X/Over Help


2001-12-27 4:01 am
The match looks good, but to calculate a x-over I need some more info:

Necessary :
- Front baffle dimensions & shape
- Position of speakers on the front baffle (just tell me the coordinates of the speakers centers, is needed to calculate the interaction in x-over range)
- Will the speakers be mounted on a flat panel or will the box be inclined or the speakers mounted on separate baffles to compensate 'mechanically' for the emission delay?

- Bass loading system (Vented, Closed, volume, tuning etc.)
- Box type (bookshelf, column, b.shelf on free air stands...)

.. I promise to come back to you in a reasonable time, it takes a while to simulate a system and this guys at SEAS measures the response in a non standard way!



2001-12-27 4:01 am
I have made a quik change to the drivers, i will be using the Audax TM025F1


A bit about the system

- Front baffle dimensions & shape
width = 216mm, height = 466mm
- Position of speakers on the front baffle
a TM config, approx 3 cms apart
- Will the speakers be mounted on a flat panel or will the box be inclined or the speakers mounted on separate baffles to compensate 'mechanically' for the emission delay?
im hoping these new drivers may take away most of the emmision delay, if not, then i may see what i can do.
- Bass loading system (Vented, Closed, volume, tuning etc.)
vented, 15 litres, 49.40Hz
- Box type (bookshelf, column, b.shelf on free air stands...)
b.shelf on free air stands

if there is any more assistance just post or email.


sandro has helped me design my crossover too. the crossover is very specific to taste.

for example I prefer low Q crossovers as I prefer a laid back sound. Hence the need for bigger inductors.

Try the crossover sandro recomends. if you find it a bit bright tinker around with the tweeter indcutor a bit (dont go too far from his recomendation) and remember that the tweeter series capacitor will have to be changed to compensate for any change to the inductor.

Sandro recomended me a crossover (in fact 2) after a lot of emails back and forth. unfortunately I am in the middle of a new job, moving house, expecting a child and quite a bit of travel for work and have not been able to build the boxes for my speakers leave alone the crossovers.

The new job is about 75-90 mins from where i live so that takes a lot of time. I have to prepare for our first baby in April and a road trip in May.

We are also looking for a apartment nearer to my work (about 45-60 mins away). In Bombay we all live in small apartments. Usually one renovates the apartment before moving in. So once we find the apartment (the area I work in all industrial so there it is not a good residential area) we have to renovate it too.

When everything settles (hopefully by June/July 2002) I will start building speakers again.
loud depends a lot on the drivers you are using.

I dont think the SEAS drivers will go too loud.

also LOUD will be limited to the upper bass after that in the low bass excursion limitation will set in unless u use a sub.

below are details of my existing system.

My project had simple parameters.

1. capable of high sound pressure levels from 40Hz - 20,000Hz. Many speakers cannot produce high sound pressure levels at low frequencies and at high frequencies tend tend to sound shrill at high SPLs I wanted to correct that.

2. Decent imaging in vertical and horizontal plane from my listening position

3. not to occupy too much space and to be flexible to be converted into other rooms.

4. Simple to build and design as I do not have any software or measuring instruments and find all that work to tedious to begin with.

5. economical.

Since I wanted a simple system I targeted building a bi amp system. The bi amp system would have 2 parts. 1 bass less than 60Hz and the 2 main system.

The bass was built using 2 12" woofers (per channel) from Audio Concepts (www.audioc.com). At that time the best woofers I could get from them were the DV12 (dual voice coil) now the SV12 is supposed to be better as well as other drivers like the Hsu and Shiva. I got these drivers in a special sale at $98 each. Each driver could operate in as little as 2.5 cu. ft but 3 cu. ft. was a bit better.

Hence I built 2 boxes 30" wide, 24" deep and 18" high. These would sit on the ground like large coffee tables. I decided to fire the woofers backwards (towards teh rear wall) so that the woofers cannot be seen. Pleae note that large woofers with heavy cones like the DV12 should not be fired downards as the spider gets streched and over time the bass limit reduces.

In between these coffee tables I have 2 chairs where I normally sit to read my newspaper, office work and other serious reading.

The box (except for the baffle) was made of a sandwich using 18mm MDF and 2mm Lead. Hence the box walls were 38mm thick. After the box was made (no baffle yet) I took fiberglass cloth and layered the box walls with cloth as poured polyester resin on this cloth to create a 3-4mm thick liner inside the box (on the walls). Hence the box walls were now 42mm thick. Then I loosley filled 5 kgs of glass wool per box. The baffle (30" x 18") was made using 30mm MDF reinforced with 50mm x 50mm solid wood 6" apart along each dimension. Hence there were 4 reinforements in one direction (30" direction) and and 2 in the other.

The 2 woofers were mounted side by side and are powered by a Carver M1.0t amp.

The concept of the main system came becuase I wanted a system to handle large ammounts of power and yet have little distortion. For simplicity the system had to be a 2 way system. hence I decided to use a MTM.

The drivers chosen were part by accident and part by looking at all specifications. A MTM is best with either 5" or 6" woofers (although I had built a MTM using 2 8" Focal woofers caleld 8N515 and a 1" Morel tweeter called MDT33).

Combinations I looked at were....
Dynaudio 15W75 and D260 (also called Gemini)
Dynaudio 17W75XL and D260
SEAS Excel 17cm and Excel tweeter
Focal 7K415 and Focal T120tdx tweeter
Scan Speak 18W8545 and 9900
Scan Speak 18W8546 and 9900

each of these were rejected because of one problem or the other. The 15W75 and SEAS Excel had limited capability in low frequency sound repssure levels (the 15W75 had limited Sd and the SEAS Excel had limited Xmax). The Dynaudio 17W75XL had some matching problems and the Focal 7K415 had uncontrolled kevlar sound in the 1-3k region. The D260 was not as good teh Scan Speak Revelator, the Focal tweeter had a foam surround giving it limited life, etc...eventually the decision came between the Scan Speak 8545 and 8546 woofer and 9900 tweeter. A friend had 4 8546 wooofers and 2 9900 tweeters he had got cheap as demos and sold them to me at $60 per tweeter and $40 per woofer. So I got lucky. For $280 I got all my main drivers.

The box on the main system was built using a home made version of B&W's matrix box. I took a box made of 18mm MDF and using 3mm MDF made a 4" x 4" mesh of wood that was inserted in the box. This is a very tedious process.

Each box was 20" high, 8" wide and 22" deep. Each box was then filled with about 1 kg of glass wool.

I built a seperate crossover box about 8" wide, 4" high and 22" deep.

The XO box was kept on the SUB and the Main speaker was kept on the XO. However I found that the system was too low. The tweeter was only 32" from the ground. by adding books from my encyclopedia collections I realised that the best height for me was withh teh tweeter 40" from the ground. Hence I rebuilt the XO Box.

I built a 12" high stand. Then placed the XO components around the vertical pillars of the stand and then built a box (5 sides) around the XO components.

The passive XO between the 8546 and 9900 is also very simple.

It has to do 4 things.

1. make impedance relatively flat
2. protect tweeter from low frequencies
3. roll of high frequncies from midbass to prevent beaming and suppress upper midrange cone resonance
4. compensate for difraction loss

the tweeter crossover is as follows.

1. series capacitor (6.7 uf using 2uf wonder cap and 4.7uf polyester cap in parallel)
2. parallel inductor (1.0mh air core inductor)
3. series resistor try values between 1 and 2 ohms
4. parallel resistor try 10 ohms - 16 ohms.

1 and 2 are the crossover you can try various values. remember if inductor is small then the Q of the corssover goes up. I prefer the sound of low Q crossovers hence I use big inductors.

you can try values between 5uf and 10uf for the capacitor and 0.4mh and 1mh for the inductor.

3 and 4 protect the tweeter and also reduce the level of the tweeter by about 2db. I use 1ohms series and 16 ohms parallel. you can try other values

The woofer corssover is as follows:

1. series inductor paralleled with resistor - 1.0mh and 3.3 ohms
2. 0.43mh series inductor
3. 10 uf polyester capacitor

1. is the difraction compensation. this system does not require too mcuh difraction compensation as the bass part (till 100hz or so) is being compensated by the sub. Again I prefer to use guidelines and work between these.

for difraction compenstion circuit you can try inductors from 0.6mh to 1.2mh and resistor from 2ohms to 4 ohms. infact I tried resistors from 1ohms to 10 ohms to see the difference.

Hope this helps.

Hi Griff,

here are some preliminary info and warnings (!)

- Volume & tuning looks perfect for the cc17
I assume 15 litres net, that means 18 'virtual' litres seen by the speaker considering medium coibentation (= all walls except front covered by a 3-4 cm of glasswool or similar dumping material)

- For simulations, I consider tweeter as close as possible to upper edge of front baffle, woofer as close as possible to tw (recommended!).
Both speakers vertically aligned at centre of baffle

Now the warnings !
- The front panel dimensions looks quite big for a 15 litres. It means that the box will be very flat, and a small depht is not optimal for bass loading. If there are no specific issues, I would suggest to review the box shape.
Normally in small speakers you have the best overall tradeoff in minimizing the front panel dimensions and smoothing all the panel corners to minimize reflections.
- 92 db are quite optoimistic! Looking at the graphs, the Audax is around 90 db until about 6khz (that is in the highest energy range) and rises to 93-94 only in the 8-12 khz range
The seas is very efficient from 700 hz, but if you want to keep basses equilibrated we probably need to attenuate it a bit.
Let's say that a TRUE 89-90 is probably the best achievable, and really it is not bad!
In terms of wattage, you double efficiency every 3 db, that is that for the same SPL you need half of the amp power.

Ok, on next lunch break I'll try to run some graph and x-over design.

there is no secret formula to building boxes.

1. make them strong and rigid
2. I prefer using the minimum dimension possible for width and then adjusting height (to ear height) and depth to match.
3. try to keep the large faces (usually the sides) non parallel. You can bend 4-6mm MDF. using 3 layers of 6mm MDF with Foam and Lead filler inbetween will given you a 25mm box that is dead.
4. I prefer sealed boxes are they are more forgiving fo design flaws or anomalies in drivers T/S specs.
if i did go sealed, i will get a F-3 of 82Hz, which isnt really what i was after, having said that, my next project will most likely be a sub. I would prefer to go ported as i have already main done the design.

is there another way to curve the box? i would rather use the 33mm MDF as it would be alot easier and save alot of time and money.

attached is a 'quick&dirty' draft of a suitable box. Is an Excel spreadsheet, sorry but I'm too lazy to learn to use a CAD. Every cell is 0.5 cm square.

Net volume is a bit bigger to account for speaker magnet, and anyway a bit of more volume is never bad!

Given the thickness of the walls and the small size, I think you can save sophisticated bracings; if it was for my use I would just connect the two lateral walls with a couple of rods (see draft).
Sole recommendation: keep all the stuff you put in the box away from the vent duct mouth.

Have no idea on how to bend 33mm MDF, I would simply round 1-2 cm the front panel wall edges with a smoother.

I will play with the x-over this weekend, but to send you the files I need an e-mail address (attachment is too big for this forum) or some hint to show bmp or jpg in this messages.
If you like to keep e-mail private, mine is [email protected]: send me a mail and i'll reply there.

Sorry for rushing away, I'm really out of time today!

PS : I had to rename from xls to txt, xls seems not allowed. I hope I didn't made something hurting the moderator.


  • griff.txt
    72.5 KB · Views: 63

I had to wait for a file before leaving, so I had time to play with your x-over, and I'e also found the way to reduce file size.

Attached the usual .xls dressed as 'txt'.

Graph 1 : wf, tw, system response . Units mounted on actual panel, free air (anechoic) room simulation.
Plus impedance module.

It looks ok, the dip at 6khz is the panel reflection. This reflections exists in EVERY box, a good hint is to use felt or other similar material to surroun the tweeter.

Graph 2 is a 1/3 oct simulation at different vertical angles and 2meters. It gives a good idea of energy emission in the environment. It looks EXTREMELY good.

Graph 3 is same as 2 but response is simulated in a room, that means here we see the contribution given by the room reflections. As you see, bass level is re-equilibrated.
This is the CLOSEST approx of real response.

NOTE : the rising response above 6 khz is due to the chosen tw.
If you like a more flat behaviour, you should pick another unit (i.e. the Seas you mentioned earlier)

The x-over is pictured too, just pay attention to connect both speakers in phase (Navin, I know that theory calls for a phase inversion in a 2nd order x-over, but that's theory !!)

Good luck !


  • griff2.txt
    53.5 KB · Views: 70
Ref; to previous project.
amp for sub: Carver M1.0t
amp for main speakers: home made, using 1200VA torroidial transformer, 60,000uf of caps per channel, amp is a souped up version of B&K ST140 but with 3 pairs of output transistors per channel and slightly higher biasing current. the amp is cooled via 2 heatsinks each 2" x 8" x 10". The heatsinks are further cooled via 4 4" fans that come on at 55deg C.
crossover between 2 amps: 6db/oct I played with a lot of caps so have forgotten the exact XO frequency but it is simple single cap XO. I am thinking of building a electronic XO using OP275. However my preamp is such a simple design (full relay switching using 2 relays per input (a la Douglas Self) to reduce crosstalk and noise and only one OP amp (I tried a variety of op amps and settlesd on the 275) that I dont want to further complicate the design. I like the signal path as simple as possible. in fact the Vol pot is a parallel pot not in series with hot leg of input. I try to avoid anything that is not absolutle ciritcal in circuit path.

CD player: Modified 63KI. New Power supply (bigger and tightly regulated and mechanically isolated), New power cable, lead sheet damping on cover, sorbathane rubbber washers on every screw that I cold find. blu tak on chassis components to reduce viabration. Totally 4 kgs added to the existing player.

Feet of player: 3 high carbide steel balls that sit in a "soup spoon" like cavity. Soup spoons sit on glass sheet that sit on 6mm rubber sheet shore hardness 60 which sits on 6mm rubber sheet shore hardness 40 which sits on 30mm MDF table.

Still exprimenting with interconnects. right now using belden copper in hot leg and silver in return leg with WBT connectors.