2445J/2380 + 2404H passive xo help

[Juntuin]

Hi everyone,

First of all, I know this is noob question..but I hope that someone would be willing to help me with this. I don't know has anyone done this as passive and got schemas already, if not, I would need something to start work with. I have 2225H on bottom and I can drive that active through miniDSP and I would like to use that as room correction for lower frequencies.

If going with 1st order xo;

2380 and 2445J driver. Crossing somewhere between 600-800hz and maybe 8000hz. So that's single cap and single coil in series.(?) But response is rising from 900hz and falling from 5000hz, that does need some kinda attention and what kinda?

2404H, crossing from 8000hz. Only single cap in series, or something more?

Then there is level matching. Simply Lpad's to tame everything down to match 2225H levels?

Am I even close on the track here?

There is always an option to sell upper end, get better horn/driver to replace 2380/2445 and make this 2-way. Don't know is there any suitable horn/driver, since there are many octaves to cover with one horn.. I used to drive this as a 2-way but I wasn't satisfied with top end.

Thanks :)

http://www.jblpro.com/pub/obsolete/2404.pdf
http://www.lansingheritage.org/images/jbl/specs/pro-comp/2225/page02.jpg
http://www.lansingheritage.org/images/jbl/specs/pro-comp/2445/page2.jpg

[Juntuin]

Hmm, funny that I haven't seen 3105 or 3106 xo's before with all this work with google. Now I found those. But I guess there are some improvements to do.. Bypass cap, where and what size? Protection cap for 2445J, I already had 47uF and I guess that'll do the trick. Any protection for 2404H? Need for zobel network?

If and when it matters, mid driver is 16 ohm and hf driver is 8ohm.

So now we got base for our xo, how to apply improvements? :)

My english isn't so good, so I would love to see some schemas!

[Ian Mackenzie]

http://www.jblpro.com/pages/pub/obsolete/3105-3115A.pdf

http://www.jblproservice.com/pdf/Network%20Schematics/3115A%20Network.pdf

I think you are saying you want a passive crossover?

If this is the case read the above links carefully for guidelines on the 4673 system

On the schematic the low filter should be fine

On the high filter the auto transformer is no longer available so take that and the 2 39R resistors out of the circuit.

I would suggest replacing the auto transformer with a switched fixed attenuation for 12 ohms (the 2445) of 6, 8, 10 dB or simply buy an 8 ohm variable Lpad and put a 20 ohm resister in parrellel with the 2445.
The bypass horn compensation CLR is still required.

That will get you started. You really need a measurement mic and analyser.

I would keep it simple and get it going as a two way first before attempting to add the 2404

[Juntuin]

Hi Ian, and thank you for reply!

What I need is passive xo ONLY between 2445J and 2404H, I have played this system active 2-way for a long time before this attempt.

Only measuremet equipment I have are UMIK-1 and REW, I know they won't take me very far, but better than nothing? :)

[Robh3606]

The 2380 cannot be used in a simple 6db network. This is a CD horn that will "not work" without a compensation network that rolls off the midrange response. Take a look at the no EQ response. Did you run a CD compensation in the active set-up? Look up the SR series 4731/32 as examples of passive networks using 2380/81 and 2404

Rob:)

http://www.lansingheritage.org/images/jbl/specs/pro-comp/2380/page4.jpg

[Ian Mackenzie]

Hi Rob

I was wondering why the need for the 2404?

The plan is not clear.

If the poster can detail specific detail of each crossover point that would make any feedback easier.

ie 500 hertz active or passive, 5000 hertz passive
CD compensation yes/no

If the horn is compensated correctly an additional HD devise should not be absolutely necessary

My impression is you are not using any compensation.

Please confirm that you have CD compensation (active EQ) on the mini Dsp.

All crossover slopes should be 12 dB or 24 dB
My suggestion is measure the horn as is and post an image.

Btw you should be able to EQ the horn flat with the mini dsp

The put in a low pass 12 slope with the mini dsp at 5000-10000 hertz

Your high pass passive filter for the 2404 can be a series capacitor and an 8 ohm
Lpad to start with

Try 1-3 uF. Look at the Jbl 8000 hertz crossovers if you need a 12 dB network

It’s a very diy space so you are going to have to work on this for a while

If you are actively crossing over the 2455 put in a 30-50 uF protection capacitor

The Jbl 5235 electronic crossover manual has specific guidelines on protection capacitors.

[more10]

The 2380 cannot be used in a simple 6db network. This is a CD horn that will "not work" without a compensation network that rolls off the midrange response. Take a look at the no EQ response. Did you run a CD compensation in the active set-up? Look up the SR series 4731/32 as examples of passive networks using 2380/81 and 2404

Rob:)

http://www.lansingheritage.org/images/jbl/specs/pro-comp/2380/page4.jpg

https://www.jblpro.com/pub/components/2380a.pdf Looks pretty flat to me below 6 kHz.

[Ian Mackenzie]

Have another look at the unequalised response.

The EQ frequency response is flat per top attachment

In the bottom attachment l have put up a simple passive network to give an indicative “voltage” drive required to EQ the horn / compression driver @ the driver terminals.

The pad is -10 dB. The high pass filter Q is manipulated to derive the boost below 1000 hertz, the LCR bypass network crests the Hf lift to compensate the CD horn.

The values are indicative only and a working schematic may require modifications with real driver loads. But it will work.

So, if you can program your mini dsp for those curves that is a good baseline to start dialing in your system.

Get that working first and then look at the 2404.

[Juntuin]

Wow, I was not expecting this much good conversation :)


The 2380 cannot be used in a simple 6db network. This is a CD horn that will "not work" without a compensation network that rolls off the midrange response. Take a look at the no EQ response. Did you run a CD compensation in the active set-up? Look up the SR series 4731/32 as examples of passive networks using 2380/81 and 2404

Rob:)

http://www.lansingheritage.org/images/jbl/specs/pro-comp/2380/page4.jpg

Hi Rob! Yes, I was using compensation for the 2445/2380, and a lot of eq otherwise too, trying to reach as close to 20khz as possible.


Hi Rob

I was wondering why the need for the 2404?

The plan is not clear.

If the poster can detail specific detail of each crossover point that would make any feedback easier.

ie 500 hertz active or passive, 5000 hertz passive
CD compensation yes/no

If the horn is compensated correctly an additional HD devise should not be absolutely necessary

My impression is you are not using any compensation.

Please confirm that you have CD compensation (active EQ) on the mini Dsp.

All crossover slopes should be 12 dB or 24 dB
My suggestion is measure the horn as is and post an image.

Btw you should be able to EQ the horn flat with the mini dsp

The put in a low pass 12 slope with the mini dsp at 5000-10000 hertz

Your high pass passive filter for the 2404 can be a series capacitor and an 8 ohm
Lpad to start with

Try 1-3 uF. Look at the Jbl 8000 hertz crossovers if you need a 12 dB network

It’s a very diy space so you are going to have to work on this for a while

If you are actively crossing over the 2455 put in a 30-50 uF protection capacitor

The Jbl 5235 electronic crossover manual has specific guidelines on protection capacitors.

Need for 2404; I don't like what 2380/2445 sounds when eq'd alot, trying to reach +16-20khz region. I have original diaphragms (ti) and would like to change them later. Sorry about my bad english, but 2445 and 2380 was eq'd quite flat and reaching +16khz freq's, but I didn't like the sound of it on higher freq's. 2404 is a nice HF driver with wide spreading horn, makes your speaker sound more "airy" on hi-fi purposes.

I already had 47uF protection cap and paralled resistor for 2445, but I don't quite know what value to use with 2404?

Targeted Xo frequencies can be found on my first post, they have been there all the time :)
2225H: - 600-800hz
2445J/2380: 600-800hz - 8000hz
2404H: 8000hz -


Have another look at the unequalised response.

The EQ frequency response is flat per top attachment

In the bottom attachment l have put up a simple passive network to give an indicative “voltage” drive required to EQ the horn / compression driver @ the driver terminals.

The pad is -10 dB. The high pass filter Q is manipulated to derive the boost below 1000 hertz, the LCR bypass network crests the Hf lift to compensate the CD horn.

The values are indicative only and a working schematic may require modifications with real driver loads. But it will work.

So, if you can program your mini dsp for those curves that is a good baseline to start dialing in your system.

Get that working first and then look at the 2404.

Thanks really much for the schema! I'm not expecting fully working xo at first, but I can order some components and take measurements outside with UMIK and let's take a look what value we have to change.

[Juntuin]

Here we go, here's another of them. 1/24 smoothing.

Looking this makes me think crossing at 600hz and 7khz? There is difference between these two invidual horn and driver, other one won't fall until +8khz

[ivica]

Here we go, both measure a little diffirent but here's another of them. 1/24 smoothing.

Hi Juntuin,

1.
I think that 600Hz is too low for the 2445ti&2380 combo.Such would produce quite large diphragm excursion and the sound can become hars, may be 800 to 1000Hz would be better. Do not forget that the membrane amplitude would correspons to 1/f^2, and tha 'standard sound ' level would correspons to about 1/f, so all together would produce diphragm amplitude displacement almost 1/(f^3), so the difference crossover frequency between 600Hz and 800Hz woud be (4/3)^3=64/27 ~ 2 times larger diaphragm amplitude displacement. May be THD measurements would be usefull.

2.
Using 2445Ti over about 12kHz, would emphasize diaphragm breakdown 'ringing', so using UHF driver over 10kHz would provide more 'fidelity sound', but using two drivers would 'introduce' some trouble round crossover frequency due to the comb-filter effects
that CAN NOT be EQ, due to the VHF abd UHF drivers spaced appart, and their distance is comparable with crossover wavelength (10000kHz wavelength 34mm).

Regards
Ivica

[Juntuin]

Hi Juntuin,

1.
I think that 600Hz is too low for the 2445ti&2380 combo.Such would produce quite large diphragm excursion and the sound can become hars, may be 800 to 1000Hz would be better. Do not forget that the membrane amplitude would correspons to 1/f^2, and tha 'standard sound ' level would correspons to about 1/f, so all together would produce diphragm amplitude displacement almost 1/(f^3), so the difference crossover frequency between 600Hz and 800Hz woud be (4/3)^3=64/27 ~ 2 times larger diaphragm amplitude displacement. May be THD measurements would be usefull.

2.
Using 2445Ti over about 12kHz, would emphasize diaphragm breakdown 'ringing', so using UHF driver over 10kHz would provide more 'fidelity sound', but using two drivers would 'introduce' some trouble round crossover frequency due to the comb-filter effects
that CAN NOT be EQ, due to the VHF abd UHF drivers spaced appart, and their distance is comparable with crossover wavelength (10000kHz wavelength 34mm).

Regards
Ivica

Hello Ivica!

I have tried many different crossover-points via miniDSP, for some reason I always ended up in 600-700hz range. I don't usually listen very loud, 85-90db range max.

Is there any matter that I was hoping to cross much lower than 10khz (8khz was something that I was aiming for), or is comb-effect something that will involve with this system no matter what?

This is classic "lets build something from what we have" rather than collect just the right needed parts for good quality system. The point is that it is extremely difficult to get your hands on vintage JBL parts in Finland. Heck, even in EU! For example I think that there's something like 5 pairs of 4341 in whole country :D

[more10]

Use the minidsp delay to handle the comb effect. If the minidsp is temporary solution you will have to deal with the combing by either accepting it or moving the 2404 backwards. Steep filters will minimize the combing if going active if analog filters are an option.

[more10]

The ti diaphragm will start to breakup at 4 kHz. Use a 2450SL diaphragm if you want to cut at 8 kHz.

[ivica]

Hello Ivica!

I have tried many different crossover-points via miniDSP, for some reason I always ended up in 600-700hz range. I don't usually listen very loud, 85-90db range max.

Is there any matter that I was hoping to cross much lower than 10khz (8khz was something that I was aiming for), or is comb-effect something that will involve with this system no matter what?

This is classic "lets build something from what we have" rather than collect just the right needed parts for good quality system. The point is that it is extremely difficult to get your hands on vintage JBL parts in Finland. Heck, even in EU! For example I think that there's something like 5 pairs of 4341 in whole country :D
Hi Juntin,

I think that 10kHz woud be better as comb filter effect woul be 'moved ' at upper frequency, so woul be less emphasized, may be even higher You can try.

Regards
Ivica

[Ian Mackenzie]

The suggestion of the SL diaphragm has merit.

Are your diaphragms the original diaphragms?

If they are or if you have not looked at them before it can be beneficial to have them serviced and the SL diaphragms installed.

[Juntuin]

I would like to get those diaphragms, is there any supplier other than random sellers in ebay? Mine are with original Ti, never serviced i guess.

[Ian Mackenzie]

http://www.audioheritage.org/vbulletin/showthread.php?3450-2450-Dia&p=100783&viewfull=1#post100783

https://reconingspeakers.com/product/jbl-d8r2450sl-diaphragm/

I recommend you do a search and read up on other users comments and suitability of these diaphragms for your specific drivers. (Make sure they fit)

I’m a Tad4003 user so l don’t use the JBL2445 driver

[Juntuin]

When and if it'll go to changing diaphragms, what about using D16R2441 and just run it as a 2-way system with some eq applied?

[more10]

The D16R2441 will start break up at 7 kHz. It is much better than the ti diaphragm.

I would not use a JBL large format driver/horn without a supertweeter. Certainly not the 2445 with its "inferior" phase plug.

Are you still thinking passive xover?

[more10]

Use what you have. You can always improve things later. 2404 can be used down to 4 kHz. If you cut at 4 kHz you will get rid of the break up modes of the 2445, it will sound just perfect.

[sguttag]

Having used countless 2380 horns (with 2245 and 2246 drivers), I would never cross it over below 800Hz, unless I had no other choice. The driver will be loaded down to 500Hz but the response and control are gone below 800Hz. I'll see if I can dig up the polar response of the 2380. Below 800Hz, it is just a circle. Now, a 2360 horn, that I would go down to 500Hz with absolute confidence.

[ivica]

[QUOTE=sguttag;416474]Having used countless 2380 horns (with 2245 and 2246 drivers), I would never cross it over below 800Hz, unless I had no other choice. The driver will be loaded down to 500Hz but the response and control are gone below 800Hz. I'll see if I can dig up the polar response of the 2380. Below 800Hz, it is just a circle. Now, a 2360 horn, /QUOTE]

Hi,

https://www.jblpro.com/pub/components/2380a.pdf

From the JBL data it is mentioned that you can use 2380a from the 500Hz, but I would suggest something like 700 Hz or more. Its polar response is quite ok from 500Hz over 10kHz.
Using 2441 AL diaphragm would give possibility to use it may be up to 15kHz, but sharp filter has to be used in lower frequency due to the aluminum diaphragm diamond shaped suspension.
Regards
Ivica

[more10]

Your initial plan was to use minidsp as crossover between bass and midrange, and passive between midrange and tweeter?

If so, the passive filter will give you nothing since the minidsp will already have "destroyed the signal" :-)

Go for a full active three or four way system. You can handle the comb filtering problem with digital delay. A small class-d amp will be cheaper than the passive filter components :.)

https://www.minidsp.com/products/minidspkits/minisharc-kit (https://www.minidsp.com/products/minidspkits/minisharc-kit?)
https://www.minidsp.com/products/minidspkits/minidac8

Create your filter with pos rePhase. pos is a member here.

[sguttag]

Its polar response is quite ok from 500Hz over 10kHz.
Ivica

I don't quite know how you got to that conclusion. It has no pattern control below 1KHz

81879
500Hz
For a 90x40 horn, that is essentially omni-directional.

81880
Its frequency response below 700Hz is pretty drastic too.

Compare that to its big brother, the 2360A:
81881

Now that has pattern control down to 500Hz though perhaps unfair to the 2380A because of its size.

So using another Lansing (Altec this time), the MRII594A which is a similar class/size to the 2380A and you can see that it has better though barely pattern control down to 500Hz and much improved pattern control at 630Hz.

81882

And the frequency response is really down to 500Hz:

81883

So I come back to, I'd avoid using the 2380A horn below 800Hz.

In actual use, this has been born out. But hey, do what you want.

[Ian Mackenzie]

As requested here is a simple passive network for crossover of 5000 hertz to the 2404 from the 2445/2380.

Looking at the response of the 2445/2380 on the basis that you use the mini dsp for the high pass crossover and EQ to flatten the response say 700-3000 hertz a simple 6 dB slope network is mostly all you require as the driver has roll off above 5000 hertz. But l have also given a 12 slope network. I have included a fixed 10 dB pad as l believe noise will otherwise be an issue.

The networks are based on 12 ohm impedance of the driver in the area of interest.

The 2404 crossover is a 12 slope as specified by Jbl and is at 5000 hertz based on a 10 ohm impedance in the area of interest. Again the response falls off below this point. I have included a 5 dB pad so the 2380 and the 2404 should be level matched. Some level adjust could be accomplished with a variable L pad if required but this will be ball park.

This crossover point should be a step in the right direction based on your experience with the 2445/2380 to date and Jbl published technical data. Using the drivers in their optimum range and using a simple crossover design usually delivers a good outcome.

These networks are real and they will work but some adjustment may be required in practice. Placement of the transducers will effect your crossover transitions and l recommend you experiment to obtain the best overall result.

Good luck with your project

Ian

[dn92]

I have impedance measurements for 2450J/2380A and 2404H, this would ease crossover simulation. Just need the time to find them and be back home to share them.
I've also on axis response but without reference timing between the two.

[Ian Mackenzie]

http://www.audioheritage.org/vbulletin/showthread.php?22936-jbl-2404-H-network-need-help

An interesting thread on the 2404 crossover

The fixed Lpad l have used will moderate impedance variations.

Maybe Rob has a schematic he can compare.

This is all tough as old boots SR drivers so it’s not like attempting a magazine curve in a HiFi magazine.

[dn92]

81891
81892

Remove ".txt" in the file name.
Response mesurement was done at 1m distance, on axis. Sound transmission delay removed (Phase aligned to minimal phase), some gating to get a clean phase response.

Files can be directly used in a crossover simulator such as XSim.

[dn92]

81893

Remove ".txt" in the file name.

[Ian Mackenzie]

Okay l will see what l can do.

[dn92]

JBL 2450J on 2380A, 1m on axis measurement file: 81898


Phase correspond to impulse response shift to 0, which means that if using it as it is it will not reflect the delay between 2445 and 2404H, especially if 2404H will be enclosure face plate aligned to the 2380A such in JBL 4750A. Nevertheless using directly as is the xover schematics of 47xx serie:
http://www.jblproservice.com/pdf/Sound%20Power%204700%20Series/4750A.pdf
http://www.jblproservice.com/pdf/Sound%20Power%204700%20Series/4755A.pdf

Just attenuating JBL 2450J by 4dB (should be 6dB, but my two measurements where not aligned in SPL), as in 475x loudspeakers there are two 2404H, and it works in simulation. Ideally, a phase shift should be added to the compression driver to reflect delay due to the depth of 2380A.
81901
81900

[Ian Mackenzie]

I unfortunately could not get the files to work in Leap.

That’s an interesting option above.

What he could do is use one of Jbls passive crossovers for the woofer and horn and use the mini dsp to evaluate different options for the 2380 crossover to the 2404.

That could include delay to the 2404.
An important point is any digital delay will only work for one 3 D point in space (X, Y, Z position).

Perhaps more important at a basic level is a smooth crossover transition on axis with the listening position as that will also be dependent on the location of the 2404 (X, Y, Z) regardless of a mini dsp or passive solution.

(This assumes the mini dsp is only two way.)

It’s a quick and easy way to assess the subjective behaviour of the 2380 and the 2404 which the graphs don’t tell him. He found that out attempting to EQ the 2380 flat for Hf extension. The graphs don’t necessarily tell you if it will be palettable to the ear.

He can then work out a passive crossover for the 2380 and 2404 and compare to the active solution and see if the mini dsp is better for use in upper crossover point or the lower crossover point.

One final point is l would use a 10 dB fixed pad or an L pad to attenuate the 2380 as at 110 dB sensitivity he is going to hear all the negative (noise and distortion) artifacts in the mini dsp and his signal path. I would also pad back the 2404 to match the 2380.

[Juntuin]

Hello everybody, and thanks for being super helpful :)

I have been busy with couple of class-A amp builds, when I get those done I will start to fabricate some passive networks you guys have kindly presented here. My ultimate goal is to run this whole system as passive but bi-amped (class-D 200w to 2225, class-A single ended 25w to upper section) and get rid of miniDSP, but I guess that 2225 is not going to be huge problem to filter compared to upper section (?). L-pads are "must" for this reason too?

If using miniDSP somewhere in this system gives great advantages, I can consider keeping it. But it's 2-channel only..

[Ian Mackenzie]

The variable L pad are initially a convenient way to match the balance of the drivers

Measure the R valued of the Loads then use fixed R values.

I l said use the mini DSP to try a few different approaches for the horn/ 2404.

When you are happy take a screenshot of the response or preferably the voltage drive.
The passive network can be then iteratively be optimised.

I don’t think this is a difficult problem.

But do NOT rely on frequency graphs alone as to the right choice.

Trust your ears, not opinions.

[Juntuin]

Ok, I'm going to start by adding 2404H in to the system and everything else as active, 4 amp channels and 2-way active xo and dsp.
Before I order anything;

- 1,5uF cap series and 0.6mH air core coil paraller to 2404H

- L-Pad for 2445J to match levels

- DSP if needed

(Also 47uF for 2445J as a protection and parallel resistor)

Everything ok here? If it sound good, I just might stick to that and forget all-passive system :D

[Ian Mackenzie]

I was able to load the data files into Leap with the following results.

The response is a good compromise with three elements c=4.7u r=4.7R r=12R

I have also shown in another curve the inclusion of L=0.20mH in series with the network to attenuate the top end of the 2380.
The network includes a passive high pass network to demonstrate effect of the mini dsp response overall.

Similarly the 2404 network can be modified by reduced L from 0.6mH to l=0.24mH. The effect is faster attenuation below 7000 hertz.

These types of series EQ circuits do however introduce some ripple in the passband due to the driver impedance modulating the voltage across the RC elements and the effect is some ripple in the response. My suggestion is use the Mini dsp for the crossover and 2380 EQ withe the passive curves as a guide.

The RLC series Trap EQ networks JBL use in its K series of consumers systems are superior in this respect.

In terms of parts I would buy both the 0.6mH and the 0.24mH to try.

[Juntuin]

Thanks alot guys :) I placed order for non-eq components to 2445 and 2404 at first, going to test that and I'll report results here as soon as I have tested it.

[Juntuin]

Ok, first listening session behind. 2404 adds some nice sparkle.
Male vocals stays too "behind" and something wrong with 2225's dsp/xo settings: system sounds somewhat noisy with complicated music, but not articulate at all and there's no impact. Shame that can't do measurements while kids are at home, they make more noise than these speakers :D but I think that upper region works quite well now.

[Juntuin]

Now I had some time to eq 2225's, definitely much, much better! Maybe I get some amp with dsp possibilities to drive bass and keep class-A to horns. Thanks alot, you guys are awesome :).

Maybe some fine tuning later, now I want to enjoy music for awhile.