Custom HT speaker project. AA1014DR2A/B - Home Theater Forum and Systems -

Thread Tools
post #1 of 3 Old 07-29-13, 06:03 AM Thread Starter
Join Date: Jul 2013
Posts: 16
Custom HT speaker project. AA1014DR2A/B

Hello HT shack. My name is Eric and I am building myself some home theater speakers this summer. If it's alright I figured I would "blog" this build in a DIY forum for folks to take advantage of. I should state up front that I have some itch to go commercial with this hobby but have a number of unknowns regarding the use of various software for profit purposes and would need to hash out a lot of details to move on that. For now, this is "hobby" and as such, I feel that the project deserves to be shared since it would not be possible without all of the amazing (free) simulation software available. If this project turns out to be successful anyone is free to replicate it for non-commercial use.

A lot of the custom speakers on the web get names. I don't have a name for these at this time so have just come up with a "model number" of sorts to keep track of the project by for now.


Allan Audio Speaker design:

AA1014DR2A-CA/B and AA1014DR2B-CA/B
(Allan Audio 10” x 14” baffle, direct radiator 2-way versions A and B, Cabernet/black finish)
Compact dedicated passband theater and critical listening loudspeakers

Design Goals:
  • Small bookshelf size speakers, no larger than ~1ft^3 OUTSIDE dimensions
  • Hardwood-Plywood construction from aesthetically acceptable but cost effective material. (probably 3/4" maple, oak, or birch from lowes, must to be relatively inexpensive)
  • Box to be finished in high gloss (maybe?) with cabernet stain, front baffle to be finished in a contrasting color/finish yet to be determined (or the inverse of this setup)
  • 2-way direct-radiator speaker system
  • Baffle size: 10”x14” maximum
  • Intended for indoor acoustic reinforcement in rooms of ~1500-3000 ft^3
  • Capable of reproducing “reference” levels(1) with exceptional linearity through the intended passband from any single speaker at the listening position without the aid of esoteric amplification levels(2).
  • 3-6dB of dynamic headroom above the linear goal stated above with acceptable distortion levels(3)
  • Version A: for traditional aesthetic placement(4)
  • Version B: for non-traditional placement(5)
  • Off axis response to provide accaptable linearity for broad range of possible seating/standing positions.(6)
  • Present an electric impedance that is compatible with traditional consumer amplification(7)

1. Defined as: 105dB @2M from ~80-16,000hz.
2. Should achieve output goals from amplification levels found in most consumer HT receivers, ~100W/ch.
3. To be determined.
4. Average Listening position matches elevation of the -middle- of the speaker. Off axis optimized for parallel speaker alignment.
5. For high and low speaker placements often associated with home theater installations. Invert speaker for “high” installations (tweeter on bottom).
6. Off axis maximum for goals defined as: +/- 15 degree in the vertical and +/- 45 degree in the horizontal axis from the design axis (Version A to have a mixed design axis, to be explained later, version B to have “design” axis set ~15 degrees off axis vertically)
7. Zmin: 6 ohm or higher. Nominal impedance: 8 ohm or higher. Assume reasonably pure voltage-gain based behavior from 6-16 ohm from most consumer HT receivers.

Considerations and contemplation's to meet design goals:

Dynamic requirements within a small size:
  • The dynamic requirements of this speaker demands the consideration of efficiency-bandwidth characteristics of drivers. The dynamic goals will be challenging. The final(1) efficiency must be ~91dB@1W1M in order to achieve 105dB@2M@100W.
  • Achieving this dynamic goal through the bottom octave of the bandwidth goal demands the use of a driver with adequate linear displacement and characteristic sensitivity.
  • WinISD has been used to great extent for quick-n-dirty simulations to feel out the bandwidth-vs-dynamic capabilities of driver options.
  • The 105dB@2M goal should be met entirely within driver linearity (mathematical Xmax limits)(2) and RMS thermal limits. Very few drivers appear suitable.

1. After BSC
2. (Hc-Hg)/2

Room, baffle and listener position relative to speaker related acoustics:
  • Simulation of the drivers on the intended baffle size show baffle step transition to 4pi space loading occuring centered at ~240hz (-3dB from “2pi space” level).
  • Boundary reflection support from the room and nearby objects (primarily the nearest wall) is centered ~160hz (+3db from 4pi space) (based on a a few rough calculations and some research into the “Schroeder frequency.”) Transition back to approximatively 2pi space loading is anticipated to occur in the bottom octave of this speakers intended bandwidth. Placement will cause this to change so there is no absolute solution here, we're shooting for a good compromise.
  • Room is anticipated to be furnished (plush), carpeted etc. upper midbass reverb related gain is not anticipated and not to be corrected for.
  • Accept that achieving flat response in the bottom octaves is highly unlikely. Trying not to overburden the process with trying to “fix” this range. Do the best we can, let audyssey do the rest!
  • Version A speakers, which are intended as the front stereo pair, while technically having a “design axis” directly in front of the middle of the front baffle, should have their response flattened based on slightly off-axis (horizontally) simulations in order to correct for the fact that when arranged parallel to the room (for aesthetics), it's impossible for any listener to be in the design axis of both speakers (the stereo pair) simultaneously. In other-words, response is optimized on version “A” for speakers arranged with NO tow-in.
  • Version B speakers (off-axis design axis) will often be located in places with more boundary reinforcement/reflection (center channel position under a television, nearby entertainment center and contents, etc.). Preserve sensitivity with less BSC. Rotate the design-axis vertically ~15 degrees to accommodate high and low speaker placement and favor midrange sensitivity over flat response. Design and response axis is “on-axis” horizontally, 15 degrees vertical. (Adjust crossover components to rotate the crossover “lobe” to an acoustically in-phase, flat response at the design axis)

Response correction related to lower frequency room/baffle effects.
  • After considering the balance of BSL and boundary gain, I intend to allow for a ~3dB roll-off in response in bottom octave to accommodate rising room boundary/support in this range and correct for baffle step losses in the octaves above this transition. There are 3 ways to achieve this depending on the driver chosen:
  • 1. Use a higher Q alignment of the low frequency to provide approximately 3dB of BSC, leave the remaining 3dB uncorrected at the bottom. (Lower EBP (~125) midbass driver).
  • 2. Only apply ~3dB of BSC to a midbass solution whose EBP provides a 2pi simulated, flat response through the passband. (EBP ~150-175).
  • 3. Apply a full or near-full BSC correction down to ~200hz, leaving intact a naturally falling response from a very high EBP (>200) driver tuned low.
  • The small box requirement and high efficiency requirement is apt to force things in the direction of #2-3 to get a desirable result. It's often more desirable to be able to “smash down” a few dB of response with the network in order to correct for peaking response issues. Options 2-3 above will have more opportunity to fine tune the response without dropping below the minimum sensitivity needs of the system.

Crossover considerations required to achieve desired horizontal axis performance (or better)
  • Simulation suggests that the highest possible crossover point to a tweeter to achieve desired horizontal off-axis performance is ~1800hz or ~2400hz, for an 8” or 6” midbass driver respectively.
  • Achieving the desired vertical axis performance with the above crossover points is not a problem provided the drivers don't share a wide band transition. A tightly controlled, steep crossover design has been simulated and has shown lots of promise for achieving the vertical axis goals with only minor changes to response.

Set a budget goal: Maximum $200 combined material cost per completed speaker (I can dream right?)
Driver maximum: $100
Crossover components, wire, terminals: ~$50
Wood, Fasteners, glue, port, stain, gloss, consumables: ~$50

Choose drivers: Must fit the physical limitations, show promise to achieve the dynamic capabilities desired while staying in budget (winISD used here for the quick and dirty to qualify/disqualify midbass drivers). Any drivers ABOVE 91dB@1W@1M efficiency should be considered if they fit the budget.

Midbass Candidates:
  • Celestion TF0818 $50. Best in group <chosen. Simulation shows linearity in this application to 117dB@1M2pi.
  • Eminence Beta 8A $60. This would be a near direct swap in this application for the TF0818, and while I like the higher thermal overhead, the celestion sims out a little better and has an easier to work with response.
  • Eminence Alpha 8A $50. This falls into the low EBP group and while workable, would not really give me the specific response profile I want on the bottom end for this project. Doesn't make sense to do combat with this driver when the TF0818 is the same price.

Lots of other drivers were looked at, including the carbon fiber 8” MCM cast frame driver, the cheap faital pro 8”, and a hand full of others. The only drivers that appear on paper to be better than the TF0818 for this project, are over $100. The RCF 8” midbass “monitor” drivers for ~$130+ have very similar characteristics to the TF0818.

Tweeter Candidates:
(I spent a lot of time looking at Zaphs measurements here, and have tried to take advantage of that information to the best of my ability)
  • Vifa DX25 $25. 0.25mm Xmax, high sensitivity, low distortion, low x-over capable with steep slopes. CHEAP!
  • Morel CAT 378 $63. Too expensive, less than half the Xmax of the DX25, neat horn load would be nice to elminate diffraction issues but I can't justify it without more information regarding how this driver performs outside the gap.
  • SB29RDC $55. This one could almost go as a direct swap for the DX25 with similar response profile, impedance, xmax, etc. Tall order distortion all lower than the DX25 but at more than double the price I'm inclined to just use the DX25 and save the bucks. If this driver were $20 cheaper and didn't come out of the package looking like a demo speaker that a kid poked I'd be all over this.
  • SB29RDCN $60. I love the small flange and insane efficiency but the distortion profile forces a higher crossover that won't meet my horizontal off-axis goals on an 8” midbass.
  • Seas 29TFFW $55. Again it's hard to justify the price increase for sonically minuscule advantages over the DX25. I prefer the lower impedance tweeter to start with as it gives more room to hammer out a flat on-baffle response.

I have chosen the DX25 and TF0818 for this project.

Note: This project will start as an “all” simulated (no measurements) design and will hopefully be measured at a later date as a way to check for the accuracy of this procedure.

The simulation procedure I have used to get to the point of simulating a crossover is as follows: (I'm writing this as "instructions" in case anyone is interested in how I like to do this).

1. Decide on a baffle/box size/shape and drivers. 10”x14” baffle, 10”x12”x14” overall dimensions, TF0818, DX25. Done
2. Begin building data files.
3. Gather data sheets, create a folder to work out of and keep organized, there will be many files produced in this process.
4. Use SPLtrace to trace impedance and response plots, note what voltage and conditions were used to produce the response plots as they may need to be normalized for use in the same simulation later. When I can tell that a response chart has been “smoothed” to excess, I may purposely exaggerate the peaks and valleys when tracing slightly.
5. Label the 4 files appropriately, I usually give these files names like “DX25 raw response trace.frd”
6. I then use hornresp to simulate the box I intend to use. In this case, I simulated a ported enclosure with a flared port. Set Hornresp to Nd single driver. Set the chamber size to the “box” volume, and build a 3 segment horn (S1-S4) with 2 exponential flared ends. I simulated one of the 3” flared ports from partsexpress and adjusted the length to get an 80hz tuning frequency. Set Eg to match the planned final simulation “level.” Calculate a combined response and export an SPL/frequency and an Impedance/frequency file.
7. Use a spreadsheet program to remove phase information from the 2 files above and replace with “0s” (3rd column should be all “zeros.” Export as a space delineated .csv file. Change file extensions manually to “.frd” and “.zma” where appropriate.
8. Open Speakerworkshop and import the traced and simulated response/impedance data for the midbass driver. Choose a place to splice the impedance together. Box simulations typically hit Re at zminimums which is probably not quite accurate and doesn't take into account the resistance in the path TO the driver from the network (upstream, so to speak). I typically transform the simulated box impedance portion upwards to meet the traced impedance data, which is apt to me more realistic from a position on the circuit that is actually being simulated. Furthermore, in a 2-way design like this, the low frequency impedance information is actually not important, I just like to do this extra step so that my final simulation shows the actual in-box impedance characteristics of the completed speaker. (you could just use the traced free-air impedance in this case and the crossover would simulate just fine).
9. To splice the impedance, you actually have to truncate the 2 zma files in speakerworkshop, then export the truncated files, then use a spreadsheet to “splice” them together manually. Again, export as a space delineated .csv and then rename to a .zma.
10. The in-box response and traced response can be spliced together right in speakerworkshop. Pick a good spot to splice the data, then export this combined response.
11. Now we have in-box “2pi” response data for the midbass and tweeter. Now lets simulate the minimum phase.
12. Open Visaton's “Boxsim.” Start a new project, adjust resolution of project to like “500” or so. Begin work on “driver 1.” Use whatever means necessary to give boxsim a complete set of T/S perimeters for the driver. Leave Re, mH empty. Select “internal enclosure,” “Use frequency response,” “use impedance response.” Import response and impedance data. Uncheck “starting around 2fc.” Select “box which is simulated.” Click “calculate Re etc....” swtich to “enclosure and impedance” tab, select “integrated enclosure.” Select “baffle and position.” uncheck “use common housing.” Uncheck all “sides.” Set diaphragm diameter and type. Hit “OK.” Edit the network and put the amp and driver on it connected directly. hit “ok” and allow the program to run calculations. Check the response chart, it should look exactly like the chart you already had for the driver. Check the phase response chart. This is your simulated minimum phase. It should look like sloped line, might wrap depending on the driver characteristics and response. Export the projects response/frequency/phase data. Use a text editor to open this exported file and replace all commas with periods, as well as remove the first line of text. Open in a spread sheet and convert to a space delineated .csv. Rename file as .frd and give it a name like “DX25 response+phase.frd.” Do this for both drivers. (you're just using boxsim for the phase simulation here, don't let it apply any box or baffle related response to the project, make sure you always do this operation with boxsim set to “internal enclosure” in both places this can be set to prevent it from trying to transform the response.
13. Now we have 2pi space data for the drivers (response, phase, impedance). Time to simulate the baffle, and, if you desire, a design axis. I use the program “edge” for this. I don't think i need to give step by step here this is a very intuitive program. Just make the baffle, set up the driver size, position, and mic position. Export the uncorrected frequency and amplitude data. Use a spreadsheet to add a column of zeros (empty phase data), and export as a space delineated .csv, then rename to a .frd. Do this for both drivers.
14. Import the baffle response .frds you just created to speakerworshop. Transform both -6dB. Combine (multiply) them with their appropriate response+phase (and response+box+phase) response files you already have created.
15. Create drivers, assign impedance and response data, create a network and start playing.

I'm sure someone has to be thinking: "WHY???? WHY all that work??? Just use the spreadsheet programs out there they will consolidate this!!! The reasoning is pretty simple....
1. I can't (no excel)
2. I can't (not going to pay for excel)
3. I can't (not going to steal excel)
4. I can't (I run linux)
5. I think this procedure may actually offer more flexibility along the way, but I have no way of knowing for sure
Interestingly enough, yes, ALL of these other windows programs work just fine in linux (wine).

There's an Image "tag" right -HERE- but it's not showing up now. Maybe I'm too virgin to be able to post images yet. Not sure.

Driver positions on baffle:
Drivers are arranged centerline on baffle, with the option to offset the tweeter 1” either direction. (There is a response blip that I could not hammer out in the crossover that would be partially solved by moving the tweeter over 1” from the center-line, either direction would work in this case).
TF0818 is centered 5” from bottom of baffle. DX25 is centered 2.75” from the top of the baffle, this should result in tight spacing (~0.1”) between driver flanges.
Port goes on the rear of box

The above simulation I really like as a starting point. The VERY steep crossovers are very much intentional here as they provide a lot of fine tuning "handles" to shape response, but more importantly, since this is a highly dynamic design with a very low xover on the tweeter, I need sudden and extreme attenuation of the tweeter in order to knock out distortion problems from the tweeter. I wound up with closer to 4dB of effective BSC which is more than I really wanted here. Raising the tweeter level to match a less compensated design causes the impedance minimums to dip lower. Current design has an impedance minimum of 5.93 ohm at 9.5K and corrects for the tweeters natural rolloff up top. I may yet (especially when I start to hash out design "B" ) tinker with the option to simply have the upper octaves roll off a bit to recover some average sensitivity.

Well, that's enough for now. I'll try to update this as the project gets some wings. I should be ordering components and making sawdust sometime in the next week or 2.


Last edited by mdocod; 07-29-13 at 07:07 AM.
mdocod is offline  
Sponsored Links
post #2 of 3 Old 07-29-13, 10:21 AM
HTS Senior Moderator
fusseli's Avatar
Join Date: May 2007
Location: The Inland NW, USA
Posts: 1,788
My System
Re: Custom HT speaker project. AA1014DR2A/B

Welcome to HTS and thanks for sharing your philosophies and the process that you use. What software is that that you've posted images from? That's too bad that you can't use the Excel options out there for design.

I do see that the impedance phase approaches -60 near 2k, which is on the high side. Most people say to try and keep it under 45. I don't see any tweeter attenuation in your crossover circuit? Are you sure that your doubled up 4th order electrical slopes will be producing acoustic slopes that will play nicely together?
fusseli is offline  
post #3 of 3 Old 07-29-13, 12:29 PM Thread Starter
Join Date: Jul 2013
Posts: 16
Re: Custom HT speaker project. AA1014DR2A/B

Hello Fusseli,

SpeakerWorkshop is used for the simulation.

I need to work on that impedance phase; Good catch I'm still going in circles making minor adjustments based on what part values are available. Maybe I can wean some of that gnarly power factor out of the design, or at least shift it to "higher ground."

The tweeter is attenuated but not with a resistor. Keep in mind that we're starting with a response curve on the tweeter that is naturally tilted downward (both from the natural response, and exaggerated further by a slight off-axis goal for response. I wanted to attenuate and SHAPE response at the same time. In the same way that big inductors are used in the low pass to tilt and shape the response to knock out the step loss and the baffle peak, the small value caps are being used in the high pass to change the slope and contour of the response.


Are you sure that your doubled up 4th order electrical slopes will be producing acoustic slopes that will play nicely together?
The simulation shows an acoustically flat, in-phase response on the design axis through the narrow x-over region and >30dB/octave acoustic slopes on either side. Hard to ask for much more than that

Last edited by mdocod; 07-29-13 at 12:58 PM.
mdocod is offline  


aa1014dr2a or b , custom , ht , project , speaker

Quick Reply

Register Now



Confirm Password
Email Address
Confirm Email Address
Random Question
Random Question #2

User Name:
Please enter a password for your user account. Note that passwords are case-sensitive.


Confirm Password:
Email Address



Activation requires you reply to an email we will send you after you register... if you do not reply to this email, you will not be able to view certain areas of the forum or certain images... nor will you be able download software.


See our banned email list here: Banned Email List

We DO NOT respond to spamcop, boxtrapper and spamblocker emails... please add @hometheatershack DOT com to your whitelist prior to registering or you will get nowhere on your registration.

Email Address:


Human Verification

In order to verify that you are a human and not a spam bot, please enter the answer into the following box below based on the instructions contained in the graphic.

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML is not allowed!
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


For the best viewing experience please update your browser to Google Chrome