August 31, 2011

Rave horn

Here is a quick design for a HUGE horn inspired by this thread:

Hornresp sim for 8 drivers loaded into a horn installed into a 1m high stage that is 7.5m wide. B&C 18TBX100

Now a tapped horn with the same driver.

Note that this is a single driver with 300w. Less power is required to find the excursion limits, but the efficiency is lower. The front horn has 107 db sensitivity while this has 100 db. This is 430L while the front horn is 7000L, in other words 16 x larger! However, the FLH doesn't increase much in size when more drivers are added. The mouth stays the same but the throat gets larger. The curious thing is that to match the output of the front horn, you need 16 tapped horns. The volume in the end is the same, but with the tapped horn you have twice as many drivers and 4.8kw of power.

So if the volume limit is the same, you can use 18 drivers + 4.8 kw of power, with 16 large boxes. Or 8 drivers into a single huge horn and 16 kw of power. 

August 28, 2011

Surround test boxes in progress

This is a test box for my 8" waveguide surround speakers

The baffle is double thickness under the driver and has a simple brace in the middle. 

Octagonal face plates are a pest! I've marked this one out in pencil but plan to flush mount using a router template. 

August 23, 2011

Drivers arrived! 2 x B&C 8PS21 & 8 x Pyle PDMR5

The drivers arrived today for my current speaker projects. Peavey Black Widow 18" shown also to give an idea of scale.

B&C 8PS21 - an 8" pro midwoofer shown in the front, Pyle PDMR5 (8x) shown in the middle along with a pair of B&C DE250 compression drivers. 

August 15, 2011

8" waveguide surrounds

I'm currently working on a very dynamic active surround speaker that will feature a waveguide for controlled directivity paired with a compression driver (B&C DE250) and 8" pro midbass (B&C 8PS21). 

B&C 8PS21 midwoofer
 Goals include:

  • THX reference levels of 105 dB at 3m from 80 Hz and upwards with significant headroom
  • low profile for on-wall mounting
  • voice matched to mains (active coaxial waveguide/horn with the same compression driver)
  • neutral and natural sound with a slight bias towards a mellow sound rather than a forward, bright or detailed presentation
  • sharp and focused imaging
The drivers

The compression driver is an obvious choice for a couple of reasons. I was able to get a pair second hand at a very attractive price and they are a perfect match to the mains. They easily meet all the goals and I've found this driver to be excellent value and sonically very good. 

 B&C DE250
 The woofer is the same one used by Geddes in his Harper speaker. It was also measured by John Krutke (Zaph Audio) and it performed very well in distortion tests. The response in his measurements is flatter than the datasheet indicates. Overall it appears well designed and the xmax is decent at 5mm.

The waveguide

The waveguide isn't yet decided. I'm considering either a custom made 8" oblate spheroid, or one based on the Pyle:

The Pyle is a low cost plastic waveguide and in previous tests at the waveguide events, it performed well both sonically and in measurements. The oblate waveguide requires a fair effort to build and has the axial dip caused by diffraction, probably a non-issue in this design where the off axis response is the focus. The Pyle does not have this issue, but it requires an adaptor for my waveguide and falls short of the standard of finish I'm aiming for. One option is to buy the Pyle and use it to create a mould, then casting the waveguide into the baffle resulting in a "one piece" look.

SPL estimate

WinISD estimate of output with different power levels - all at 1m

Cone excursion is kept under control by a high pass at 80 Hz. Even at high power levels, excursion is well within limits with a high pass filter in place.

At 3m, the output is reduced by ~4.5 dB, so to reach my goal of 105 dB, around 110 dB is required. Around 40W is needed to achieve this level and at the maximum simulated output, there is around 7 dB of headroom. This design will easily reach the targets.

Coming up:

1. Measurements when the drivers arrive
2. Initial prototype version
3. Experiments with placement and toe down

August 14, 2011

Black widow prototype - more measurements

Nearfield measurements
Step response
Magenta: Rythmik sealed
Grey: Peavey driver on an open baffle
Red: BP6 rear port
Blue: BP6 front port

Group delay

Phase response

August 13, 2011

Black widow prototype - 6th order bandpass

Before building the final version of my black widow active woofers, I decided to build a test version. I didn't fully trust the WinISD simulation of band pass boxes as it seems buggy and incomplete. This box was built, tested and then enjoyed in the one day. Many thanks to Antripodean who helped with building the box.

The box is 550mm wide x 580 deep x 620 high. The top of the box behind is about 900mm or seated ear level and you can see the 6" waveguide with foam in it at the top of the shot. The test box is similar in size to a 2 draw filing cabinet.

1. Testing driver parameters with the Dayton Woofer Tester 3

I did a quick test to make sure the parameters were close enough to the model.

Peavey Black Widow 18" (specs from Peavey in brackets)

fs: 39 Hz (36)
re: 6.5 (5.8)
Qts: 0.43 (0.37)
Le: 2.94 (2.07)

2. Simulation in WinISD

As you can see, there is a substantial difference in parameters, so I checked the simulation to see if it meant a different design. The two were quite close.

You can see a big peak from the front chamber port. This means WinISD thinks the port is too efficient, ie tuned too high for it's size, or the volume too large for the tuning point, or both. Reduce the front chamber size and you get a big dip in between the two humps. My plan was to EQ out the peak carefully, then yielding a flatter response than otherwise possible. However, the test box revealed that WinISD didn't really predict things too well.

3. Building the prototype

This is a simple unbraced box with an estimated net volume of 124L (12L for the front) with a shelf port for each. You can see the front chamber which is not much bigger than the driver:

The front chamber can be removed easily.

There isn't too much you can show when building a very simple box like this. Bunnings were out of 18mm MDF so this was built with 16mm. Not too much different, but given that this is a big unbraced undamped box, it rings like a hollow el cheapo boombox. It often can be a surprise how much you can get away with ...

Surprisingly I didn't quite manage to build the whole thing with one sheet. I had to add some particleboard flooring on the front chamber. Being a timber hoarder does have advantages.

Continue to measurements and listening impressions >

Black widow test box measurements and impressions

< Project intro

The measurements were surprising, better than expected based on the sims. 

What is a 6th order bandpass? Think of it as a vented box (the rear chamber and port) that has an acoustic front chamber added. This chamber creates another tuning point that sets the upper bandwidth limit. It filters harmonic distortion as well. 

Nearfield measurements

Red: rear port
Blue: front port

The front port does not have the predicted peak, it is quite flat. I also discovered that I had tuned it too low hence it doesn't reach the 300 Hz target. You can see the front port has surprising extension and the rear port

Electrical crossover filters have been added. 

The box was measured with the box in the middle of the main speakers. Hence there are some modes that haven't been tamed. 

Listening impressions

It's hard to comment too much because, as you can see above, there are some room modes in the response that normally I would have dealt with. The peak around 44 Hz is probably related to the sense of the bass being much more tactile than usual. It's fun and novel for a short burst, but would probably bother me over time because everything takes on a certain "sameness." However, for a thrown together box not fully integrated, all the signs are there that this is going to work well. 

I have a couple of specific goals with this box:
  • cover 40 - 300 Hz using bandpass to increase the efficiency at the top end
  • effectively meet the horn at 300 Hz with the acoustic source close to the horn (achieved with the slot port at the top)
  • reduce the "thermal load" on the subs with music
  • fit inside the large corner bass traps I have planned
So far it appears that this project will work out well. 

August 7, 2011

Point source horn driver selection

For my upcoming point source horn project, I'm considering two drivers. Misco JC5RTF and Pyle PMDR5. $500 for 8 of the Misco, and about $200 for the Pyle.

Both are similar in their response, but the Misco is more efficient and has higher maximum SPL. More than enough in both cases. My target is 105 dB in the listening position for midrange, at a distance of 3m with 160w. Working backwards from the target, SPL at 1m should be 4.5 dB greater to allow for distance, hence 110 dB rounded up. This target is exceeded by 10 dB at 1k, so there is considerable headroom.

The holes have all been combined into one for the model.

Hornresp settings used:

Only the driver input settings are different.

August 1, 2011

Peavey 18" active woofers

I'm currently nutting out the final details for pair of 18" active woofers. It will cover around 40 - 300 Hz with 96 dB 1W1m. The prototype will be a vented box of around 150L. Max SPL is estimated at about 120 dB in the listening position. Even at that level, there would still be considerable excursion headroom. At THX reference levels of 115 dB it will be cruising.

The driver

A nice deal came up on a Peavey Black Widow 18". The new version on Parts Express costs about $570 for a pair shipped to Australia. As I bought these second hand, the cost was considerably less.

fs: 36
SPL: 96 dB 1W1m
Le= 2.073 mH
VAS 230L
Qts 0.37
xmax 7mm
PE 650W

These will be powered with a Behringer Europower EP2500 which can put out 450w into their 8 ohm load. Turns out that much power is "just right" for this driver.

The nearfield response shown above was measured on an open baffle. You can see the response is quite smooth, but high inductance rolls of the top end early. At 200 Hz the sensitivity has dropped by 2 dB to 94 dB and at 300 Hz it is only 90 dB.

The box

The numbers work quite well for a 90L sealed box or 170L vented. The design that I have in mind will be more like 150L. The reduction is not significant and saves buying more sheets.

It will feature a composite of 18mm MDF and 17mm form ply, with a thicker front baffle. The sides and rear will be constructed out of one long continuous curve similar to the JL Audio Gotham.


These will be placed in the corners, installed flush inside huge bass traps so their size is less critical than usual. In that position SBIR is reduced and the overall room response is quite good. I will need to EQ out a large 15 dB peak centred at 44 Hz that is room related. This room mode has caused me to consider a sealed box, since the mode actually behaves like a vent port already. I will be experimenting with a sealed and vented version in the prototype. The bottom end will be augmented by a pair of Rythmik Servo subs. One of the goals with this project is to increase bass headroom to 40 Hz. Like most subs, the Rythmiks are thermally limited. It might seem crazy to have 18" woofers down to 40 Hz then 12" sealed subs below, but my future plans include tapped horn subs. Initial simulations indicate a pair of Rythmik 12" subs in a tapped horn will achieve about 98 dB 1W1m down to 20 Hz. Two of them (4 x 12") installed under the floor have an estimated max SPL of 126 dB. With the current pair I expect to achieve THX reference (115 dB but with greater extension). With a sealed box and the driver running at 36mm excursion, the output is 101 dB @ 20 Hz. A vented box offers around 106 dB.


The design has evloved, here is the bracing in 3D:

You can see the slot port at the bottom needs some work. The bracing has a H core where the curved sides aren't braced to each other. The idea with the bracing is to tie the flat surfaces together while still being open enough to be able to apply damping to the inside surfaces of the box.

Speaker box damping and composites

MDF is a cheap and easy to use box material. It works quite well for sub boxes when adequately braced, but it’s far from ideal for full range speakers due to the resonances around 300 Hz. Plywood is stronger and lighter and is considered by many to have better acoustic properties, but it can tend to ring.

The ideal enclosure for a full range speaker box is first of all very strong and stiff. This pushes resonances up higher in frequency where they can then be damped and more easily reduced.

Here is a quick ‘n dirty test of some different composite options, each evaluated by the sound of a small panel when knocked. This isn’t the ultimate test, but it does give some indication of performance.

The MDF and ply samples both had an obvious ring in their raw state, the ply actually being slightly worse in that regard.

Best option
4 layers of 3mm MDF were curved and laminated together with liquid nails to a total thickness of 15mm. Cornice adhesive reinforced with perforated metal was added to half of the sample.

Knock test:

MDF and ply were sandwiched together. In one part of the sample, cornice adhesive and pegboard were added instead of the MDF. Performance was about the same.

Pegboard and cornice adhesive was added to ply.

A composite was made with a polystyrene core connected to an outer layer of pegboard with cornice adhesive and perforated metal reinforcement.

 MDF and ply made a good combination, a dramatic improvement over each on their own and also better than a double thickness of each. Their different properties appeared to combine well. A small amount of damping is offered by cornice adhesive applied to ply sufficient the damp the ringing. The polystyrene core composite didn't perform well with pegboard - it lacked inherent rigidity. Curve laminated MDF with cornice adhesive was the best option. 


The next day I discovered the cornice adhesive bonds poorly in all cases. I may have been a little impatient since this stuff needs time to fully harden when thick, but I found that in all cases it didn't hold very well. So in the end the most effective turned out to be the ply/MDF combination. I will be investigating other damping methods.