PDA

View Full Version : Is this doable?



pangea
11-02-2003, 02:27 AM
I've got so many great replies on my questions, that perhaps my main remaining question will be lost, so I hope you don't mind that I put it forward again and at the same time I have tried to isolate the most important question and also tried to clarify my question/problem.

A few days ago I received a suggestion on a first order x-over from an expert, which he said, would work and at the same time flaten out the frequency response on the 2445J driver

So, I'm asking for your help on this somewhat unorthodox x-over.

Will it work and will it reduce the hump and even out the frequency response between mid and high on the 2445J? :confused:

BR
Roland

scott fitlin
11-02-2003, 10:14 AM
Giskard is the man for this question. I personally would not use less than 12 Db slopes on a compression driver. And not higher than 18 DB slopes.

First order filters are known for their excellent time domain characteristics, and excellent impulse response. But 6 DB slopes with a compression driver might be a bit shallow at the lower end of the operating range.

Alex Lancaster
11-02-2003, 11:36 AM
And if You are planning on a cone speaker below the horn, why not xover at 1500 Hz, 6dB?, it would be even flatter, if the cone is flat that high.

Alex.

pangea
11-02-2003, 11:48 AM
Originally posted by scott fitlin
Giskard is the man for this question. I personally would not use less than 12 Db slopes on a compression driver. And not higher than 18 DB slopes.

First order filters are known for their excellent time domain characteristics, and excellent impulse response. But 6 DB slopes with a compression driver might be a bit shallow at the lower end of the operating range.

Thanks, but could you pleas explain what you mean by "shallow"? I'm from Sweden you know!

BR
Roland

scott fitlin
11-02-2003, 11:54 AM
6 db per octave is not as steep a rolloff as 12 dbb per octave, and you will have considerable overlap between the woofer and your horn. You are using a compression driver and since the rolloff is not that steep, if your crossover point is say, near the lower frequency limit of the driver, you might get too much low frequency below what the driver is rated for.

As Alex had said, with 6db slopes a higher crossover point might be a good idea, if you have a woofer that can go up high enough.

pangea
11-02-2003, 12:03 PM
Originally posted by Alex Lancaster
And if You are planning on a cone speaker below the horn, why not xover at 1500 Hz, 6dB?, it would be even flatter, if the cone is flat that high.

Alex.

The x-over I was told, had the x-over point at around 5 kHz, because it would take away the big hump on the drivers mid section, but I don't know if it really does just that.

IF it does that, I guess I could insert a midbass driver just about at any x-over point.
I was thinking that maybe a, 2123H, 2118H or E 110 would work just fine in my small sub chamber up to around 3500 Hz, or am I thinking wrong here!?

BR
Roland

pangea
11-02-2003, 12:09 PM
Originally posted by scott fitlin
6 db per octave is not as steep a rolloff as 12 dbb per octave, and you will have considerable overlap between the woofer and your horn. You are using a compression driver and since the rolloff is not that steep, if your crossover point is say, near the lower frequency limit of the driver, you might get too much low frequency below what the driver is rated for.

As Alex had said, with 6db slopes a higher crossover point might be a good idea, if you have a woofer that can go up high enough.

You think that it still would be a problem with a 6 dB slope when the x-over point is as high as 5000 Hz or higher?

BR
Roland

scott fitlin
11-02-2003, 12:17 PM
Originally posted by pangea
You think that it still would be a problem with a 6 dB slope when the x-over point is as high as 5000 Hz or higher?

BR
Roland No at 5000 hz it will not be a problem at all. But I dont think that using a 2445 from 5K on up is the best use, for this driver.

pangea
11-02-2003, 12:35 PM
Originally posted by scott fitlin
No at 5000 hz it will not be a problem at all. But I dont think that using a 2445 from 5K on up is the best use, for this driver.

According to the person that designed this x-over, who stated that there would be a lot of energy left quite a bit down, because of the big hump on the 2445's frequency response. At least down to the one or two kHz region, that is IF this persons theory works as he said it would in this particular design.

BR
Roland

Alex Lancaster
11-02-2003, 12:54 PM
According to the person that designed this x-over, who stated that there would be a lot of energy left quite a bit down, because of the big hump on the 2445's frequency response. At least down to the one or two kHz region, that is IF this persons theory works as he said it would in this particular design.

"the person" seems, IMHO, to have smoked crops sprayed with paraquat.

About Your Philosophy, no offense, Welcome to the planet!.

Alex.

pangea
11-02-2003, 01:40 PM
Originally posted by Alex Lancaster
According to the person that designed this x-over, who stated that there would be a lot of energy left quite a bit down, because of the big hump on the 2445's frequency response. At least down to the one or two kHz region, that is IF this persons theory works as he said it would in this particular design.

"the person" seems, IMHO, to have smoked crops sprayed with paraquat.

About Your Philosophy, no offense, Welcome to the planet!.

Alex.

I don't know if he has smoked all that sprayed stuff you're suggesting, what I do know is that he has worked with the swedish JBL representative a few decades, so he should know what he's talking about. I just wanted a second opinion. I figured this would be the place to ask for it.

Why would you want to, or even be in the position to welcome me, to a planet on which I have already been a resident for over fifty years? No offense!

BR
Roland

4313B
11-02-2003, 02:05 PM
Originally posted by pangea
Will it work and will it reduce the hump and even out the frequency response between mid and high on the 2445J?
What horn are you using? Flat-front bi-radial? Exponential as mentioned in your other thread? What is the model? Did the guy in Sweden come up with his solution based on a specific horn?

pangea
11-02-2003, 02:21 PM
Originally posted by Giskard
What horn are you using? Flat-front bi-radial? Exponential as mentioned in your other thread? What is the model? Did the guy in Sweden come up with his solution based on a specific horn?

I'm using a Brasilian made exponential horn made by Selenium model HL 14-50 it is 131 mm long, 160 mm mouth.

I told this person I'm using an exponetial horn, but I'm not sure if we discussed specifically which horn I was using.

BR
Roland

Robh3606
11-02-2003, 03:00 PM
Hello Pangea

All of the biradials like the 2360 and 2380 series require compensation to flatten the curve. Basically reducing output in the lower range of the curve. The 4430/4435 used passive compensation with the 2344 and the 4425 with the 2342 horn. All these horns give you flat/uniform power response as such they must be equalized electrically to be flat on axis or anywhere else for that matter. The directivity remains constant with freguency. Take a look at the horn curves WITHOUT eq./compensation, Also look at the directivity curves.

An exponetial horn will compensate the driver so on axis it is flat but the power response is not. The directivity changes and gets higher the higher in frequency you go. Basically you go from wide beam to spotlight.

The idea about the first order network is theoretically it can match the roll off of the driver. So depending on where you choose to make the hinge point on the crossover you will reduce the lowend response at a rate that closely follows inversely what the driver is doing. This will work with a CD horn.

Take a look at the 4430/4435 crossover. Then take a look at the 4722 crossover. The 4722 has the idea in it purest form. Please note that these are 12db networks not 6db

You have to know what type of horn you have and what its directivity is to do the compensation. Any published specs on the horn you are using??

Rob:)

Earl K
11-02-2003, 03:20 PM
Hi Pangea

I use that horn. For my purposes, I find it very nice with no horn coloration that I can make out. I use 2441 diaphragms in 2440 magnets some parts of the week - and other times a 2450SL diaphragm in one of those 2440 magnets. Right now I'm listening to one of each. The 2450SL is more extended with a nimbler / tighter response all through its' response curve. My second 2450SL diaphragm has a fit problem that effects its response curve - I may have a defective diaphragm , I'm slowly researching this ) .

I biamp my system, crossed at about 750 hz ( LR, 24db per octave ). I use a simplistic RC network to bump-up the HF response a little bit. A simple 1.5uf cap strapped across a 16 ohm resistor is all the EQ I use. I'm not after ruler flat response - and - well, this RC network just won't deliver "flat" - but it is very musical sounding . I've tried 3uf and larger but those weren't to my liking.

Today, I just tried a 12ohm resistor strapped across the driver with a 2.35uf cap inline ( as in your drawing ). That setup, also won't deliver "flat" HF response. Also, it's not as musical or as open sounding as the simple RC network ( at least with my Bryston 2B-LP ) . Though that setup does have a significant db drop or padding effect - perhaps more than you're after. A tone generator and a cheap SPL meter would tell you all this info .

For the $ 5.00 it'll cost you to buy the 2.2uf cap and resistor - try this out yourself - you won't blowup your amp as long as it can handle a 6 ohm load ( and if you don't create any short-circuits when you're messing about ).

The 2441 driver on that Selenium horn has an AC impedance of about 12 ohms across most of the range in question ( actually 10.5 to 12.5 ohms ). The paralleled 12 ohm resistor gives a working circuit impedance of 6 ohms. The 2.2uf cap "sees" or is reacting with the 6 ohms of AC impedance . After a bit of my math, I figure that gives a cornering frequency for that cap of about 12000 hz ( with 6 db per octave rolloff for the frequencies below ). I also assume that horn with a JBL driver is about 109 - 110 db (with 1 watt at 1 meter). Selenium rates it at 108 db with their older line of drivers - but these drivers were always a bit less effiecient than JBLs .

I'd ask your JBL contact to redesign the HF portion of the network. Offer him some money to do it. Ask him to incorporate a 16 ohm Lpad for some level matching. ( An Lpad in the 2123h portion of the circuit would likely be useful ). Suggest he look at the S3100 HF circuit for some "ideas & inspiration" in circuit design. That S3100 is designed for a CD style horn driven by an 8 ohm nominal ( 3.2 dcr )2426h driver - so he will have some scaling of values to accomplish . Still it's a nicely designed circuit for the DIYer .

That's what I suggest .

regards <> Earl K :)

pangea
11-02-2003, 03:32 PM
Originally posted by Robh3606
Hello Pangea

All of the biradials like the 2360 and 2380 series require compensation to flatten the curve. Basically reducing output in the lower range of the curve. The 4430/4435 used passive compensation with the 2344 and the 4425 with the 2342 horn. All these horns give you flat/uniform power response as such they must be equalized electrically to be flat on axis or anywhere else for that matter. The directivity remains constant with freguency. Take a look at the horn curves WITHOUT eq./compensation, Also look at the directivity curves.

An exponetial horn will compensate the driver so on axis it is flat but the power response is not. The directivity changes and gets higher the higher in frequency you go. Basically you go from wide beam to spotlight.

The idea about the first order network is theoretically it can match the roll off of the driver. So depending on where you choose to make the hinge point on the crossover you will reduce the lowend response at a rate that closely follows inversely what the driver is doing. This will work with a CD horn.

Take a look at the 4430/4435 crossover. Then take a look at the 4722 crossover. The 4722 has the idea in it purest form. Please note that these are 12db networks not 6db

You have to know what type of horn you have and what its directivity is to do the compensation. Any published specs on the horn you are using??

Rob:)

I simply know that something is missing, because when I add the 2402 it sounds much better all of a sudden.

I'll attache the specs on the Selenium horn. Hopefully you'll be able to read it.

I'll also check out the other x-overs.

BR
Roland

pangea
11-02-2003, 04:11 PM
No luck in finding the x-over 4722 though.

Any suggestion where to find it?

BR
Roland

Robh3606
11-02-2003, 05:39 PM
OK its an exponential. Look at the polar response graph on the bottom right hand side of the sheet. Look at 1k wide pattern then see how directional it gets and lastly at 10K??, hard to read, you have to be looking down the mouth at ear level or you will be outside the narrow beam.


The 4722 crossover link

http://www.jblproservice.com/pdf/SR-Series/SR4722.pdf


Since it is a exponential all you should have to do is set up the crossover point you need and pad to level match. Looks like the horn will work with an 800hz crossover point. The high frequency can be improved as Earl described with the small series cap.

Rob:)

pangea
11-03-2003, 04:35 AM
Originally posted by Earl K
Hi Pangea

I use that horn. For my purposes, I find it very nice with no horn coloration that I can make out. I use 2441 diaphragms in 2440 magnets some parts of the week - and other times a 2450SL diaphragm in one of those 2440 magnets. Right now I'm listening to one of each. The 2450SL is more extended with a nimbler / tighter response all through its' response curve. My second 2450SL diaphragm has a fit problem that effects its response curve - I may have a defective diaphragm I'm slowly researching this ) .

I biamp my system, crossed at about 750 hz ( LR, 24db per octave ). I use a simplistic RC network to bump-up the HF response a little bit. A simple 1.5uf cap strapped across a 16 ohm resistor is all the EQ I use. I'm not after ruler flat response - and - well, this RC network just won't deliver "flat" - but it is very musical sounding . I tried 3uf and larger but those weren't to my liking.

I just tried a 12ohm resistor strapped across the driver with a 2.35uf cap inline ( like your drawing ). That setup, also won't deliver "flat" HF response Also, it's not as musical or as open sounding as the simple RC network ( at least with my Bryston 2B-LP ) . Though that setup does have a significant db drop or padding effect - perhaps more than you're after. A tone generator and a cheap SPL meter would tell you all this info .

For the $ 5.00 it'll cost you to buy the 2.2uf cap and resistor - try this out yourself - you won't blowup your amp as long as it can handle a 6 ohm load ( and if you don't create any short-circuits when you're messing about ).

The 2441 driver on that Selenium horn has an AC impedance of about 12 ohms across most of the range in question ( actually 10.5 to 12.5 ohms ). The paralleled 12 ohm resistor gives a working circuit impedance of 6 ohms. The 2.2uf cap "sees" or is reacting with the 6 ohms of AC impedance . After a bit of my math, I figure that gives a cornering frequency for that cap of about 12000 hz ( with 6 db per octave rolloff for the frequencies below ). I also assume that horn with a JBL driver is about 109 - 110 db (with 1 watt at 1 meter). Selenium rates it at 108 db - but their drivers are always a bit less effiecient .

I'd ask your JBL contact to redesign the HF portion of the network. Offer him some money to do it. Ask him to incorporate a 16 ohm Lpad for some level matching. ( An Lpad in the 2123h portion of the circuit would likely be useful ). Suggest he look at the S3100 HF circuit for some "ideas & inspiration" in circuit design. That S3100 is designed for a CD style horn driven by an 8 ohm nominal ( 3.2 dcr )2426h driver - so he will have some scaling of values to accomplish . Still it's a nicely designed circuit for the DIYer .

That's what I suggest .

regards <> Earl K :)

Hi Earl!

Lot's of info in your post, but I'm an amateur on this, so I'm still not sure what goes where and why. :confused:

Therefore I have put together two different schematic's and hoping you could tell me if anyone of them is coming close to what you're telling me and perhaps you could fill in the "blanks" for me?

I would be ever so greatful!

Interesting that you use the same horns, I bought them because a friend of mine has them and they sonded great at his place. Also interesting that you're listening to one phragm of each kind.

Then I was surprised to read you're using 24 dB slope. Why? I thought less is more, all litterature I have read say 6 dB slopes sound better than 12 dB and 12 dB sound better than 24 DB, making a long story short.
Am I missing something? What are your reasons for using 24 dB slopes?

I'm not a flat response addict either. I want it to sound great too, but I thought the hump in the mid-section was causing the "hole" in the high notes, since the insertion of a 2402 managed to repair it. So, obviously something had to be wrong in my crossover.

You think the 2123H is a good choice for the midbass section from 300 Hz to around 3500 Hz, even if my sub chamber is only 6,5 L?
What about 2118H or the E110? What do you think of those in very small enclosures?

Greatful for any thoughts you could share!

BR
Roland

boputnam
11-03-2003, 08:15 AM
Originally posted by pangea
I simply know that something is missing, because when I add the 2402 it sounds much better all of a sudden. As it should. I suggested this (actually, the 2405...) in a reply to your OTHER thread on this, but you might have missed it. But I can't tell for certain if you've got a three-way or four-way set-up. ??


Originally posted by pangea
I thought the hump in the mid-section was causing the "hole" in the high notes, since the insertion of a 2402 managed to repair it. So, obviously something had to be wrong in my crossover... I just don't think that conclusion follows the evidence. Just MHO.

You will improve the frequency response of your system by having a UHF ring radiator ("tweeter") atop the horn.

Bye now... :wave: :slink:

pangea
11-03-2003, 08:50 AM
Originally posted by boputnam
As it should. I suggested this (actually, the 2405...) in a reply to your OTHER thread on this, but you might have missed it. But I can't tell for certain if you've got a three-way or four-way set-up. ??

I just don't think that conclusion follows the evidence. Just MHO.

You will improve the frequency response of your system by having a UHF ring radiator ("tweeter") atop the horn.

Bye now... :wave: :slink:


I saw it, I just couldn't believe it, then.
Now I'm inclined to take your word for it!

My bass drivers are Bi-Amped, the rest is a mix of two- and three way systems. A total hybrid right now, until I can get a pair of 2405's and a pair of 2123's. But this will have to do for now, if only I could decide which x-over to use for the 2445J. :confused:

BR
Roland

boputnam
11-03-2003, 08:58 AM
Originally posted by pangea
I'm inclined to take your word for it! Take nobody's word for nuthin'. "Believe your ears" - paraphrased teaching from a one Giskard... ;)

pangea
11-03-2003, 10:25 AM
-Bo

You're right, Giskard seems to be a very wise teacher!:)


-Earl

I just saw that you also asked me if the last graph showed the 10 kHz range. It doesn't, it shows the 16 kHz range.

http://www.selenium.com.br/sitenovo/admin/Documentos/HL14-50_i.pdf

BR
Roland

Earl K
11-04-2003, 06:43 PM
Hi Roland

Greatful for any thoughts you could share!

• What is the crossover frequency supposed to be for those two schematics you posted ?

• What formulas are you using to derive those values ?
- The values I get ( with my rough math ) for a 800 hz crossover point working with a 12 ohm load for the 2441 driver / giving a Bessel transform / are @ 9.4 or 9.5uf for the cap and @ 4.15 or 4.2 mh for the inductor. For a 16 ohm load ( maybe working against some Lpads constant resistance ) , I get @ 7uf for the cap and @ 5.5 mh for the inductor.

• NOTE: for the RC "midrange suppression / HF boost " circuit to work, some resistance ( over & above the drivers nominal value ) has to be added into the circuit with the "boost" cap so as to offer a path of "least resistance" to the frequencies above the capacitors turn over frequency. I'm not sure if you understand that from what I see in your two presented schematics. For instance; I parallel a 16 ohm resistor with the 1.5uf cap and put the pair in series with the driver. Below the roll-off frequency of the 1.5uf cap the amplifier eventually "sees" @ 30 ohms as a load ( 16 - 17 ohms from the resistor and approx. 10 - 13 ohms from the driver ). "Watts" are dissipated as heat for all the frequencies below about 8000 - 9000 hz by the 16 ohm resistor - that is acting as a voltage divider . The "bypass" cap allows all available power to pass to the driver through this 1.5uf cap. The schematics you present ( while incomplete ) don't seem to reflect this concept of doubling the impedance load to create the necessary conditions for a midrange suppression/HF boost .

• Now, obviously if this RC "boost circuit is used it needs to be worked into the circuit with some care. Right now , if it was located where you have your Lpad - the preceeding LC crossover components would have to be reworked to work with a 30 ohm load. If you do the math you'll see this creates some whopping huge inductor values. So, you are better off ( if you are going to use the RC circuit ) to make it the first thing the amplifier sees . Then place the Capacitor & Inductor into the circuit and then a variable 16 ohm Lpad if used / finally the driver.
- Alternately ; the order could be; RC boost network, variable 16ohm Lpad, Cap & Inductor / & finally the driver. This gives a smaller ( cheaper ) inductor value since the value for the cap and inductor are figured for 12 ohms ( not the 16 ohms of the Lpad ) . Cap size is increased some .

• I haven't built all these circuit scenarios to give you the "Sonic" pros and cons - but I have heard a coil directly shunted across a driver and it does offer a very well damped sound ( perhaps a bit overdamped ). This could be useful and is something to consider when using a 2445 which in my opinon has a very "underdamped" sound to it .

• It might be a good idea to buy yourself some crossover design software to help you through this. I don't have any purchase recommendations so hopefully others here might.

regards <> Earl K :)

pangea
11-05-2003, 03:30 AM
Originally posted by Earl K
Hi Roland


• What is the crossover frequency supposed to be for those two schematics you posted ?

• What formulas are you using to derive those values ?...


I have been using JBL Speakershop and another programme which I got from my local supplier.

I have set the crossover at 1100 Hz. Altough now I'm not sure that that is what I got, if I have understood what you're telling me. Somewhat :confused:

Does at least my iterpretation as shown below resemble what you're saying? Here I've used a 8 + dB fixed L-pad cause my supplier doesn't have a 16 Ohm variable L-pad. :(

All this has become much more complicated than I thought it would be.
Perhaps it would be easier, to simply cross over at around 7000 Hz and let the "bullet" take over from there, since they have just about the same sensitivity, or would it?

BR
Roland

Earl K
11-05-2003, 06:12 AM
Hi Roland

Yes, you understood me correctly, regarding the placement of the RC ( boost/suppression) network.

I just built up your fixed Lpad (using the (2) 10ohm resistors), put it into place across my 2441 and made AC ( motional ) impedance measurements in the 1100 hz area. The reflected impedance back to the amplifier ( or anything preceeding the driver and Lpad ) is about 14 ohms . After doing more calculations , that essentially gives you a "transform function" ( for your given LC values of 4mh & 5.2uf ) , as being very close to a classic L-R type. A Bessel transform at 14 ohms , according to my math , gives values of about 5.9uf for the cap and 3.52 mh for the coil .

I haven't yet measured the padding effect ( in real db ) since it's still too early here to spark up the tone generator . But I do need to mention that the resistor in the RC network also drops the level in db ( as it sucks in voltage and dissipates it as heat ) . Your final drop in db is going to be more like 10 to 12 db with the present fixed Lpad values .

Moving your crossover point higher & higher ( such as 3000hz ) really defeats the purpose of owning a large diaphragmed compresson driver like a 2441 . They excel in midrange clarity which is why people use them with lower crossover points. I'd try to get your crossover point back down to 800 hz .

regards <> Earl K

Earl K
11-05-2003, 08:15 AM
Hi Roland

I just measured the db drop at 2K, it was @ 11.2 db .

And ,,, oops,, I was wrong to suggest putting the LC components right after the RC network. The reason is that those two caps (1.5uf & 5.2uf ) back to back in that fashion will give a different value from their assumed values.

So , that implies the LC portion of the network is best placed behind the Lpad - closest to the driver side where it's "protected" from the RC cap value by the Lpad resistors. It's quite possible that the Lpad needs to become a balanced Tpad to present the LC ( cap & coil ) with a consistent AC resistance. Perhaps moving the coil between the two caps will solve this problem - unfortunetly I don't have the coil to try this solution .


All the above mucking about ( tail chasing ) is why I initially suggested you have someone scale the S3100 HiPass network for a 2445 ( as a 14 ohm load ) since that network takes a different ( somewhat easier ) approach to the implementation HFboost/midrange suppression . All reports are that its a very successful circuit . If you want the schematics on pdf , send me a private message with your home email address.

regards <> Earl K

pangea
11-05-2003, 08:33 AM
Originally posted by Earl K
Hi Roland

I just measured the db drop at 2K, it was @ 11.2 db .

And ,,, oops,, I was wrong to suggest putting the LC components right after the RC network. The reason is that those two caps (1.5uf & 5.2uf ) back to back in that fashion will give a different value from their assumed values.

So , that implies the LC portion of the network is best placed behind the Lpad - closest to the driver side where it's "protected" from the RC cap value by the Lpad resistors. It's quite possible that the Lpad needs to become a balanced Tpad to present the LC ( cap & coil ) with a consistent AC resistance. Perhaps moving the coil between the two caps will solve this problem - unfortunetly I don't have the coil to try this solution .


All the above mucking about ( tail chasing ) is why I initially suggested you have someone scale the S3100 HiPass network for a 2445 ( as a 14 ohm load ) since that network takes a different ( somewhat easier ) approach to the implementation HFboost/midrange suppression . All reports are that its a very successful circuit . If you want the schematics on pdf , send me a private message with your home email address.

regards <> Earl K

You mean like this?

BR
Roland