-Way
- The 'Circles of Doom’... Open baffleless full range speakers