P.O.C. #2 - Compact Tapped Horn

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi Everyone:

Please critique, thanks:

Driver params:
Sd: 855.30 cm^2
Cms: 1.73 E-04
Mmd: 88.63
Re: 6.20
BL:20.50
Rms:3.92
Le:2.49
Xmax=6mm

Tapped Horn:
S1=550 cm^2
S4=2000 cm^2
L12=22 cm
L23=239 cm
L34=22 cm

Folding: similar to http://www.diysubwoofers.org/projects/other/tapped-pipe/construction.html
Net depth works out to about 72 cm or so. Height and width depend on what dimensions I choose for the cross-section, but should be somewhere around 80 * 64 cm (net)

First two folds stuffed with polyester fiberfill. Speaker mounted with magnet facing into mouth (but I will experiment)

Predicted Result ~ 120dB @ 100W from 40 Hz to 100 Hz without exceeding Xmax. Stuffing should actually improve the response above 100 Hz (based on P.O.C. #1 results). Will likely experiment with the amount used - the removable side panel gives me access to all sections of the "horn".

Use - stacks of 2, 4 or even 8 :), to provide everything below 100 Hz . Small club, or back of truck, LOL. Will likely adjust dimensions to be able to get 4 across back of a flatbed truck, if possible.

Note: vented 3.6 cu.ft. will get a dB or two higher above 60 Hz, but at double the power (and I'm going to bet that power compression is going to remove that advantage anyway).

Any suggestions welcome (including "you idiot - don't do that!, LOL)
 
Please critique

Hi Everyone:

Please critique, thanks:..

Any suggestions welcome (including "you idiot - don't do that!, LOL)

Hi Brian,

:) :up: You won't get any negative critique from me other than a reminder that 12 dB HP filter is needed at 30 Hz.
This TH, easy to integrate to most main speakers would work IMO nice at 70 W input producing more than 117 dB SPL above 40 Hz. :D

The driver brand?

b
 
Hi Brian,

:) :up: You won't get any negative critique from me other than a reminder that 12 dB HP filter is needed at 30 Hz.
This TH, easy to integrate to most main speakers would work IMO nice at 70 W input producing more than 117 dB SPL above 40 Hz. :D

The driver brand?

b

Thanks.. :)

The driver brand? I'm going to keep that secret for now - but will disclose later, if this project goes forward. It's a reliable brand - I'll leave it at that :).

I forgot to mention that I'm considering implementing this as a "stepped" horn, i.e. four sections of constant cross-section, rather than a conical taper from S1 to S4. If I've understood things correctly, this would likely introduce response abberations, but above the TH's cutoff frequency (I hope!). To do this, I'll basically be dividing the box into 4 segments of increasing volume from throat to mouth, like indicated in the diagram below (ignore the dimensions given).
20090830-th-idea.gif


I'm not sure how to model this in HornResp though.
 
No way to model it in HR, it doesn't have enough segments to create the transitions, but AkAbak can AFAIK plus it can do stuffing effects. Anyway, even with only two usable octaves, IME you'll need at least five pipe segments, three for the ~120-60 Hz octave and two for the ~60-30 Hz one.

Then again, my experience is with higher BW compression horns where our hearing acuity is better, so looking forward to your results.

GM
 
No way to model it in HR, it doesn't have enough segments to create the transitions, but AkAbak can AFAIK plus it can do stuffing effects. Anyway, even with only two usable octaves, IME you'll need at least five pipe segments, three for the ~120-60 Hz octave and two for the ~60-30 Hz one.

Then again, my experience is with higher BW compression horns where our hearing acuity is better, so looking forward to your results.

GM

I'm unfamiliar with Akabak - looks like another tool I'll have to start getting familiar with :)

Each segment in this "horn" is going to be around 71 cm. Some notes from Danley (copied here - http://techtalk.parts-express.com/showthread.php?t=212802) suggest that 1m segments start having a serious impact around 200 Hz, so I'm thinking that the impact of these 71 cm long "steps" might occur a bit higher - and out of the TH's passband.

I've run into another "issue" though - it looks like a simple EBS design in a smaller box (~ 5.5 cu.ft.) with the same driver can produce the same low-end response at similar theoretical SPL levels, albeit at lower efficiency (220W needed).

I might still build P.O.C#2, but I'm now wondering about its practicality....
 
Here's an Akabak script and SPL graph for your design, for both the theoretical (conical) model and the stepped model with 4 segments. There's not a lot of difference at the frequencies of interest. Note the deeper notches at high frequencies in the stepped version, as predicted.



Def_Driver 'Driver'

Sd=855.30cm2
Bl=20.50Tm
Cms=1.73E-04m/N
Rms=3.92Ns/m
fs=37.7004Hz |Mmd = 88.63g not recognised by AkAbak, fs calculated and used instead
Le=2.49mH
Re=6.20ohm
ExpoLe=1


System 'Conical'

Driver Def='Driver' 'Driver_C'
Node=1=0=12=14

Waveguide 'W1'
Node=11=12
STh=550cm2
SMo=912.5cm2
Len=71cm
Conical

Waveguide 'W2'
Node=12=13
STh=912.5cm2
SMo=1275cm2
Len=71cm
Conical

Waveguide 'W3'
Node=13=14
STh=1275cm2
SMo=1637.5cm2
Len=71cm
Conical

Waveguide 'W4'
Node=14=15
STh=1637.5cm2
SMo=2000cm2
Len=71cm
Conical

Radiator 'RC'
Node=15
SD=2000cm2


System 'Stepped'

Driver Def='Driver' 'Driver_S'
Node=1=0=22=24

Duct 'D1' Node=21=22
SD=550cm2 Len=71cm

Duct 'D2' Node=22=23
SD=1033.3cm2 Len=71cm

Duct 'D3' Node=23=24
SD=1516.6cm2 Len=71cm

Duct 'D4' Node=24=25
SD=2000cm2 Len=71cm

Radiator 'RS'
Node=25
SD=2000cm2
 

Attachments

  • POC-2.PNG
    POC-2.PNG
    25.4 KB · Views: 547

GM

Member
Joined 2003
Some notes from Danley........

I've run into another "issue" though - it looks like a simple EBS design in a smaller box (~ 5.5 cu.ft.) with the same driver can produce the same low-end response at similar theoretical SPL levels, albeit at lower efficiency (220W needed).

I might still build P.O.C#2, but I'm now wondering about its practicality....

His example is a much faster expanding expo flare whereas yours is very slow expanding, so kind of surprised that AkAbak predicts minimal variance in the intended pass-band.

Yeah, driver specs dictate different optimal pass-bands for different loading schemes, so if it performs best overall in an EBS........... Still, less power = lower distortion if the intended app requires high SPL and once damped, the TH will have a flatter phase response which may/may not be a plus depending on the details of the app.

GM
 
Here's an Akabak script and SPL graph for your design, for both the theoretical (conical) model and the stepped model with 4 segments. There's not a lot of difference at the frequencies of interest. Note the deeper notches at high frequencies in the stepped version, as predicted.



Def_Driver 'Driver'

Sd=855.30cm2
Bl=20.50Tm
Cms=1.73E-04m/N
Rms=3.92Ns/m
fs=37.7004Hz |Mmd = 88.63g not recognised by AkAbak, fs calculated and used instead
Le=2.49mH
Re=6.20ohm
ExpoLe=1


System 'Conical'

Driver Def='Driver' 'Driver_C'
Node=1=0=12=14

Waveguide 'W1'
Node=11=12
STh=550cm2
SMo=912.5cm2
Len=71cm
Conical

Waveguide 'W2'
Node=12=13
STh=912.5cm2
SMo=1275cm2
Len=71cm
Conical

Waveguide 'W3'
Node=13=14
STh=1275cm2
SMo=1637.5cm2
Len=71cm
Conical

Waveguide 'W4'
Node=14=15
STh=1637.5cm2
SMo=2000cm2
Len=71cm
Conical

Radiator 'RC'
Node=15
SD=2000cm2


System 'Stepped'

Driver Def='Driver' 'Driver_S'
Node=1=0=22=24

Duct 'D1' Node=21=22
SD=550cm2 Len=71cm

Duct 'D2' Node=22=23
SD=1033.3cm2 Len=71cm

Duct 'D3' Node=23=24
SD=1516.6cm2 Len=71cm

Duct 'D4' Node=24=25
SD=2000cm2 Len=71cm

Radiator 'RS'
Node=25
SD=2000cm2

Thanks for this Don.

I've got two questions though....

1. How did you come up with the equivalent duct sizes? I just averaged the volumes for each section, and came up with the following: S1=731.5 cm^2, S2=1093.75 cm^2, S3=1456.25 cm^2 and S4=1818.75 cm^2. Your method looks like it's obviously more accurate, as the resulting response looks pretty close to the unstepped horn.

2. The low frequency response predicted by AkAbak seens a bit different to the HornResp model - see attachment. Any ideas why, and if so, which is more reliable? If it helps, I'm using 2*Pi for the HornResp model.

Thanks again...
 

Attachments

  • 20090902-thfr.gif
    20090902-thfr.gif
    24.9 KB · Views: 554
Yeah, driver specs dictate different optimal pass-bands for different loading schemes, so if it performs best overall in an EBS........... Still, less power = lower distortion if the intended app requires high SPL and once damped, the TH will have a flatter phase response which may/may not be a plus depending on the details of the app.

GM

I'm having a hard time deciding what's best here. The actual subs, if they actually ever get built, and I'm thinking of building several, will be used to provide bass for small club (type of music played is typically hip-hop, R&B, that sort of thing). Not sure of the floor space, but it can fit up to 350 people comfortably. There's also a good chance that they'll be dragged onto a flatbed truck at least once a year for a more, um, "mobile" experience :).

I like the idea of a lower Pe requirement. OTOH, the TH takes up about 13 ft. net - enough space to use two of these 15" drivers in an EBS alignment. The driver in question is about $140 US, to it might make sense to just spring for the extra $140, ramp up the watts (and watts are cheap these days) and enjoy 6dB more output capability in something that not only takes up less space than the TH, but has a greater bandwidth as well.
 
Thanks for this Don.

I've got two questions though....

1. How did you come up with the equivalent duct sizes? I just averaged the volumes for each section, and came up with the following: S1=731.5 cm^2, S2=1093.75 cm^2, S3=1456.25 cm^2 and S4=1818.75 cm^2. Your method looks like it's obviously more accurate, as the resulting response looks pretty close to the unstepped horn.

2. The low frequency response predicted by AkAbak seens a bit different to the HornResp model - see attachment. Any ideas why, and if so, which is more reliable? If it helps, I'm using 2*Pi for the HornResp model.

Thanks again...

1. You said 4 sections. Throat is 550cm2 and mouth is 2000 cm2.
Therefore first section is 550cm2, last section is 2000 cm2. Middle 2 sections:
2000 - 550 = 1450
1450 / 3 = 483.3
550 + 483.3 = 1033.3cm2
2000 - 483.3 = 1516.6cm2


2. Ah, you spotted my deliberate mistake - well, that's my excuse anyway. I messed up, I put the driver 71 cm in from the ends instead of 22 cm.
Here is the revised script and SPL graph:

Def_Driver 'Driver'

Sd=855.30cm2
Bl=20.50Tm
Cms=1.73E-04m/N
Rms=3.92Ns/m
fs=37.7004Hz |Mmd = 88.63g not recognised by AkAbak, fs calculated and used instead
Le=2.49mH
Re=6.20ohm
ExpoLe=1


System 'Conical'

Driver Def='Driver' 'Driver_C'
Node=1=0=12=13

Waveguide 'W1'
Node=11=12
STh=550cm2
SMo=630.29cm2
Len=22cm
Conical

Waveguide 'W2'
Node=12=13
STh=630.29cm2
SMo=1854.85cm2
Len=239cm
Conical

Waveguide 'W3'
Node=13=14
STh=1854.85cm2
SMo=2000cm2
Len=22cm
Conical

Radiator 'RC'
Node=14
SD=2000cm2


System 'Stepped'

Driver Def='Driver' 'Driver_S'
Node=1=0=22=26

Duct 'D1a' Node=21=22
SD=550cm2 Len=22cm

Duct 'D1b' Node=22=23
SD=550cm2 Len=49cm

Duct 'D2' Node=23=24
SD=1033.3cm2 Len=71cm

Duct 'D3' Node=24=25
SD=1516.6cm2 Len=71cm

Duct 'D4a' Node=25=26
SD=2000cm2 Len=49cm

Duct 'D4b' Node=26=27
SD=2000cm2 Len=22cm

Radiator 'RS'
Node=27
SD=2000cm2


----------------------
 

Attachments

  • POC-2.PNG
    POC-2.PNG
    16.6 KB · Views: 529
I think I've come up with a smaller version of s segmented TH for the same driver:


S1 600 cm^2
S4 1300 cm^2
L12 18 cm 7.1 in
L23 204 cm 80.3 in
L34 18 cm 7.1 in

Length 240 cm 94.5 in
Segments 4
Net Volume 228000 cm^3 8.05 cu.ft.
Fc 36 Hz


I gave up a bit @40 Hz, but ended up with a bit more above 100 Hz, which might be a better fit for the intended purpose of this box. Passband SPL remains about the same (~120dB). Box size is quite smaller too (version #1 had a net volume of 12.75 cu.ft.). Net dimensions work out to 31.5" x 18.75" x 24", which seem reasonable I think.

Shoe-horning a 15" in driver into this smaller TH might be a problem though! It might not be possible to mount the driver so that the magnet structure is facing into the start of the line, as the depth at that point might be too low (6 in.!). OTOH, if I mount it so that the magnet is facing into the mouth, it might likely shade the mouth too much - some ad-hoc experiments I've done with P.O.C #1 suggest that even a small amount of shading can drastically change the response. I might settle for a version that's a bit wider to avoid these problems.

Now I just need to get my hands on one of these drivers: http://www.parts-express.com/pe/showdetl.cfm?Partnumber=295-080.... :)
 
Why? It's a worse choice for this app than the POC. You want to be moving to smaller, high Xmax drivers to keep size reasonable in this BW or at least ones with a lower Vas, Qts.

GM

The parameters of that driver were used for modelling P.O.C. #2 :)

Efficiency of this model works out to around 99dB/1W/1M to 50 Hz.

Any suggestions for smaller, high Xmax drivers that can be used instead? I've put together a spreadsheet that basically chucks out the box dimensions when I put in the TH specs, it should easy to compare a few options.
 
Hmm, I use compression horn theory, so for the POC I get a 21.11-209.44 Hz BW to load it from its Fs - up with a conic expansion and just like in any cab alignment, the lower the Vas and/or Qes the lower its acoustic bulk, so the Dayton is going in the wrong direction WRT bulk. Plus, it has a lower Fs, necessitating a longer axial length to a lower HF corner, further increasing its bulk as it shifts its pass-band lower, hence my remark.

Well, in the big TH threads the MCM 55-2421, its TB variants as well as the CSS TRIO8 comes to mind. For some really serious SPL (up to 135+ in <12 ft^3 in a sim) there's the B&C 15TBX100 that DSL uses in at least one of their products if you still prefer a large driver, but multiple small drivers is the future for HIFI/HT/small prosound apps whether in a single cab or spread across multiple small horns that you can carry one on each shoulder.

GM
 
Hmm, I use compression horn theory, so for the POC I get a 21.11-209.44 Hz BW to load it from its Fs - up with a conic expansion and just like in any cab alignment, the lower the Vas and/or Qes the lower its acoustic bulk, so the Dayton is going in the wrong direction WRT bulk. Plus, it has a lower Fs, necessitating a longer axial length to a lower HF corner, further increasing its bulk as it shifts its pass-band lower, hence my remark.

Well, in the big TH threads the MCM 55-2421, its TB variants as well as the CSS TRIO8 comes to mind. For some really serious SPL (up to 135+ in <12 ft^3 in a sim) there's the B&C 15TBX100 that DSL uses in at least one of their products if you still prefer a large driver, but multiple small drivers is the future for HIFI/HT/small prosound apps whether in a single cab or spread across multiple small horns that you can carry one on each shoulder.

GM

I think that so much is different between a TH and "true" horn that it's dificult to apply the same theories to both. Look at mouth size, for example. The mouth of my POC is only 1300 cm^2, but the sims suggest response down to 40 Hz.

One thing that sticks out though from the sims I've done so far is that it's difficult to get an efficient TH with a smooth passband that goes above 100 Hz if you start with a driver that has a low Fs. All the sims I've tried so far suggesting a starting point for Fs of ~ 40 Hz. Then the Qes can't be too low either - this seems to affect the S1:S4 ratio, and if it's too small, the throat gets almost impossible to implement in an actual box (some designers are opting to place the magnet structure in the mouth to get around the problems with small S1, but this can introduce response problems because the magnet structure is now obstructing an area where velocity is at or near maximum). Finally, Vd is also important for peak output, and if you start with a small driver, it either has to have massive excursion potential or you have to use multiples of them, driving up cost.

The sim suggests a clean 120dB in the passband for a box that's a bit above 8 cu.ft. net for this 15" driver. I'm going to try out one or two of those MCMs in a sim, but I'm going to bet it's not going to approach that, at least not for the same passband. I could also swap the Dayton driver for a Kappalite 15, increase the box size a bit and get another 5dB of output for the same passband (I might even opt for that :)).
 
There's at least two valid ways to mathematically approach designing a TP/TH, either as a segment of a much larger horn and/or as a 6th order BP taken to its logical extreme and in either case you're always trading cab efficiency for BW and vice versa same as any other speaker alignment regardless of the driver's size.

Anyway, when I use TL/horn theory as a basis to design TPs/THs I get 'close enough' simmed matches to the DSL products that I have enough tech data on to compare them (once damped) that I feel I'm in good enough 'company' not to be swayed by your 'argument'.

Regardless, thanks for sharing your 'adventures' in (TP/TH/BP6) horn design. Looking forward to what you find out and why/what you ultimately choose for the app.

GM
 
There's at least two valid ways to mathematically approach designing a TP/TH, either as a segment of a much larger horn and/or as a 6th order BP taken to its logical extreme and in either case you're always trading cab efficiency for BW and vice versa same as any other speaker alignment regardless of the driver's size.

Anyway, when I use TL/horn theory as a basis to design TPs/THs I get 'close enough' simmed matches to the DSL products that I have enough tech data on to compare them (once damped) that I feel I'm in good enough 'company' not to be swayed by your 'argument'.

Regardless, thanks for sharing your 'adventures' in (TP/TH/BP6) horn design. Looking forward to what you find out and why/what you ultimately choose for the app.

GM

Actually I haven't looked at the mathematics of tapped horns at any serious level yet. I'm basing my observations solely on the HornResp sims that I've done (or attempted to do) since building POC #1 for a number of drivers, including the suggested MCM driver (added that to the sim database yesterday). The results indicate path lengths that are similar to horns, but mouth sizes that are not. I did come up with a TH with four of the MCM drivers that in the HornResp sims appears to be more efficient in the passband I'm aiming for, but with S1=136cm^2 and S4=900 cm^2, it's probably going to be impossible to implement in any meaningful way. I also couldn't come up with something using the Trio8 that I was really happy with. The B&C driver looks interesting, but costs are bit out of my budget.

Thanks for your help so far - very much appreciated. I'm certainly going to review a few more options before settling on a particular driver for POC #2.
 
Hmm, I use compression horn theory, so for the POC I get a 21.11-209.44 Hz BW to load it from its Fs - up with a conic expansion and just like in any cab alignment, the lower the Vas and/or Qes the lower its acoustic bulk, so the Dayton is going in the wrong direction WRT bulk. Plus, it has a lower Fs, necessitating a longer axial length to a lower HF corner, further increasing its bulk as it shifts its pass-band lower, hence my remark.

Well, in the big TH threads the MCM 55-2421, its TB variants as well as the CSS TRIO8 comes to mind. For some really serious SPL (up to 135+ in <12 ft^3 in a sim) there's the B&C 15TBX100 that DSL uses in at least one of their products if you still prefer a large driver, but multiple small drivers is the future for HIFI/HT/small prosound apps whether in a single cab or spread across multiple small horns that you can carry one on each shoulder.

GM

I've noticed that Danley begain using multiple woofers in many of his tapped horns. I think this is to smooth the response. Or at least Akabak indicates that it will.

The glue is drying on a new design, with three of the ubiquitous MCM 55-2421 drivers. I've done some preliminary measurements, and it has usable response over three octaves (40 -320hz.) There's a notch around 180hz, but it's narrow, probably inaudible.

This one should be a good all-around design, particularly compared to the two I did previously. The first one basically didn't work. And some people on here warned me it wouldn't (including you!) The second one, based on the TH-Mini works great. But some whiners complained that 4.5cf was too big for a single 12" :D

So the new box trades some efficiency for box size, and uses woofers with much higher xmax to increase maximum output. It's harder to build, but it's cheaper.

I'll have some pics and measurements posted soon.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.