3-way vs. 4-way

[dn92]

Given this is an idea for a friend and if you don’t have the drivers or the horn your questions are just tempting conjecture

Without a working knowledge of the complexities involved it’s guess work.

Without some real data it’s a guess on what may or may not be a preferred option until such time as you have the horns and some test equipment to measure some approaches to implementing it

If you are looking for more certainty pre purchase of components other than bouncing off ideas it gets back to following a known design.

The closest design is the Tad 2 way

Follow that closely and you will have a higher likelihood of success

The hypex horn allows good loading down to the desired cross point @650 hertz in a 90x40 angle

Yuichi points to some of the compromises discussed such as path length delays on his website

Note the virtual acoustic centre of each driver is seldom the location of the voice cool with the inclusion of the crossover filters

The phase response includes both the drivers physical and electrical characteristics and the crossover filters electrical characteristics

Without some advanced test equipment and powerful sumulator like LinearX LMS and LEAP 5 you would find it difficult if not impossible to implement an optimal design

So what’s the point ?

That’s why you pay the big bucks for a commercial product

Some insights

The 650 Hertz low pass crossover point and a steep crossover point offers several benefits

Lower vertical nulls or lobbing in the crossover region
Minimise the impact of cone breakup beyond piston range of the woofer

Attempting more than a 50mm recess of the woofer is not ideal

Typically a steep 36 dB crossover may incorporate some acoustic roll off from the woofer and group delay.

This under simulated conditions provides phase compensation so the crossover transfer function is close to optimal

The assymetric 12 dB high pass filter in combination with the horns natural high pass function provides the optimal final acoustic Response

Typically this can be validated by reverse of the driver phase and with fine adjustment a uniform deep null > 20 dB as measured at the crossover is visible

If you use a digital amp with inbuilt dsp a lot of the hack work problems above can be resolved for the diy builder BUT you still need to be able to measure it properly

Th-4001 can be loaded lower than 650Hz. There have been several DIY loudspeakers inspired by Exclusive 2402 or Rey Audio loudspeakers, using JBL 2446 or 2450 and crossed at 480/500Hz. It is realy working well, I have some friends who have this configuration. The trick is to do a 24/12dB cross over searching for phase aligment between LF unit and horn/compression driver unit.

[Ian Mackenzie]

A shift in frequency (spread) is another method of phase alignment

In fact this is used in a new version of the Altec 604/605 by someone

As l mentioned earlier the newer digital diy modules have adjustment for active crossover networks that help the diy person though the maze of crossover optimisation

That assumes you are comfortable with the signal in the digital space

Passive crossover network design is very complex and time consuming

Active is a better way but still requires optimisation (as text book filters do not account for the characteristics of individual drivers)

I hope to obtain some Tad / Yuichi) horn clones in the new year (given the interest) and work up a validated passive and active network

I realised this may appear over kill here but looking at other diy forums such as the seos wave guide diy space there are a number of people doing optimised crossover networks. That is with cheap drivers compared to Jbl

[dn92]

There is a 24/12 passive crossover at 480Hz here: http://www.cinetson.org/phpBB3/viewt...?f=113&t=38379 made by two skilled former sound engineers, that were used to work with Rey Audio loudspeakers, one at Radio France and the other one on some well known recording studios in Paris.
LF speakers were modified (lower fs) versions of Davis 40rca15, and hf was JBL 2450J in IWT TH-4001 clones.
As stated in the picture, there is a copyright on it and it cannot be used for making money with it.

This passive crossover is easily convertible in an active version, but using digital crossover, especially for crossing around 500Hz symetrical filter orders can be used adding a delay for the LF, as well as FIR filters can be used.

Another possibility is to do a passive correction of the horn response, such it was made in TAD TN-3, and let the crossover part in active, either 36/12, 24/12, 24/24 with delay.

[Robh3606]

Another possibility is to do a passive correction of the horn response, such it was made in TAD TN-3, and let the crossover part in active, either 36/12, 24/12, 24/24 with delay.

That can work very well and what JBL did for the Bi amp version of the 4430/4435. I have done that as well with my active set-up using more modern wave guides. Makes things a bit easier but you need measurement and simulation software to really get it right as you would also need using DSP.

Rob

[dn92]

If the driver is a TAD 4001 or 4002, just have to clone TN-3, or better a TN-1 or TN-2 HF but without the HF filter part. You benefit from the level attenuation and therefore less noise from the amp. Valeus for TN-1, TN-2 are available at least by buying the service manual, TN-3 can be estimated from them and some available photos on Internet.

If the driver is a JBL the EQ is a bit different, but can be inspired from the schematic published in the French forum for which I gave the link. In this French forum thread there is another one "à la JBL", only with serie RLC for the EQ, located after the LPAD.

With a passive crossover simulation software such as XSim it is easy to simulate the EQ, and the search for the optimal crossover can be made with an active crossver testing several combinations.

[bldozier]

i am far from a professional stereo designer, however my feeling if the sub is forward firing to the listing position
and playing sting instrument music, with heavy vocal. my opinion would be for the 3way system.

if crossing over at 800hz with a short playing distance i believe the system db level should blend nicer.
with no live music or classical, i don't think at that distance, you would miss the 2123, tied to a 2234H.

if you where playing per say, led zeppelin, with long introductions an a hard drum, i think the 2123 would benefit your
landscape, especially at a closed distance as you described, but i would also like to recommend a 4 way with the 2123.

my feeling is with drum and other percussion instrument at a 300hz cutoff a 21xx is needed to blend the sub, especially if the cabinet
is 43xx by design an not L212 standing, the placement of the bass energizer, tied with the forward firing 21xx should blend that 4 way well.

[Dr.db]

Thanks a lot for all the inspiring answers!!

To summarize this, there seem to be two major opinions.
Most prefer the 3-way setup because of better driver integration and a more harmonious playing of all drivers together.
On the other hand, some favour a 4-way because of the midrange punch of the additional midcone.

The filter-desgin:
Tad used a asymetrical 36db/12db filter for the woofer/horndriver. But what about JBL ? Most networks I´m aware of are symetrical 12db/12db-designs...
Concerning the crossover frequencies; TAD crosses at 650hz, JBL crosses most times at ~850hz (like Everest 66000...). Why does JBL cross that high even with 2"-drivers on big horns?

Dispersion match-up:
We talked about matching the dispersion of the woofer with that of the horn. The 90° horizontal dispersion of the yuichi matches the dispersion of a 15" cone very well.
But how about the horn matching the slot-tweeter?
The JBL 2405 has a 140° dispersion that matches perfectly with the 2397 at 10khz. But the yuichi´s 90° wouldn´t match well with 140° of the 2405!!

@Ian:
Which drivers are you going to test with the yuichi?

[Ian Mackenzie]

I don’t have a Yuichi horn or a two inch driver at present

If anyone following this thread has a similar interest and a Yuichi horn
in mainland Australia and can make the horn available l would be happy to solution a network and respond to your questions

As mentioned in my previous posts you could incorporate an analoge active crossover with “simplified “ passive response equalisation.

This will get you close for diy purposes and save you a lot of time.

Unfortunately passive network design is NOT a guess or a cut and paste of different crossover schematics.

It requires a lot of trial and error experimentation and measuring equipment to get it right. This is what Jbl do.

Given the diy investment involved the above is a prerequisite for the payoff.

Assuming you don’t have an active crossover or measurement tools then my recommendation is pursuing the cloning or diy a Jbl design such as the 4430/4435 or similar.

People like Marchland have suitable active crossover for diy people

[Dr.db]

Thanks a lot for your help so far!

If using an active crossover, I would have to use a single capacitor in series with the horn-drivers for protection purpose (turn-on pops etc..).
I remember to use rather big capacitors to inshure their operating range to be a lot lower than the actual active crossover-point...
Which values for the capacitors would I have to use?
2441 will be crossed actively at 700-800hz
2405 will be crossed actively at 10.000hz


If going with the Marchand active-crossover, should I use asymetric filters?
For example the crossover between 2441 and 2405 in the 4355 monitor was at ~10khz with a 12db-lowpass for the mid and 18db-highpass for the tweeter.
Should I copy this asymetrical crossover-sheme?
Did JBL use the 18db-highpass for the 2405 due to more protection from lower frequencies or rather due to a better frequency-response match-up with the midhorn?

[dn92]

TAD asymetrical slopes crossover allows to compensate delay between bass unit and horn compression driver. Not sure if Arai290 will give the same delay as TH-4001.

To integrate the tweeter you should align it in time with the compression.

One possibility for using an analog active crossover would be using an Rane ac23, a JBL M553 or a JBL 5235 with the active 53-3333 filter card, as they have configurable delay for the Lf channel, and in addition use a passive EQ such as the TAD TN-3 was.

The simpler way would be using a digital crossover.

Of course for all of this you need tools for doing your own measurements

[Dr.db]

I tried to calculate the delay caused by TAD´s 36db-Lowpass...

The midhorn uses a 12db-highpass that causes a delay as well for this middriver. So the woofer just has 24db (not 36db) more of filter-delay than the middriver, right?
Considering the speed of sound(343m/s) this additional 24db-filter @650hz (0,7m/s) for the woofer should result in aprox. 20cm (8inch) physical offset, right

[Ian Mackenzie]

That’s not correct

A visual of the relevant Tad network indicates a 36 dB low pass network

To be exact you need to take a spl frequency measurement and perform minimum phase transform function of the low and high pass using this particular horn/ driver you plan to use to determine the group delay from a reference point ie the front baffle

Ideally you need to do that with your Yuichi horn and proposed drivers.

I have a simulation of the simplified Tad network on my Soundeasy program
But you need to acquire a good measurement kit and simulator software if you intend to pursue a full passive network

[dn92]

This will be also wrong

in low frequencies (below 1kHz) TH-4001 has also some group delay. For TH-4001 minimum phase matches real phase only above 400Hz.

I did some mesurements of TN-2 crossover. LP filter is closed to a Linkwitz Riley 6th order and HP filter close to Butterworth 2nd order. Raw delay between hf and lf without crossover is around 0.8ms (but not constant along the complete bandwidth.

[Ian Mackenzie]

“To be exact you need to take a spl frequency measurement and perform minimum phase transform function of the low and high pass using this particular horn/ driver you plan to use to determine the group delay from a reference point ie the front baffle

Ideally you need to do that with your Yuichi horn and proposed drivers.”

I did say the low and high pass filter functions

At any rate with due respect to Dr.db this discussion is academic at this point and l think it’s is inappropriate to make it appear something the can be simply understood in few forum posts and easily solved in a few minutes.

Unfortunately it goes on and some members feel they have struck pot luck without having any technical skill or background on the specifics of the topic.

I am only providing high level insights. A meaningful discussion requires that the recipient be full conversent with not only the theory and technical problem solving but also the practical application and measurement requirements of this kind of topic.

It isn’t easily understood and not something that can be learnt by googling.

In the end it comes back to the reality of going and doing some practical evaluation of the drivers and horns you have on hand.

This post is an excellent account of the choices and compromises between doing your own diy design and closely following an existing Jbl design .

http://www.audioheritage.org/vbullet...l=1#post397052

[Dr.db]

The longer I´m talking to you guys, I believe it would be the smartest idea for me to use an active-crossover instead of tinkering with passive crossovers. Of course you are right, that measuring equipment is essential to achieve best results. But as my knowledge isn´t up to that procedure by now, I´d be very glad to hear your suggestions on crossover freq. and slopes.

Ian, you have suggested aprox. 700hz for low to mid... Would you try 36db-lowpass on the 2234 and 12db-highpass on the 2441 to start with?
Should I cross at 10khz between mid to high with 12db-lowpass on the 2441 and 18db-highpass on the 2405 as in the 4355-monitors?


If going active, what size e.g. value of protection capacitors do I have to use infront of 2441 and 2405 driver?


We talked about matching the dispersion of the woofer with that of the horn. The 90° horizontal dispersion of the yuichi matches the dispersion of a 15" cone very well.
But how about the horn matching the slot-tweeter?
The JBL 2405 has a 140° dispersion that matches perfectly with the 2397 at 10khz. But the yuichi´s 90° wouldn´t match well with 140° of the 2405!!