simulating "Deep Frame Dipoe" (DFD) Woofers, AkAbak?

mige0 · 2010-12-18 7:41 pm

Hi folks

After having arrived at a "new" speaker concept :

http://www.diyaudio.com/forums/subwoofers/172553-amt-sub-using-dynamic-woofer.html#post2399769
(roughly down from posting #4)

I'd thought, it would be handy to have a somehow accurate simulation tool to have an easy look at possible variants and optimizations.
I've been advised to possibly give AkAbak a try, after Hornresp and AJHorn did not seem to be a good choice.

Needed is a possibility to simulate a *distributed speakers arrangement* in pipes - and possibly add dipole behavior.

I just download AkAbak:
Akabak

and had a look through a few pages:
http://www.diyaudio.com/forums/multi-way/90362-akabak-simulator.html

Thats the starting point.
Any input appreciated.

Michael

mige0 · 2010-12-19 7:39 pm

After some reading, AkAbak starts to make sense.
First, a most simple simu of the H-frame with only the front duct being modelled - looks definitely promising with respect to the measurements already taken (red trace):

Measurement:

Akabak simu:

This is the code used:

Code:

System 'H-Frame Part1' |Anything behind a line is for humans only


Def_Driver  '15PW5'
  dD=33cm   		|Piston
  fs=40Hz  
  Mms=84.6g  
  Qms=16.8
  Qes=0.41  
  Re=5.5ohm  
  Le=1mH  
  ExpoLe=0.618
  

Def_Const |This sets the units within to all be Meters
{ 			|Starts
Vrc= 9000.00e-3; 	|Rear chamber volume (litres)
Lrc= 1.00e-2; 		|Rear chamber average length (cm)

Sd = 860e-4; 		|Diaphragm area (sq cm)
Sdd = 1600E-4;		|Duct Outlet area (sq cm)
} 			|Ends 


Driver Def='15PW5''15PW5'
Node=1=0=2=3


Enclosure 'Rear chamber' 
Node=2
Vb={Vrc} 
Lb={Lrc}


Duct    'Du1'  
Node=3=33
dD=45cm  
Len=40cm  


Radiator 'DUCT outlet'
Node=33
SD={Sdd}
Label=2

Not too bad for a first lift off, I'd say

Michael

soho54 · 2010-12-20 10:25 pm

Interesting experiment.

I have a few ideas, but some questions first. Can you post the exact measurements of the test enclosure, external is fine, and wood thickness. Was the driver mounted as shown in the 3d shot? Where was the mic placed?

soho54 · 2010-12-21 1:41 am

I got bored.

Try this setup out for starters.

Code:

System 'H-Frame Full' 


Def_Driver  '15PW5'
  dD=33cm   |Piston
  fs=40Hz  Mms=84.6g  Qms=16.8
  Qes=0.41  Re=5.5ohm  Le=1mH  


Driver Def='15PW5''Driver'
Node=1=0=2=3

Duct    'Rear'  
Node=2=22    
WD=40cm  HD=40cm  Len=40cm |End Correction is not needed, as the attached Radiator handles it.

  Radiator  'Rear Outlet'  Def='Rear'    
  Node=22
  x=0  y=0  z=-40cm  HAngle=180°  VAngle=0
  WEdge=43.6cm HEdge=43.6cm
  Label=2 
 

Duct    'Front'  
Node=3=33    
WD=40cm  HD=40cm  Len=40cm |End Correction is not needed, as the attached Radiator handles it.


  Radiator  'Front Outlet'  Def='Front'    
  Node=33
  x=0  y=0  z=40cm  HAngle=0  VAngle=0  
  WEdge=43.6cm HEdge=43.6cm  
  Label=1

Just add "all" and labels 1 & 2 when simulating. Don't forget to play with the "mic/listener" distance as well.

soho54 · 2010-12-21 1:51 am

You should end up with something like this, with the total FR, and both ends separately.

An externally hosted image should be here but it was not working when we last tested it.

mige0 · 2010-12-21 6:20 am

soho54 said:
Interesting experiment.

I have a few ideas, but some questions first. Can you post the exact measurements of the test enclosure, external is fine, and wood thickness. Was the driver mounted as shown in the 3d shot? Where was the mic placed?

Have arrived at something close to your script yesterday (except the "edge" diffraction statements I have not looked into yet) - but are still a little bit slow in verifying

Thanks for your input - very much appreciated!

regardig "wood thickness" : don't bother, there is no need to get "precise" results - all this simus are meant to just get the trend
regarding "driver mounting" : yes exactly like shown (I have added cone depth in my simu but it does not make the heck of a difference - as expected)
regarding "mic placement" : measurements (right a mouth) also were done just to catch the trend of different arrangements - so no claim to be "absolutely precise"

I will be back with some plots and specific questions soon...

Nice overlay plot BTW, I have to learn how to do this easily...

Michael

soho54 · 2010-12-21 2:38 pm

You're welcome.

This should be close enough then. I find that adding a conical diaphragm isn't necessary below ~1kHz.

The overlay is easy to do, look at the lower left corner here.

As you only have one System <all> and H-Frame Full are the same.

mige0 · 2010-12-22 6:21 am

Well,I added conical mainly because it adds some un-symmetry in case of H-frame.

For doing the overlay "right", I'm not clever enough as it seems.

As far as I understood there is a max of thee traces to be overlaid. Works if I drag and drop from one plot into the other, but does not work to immediately simulate into one plot only and - also does not work to get / keep the colors of traces adjusted.

Another problem I've come across - seems if I simu close to mouth - approaching 40cm in above example - SPL level sky rocks to several 100dB. Any idea?

Michael

mige0 · 2010-12-22 4:55 pm

mige0 said:
For doing the overlay "right", I'm not clever enough as it seems.

Boy this took me time !

I always tried to *add* traces to a given plot window - better to just choose / select the scripted labels and every trace is then put into the upcoming plot at once...

Uff!

Michael

mige0 · 2010-12-22 5:14 pm

Why writing:

Driver Def='15PW5''Driver'
Node=1=0=2=3

and not :
Driver Def='15PW5'
Node=1=0=2=3

is there a difference?

Michael

mige0 · 2010-12-22 8:06 pm

Soho54,
I gave the dual chassis N-frame a try.

this is how it measures (red trace) directly at mouth:

Could you have a look if code is ok so far?
I switched / simplified to „dEdge=1mm“ as I found this in the manual to force Akabak to do diffraction calculation as well.
What I'm not really happy with, is to use a „duct“ for the closed end piece.“Waveguide“ seems not to allow for closing one end with a „0-node“ ??
Any suggestions?

Code:

System 'N2-Frame' 

Def_Driver  '15PW5'
  dD=33cm   |Piston
  fs=40Hz  Mms=84.6g  Qms=16.8
  Qes=0.41  Re=5.5ohm  Le=1mH  


Driver Def='15PW5''Driver'
  Node=1=0=2=3

Driver Def='15PW5''Driver'
  Node=1=0=200=300

  
  
  
Duct    'DU-r'  Node=200=0
  dD=10cm  Len=20cm
     
Waveguide    'Rear1'  Node=200=2    
  dTh=10cm  dMo=30cm
  Len=40cm  Conical		
 
Waveguide    'Rear'  Node=2=22    
  dTh=30cm  dMo=45cm
  Len=20cm  Conical		
  
Radiator  'Rear Outlet'  Def='Rear'    
  Node=22
  x=0  y=0  z=-40cm  HAngle=180°  VAngle=0
  dEdge=1mm
  Label=2 
 
 
 
 
Duct    'DU-f'  Node=300=0
  dD=10cm  Len=20cm

Waveguide    'Front1'  Node=300=3    
  dTh=10cm  dMo=30cm
  Len=40cm  Conical		
 
Waveguide    'Front'  Node=3=33    
  dTh=30cm  dMo=45cm
  Len=20cm  Conical		

Radiator  'Front Outlet'  Def='Front'    
  Node=33
  x=0  y=0  z=40cm  HAngle=0  VAngle=0  
  dEdge=1mm
  Label=1

here is the simulation result of above code at 1m mic distance:

Quite some room for improvement I'd say - but - at least the deep null seems to be at the right place - as seems to be the up-shift of the first peak captured - as seems the SPL increase at the bottom end to be captured..

Michael

mige0 · 2010-12-22 8:48 pm

mige0 said:
Soho54,
I gave the dual chassis N-frame a try.
...

What I'm not really happy with, is to use a „duct“ for the closed end piece.

Composed the conical end piece out of „waveguide“ and „duct“

Code:

System 'N2-Frame' 

|-------------- driver ----------------
Def_Driver  '15PW5'
  dD=33cm   |Piston
  fs=40Hz  Mms=84.6g  Qms=16.8
  Qes=0.41  Re=5.5ohm  Le=1mH  


Driver Def='15PW5''Driver'
  Node=1=0=2=3

Driver Def='15PW5''Driver'
  Node=1=0=200=300
|-------------- end driver ----------------  



|-------------- rear firing pipe ----------------  
  
Duct    'DU-r'  Node=201=0
  dD=1cm  Len=1cm

Waveguide    'Rear2'  Node=201=200    
  dTh=1cm  dMo=10cm
  Len=19cm  Conical		
    
Waveguide    'Rear1'  Node=200=2    
  dTh=10cm  dMo=30cm
  Len=40cm  Conical		
 
Waveguide    'Rear'  Node=2=22    
  dTh=30cm  dMo=45cm
  Len=20cm  Conical		
  
Radiator  'Rear Outlet'  Def='Rear'    
  Node=22
  x=0  y=0  z=-40cm  HAngle=180°  VAngle=0
  dEdge=1mm
  Label=2 
|-------------- end rear firing pipe ---------------- 
 
 
 
|-------------- front firing pipe ----------------

Duct    'DU-f'  Node=301=0
  dD=1cm  Len=1cm

Waveguide    'Front2'  Node=301=300    
  dTh=1cm  dMo=10cm
  Len=19cm  Conical		

Waveguide    'Front1'  Node=300=3    
  dTh=10cm  dMo=30cm
  Len=40cm  Conical		
 
Waveguide    'Front'  Node=3=33    
  dTh=30cm  dMo=45cm
  Len=20cm  Conical		

Radiator  'Front Outlet'  Def='Front'    
  Node=33
  x=0  y=0  z=40cm  HAngle=0  VAngle=0  
  dEdge=1mm
  Label=1
|-------------- end front firing pipe ----------------

here is the simulation result of above code at 1m mic distance:

Slightly more elegant this way, though the deep null seems to be less correct

Michael

soho54 · 2010-12-22 10:36 pm

mige0 said:
Another problem I've come across - seems if I simu close to mouth - approaching 40cm in above example - SPL level sky rocks to several 100dB. Any idea?

AkAbak doesn't like the listening point to be inside the model. You will also find that you have to place it a bit back from the Radiator as well do to the end correction it adds.

mige0 said:
Why writing:

Driver Def='15PW5''Driver'
Node=1=0=2=3

and not :
Driver Def='15PW5'
Node=1=0=2=3

is there a difference?

The 'Driver' added is the name of the component. I just didn't feel very creative.

You don't have to add it. It helps more with multi driver scripts when you are checking out different graphs.

soho54 · 2010-12-23 12:47 am

You don't have to ground the end of a waveguide really.

This is how I would start off. Note that when you start using Waveguides you have to have more exact placements.

Here is a guesstimate.

Code:

System 'N2-Frame' 

Def_Driver  '15PW5'
  dD=33cm   |Piston
  fs=40Hz  Mms=84.6g  Qms=16.8
  Qes=0.41  Re=5.5ohm  Le=1mH  


Driver Def='15PW5''Driver 1'
  Node=1=0=44=3

Driver Def='15PW5''Driver 2'
  Node=0=1=4=33

   
     
Waveguide    'Rear A'  Node=22=33
  WTh=5mm  HTh=36.4cm  WMo=8.4cm  HMo=36.4cm
  Len=18.2cm  
  		 
Waveguide    'Rear B'  Node=33=44    
  WTh=8.4cm  HTh=36.4cm  WMo=26.6cm  HMo=36.4cm
  Len=39.3cm    	
  	
Waveguide    'Rear C'  Node=44=55    
  WTh=26.6cm  HTh=36.4cm  WMo=36.4cm  HMo=36.4cm
  Len=21.5cm          
  
Radiator  'Rear Outlet'  Def='Rear C'    
  Node=55
  x=0  y=0  z=-40cm  HAngle=180°  VAngle=0
  WEdge=40cm HEdge=40cm
  Label=2 
 
 
 
 
Waveguide    'Front A'  Node=2=3
  WTh=5mm  HTh=36.4cm  WMo=8.4cm  HMo=36.4cm
  Len=18.2cm  
  		 
Waveguide    'Front B'  Node=3=4    
  WTh=8.4cm  HTh=36.4cm  WMo=26.6cm  HMo=36.4cm
  Len=39.3cm    	
  	
Waveguide    'Front C'  Node=4=5    
  WTh=26.6cm  HTh=36.4cm  WMo=36.4cm  HMo=36.4cm
  Len=21.5cm        
  
Radiator  'Front Outlet'  Def='Front C'    
  Node=5
  x=0  y=0  z=40cm  HAngle=0  VAngle=0
  WEdge=40cm HEdge=40cm
  Label=1

mige0 · 2010-12-26 8:12 pm

Thanks soho54

good hint regarding theres no need for closing a WG.

Your code sim's like that:

not that bad - but - what you think could achieve even better correlation to measurements?

I messed around with the idea of distributing the 15" chassis into a hex arrangement to come closer (in simulation !) to the fact that the 15"ers actually do *not* radiate into the wave guide at *exactly* one point only.

This "distributed" radiation into the WG must IMO result in a strong FR difference, as "resonance modes" are excited in a pretty different manner that way.

Simu I got - sadly - are even less "the real thing":

Peaking is way too hot and not a t the right place as well - also deep null (400-500Hz) filled too much
here is the code I used:

Code:

System 'N2-Frame' | N-frame Deep Frame Dipole  / outer dimension 40x40x80cm / 2x15" arranged distributed in "hex+centre"  


|-------------- driver ----------------
Def_Driver  '15PW5'
  dD=12.5cm   |Piston
  fs=40Hz  Mms=84.6g  Qms=16.8
  Qes=0.41  Re=5.5ohm  Le=1mH  



Driver Def='15PW5''Driver 1_1'		| "distributed" driver "hex+centre"  arrangement
  Node=0=1=310=201

Driver Def='15PW5''Driver 1_2'
  Node=0=1=309=202

Driver Def='15PW5''Driver 1_3'
  Node=0=1=309=202

Driver Def='15PW5''Driver 1_4'
  Node=0=1=308=203

Driver Def='15PW5''Driver 1_5'
  Node=0=1=307=204

Driver Def='15PW5''Driver 1_6'
  Node=0=1=307=204

Driver Def='15PW5''Driver 1_7'
  Node=0=1=306=205




Driver Def='15PW5''Driver 2_1'		| "distributed" driver "hex+centre"  arrangement
  Node=1=0=206=305
  
Driver Def='15PW5''Driver 2_2'
  Node=1=0=207=304

Driver Def='15PW5''Driver 2_3'
  Node=1=0=207=304

Driver Def='15PW5''Driver 2_4'
  Node=1=0=208=303
  
Driver Def='15PW5''Driver 2_5'
  Node=1=0=209=302
  
Driver Def='15PW5''Driver 2_6'
  Node=1=0=209=302
  
Driver Def='15PW5''Driver 2_7'
  Node=1=0=210=301
  

|-------------- end driver ----------------  

 
 
|-------------- front firing pipe ----------------


Waveguide    'Front A'  Node=211=210    
  WTh=1mm  HTh=40cm  WMo=2cm  HMo=40cm
  Len=4cm  Conical		

Waveguide    'Front B1'  Node=210=209    
  WTh=2cm  HTh=40cm  WMo=6cm  HMo=40cm
  Len=8cm  Conical		

Waveguide    'Front B2' Node=209=208    
  WTh=6cm  HTh=40cm  WMo=10cm  HMo=40cm
  Len=8cm  Conical		
    
Waveguide    'Front B3'  Node=208=207
  WTh=10cm  HTh=40cm  WMo=14cm  HMo=40cm
  Len=8cm  Conical		
 
Waveguide    'Front B4'  Node=207=206
  WTh=14cm  HTh=40cm  WMo=18cm  HMo=40cm
  Len=8cm  Conical	
 
Waveguide    'Front C'  Node=206=205
  WTh=18cm  HTh=40cm  WMo=22cm  HMo=40cm
  Len=8cm  Conical 
 
Waveguide    'Front D1'  Node=205=204    
  WTh=22cm  HTh=40cm  WMo=26cm  HMo=40cm
  Len=8cm  Conical 
 
Waveguide    'Front D2'  Node=204=203    
  WTh=26cm  HTh=40cm  WMo=30cm  HMo=40cm
  Len=8cm  Conical 

Waveguide    'Front D3'  Node=203=202
  WTh=30cm  HTh=40cm  WMo=34cm  HMo=40cm
  Len=8cm  Conical 

Waveguide    'Front D4'  Node=202=201    
  WTh=34cm  HTh=40cm  WMo=38cm  HMo=40cm
  Len=8cm  Conical 

Waveguide    'Front E'  Node=201=200
  WTh=38cm  HTh=40cm  WMo=40cm  HMo=40cm
  Len=4cm  Conical		

Radiator  'Front Outlet'  Def='Front E'    
  Node=200
  x=0  y=0  z=40cm  HAngle=0  VAngle=0  
  dEdge=1mm
  Label=1
|-------------- end front firing pipe ----------------



|-------------- rear firing pipe ----------------  
  
Waveguide    'Rear A'  Node=311=310
  WTh=1mm  HTh=40cm  WMo=2cm  HMo=40cm
  Len=4cm  Conical		

Waveguide    'Rear B1'  Node=310=309
  WTh=2cm  HTh=40cm  WMo=6cm  HMo=40cm
  Len=8cm  Conical		

Waveguide    'Rear B2'  Node=309=308
  WTh=6cm  HTh=40cm  WMo=10cm  HMo=40cm
  Len=8cm  Conical		

Waveguide    'Rear B3'  Node=308=307
  WTh=10cm  HTh=40cm  WMo=14cm  HMo=40cm
  Len=8cm  Conical		

Waveguide    'Rear B4'  Node=307=306
  WTh=14cm  HTh=40cm  WMo=18cm  HMo=40cm
  Len=8cm  Conical		

Waveguide    'Rear C'  Node=306=305
  WTh=18cm  HTh=40cm  WMo=22cm  HMo=40cm
  Len=8cm  Conical		
    
Waveguide    'Rear D1'  Node=305=304
  WTh=22cm  HTh=40cm  WMo=26cm  HMo=40cm
  Len=8cm  Conical		

Waveguide    'Rear D2'  Node=304=303
  WTh=26cm  HTh=40cm  WMo=30cm  HMo=40cm
  Len=8cm  Conical		
 
Waveguide    'Rear D3'  Node=303=302
  WTh=30cm  HTh=40cm  WMo=34cm  HMo=40cm
  Len=8cm  Conical		
 
Waveguide    'Rear D4'  Node=302=301
  WTh=34cm  HTh=40cm  WMo=38cm  HMo=40cm
  Len=8cm  Conical		
  
Waveguide    'Rear E'  Node=301=300
  WTh=38cm  HTh=40cm  WMo=40cm  HMo=40cm
  Len=4cm  Conical	
 

Radiator  'Rear Outlet'  Def='Rear E'    
  Node=300
  x=0  y=0  z=-40cm  HAngle=180°  VAngle=0
  dEdge=1mm
  Label=2 
|-------------- end rear firing pipe ----------------

Measurement at mouth (red trace)

Michael

mige0 · 2010-12-26 8:32 pm

Ahh - here is the plot (two 15" *not* "distributed") with only the front radiator active to allow for tighter conditions of measurement versus simu:

The upper part of bandwidth (till 300Hz) is not bad - but the the low frequency slope comes in too high - at roughly 90Hz instead of around 40-50Hz

Michael

mige0 · 2010-12-26 8:42 pm

Ahh - here is the plot (two 15" *not* "distributed") with the rear radiator having "edge / diffraction" feature disabled:

The low frequency slope is where it should be (no more closed volume

like in my previous post) but upper part (till 300Hz) is not "as expected" - not at all

Michael

soho54 · 2010-12-28 5:41 pm

In the pic of my sim, we seem to be looking at the whole system. I would stick with Label 1 to compare to your measurements.

Once you start using waveguides placement is paramount. I just guessed at some simple figures for the enclosure and driver placements, so there is an error factor right there.

The other problem comes from using a waveguide in a sim. The sim only calculates the waveguides contribution, and not the extra direct sound from the woofer. It is the same condition that produces the early roll-off up top in a mid-horn sims, and the "hybrid" horn roll-off down low.

I have a couple of ideas on how to get around it, but I haven't had time to play with it yet.

mige0 · 2010-12-28 5:56 pm

soho54 said:
In the pic of my sim, we seem to be looking at the whole system. I would stick with Label 1 to compare to your measurements.

Yeah we can try...

------------------

I have tried to bring the rear radiating SPL contribution out of picture by:

- adding a quasi IB (short but very wide WaveGuide) to the rear outlet
- setting the distance for the rear outlet way back (some -10m)

Both attempts actually did not work, so, I'm a little bit lost to which point AkAbak sims correctly regarding edge / 180deg diffraction - besides all quirks with the "distributed speaker" alignment in the pipe.

Michael

soho54 · 2010-12-29 5:00 am

All you have to do is use the Labels section in the SPL simulation setup screen.

If you check a box the graph will only be from the output Label you picked. If you pick Label 1 you will only get the output from the forward opening. If you pick Label 2 you get just the rear. Selecting All or N2-Frame you get the combination of both sides.

Just selecting Label 1 is like getting a near field measurement. This is how you get around the program going a little nuts from having the listening position to close to it, and the SPL shooting way up.

Either way this is still only going to show the waveguides contribution, when in this case you have to combine the waveguides output with the traditional IB output as well.

simulating "Deep Frame Dipoe" (DFD) Woofers, AkAbak?

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member