I made a need some help fixing a 1990's design I made that turned out so-so:
Dynaudio D-28AF 1" dome tweeter
Dynaudio D-76 3" dome midrange
JBL 2235H 15" woofer
vented 6.2 cu. ft/175L cabinet
Xover is a clone of Profil 4
I used up all my energy on the project getting the low end great- making 4 enclosures and dozens of ports to get there as the TS modeling didn't pay off.
But my main complaint is the midrange is poor- image is hard to find, mids are distant, washed out and in the background, more H&V lobes than I'm used to. I think it's the big difference in cone centers (2.3") between the mid/woof on the flat baffle giving poor time alignment but the odd crossover-mid phasing network I thought would help.
I'm open to suggestions here. At the time Audax PR17HR70 midrange was recommended (obsolete now) and I'm thinking the midrange or crossover need to be redone.
Dynaudio D-28AF 1" dome tweeter
Dynaudio D-76 3" dome midrange
JBL 2235H 15" woofer
vented 6.2 cu. ft/175L cabinet
Xover is a clone of Profil 4
I used up all my energy on the project getting the low end great- making 4 enclosures and dozens of ports to get there as the TS modeling didn't pay off.
But my main complaint is the midrange is poor- image is hard to find, mids are distant, washed out and in the background, more H&V lobes than I'm used to. I think it's the big difference in cone centers (2.3") between the mid/woof on the flat baffle giving poor time alignment but the odd crossover-mid phasing network I thought would help.
I'm open to suggestions here. At the time Audax PR17HR70 midrange was recommended (obsolete now) and I'm thinking the midrange or crossover need to be redone.
I have used the Audax PR170M with very good success with the 2235H.
I can't really tell, but it looks like the Dynaudio mid may not have enough SPL compared to the 2235H. It may be a few dB short and might explain some of the dullness.
You really need to provide complete details for the total design before anyone can meaningfully provide answers to your questions.
I think your crossover is not done correctly. Add a 7" is a good idea but not the PR17, 600Hz crossover frequency, too high. It only replace the 3"mid. Go to a 4 way is not a good idea.
I think you just need to redo the crossover, you don't need to do a lot of thing.
Do you have a photo of your baffle ?
If your baffle looks the profil 4, it's OK.
About the crossover, it should be very simple.
Can give us your schema ?
According to the crossover schema of the profil 4
http://img69.imageshack.us/img69/7225/duennaudiokn5.jpg
The transition between the 3" and 1" is easy and very simple, keep it.
The only problem i see is the transition between the 15" and the 3" because the woofer is different, the crossover should be different.
Do you keep the delay line Lp/Cp ? you can remove it because it affects only the mid, a source of problem.
You should only play with L1/C1/R1 ...
I think you could have problem with the polarity of the drivers, wait for your crossover schema...
edit : this a similar project : http://www.parts-express.com/projectshowcase/indexn.cfm?project=AngelF1
I like the fact he have used the same capacitor for mid and tweeter.
I think you just need to redo the crossover, you don't need to do a lot of thing.
Do you have a photo of your baffle ?
If your baffle looks the profil 4, it's OK.
About the crossover, it should be very simple.
Can give us your schema ?
According to the crossover schema of the profil 4
http://img69.imageshack.us/img69/7225/duennaudiokn5.jpg
The transition between the 3" and 1" is easy and very simple, keep it.
The only problem i see is the transition between the 15" and the 3" because the woofer is different, the crossover should be different.
Do you keep the delay line Lp/Cp ? you can remove it because it affects only the mid, a source of problem.
You should only play with L1/C1/R1 ...
I think you could have problem with the polarity of the drivers, wait for your crossover schema...
edit : this a similar project : http://www.parts-express.com/projectshowcase/indexn.cfm?project=AngelF1
I like the fact he have used the same capacitor for mid and tweeter.
Last edited:
Finally, i looked the profil 4 crossover. I manage to do something like this:
If you have L1 = 3.9mH, C2 becomes 28uF.
I suppressed the delay line and inverted the midrange. Midrange inversion is necessary because acoustic slopes are 12dB.
Crossover frequencies are 700Hz/4kHz. The overall sensitivity will be 93dB/2.83V.
Let us know how the sound changes.
Enjoy
If you have L1 = 3.9mH, C2 becomes 28uF.
I suppressed the delay line and inverted the midrange. Midrange inversion is necessary because acoustic slopes are 12dB.
Crossover frequencies are 700Hz/4kHz. The overall sensitivity will be 93dB/2.83V.
Let us know how the sound changes.
Enjoy
Attachments
more info and pics
Here is a pic and xover details (old pic missing woofer RC). Solen parts.
Cabinet outside dimensions are 45.5"H x19.5"W x 18"D (3/4" MDF) - Driver centers (from the floor up) measure 22.75", 34.75", 40" (tweeter)
The front baffle is one piece, just two-tone paint.
Here is a pic and xover details (old pic missing woofer RC). Solen parts.
Cabinet outside dimensions are 45.5"H x19.5"W x 18"D (3/4" MDF) - Driver centers (from the floor up) measure 22.75", 34.75", 40" (tweeter)
The front baffle is one piece, just two-tone paint.
Attachments
Thank for the pictures.
You have to tweak the crossover
-First remove the delay line (0.8mH 24uF), connect the mid inverted and listen. if you want to keep it, you should add the same on the tweeter but i think it is a bad idea.
- The 39uF is too much but you can tune it by ears.
- Perhaps the 47uF is too much but not vital.
You can fine tune the crossover after doing the first tweak.
A comment on the box, you must flush mount tweeter and midrange. It could be better to offset them, see with edge Home of the Edge
You have to tweak the crossover
-First remove the delay line (0.8mH 24uF), connect the mid inverted and listen. if you want to keep it, you should add the same on the tweeter but i think it is a bad idea.
- The 39uF is too much but you can tune it by ears.
- Perhaps the 47uF is too much but not vital.
You can fine tune the crossover after doing the first tweak.
A comment on the box, you must flush mount tweeter and midrange. It could be better to offset them, see with edge Home of the Edge
For drivers placement take a look to ATC SCM150
ATC Loudspeakers. ATC Monitors. ATC SCM. ATC Speakers. Manufacturers of speakers, monitors, drive units and amplifiers.
You could have a smoother response.
ATC Loudspeakers. ATC Monitors. ATC SCM. ATC Speakers. Manufacturers of speakers, monitors, drive units and amplifiers.
An externally hosted image should be here but it was not working when we last tested it.
You could have a smoother response.
The resistance of 6.8 in series with the 47 µF cap shall be reduced to 1-2 Ohm . And , can the delay line before the dome mid be thinked like a notch filter with wide band ?!
Also ,driver offset can be corrected by tilting the cabinet about 10° .
Last edited:
Hi picowallspeaker, I really don't think he wants to reduce the 6.8 ohms in series with the 47uF to 1-2 ohms. As far as I can see that is a zobel and doing so will result in an impedance of 1-2 ohms above a certain (I suspect quite low) frequency for the woofer part of the circuit. As the speakers are in parallel, this will result in an impedance of 1 ohm or less above whatever that frequency is presented to the amp.
Prairiemystic I've not seen a network like that before on the midrange with reversed components in the negative and positive legs of the driver. Was the crossover you used designed for the same drivers with the same baffle dimensions?
Your best bet is to do some acoustic measurements of the drivers on the baffle and also impedance measurements of the drivers on the baffle. Then you can put them into something like Jeff Bagbies passive crossover designer or speaker-workshop and tweak away. Also if you do a measurement using the the current crossover as well, you can hopefully see what is going on.
Tony.
Prairiemystic I've not seen a network like that before on the midrange with reversed components in the negative and positive legs of the driver. Was the crossover you used designed for the same drivers with the same baffle dimensions?
Your best bet is to do some acoustic measurements of the drivers on the baffle and also impedance measurements of the drivers on the baffle. Then you can put them into something like Jeff Bagbies passive crossover designer or speaker-workshop and tweak away. Also if you do a measurement using the the current crossover as well, you can hopefully see what is going on.
Tony.
Last edited:
I'd ditch the D76 and get a B&C 6MD38. You'll need to re-do the xover though.
I analyse this schema, crossover frequency 500Hz and 4kHz. 39uF & 47uF are finally OK for me. You just need to remove the delay network (0.8mH/24uF) and connect the midrange inverted.
This schema is very classical and work very well but with a delay network it is a disaster. The difference of offset between the 15" woofer and the 3" mid is at least 4", 10cm, it's very hard to compensate physically.
You should have some diffraction problems, could be resolve by flush-mounted the mid-tweeter and offset them on the panel. It could be a second step of improvement.
Note a measurement can tell what's happen but it is an investment.
I heard this speaker in the late 80s. Very nice sound, but lacking a bit in midbass dynamics and very "polite" sounding.
3 major issues I see -:
1 - 500Hz is too low to cross to the D76AF, especially first order. Domes don't have the Xmax and dynamics below 500Hz.
2 - You have no resonance compensation network on the D76, effectively making the crossover freq more like 300Hz which is WAY too low for any dome.
3 - The ladder network may have been fine to merge with the Dynaudio 30W54 woofer, but it's unlikely to be suitable for your JBL.
Options -:
a) Raise the crossover point to 800Hz and add resonance compensation across the D76.
or
b) Add a 7-8" midbass driver, handling 200-800Hz.
or
c) Replace with D76 with a good 4-5" midrange driver.
Assymetry in drive placement would probably help your imaging situation.
3 major issues I see -:
1 - 500Hz is too low to cross to the D76AF, especially first order. Domes don't have the Xmax and dynamics below 500Hz.
2 - You have no resonance compensation network on the D76, effectively making the crossover freq more like 300Hz which is WAY too low for any dome.
3 - The ladder network may have been fine to merge with the Dynaudio 30W54 woofer, but it's unlikely to be suitable for your JBL.
Options -:
a) Raise the crossover point to 800Hz and add resonance compensation across the D76.
or
b) Add a 7-8" midbass driver, handling 200-800Hz.
or
c) Replace with D76 with a good 4-5" midrange driver.
Assymetry in drive placement would probably help your imaging situation.
Last edited:
From my experience with the 2235H it will not play well above 500 Hz due to some nastiness with cone breakup. 800 Hz may be pushing it depending on the slope.
Ouch
To improve the crossover, measurements are needed. A replacement of the 3" dome could be a better solution.
To raise the crossover frequency to 700Hz, just replace 47uF and 39uF by the 24uFs.
I think the OP put the cart before the horse. You can't expect an off the shelf crossover to work with any drivers you throw at it. You must first know each drivers output characteristics. Only after that can you design a crossover that makes the drivers work as a complimentary set to reach a desired goal.
Bear with me, I'm a bit overwhelmed digging into this. Three things I tried to do originally, that did not work out:
1. I wanted a low xover freq 300-400Hz since the woofer is large; looks like I should move up to 500-800Hz (with the 6dB/oct xover), otherwise the mid gets pushed too low.
What I have now I measured the woofer 1/2 voltage at 500Hz (without the Zobel). Inductor DCR=0.3R and L=3.9mH; woofer DCR=6.0R and Le=1.2mH and my math is out to lunch fc 198Hz (6.3R+(1.2mH+3.9mH)). Zobel should be ~6R/33uF. Does it cancel Le out? Even then fc=257Hz.
(side note, the JBL 4430 and 4435 were two-way systems using the max. recommended 1kHz xover freq... 4430/35 and http://www.jblpro.com/pub/obsolete/443035.pdf)
2. The 5dB mismatch between the mid and woofer efficiencies. The 2235 is 93db/1W/1m and D76 88db/1W/1m? and the delay network seems to lose a few more dB (measuring voltage on the driver).
I'd have to go to a better matched midrange.
3. The time alignment of the drivers
I thought the delay network would compensate for some of the difference in centers.
Otherwise, angle the front baffle or maybe a satellite enclosure on top with tweeter/mid.
Next I tried jerome69's suggested xover changes - I removed the delay network and went to a 24uF cap. It's a big difference, at least there's an image now and the mid's are upfront. What I really notice is the vertical lobes (dead spots) are gone. I'm gonna try it for a while but this still feels like a mess.
1. I wanted a low xover freq 300-400Hz since the woofer is large; looks like I should move up to 500-800Hz (with the 6dB/oct xover), otherwise the mid gets pushed too low.
What I have now
I measured the woofer 1/2 voltage at 500Hz (without the Zobel). Inductor DCR=0.3R and L=3.9mH; woofer DCR=6.0R and Le=1.2mH and my math is out to lunch fc 198Hz (6.3R+(1.2mH+3.9mH)). Zobel should be ~6R/33uF. Does it cancel Le out? Even then fc=257Hz.(side note, the JBL 4430 and 4435 were two-way systems using the max. recommended 1kHz xover freq... 4430/35 and http://www.jblpro.com/pub/obsolete/443035.pdf)
2. The 5dB mismatch between the mid and woofer efficiencies. The 2235 is 93db/1W/1m and D76 88db/1W/1m? and the delay network seems to lose a few more dB (measuring voltage on the driver).
I'd have to go to a better matched midrange.
3. The time alignment of the drivers
I thought the delay network would compensate for some of the difference in centers.
Otherwise, angle the front baffle or maybe a satellite enclosure on top with tweeter/mid.
Next I tried jerome69's suggested xover changes - I removed the delay network and went to a 24uF cap. It's a big difference, at least there's an image now and the mid's are upfront. What I really notice is the vertical lobes (dead spots) are gone. I'm gonna try it for a while but this still feels like a mess.
It has took to me a lot of time before understand something in electro-acoustic. You have to deal with electrics and acoustics
I can give you a response for
1. The response is not simple. You can't add L+Le, it is not correct. The Le of the driver limits the treble. The L you put to filter do something but depend on the diffraction of the baffle (baffle step). Here there is no baffle step with the 2235H, the L gives a slope and the R-C does the rest.
You can read this :
epanorama.net/Speaker impedance
http://users.ece.gatech.edu/mleach/papers/vcinduc.pdf
2. No the midrange is already attenuate ! you don't need to change it !
For me the sensitivity of the mid is near 90dB. The contribution of the midrange is lower. Take a look on one of my design with 12dB acoustic slopes :
3. In your case "time alignment" is not necessary, i don't think you will have a better sound. In my schema above, the difference of acoustic center between the mid and the woofer is 10cm at 1m... In your design you should push back the mid and tweeter to 10cm too.
The delay network on the mid isn't a good idea because the tweeter contribute a lot as seen and you don't delay the tweeter !
I think it's better to make a good placement of your mid-tweeter to minimize diffraction and smooth the overall response.
You just need make fine tuning, making it sounds very realistic. Old design school, old method of works !
If the JBL2235H is a good driver, the sound must be very good !
I can give you a response for
1. The response is not simple. You can't add L+Le, it is not correct. The Le of the driver limits the treble. The L you put to filter do something but depend on the diffraction of the baffle (baffle step). Here there is no baffle step with the 2235H, the L gives a slope and the R-C does the rest.
You can read this :
epanorama.net/Speaker impedance
http://users.ece.gatech.edu/mleach/papers/vcinduc.pdf
2. No the midrange is already attenuate ! you don't need to change it !
For me the sensitivity of the mid is near 90dB. The contribution of the midrange is lower. Take a look on one of my design with 12dB acoustic slopes :
An externally hosted image should be here but it was not working when we last tested it.
The sensitivity is 85dB but the mid contribute only 82dB. The tweeter and the woofer are important ! In you design the mid contributes to 90dB but you need to attenuate it because you have a boost from the 6dB high pass.
3. In your case "time alignment" is not necessary, i don't think you will have a better sound. In my schema above, the difference of acoustic center between the mid and the woofer is 10cm at 1m... In your design you should push back the mid and tweeter to 10cm too.
The delay network on the mid isn't a good idea because the tweeter contribute a lot as seen and you don't delay the tweeter !
I think it's better to make a good placement of your mid-tweeter to minimize diffraction and smooth the overall response.
I don't understand ? The loudspeaker seems to play better now ?
You just need make fine tuning, making it sounds very realistic. Old design school, old method of works !
If the JBL2235H is a good driver, the sound must be very good !
jerome, thanks for your help! It does sound much better The mid does not sound odd, it sounds fine now (without the delay network). Matching theory and design is much harder with loudspeakers and lack of knowledge. Do you recommend any S/W for crossover design?
Eons ago, I had a summer job with a big telecom company in their acoustic lab. There was $10,000's in Brüel & Kjæ equipment, an anechoic chamber, and MLSSA on a PC. With all that space and equipment, it was still hard to make good acoustic measurements. I used it on the woofer/cabinet/port work. Then went back to school and could not use for the mid/tweeter.
I will revamp the crossover and if I am still not happy, try a cone midrange or mid-bass add on.
The JBL 2235H is an dynamite driver. I fell in love with it in the 1980's (when things were big) and reminds me of old Altecs in sound.
The mid does not sound odd, it sounds fine now (without the delay network). Matching theory and design is much harder with loudspeakers and lack of knowledge. Do you recommend any S/W for crossover design?Eons ago, I had a summer job with a big telecom company in their acoustic lab. There was $10,000's in Brüel & Kjæ equipment, an anechoic chamber, and MLSSA on a PC. With all that space and equipment, it was still hard to make good acoustic measurements. I used it on the woofer/cabinet/port work. Then went back to school and could not use for the mid/tweeter.
I will revamp the crossover and if I am still not happy, try a cone midrange or mid-bass add on.
The JBL 2235H is an dynamite driver. I fell in love with it in the 1980's (when things were big) and reminds me of old Altecs in sound.
Yes i use S/W a lot, it misses some functionalities but it's free. You have an other simulator with excel sheet : PCD, not use it but have some interesting functionalities.jerome, thanks for your help! It does sound much better
Yes measure bass is very difficult. But we can do now very good gated measurement at home. The validity of this measurement is above 200-300Hz. If you have large space, you can go lower.
Ok !
I verified with edge your driver placement, i didn't see big problem.
Do extensive listening
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.