Dipole design -will this work?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello All,
I can't stand it any longer - after what I've read here (and other places) I've just got to try building a dipole and hear what it's all about. The way I though of is this. I'll remove the drivers from my existing speakers (HM170Z18 and SEAS 27TFFC) plus existing x-o, buy a veneered shelving board (15 inches wide) from local DIY store, then get 4 Audax AP210M0 from Maplin (Fs=46, Qt=0.72, S=90.2db, Pow=50W). The HM170 and tweeter will be driven through the existing x-o by my main amp. From the pre-out I'll use a line level low-pass passive x-o into an old NAD receiver which will drive both Ap210 in parallel to reinforce below where dipole roll-off begins (around 400Hz? - opinions seem to vary here lately!) Having 2 per channel should ensure reasonable cone area and power handling. I'll have to tweak the low-pass filter (in fact this is the greatest uncertainty I can see), but the freq response of the AP210 gets ragged above 1kKz so I may need a 2-stage filter. Is there anything crucial that I've missed here?
Thanks
IJ
 
Konnichiwa,

Ian J said:

The way I though of is this. I'll remove the drivers from my existing speakers (HM170Z18 and SEAS 27TFFC) plus existing x-o, buy a veneered shelving board (15 inches wide) from local DIY store

Beware young Will Robinson.

The Audax Driver you use has a Qt that is way too low to be used on a small open baffle without adding equalisation. You can do the EQ passive (see Siegfried Linkwitz small passive dipole project) but you must use EQ or alternatively use a different driver, like one suited to use in "open air" (Qt > 0.5 and < 1.5).

You can use the added dual 8" drivers to fill in some of the lower midrange/upper baffle loss from the too narrow baffle and way too low Qt of the Driver but it is trickey to make that work well.

Sayonara
 
Well, you'll want some eq where you start getting baffle cancellation. The amount and frequency will be dependent on baffle geometry. But it will be pretty straightforward.

You might consider getting a second pair of tweeters, connect them antiphase to the already-existing ones, and mount them on the back of the baffle.
 
The Audax Driver you use has a Qt that is way too low to be used on a small open baffle without adding equalisation.
I dunno. Looks to me like it would work fine as a mid above where the dipole rolloff starts. That's way above the frequency where Qt has any relevance. The dipole rolloff can act as the first pole of the highpass filter. Getting the mid bass drivers equalized flat is another matter but it should be doable with a little experimentation.
 
KYW
Won't the HM170 just become a mid-range in this arrangement with an acoustic -6db/octave roll-off starting at ~400Hz? Why is the low Qt of this driver important? It is following your recent posts that have really given me the impetus for this, and is why I have sought a higher Qt driver for the other drivers (well admittedly not that high, but this is an experiment so I don't want to spend loads yet). I can see the problem in blending the lower with the mids, and that is where the challenge will lie isn't it?
Many thanks
IJ
 
Konnichiwa,

catapult said:
I dunno. Looks to me like it would work fine as a mid above where the dipole rolloff starts.

In which case you need a much narrower baffle than 15".

I'll repeat it the 10,000th time - do not bother with Olsons "ballfe" equations, they do not work. Get into the late 20't or early 21st century by using at least Backmans Math.

Ian J said:
KYW
Won't the HM170 just become a mid-range in this arrangement with an acoustic -6db/octave roll-off starting at ~400Hz?

Yes, but it will do so not because of the narrow baffle (primarily) but because of the low Q.

Ian J said:
Why is the low Qt of this driver important?

Because using a low Q driver on an open baffle means operating a driver intentional designed for use with a rear acoustic load to raise the system Q to a sensible value without the rear load. Or in other turns with a severe LF Rolloff (LF implying a fairly low frequency, not neccesarily < 100Hz) which is not due to the Baffle, but due to using an inapproriate driver.

As said, this CAN be compensated and quite easily (Linkwitz) but you MUST be aware of it. For What It's worth, I'll be using a driver with a Qt of 0.8, but due to tyhe source impedance of a non fedback SE Amplifier raised to >1.2 on a baffle considerably wider than yours. The difference in SPL (theoretical on Backmans and other peoples modern equations) will be in the region of 6 - 8db @ 100Hz.

I realise that you want to use seperate LF drivers, but if you wish to place a driver "open air" without active EQ, it should have a sufficiently high Qt. Otherwise you need to Equalise your "LF" section to have more upper midrange output than can be achieved by a 1st order filter.

Ever which way, you will need to equalise something, somewhere.

Sayonara
 
I'll go one step further than KYW's useful suggestions and propose that you build the dipoles and measure them before deciding on the EQ curve. The equations will get you in the ballpark, but won't help you find your seat.

EQ is an easy thing, once you know the starting point and the target.
 
Konnichiwa,

SY said:
I'll go one step further than KYW's useful suggestions and propose that you build the dipoles and measure them before deciding on the EQ curve. The equations will get you in the ballpark, but won't help you find your seat.

I completely agree on the "ballpark" thing. I would not use basic equations to give me an "authorative" EQ curve. What I use them for in my spreadsheets is to assess the feasibility of building something and if it needs and if then about how much.

So yes, for the final EQ measurements are a neccessity.

Sayonara
 
I'll repeat it the 10,000th time - do not bother with Olsons "ballfe" equations, they do not work. Get into the late 20't or early 21st century by using at least Backmans Math.

Well, thanks for the condescending remarks, Thorsten. :confused: If you don't mind, I'll stick to measuring the actual response of real drivers on real baffles rather than presenting myself as some sort of scholar who possesses the one and only true path to enlightenment.

Ian, it may well work out that the baffle rolloff (starting around 3-500 Hz ? - measure it) will create a 1st order highpass on the mid without any extra EQ or crossover components needed. Since you are biamping, you can do a passive 1st order RC lowpass between the preamp and woofers' power amp. Set Fc down around the lowest frequency you want to play. That should flatten the response in the lower freq and start rolling it off at 6dB when it gets up to the baffle frequency. By adjusting the gain and Fc, you should be able to get a good 1st order acoustic crossover between the woofs and the mid with one resistor and one capacitor.
 
If you don't mind, I'll stick to measuring the actual response of real drivers on real baffles...

Pardon me, but it seems to me that whoever decided baffle width was the deciding factor in dipole bass response, either used circular baffles, or never measured the speakers. As far as I can tell, everything we've seen here on this board has laughed in the face of baffle width, giving deep and clear bass, in spite of what the simulations say shoud happen. In my own experience, I've gotten deeper and louder bass out of dipoles than anything else I've tried. This seems to be the general consensus of people who have actually built and experimented with dipoles.

Perhaps you should measure reality instead of mathematical models?

(/hijack)
 
Perhaps you should measure reality instead of mathematical models?

Isn't that what I just said? Forget theory. Put the driver on the baffle, and measure the sucker. :nod:

PS - I can assure you that Sigfried Linkwitz's designs are based on considerable hands-on experimentation. He was designing and building critically acclaimed dipoles years before most of us ever considered the notion. Dismissing his years of work as "mathematical models" is insulting. He walks the talk.

About Thorsten's notion (theory) of building a "proper" dipole, I find the bass resonance of underdamped, high-Q drivers, driven by low-damping-factor amps to be annoying. Been there, done that, moved on. But, that's just me. Whatever floats your boat...... ;)
 
Konnichiwa,

catapult said:

About Thorsten's notion (theory) of building a "proper" dipole, I find the bass resonance of underdamped, high-Q drivers, driven by low-damping-factor amps to be annoying.

How do you POSSIBLY arrive at that conclusion? What I propose is not to use "High Q" Driver, instead I propose to use APROPRIATE Q Drivers or failing the procurement to make such by adjusting the amplifiers output impedance. If you have to equalise both the baffle rolloff (if you make the baffle bvery narrow) and a low Q driver you need a lot of EQ and a slope that is steeper than 1st order.

You sould see what THAT does do to groop delay, impulse response etc. Been there, done that - decided to use the right driver for the right job is better than trying to force the wrong driver to a job it is not designed to do. But that's just me and what the flying french coitus do I know anyway....

Sayonara
 
All that matters is the final transfer function of the amp/wire/driver/baffle/EQ combination. Add all the mechanical and electrical stuff together - measure the acoustic output - and see what you've got. A final Q of .5-.6 will have good impulse and group delay characteristics (and sound good to my ear.) Your proposed final Q of >1 has consequences that are measurable (and audible to me, though not to everyone.)

Edit: SL explains it more eloquently than I.....

http://linkwitzlab.com/faq.htm#Q34

The effect of using high Q for equalization is the frequency rolloff below resonance is greater than second order. Just guessing, it would probably approach 4th order initial rolloff with Thorsten's proposed high-Q amp/driver combination an no additional EQ. The steep slope of the rolloff is what messes up the transient response. Kinda like a ported box vs. a sealed box.
 
Konnichiwa,

catapult said:
A final Q of .5-.6 will have good impulse and group delay characteristics (and sound good to my ear.)

Yup. And that is exactly what I'm shooting for. A system Q of 0.5, namely one where the output at resonance (50Hz) is 6db down on the output at higher frequencies. And as the baffle has a very low Q (< 1), so the net Q approaches closesly 0.5, which can be seen from the final transfer function which is 3rd Order below resonance, just as expected for an open baffle and driver....

Shall we agree that Q as such is mostly meaningless? The transfer function determines the impulse response (and visa versa).

Still, if you use a driver from the beginning designed for open baffle operation you do not require a lot of EQ (normally non) and (I have done it both ways before) the effective response (impulse, SPL/THD etc) at low frequencies is much better than forcing a low Qt driver via EQ to do what it cannot.

A driver with a Qt of 0.2 - 0.3 belongs in a horn, a driver with Qt of 0.3-0.6 into a reflex or sealed enclosure and one with a Qt of > 0.6 belongs into a Dipole of a size suited to it's Qt (higher Qt smaller baffle for same Fs). You can of course do anything else by adding electronic equalisation.

As for all the rest, decades before Mr. Linkwitz companies in Europe build "open air" and thusly even in their "winged" form dipole woofer system for sound reinforcement. They knew very well what they where doing and choose to use a sensible Qt for the Woofer (except it was not called Qt then because everyone still used fundamental parameters to calaculate, a much more lengthy but equally more relevant way of calculating Sound output from a dynamic driver than T/S).

Sayonara
 
Yup. And that is exactly what I'm shooting for. A system Q of 0.5, namely one where the output at resonance (50Hz) is 6db down on the output at higher frequencies. And as the baffle has a very low Q (< 1), so the net Q approaches closesly 0.5, which can be seen from the final transfer function which is 3rd Order below resonance, just as expected for an open baffle and driver....

Gotta look at the shape of the whole curve. If you can really get a Q=.5 rolloff from your Q=1+ driver/amp, I'm impressed. Then there's still the issue of the slope. 2nd order (Linkwitz' goal) is better than 3rd.

Got any measured curves to share? I'm always ready to learn something and eat humble pie if I'm wrong. :)
 
Konnichiwa,

catapult said:


Gotta look at the shape of the whole curve. If you can really get a Q=.5 rolloff from your Q=1+ driver/amp, I'm impressed. Then there's still the issue of the slope. 2nd order (Linkwitz' goal) is better than 3rd.

You can't get a 2nd order slope below resonance with an open baffle. It's physics. It must be the 2nd order LF rolloff of the driver added to baffle slope 1st order, ergo 3rd order no matter which way you cut it. The main difference is around the actual resonance considerably below and above that the Q matters not.

catapult said:

Got any measured curves to share?

They where up a year or two ago when a friends system included such an open baffle. Full ETF, I did not keep copies.

Sayonara

PS, when the metalwork for my baffles arrives (Acrylic is already here) I'll bolt them together and measure.... Metalworker say 2 Weeks :mafioso:
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.