JBL 2226J 2445J 2405 active DSP crossover help

[danvprod]

I've posted about this design before, but am looking for help as I continue to refine the sound and design a crossover topology that best takes advantage of the components in the system that I've built.

72908

Design goals:

Build a speaker that sounds great with these great components that I can have for years without the urge to upgrade.
Build a sensitive speaker that can be powered, in part, to adequate volumes with my 300b amps and sound good in my small listening room.
Build a speaker that takes advantage of the MiniDSP for cross-over/EQ duties.


Background:

I had the opportunity to purchase a number of excellent JBL parts for a very reasonable price. These include 2 x 2226J 16 ohm woofers, 2 x 2445J 2" Compression Drivers, 2 x 2405 super tweeters and 2 x 250 Hz tractrix waveguides.
I also picked up a miniDSP 4x10 HD for my desire to not spend a fortune on crossover components while I tweak the crossover.
Amplifiers include a pair of 300B monoblocks, which power the 2445s, and 2x Crown D45 amps, which power the 2226s and 2405s.
I've been living with 2226s built in what I would call "draft" enclosures: ~ 4 cu ft. vented MDF enclosures tuned to 45 Hz with the horns/waveguides and super tweeters mounted on the top on the enclosures. I lift the bass with DSP.
They sound pretty good as-is and give me the sense that these components could sound exceptional with the right crossover configuration.


So that is what the thread is about -- the next steps in putting together this system.

The drivers:

The JBL 2445J is a 2" compression driver with a titanium diaphragm. Part of this build will be to replace the stock diaphragms with the Truextent Beryllium replacement diaphragms. I have these coupled to 250 Hz tractrix waveguides, and mounted on aluminum stands.
For woofers, I have the JBL 2226J 15" 16 Ohm drivers. I am pretty happy with their performance in the 4 cu. ft. enclosures, but they could be better -- I think giving them a bigger enclosure, and tuning them deeper could improve their performance. I also have the MiniDSP available to me for tweaking.
The super tweeters are 2405s. They are crossed over at 8 kHz and I find them kind of harsh. They may stay in the system or may not depending on the performance of the 2445Js with new diaphragms.


I am looking for help on how best to measure this system and design the crossover using the MiniDSP.

Right now I have my crossover is configured as follows.

Woofers:
20 Hz LR4
500 Hz LR4
35 Hz +8 dB Q 0.8

Midrange:
500 Hz LR4
8000 Hz LR4
1000 Hz -3 dB Q 1
400 Hz -4 dB Q 2
.3 ms delay

Tweeters:
8000 Hz LR 4
1.05 ms delay

I also have a protection cap network for the MF and HF drivers.
72909

So anyways -- looking for some inspiration, ideas, and feedback. If you are interested on where I am taking this, please feel free to follow along. Hopefully I can turn this system into something special.

[DingDing]

Your system looks good already. Much better than my random stack :)

Have you considered throwing a pair of JBL 2123's in there? They're cheap and great. Maybe have them run from ~300-1.2kHz.

What kind of hardware and software will you be using to take measurements and will you be measuring inside or outside?

What diaphragms will you be getting for your 2445's?

[danvprod]

Your system looks good already. Much better than my random stack :)

Have you considered throwing a pair of JBL 2123's in there? They're cheap and great. Maybe have them run from ~300-1.2kHz.

What kind of hardware and software will you be using to take measurements and will you be measuring inside or outside?

What diaphragms will you be getting for your 2445's?


Thanks! I'm sure your random stack is nice too! Interesting thought on the 2123s. Mid-coupler might help the crossover between the 15" and the horn for sure. I could put it in a nice, damped sealed box or tube. I'll investigate.

I have REW for measurement and nice measurement mic and a Focusrite scarlett 2i2 for measurement. I'll plan on doing some outdoor measurements after it cools down a bit here. I've done some measurements inside.

These are in-room measurements, with crossovers as described in the first post.
2226:
72911
2445:
72913
2405:
72912
L/R FR
72914

[danvprod]

Here is an annotated picture of my protection crossover. The 2445 is 1.5 octaves below my crossover frequency. The 2405 is 2 octaves below. Both caps are bypassed and there is a 47 Ohm resistor placed parallel to the input, an idea I got from several other threads here.
72923

[ARMED]

you can start from getting rid of that white horn, and selling it to me:D

[DingDing]

About the 2123, yes, you should def. investigate. It's very efficient and can handle 250W too. The box can be very small. My system is fully active (and I know next to nothing about passive crossover theory). The 2123 is between 2226 & 2446. BW 24dB/oct filters + correction filters and time alignment.

I actually made a thread asking about advice on protection networks yesterday, so the annotated picture of your network was a positive surprise. What is the point of the 47R resistor and the 0.01mF cap, and what voltage rating does that chunky 50mF resistor have, please?

[danvprod]

About the 2123, yes, you should def. investigate. It's very efficient and can handle 250W too. The box can be very small. My system is fully active (and I know next to nothing about passive crossover theory). The 2123 is between 2226 & 2446. BW 24dB/oct filters + correction filters and time alignment.

I actually made a thread asking about advice on protection networks yesterday, so the annotated picture of your network was a positive surprise. What is the point of the 47R resistor and the 0.01mF cap, and what voltage rating does that chunky 50mF resistor have, please?

Do you have a thread about your build? I'd be interested in seeing what you have developed. I'm glad the protections network diagram helps.

Here was the thread I found the crossover idea on: http://www.audioheritage.org/vbulletin/archive/index.php/t-19613.html

My 50 uF is 250v btw. The 47 Ohm resistor is to cut the impedance peak at 750 Hz; it's suppose to be a damping resistor. The 0.01 uF cap is a bypass cap.

Here is the info about the bypass caps: http://www.hawthorneaudio.us/forums/viewtopic.php?f=1&t=3763

[danvprod]

you can start from getting rid of that white horn, and selling it to me:D

:) They are nice. 250 Hz Tractrix from StereoLab

[DingDing]

Do you have a thread about your build? I'd be interested in seeing what you have developed. I'm glad the protections network diagram helps.

Here was the thread I found the crossover idea on: http://www.audioheritage.org/vbulletin/archive/index.php/t-19613.html

My 50 uF is 250v btw. The 47 Ohm resistor is to cut the impedance peak at 750 Hz; it's suppose to be a damping resistor. The 0.01 uF cap is a bypass cap.

Here is the info about the bypass caps: http://www.hawthorneaudio.us/forums/viewtopic.php?f=1&t=3763

Thank you! :)

Unfortunately, there's no thread for them as of yet, but I did some research on here a while back which may be of interest to you as it is fairly close to what you're doing. Please look at the DingDing goes DIY thread (http://www.audioheritage.org/vbulletin/showthread.php?37404-DingDing-goes-DIY-JBL-2404-mounted-inside-or-outside-2360A-and-one-or-two-2123) here.

It's still a work in progress, but the system is basically 2404, 2446 in 2360A, 2123 and 2226 + subs for the lows. I tried without the 2123, but I could never really get the magic happening from 250-300ish up to around 1kHz until the 2123 was added. That is not to say it's not possible! It just means I couldn't do it, and my experience is not that vast. I've been so happy with the performance that development went sluggish for a while and I've just recently started picking up on it again. :D

[danvprod]

BTW -- I am using Lee in Montreal's suggestions for Bass EQ on the 2226s. It sounds great:

"Basically, you will need two filters on the woofers. Try the following and adjust to your own tastes.
a- 12db bandpass at 40Hz w/ Q=1.4
b- 6db bandpass at 65/70Hz w/ Q=1.1"

I have my 2226s rolling off at 800 Hz LR, and my 2445s rolling off at 500 Hz, for approximately a 600 Hz LR 4 electrical crossover.

[berga12]

Hi,
I had more or less a similar setup, starting with my personal suggestions:

- 2226J --> 16ohm require a HUGE amplifier to have good amount of current, so run up to 500-600W - 8ohm amplifier or a Class D, tuning the 2226 to 38-40-42-45hz is making more ore less the same brake up freq at 40hz, if you want something nicer you can try to go down to 35hz and put a subsonic filter 24db to 30-32hz.

- 300b --> use them for the FULL high session placing a passive filter between 2445 and 2405 (use it around 9-10khz...), use good component, paper oil capacitors, flat wire coil and autotransformer to attenuate 2445.

- JABO KH55 --> I don't like this horn, I had for few months used with a 2441.......they sound good but too "wide" and coloured, I prefer JBL2350 style horn, but "degustibus" this is not a "back and white" factor, Just use them, if you can put some damping material on the back (like BRAX noise stopper etc), I've done it and it improves a lot the sound, keep this in mind.

- Crown D45 --> don't use them on 2405...my good.... maybe a small tube amp like 0,5-1W with Real quality, or passive network with 300b like I suggest above.

- SELL --> both Crown to have a BIGGER AMP (read: more power...) for low bass, Maybe MC2300, you have tons of choice.


Wonderfull setup, a friend of mine is using 2226h in 90L 45hz, B&C 2" driver in a circular horn and Fostex T500, superb setup, with 300b tamura amp and Hypex UCD400 for low bass.

below my Old setup, see the Horn treatment.

7299372994

[danvprod]

Hi berga12

2226 -> Thank you for the detailed response here. This is quite helpful. Good idea on a bigger amp for the 2226. I think my tuning ends up coming up at 40 Hz (both modeled and also measured in-room) and then I end up lifting the response a bit in the bass (which is all the more reason for a bigger, higher-current amp). Right now I have a sub-sonic at 20 Hz, but I might lift that to 30-32 Hz as you've said and experiment with that.

300B --> Also a good idea here. Are you suggested a 12 dB/octave filter at 10 kHz, for example? Or let the 2445 run all the way unfiltered. This certainly would make a simpler wiring setup, with one fewer amp.

JABO KH55 --> Thanks for the thoughts here. I've often considered exploring a different horn, maybe a wooden smith horn. Good idea on the noise-stopper. I'll explore that.

Agree with you about selling the Crown Amps and getting something bigger, and looking at a 2-amp vs. 3-amp setup. Should be a nice improvement.

If you have any ideas for the XO between the 2445 and 2405, that would be helpful.

BTW, I did experiment a bit more with the MiniDSP last night. This is where I am at. I decided to loosen the slope on the 2445 to 12 dB an octave. That seems to integrate better with the 24 dB/octave on the 2446 (being that the horn is already falling off pretty quick at about the crossover frequency). I do have a weird bump at 700-800 Hz with the 2226, but I think that is a floor bounce or sidewall bounce. I change the rolloff of the 2445 to 12 dB at 8kHz as well and that seems to integrate pretty good with the 2405.

72995

And thanks for the example system. It's very nice.

[ivica]

Hi,
I had more or less a similar setup, starting with my personal suggestions:

........

- JABO KH55 --> I don't like this horn, I had for few months used with a 2441.......they sound good but too "wide" and coloured, I prefer JBL2350 style horn, but "degustibus" this is not a "back and white" factor, Just use them, if you can put some damping material on the back (like BRAX noise stopper etc), I've done it and it improves a lot the sound, keep this in mind.
.......

Hi berga12,

What about YUICHI-A290 horn...

regards
ivica

[berga12]

Yes YUiICHI or IWATA project I prefer insthead of JABO KH55, but this is Just for my personal Taste

I still Own a pair of JBL2350, and I would keep them, IWATA I listen several time and they are state of the art for 2" drivers...IMHO.



About crossover, if you want to try quickly you can easily find 2405 12db schematics looking at every JBL 43XX studio monitor, the filter is more ore less the same for everyone.

for the 2445 I don't know about their impedance if it's similar to other drivers but I would place a Low Pass 12db at around 8-9Khz (for the 2441 it's around 0,48mH series and 1uF parallel, then you place an LPAD or autotransformer...pay attention with this one, it change CX point)
for 2405 around 1,5uF series, 0,16mH parallel, 1,5uf series (18db)

then you can play around with value.

Low pass for 2226 and hipass for 2445 use the Mini Dsp (Just place a big cap as protection, around 16-20uF series to 2445)

or you can also decide to have a passive 12db high pass, but the cost will not be so cheap, consider a 16-20uF if i'm not wrong and a 6,6mH parallel .....

use the DSP for the moment.

[berga12]

Here is an annotated picture of my protection crossover. The 2445 is 1.5 octaves below my crossover frequency. The 2405 is 2 octaves below. Both caps are bypassed and there is a 47 Ohm resistor placed parallel to the input, an idea I got from several other threads here.
72923



In case your DSP fail with filter, and you are running good amount of power you will DESTROY your diaphagram with those value.

2405 is not happy below 6,5khz 18db (not 6!!!!)
2445 is really not happy below 200hz!!!!!

pay attention, diaphagram for 2405 are nearly not existing anymore....

[danvprod]

72998

I found the 4343 reference thread and also this thread (http://www.audioheritage.org/vbulletin/showthread.php?21976-crossover-thoughts-for-JBL-2405-compression-tweeter) on 2405 crossovers. Above the resulting design.

On the 4343, there is an additional 10 ohm resistor parallel to the terminals. And note, the choke should be .12 mH, not 1.2, it's incorrect in my sketch.

This would give me 18 dB/octave and having the l-pad will let me tweak the level. That may be a good first step here.

1) Build crossover network for 2405, place in parallel with 2445 on 300B amps.
2) Sell both Crown D45 amps, purchase something new for the 2226s.
3) Investigate a replacement horn for the 2445 (or add damping to the current horn).
4) Add 16 uF protection HP filter to the 2445.
5) Finish crossover for 2445 (6-12 dB/octave LP filter; l-pad). Should the l-pad for this be 16 ohm? -- I'm looking at the JBL 4343B schematic here (http://www.audioheritage.org/vbulletin/showthread.php?16951-4343-References-Thread).

[danvprod]

I did find all of the network schematics here (http://www.jblproservice.com/navigation/Network%20Schematics.html). And also learned something that I want to verify:
1) There really is only one version of the 2405, and it's not really 8 or 16 ohms, it is more like 10, even though it says 16 ohms.

Looks like there is 3 crossovers to look at for the 2405:

3155: http://www.jblproservice.com/pdf/Network%20Schematics/3155%20Network.pdf
3145: http://www.jblproservice.com/pdf/Network%20Schematics/3145%20Network.pdf
3107: http://www.jblproservice.com/pdf/Network%20Schematics/3107%20Network.pdf

The 3107 seems to be the most straight-forward. I don't know what the variable inductors are in the 3155 and 3145. But the 3107 seems to be closest to my previous post. 1.0 uF amp-side, .16 mH inductor and 1.5 uF speaker side with an 8 ohm l-pad.

3145/55 have bypass caps on the two caps, more resistors parallel to the load and what looks to be an adjustable inductor. Sorry I am a novice at this stuff. I appreciate all the help.

[grumpy]

I don't know what the variable inductors are in the 3155 and 3145

Fixed/tapped, not variable. Acts as an inductor element, impedance matching transformer/level shifter.

Hard to obtain exact replacements, and as you've noticed, alternatives exist :)

Difficult to DIY without a lot of learning about core materials, a winding jig, and some test equipment
(not on my bucket list).


... and yes, just the one 077/2405 at 10-12 ohms where it matters.

[berga12]

Dear Danv, pay a lot of Attention on "copying networks around".


for the 2405, do Streight forward, every schemathics is similar on every crossover, some use tapped autotransformer, some LPAD (see 3155 equivalent), but more ore less are similar, I use 1,5uF - 0,14mH - 1,8uF and my L pad is 4.3ohm series and 36ohm parallel.

for the 2445 you can concentrate on Low Pass, as I told you value like 0,48mH and 1uF could fit for a cut on top about 8-10Khz

for the High pass If you want to Run PASSIVE high pass, remember that every studio monitor cut the driver at 1,2-1,5Khz, and you need something around 400hz (theoretically 2xHorn Fc, means 250hzx2= 500hz....)

and for 400-500hz the value Are much bigger than the 4343 or 4355 filter... (notice that only 4350-55 have 2" Driver!!!! like 2445)

Yes you need an LPAD setup on 16ohm for 2445, Just search LPAD calculation and place 16ohm and the required Db attenuation -3 -6 -9 -12 etc... between 2405 and 2445 should be around 6/9db of difference.

This is always Just my opinion! Not a Must.

[danvprod]

Thank you, grumpy for the information on the tapped inductor. Yes, not going to deal with all that. The l-pad will work. I did noticed that there is a 20 ohm resistor in parallel with the 2405 in some of the schematics -- is that just for impedance control?

@berga12 -- thank you for your continued support and ideas. For the 2405 1,5uF - 0,14mH - 1,8uF is where I will go here. I'l likely do the variable l-pad, just to dial in the attenuation.

I'll do the protection cap on the 2445 (16-20 uF) and the 12 dB LP as you've suggested -- 0,48mH and 1uF

Can you expand on this set of filters? "cut the driver at 1,2-1,5Khz, and you need something around 400hz (theoretically 2xHorn Fc, means 250hzx2= 500hz....)" What do these filters look like? I could do this in my MiniDSP, likely. I don't want to go down the route of a passive HP and filters, just because of the price as you've stated. If I have some ideas about the Q, dB and center frequency for the 1,2 - 1,5 cut, and what to do around 400 Hz, that would be helpful. I could try that tonight in the MiniDSP.

Regarding the l-pad between the 2405 and 2445, are you thinking that the 2445 needs to be padded down more than the 2405 or the other way around? Either way, I'll likely get the variable l-pad at 16 Ohm for the 2445 and 8 Ohm for the 2405.

Many thanks again, and I appreciate you answering all my questions.

[berga12]

Just in my mind, if I remember well, 2445+Horn(kh55) is about 110-111db/W/m, the 2405 is about 105-106db/W/m realistic...

so depending on your ear sensitivity you have around 6db between 2445 and 2405. With a biamp in your case it could be that you don't need attenuation on the 2405 and you can go down with full mid high using your DSP and adjust level between the mid and super tw. Actually most of audiophile, or like Klipsh filter uses variable Autotrasformer only on Mid Driver.



You need an HIGH PASS done via your DSP and a protection cap, Around 400-500hz (theoretically 500hz, 2 times horn Fc)

___________________________________

JBL studio monitors (like 4350-55) uses the 2" driver from 1,2Khz to 10Khz, but this is not your case, so forget about this sentence and don't look at HP session in JBL studio monitors filter schematics.

___________________________________

I'm shure you will be happy with That, if possible in the meantime, use Crown D45 bridged on 2226J

[ivica]

Thank you, grumpy for the information on the tapped inductor. Yes, not going to deal with all that. The l-pad will work. I did noticed that there is a 20 ohm resistor in parallel with the 2405 in some of the schematics -- is that just for impedance control?

@berga12 -- thank you for your continued support and ideas. For the 2405 1,5uF - 0,14mH - 1,8uF is where I will go here. I'l likely do the variable l-pad, just to dial in the attenuation.

I'll do the protection cap on the 2445 (16-20 uF) and the 12 dB LP as you've suggested -- 0,48mH and 1uF

Can you expand on this set of filters? "cut the driver at 1,2-1,5Khz, and you need something around 400hz (theoretically 2xHorn Fc, means 250hzx2= 500hz....)" What do these filters look like? I could do this in my MiniDSP, likely. I don't want to go down the route of a passive HP and filters, just because of the price as you've stated. If I have some ideas about the Q, dB and center frequency for the 1,2 - 1,5 cut, and what to do around 400 Hz, that would be helpful. I could try that tonight in the MiniDSP.

Regarding the l-pad between the 2405 and 2445, are you thinking that the 2445 needs to be padded down more than the 2405 or the other way around? Either way, I'll likely get the variable l-pad at 16 Ohm for the 2445 and 8 Ohm for the 2405.

Many thanks again, and I appreciate you answering all my questions.



Hi
I think that our forum member 4313B has shown several solutions:

http://www.audioheritage.org/vbulletin/showthread.php?24283-4344-Mk-II-network&p=247755&viewfull=1#post247755

I have been experimenting with the attached solutions with the following (Radian diaphragm)
2441Rad & 2311 & 2308 + 2405

Regards
ivica

[4343]

Are the TruExtent Be diaphragms available in 16 ohms?

Personally, I would double the 2226J for 8 ohm bottom and use 8 ohm diaphragms from the start, but you have what you have...

[danvprod]

Are the TruExtent Be diaphragms available in 16 ohms?
I do believe they are:
http://reconingspeakers.com/product/bex4016-truextent-beryllium-diaphragm-16-ohm-copy/
$650/each though. Ouch.

http://www.parts-express.com/radian-1245-16-diaphragm-fits-most-jbl-15-and-2-16-ohm--294-728
Radians may be another option, at $108/each.


Personally, I would double the 2226J for 8 ohm bottom and use 8 ohm diaphragms from the start, but you have what you have...
re: components -- I purchased all of this as a group -- the three sets of drivers and the horns. So I am not against upgrading certain items, but I largely have what I have as you've said and I want to optimize the components I have rather than replacing a lot of stuff.

[danvprod]

73032
I put my 12 dB/octave crossover back on my 2405 this evening. This is a .3 mH inductor and 1.5 uF cap. These are both running on the 300B amps now.

What do you think about a Crown D150A Series II for the 2226? Seems like those are pretty reasonably priced and would work reasonably well. Also contemplating an Adcom GFA 555 or similar.

Emotiva A-300 may be a good option as well: http://emotiva.com/products/amplifiers/300

[berga12]

Just pay a lot of attention on 2405, ok 300b are around 4-5Watt, but in general 18db is suggested for 2405, you can try with another 1,5uF after the coil, 0,3mH seems a bit to high, maybe is just something on Q factor.


About low pass I would consider a minimum 400-500W into 8ohms, so you will have 200-250W on 16ohm, that's why I ask you if Crown D45 can be placed in bridge mode.

consider 2226h are 96db
mid high around 106db average

so you have to gain around 10-12db it means compared to 300b amp (if I consider 5W) that you need around 80-100W for the low, but If you think that normally we want a bit more bass due to ears sensitivity and due to Room loss, impedance curve,different input sensitivity of the amplifiers etc etc... Keep in mind good 200Wrms :)


how it sound now?

[danvprod]

Just pay a lot of attention on 2405, ok 300b are around 4-5Watt, but in general 18db is suggested for 2405
Yes, I plan on rebuilding this with the correct components. This was what I had on-hand to test out. I'll build the correct network next.



Crown D45 can be placed in bridge mode.
I can't bridge them -- I believe there is something you have to do internally to enable that. I'm not going to mess around with. I'd rather just sell them and get something new. I actually found a really excellent condition D150 locally for a quite good price. I am going to pick it up and give it a try so I can sell my D45s. I can always use that amp for my garage speakers if I get something bigger for the 2226s



how it sound now?
It sounds OK. I put a -3 dB cut @ 1.2 kHZ Q1 for the MF and am using a 500 Hz 24 dB/octave HP filter in the MiniDSP for the 2445. Still more work to do.

Best,
Dan

[Lee in Montreal]

Maybe a dumb question. But why not use a 24bd/octave filter which would allow to keep phase between drivers and perhaps even better protect the 2405 diaphragm?

Oups. Edit. The last reply mentions 24db filter. Sorry. ;-)

[danvprod]

73045
Found this guy locally on craigslist today. Good shape and was able to get it for a really good price. Should hold me off for a while. Listen to Random Access Memories right now and it is definitely a nice upgrade over the D45. No trim being applied to either the 300B or the D-150 so that seems to be a pretty good match.

[ivica]

Maybe a dumb question. But why not use a 24bd/octave filter which would allow to keep phase between drivers and perhaps even better protect the 2405 diaphragm?

Oups. Edit. The last reply mentions 24db filter. Sorry. ;-)
Hi




Fourth Order Linkwitz-Riley (http://www.audioheritage.org/../../../Documents%20and%20Settings/ivicai/Local%20Settings/Temp/schemo/4P2way.html)



Fo=9000Hz








UHF (10 Ohms)- High Pass Section




C1=
0.94
uF




L1=
0.11
mH




C2=
1.88
uF




L2=
0.50
mH




VHF (12 Ohms) - Low Pass Section





L3=
0.40
mH




C3=
2.34
uF




L4=
0.20
mH




C4=
0.52
uF











Sum at Fc=
0
dB




Tweeter Polarity=
Normal

[pos]

A textbook electrical LR 24dB/oct filter is unlikely to produce an acoustical LR 24dB/oct filter, especially for a HP filter on a horn.

[ivica]

A textbook electrical LR 24dB/oct filter is unlikely to produce an acoustical LR 24dB/oct filter, especially for a HP filter on a horn.

Hi POS,

So what would be Your suggestion
if ONLY passive filters have to be applied?

Not to mention geometry, phase response,....

Regards
Ivica

[berga12]

Ivica, you should measure the Loudspeaker in real condition to check Impedance curve response, then calculate the filter based on Real Impedance at desired frequency, A nominal Freq can't be used, especially if he would like to cut at 500hz, but that's no problem because he will use the DSP for HP.

In general the value you suggest for the 24db are quite fitting with reality because you are cutting Far Far away to Resonance freq and horn resonance so impedance value are closer to nominal.

A check with Loudspeaker with an instrument like CLIO board or DATS or and Impedance measurment during frequency operation should be a must, especially on old loudspeakers.

[ivica]

Ivica, you should measure the Loudspeaker in real condition to check Impedance curve response, then calculate the filter based on Real Impedance at desired frequency, A nominal Freq can't be used, especially if he would like to cut at 500hz, but that's no problem because he will use the DSP for HP.

In general the value you suggest for the 24db are quite fitting with reality because you are cutting Far Far away to Resonance freq and horn resonance so impedance value are closer to nominal.

A check with Loudspeaker with an instrument like CLIO board or DATS or and Impedance measurment during frequency operation should be a must, especially on old loudspeakers.

Hi berga12,

I can understand all the mentioned, but if someone want to put 2405 with 244x to improve UHF and dispersion response what would You suggest him, if only passive filters 'are on the table'?

I do not understand what is so special with "...especially on old loudspeakers...", is it their age or technical operating condition, I think technical operating condition of newer drivers can involve some troubles if were treated unappropriated.

If we are talking about DSP driven FIR filters speaker active networks, then appropriated spacial measurements have to be done in order to compensate most of the drivers , or even ambience, "irregularities" , but to do that a person has to be properly educated with the experience in that fields.

Try to do high resolution measurements of the F/R of the speakers that are on Your avatar, and try to move the measurements mic 50cm left or right and make another measurements and compare them, then up and down, and compare all, after that if You decide to "throw them away" please INFORM us before, where would be that place, so we can ...... :-))

regards
ivica

[berga12]

My suggestion is few reply below, but also yours it's worth a try ;)

I'm not a fan of 24db on mid high, I do prefer 6-12-18, if possible 6 and 18db order.

I prefer 24db in the separation with Low-sub and for low freq.

[Lee in Montreal]

I don't understand what's not to like with a LR 24db/oct. filter. ;)

You get a flat response (no hump at crossover point) and get to keep the phase (well. maybe a single cycle off - yet still matching). Sure one can argue that the impedance needed for calculation might be the actual measured impedance, not the nominal one, but hey, this is nothing more than a detail in a principle. :D

[berga12]

for my taste a smoother transition between the mid-high way and a bit of overlap is preferable.

I've this rule in my head since I was teenager to use an odd order filter especially for the tweeter :D and as soon as you can, avoid parallel filter before driver.

In reality we should use a passive New on the imput low signal before the amplifier and remove passive networks.

Or a tube passive crossover (line driver)

[ivica]

for my taste a smoother transition between the mid-high way and a bit of overlap is preferable.

I've this rule in my head since I was teenager to use an odd order filter especially for the tweeter :D and as soon as you can, avoid parallel filter before driver.

In reality we should use a passive New on the imput low signal before the amplifier and remove passive networks.

Or a tube passive crossover (line driver)

Hi berga12,
Yes You are right, but when using 2405 and 244x&2311&2308 due to the time not-alignment a huge comb filter effect would be present

regards
ivica

[pos]

Hi POS,

So what would be Your suggestion
if ONLY passive filters have to be applied?

Not to mention geometry, phase response,....

Regards
Ivica
I was thinking about active filtering, but passive also adds the complication of having to deal with impedance curves as berga12 noted.

But impedance aside (easy enough to linearize using a pad), and even with active filtering, you have to take the natural rolloff (and general response) of the driver when building the eletrical filter.
What matters is the acoustical result, which is the "addition" of the electrical and "natural" filters.

If you want flat on axis summation, predictable off axis behavior, and all the good things that come with properly done LR filtering then you need to obtain an actual acoustical LR, and use whatever electrical (passive, active, a combination, or whatever) filters needed to reach that goal.
For a horn that typically implies using a lower order filter, and most likely at a different frequency.

As an example the M2 has a passive circuit that linearizes the impeance (lpad) and applies a 6dB/oct filter at 1kHz, then an active set of EQs and another 6dB/oct filter at around 800Hz.
The combination of the passive network + active settings (EQ, shelving and filter) + natural response of the driver/horn combo achieves an almost perfect LR 36dB/oct acoustical filter at around 800Hz. That means that the steepness natural filter of the driver/horn combo is around 24dB/oct. That would be hard to ignore...

Even when aiming a crossover frequency far enough from the natural cutoff of the horn it cannot be ignored, especially as far as the phase is concerned.

[danvprod]

Thank you, pos et al. I am going to try and bring my 2445 and horn outside this weekend and capture a gated outdoor measurement. That will be a good starting point. Is there a good tool you know of where I can simulate what the ideal response should be -- say 24 dB octave at 500 Hz between the LF and MF and then see what combination of natural roll-of from measurement and electrical filter vis DSP I can provide (plus passive protection cap) might get my response close to this ideal filter?

This thread comes to mind: http://www.diyaudio.com/forums/multi-way/231594-quasi-optimal-crossover-high-efficiency-loudspeaker-system.html

[danvprod]

73102
I took the 2445 and 250 Hz horn outside and did a gated measurement to get a better sense of the frequency response without the impact of my room. I also put together a "Quasi-Optimal" crossover topology between the LF and MF speakers (http://www.diyaudio.com/forums/multi-way/231594-quasi-optimal-crossover-high-efficiency-loudspeaker-system-6.html) in the MiniDSP.

Measurements above are taken elevated 2m (on top of a 6' extension ladder) at a microphone distance of 1m. This is on grass with the ladder wrapped in a fleece blanket. I'm able to get 4ms of clean decay before the first reflection off my garage and my neighbor's garage. It's not perfect but not bad.

I took a 0 degree, 30, 60 and 90 degree measurement. (I don't have a turntable so these are approximate). 8 kHz is definitely a good handoff frequency between the horn and the 2405. Using this measurement as a guide, I put a -2.5 dB cut @ ~950 Hz with a Q of 1.5 to tame that hump in the response. The peaks at 2.5 kHz and 5.5 kHz are interesting. I don't know what they are from. Any ideas? likely reflections inside the horn, perhaps?

Also, based on this measurement, I decided on a new XO frequency of 700 Hz. I adopted one of the quasi-optimal topologies this evening to see how it sounds.

Fx = 700 Hz

2226 (Crown D-150A Series II amp -0 dB):
6th Order LR @ Fx * 1.06 (+) polarity = 742 Hz.
2nd Order LR @ 30 Hz HP.
+8 dB 40 Hz Q 1.4 "Lee" filter and +3 dB 70 Hz Q 1.1 "Lee" filter

2445 (300B amp)
3rd order Butterworth @ Fx * 1.13 (-) polarity = 791 Hz.
968 Hz -2.4 dB Q 1.56
8.5" offset delay introduced by the filter -- .445 * 1100/700 = .69 feet = 8.5"
50 uF protection cap on 2445 still. <-- Need to address.
12 dB/octave filter on 2405. <-- Need to address.

Subjectively, the 2445s sound much less strained @ 700 Hz. And better integration between 2226 and 2445. Off to go order some passive XO parts.