Crossover Design Changes Based on Horn Dispersion

[toddalin]

I'm designing the crossovers for the surround mini-monitors that will use a W10GTI and a 2425 on a 2370.

They will be based on the "keeper" crossovers that show OUTSTANDING results using the W10GTI, LE175/HL-91, 075.

But the 2370 horn would have different dispersion characteristics than the HL-91. Apparently, this then would change aspects of the crossover???

Radian offers two crossovers that appear identical EXCEPT one is listed for a 60x40 horn and one is for a 90x40 horn.

So, what changes would have been made to the components to compensate for the horn dispersion if I wanted to implement them???

Thanks

http://www.usspeaker.com/radian%20324-1564-1.htm

http://www.usspeaker.com/radian%20324-1594-1.htm

[Earl K]

Hi Todd,




- I'm designing the crossovers for the surround mini-monitors that will use a W10GTI and a 2425 on a 2370.

- But the 2370 horn would have different dispersion characteristics than the HL-91. Apparently, this then would change aspects of the crossover???

- So, what changes would have been made to the components to compensate for the horn dispersion if I wanted to implement them???

- It's generally accepted that it's good design practice, to match the horizontal beam-width of the woofer ( in the crossover region ) to the horizontal coverage of the chosen horn.

- Therefore, if your chosen horn has 90° H. coverage, one would choose a crossover region where the woofers' beam-width has narrowed down to around 90° .

- Different diameter woofers all have different 90° ( 6 db down point vs frequency ) beam-widths .

- You'll need to measure your new tens to find the frequency range where they narrow down to about 90° ( assuming a 90° horn ) . A 10" woofer ( with a 8.5 " effective piston diameter ) is going to narrow to around 90° ( conical ), somewhere around 2000 hz .



I'm designing the crossovers for the surround mini-monitors that will use a W10GTI and a 2425 on a 2370.

They will be based on the "keeper" crossovers that show OUTSTANDING results using the W10GTI, LE175/HL-91, 075.

- I saw the pic of your RTA showing the net results. It did have better midrange blending .
- It's still somewhat a mystery to me how you got "Outstanding" results with a crossover that was designed around an 8 ohm woofer . If I remember correctly, you substituteded in a 3 ohm woofer load ( or 12 ohm depending on how you wired the dual voice coils ) . Why it all worked out for you, is now anyones' guess . :)


Radian offers two crossovers that appear identical EXCEPT one is listed for a 60x40 horn and one is for a 90x40 horn.

- How Radian offers two crossovers that are advertised as having the same crossover point ( and slopes ), yet are optimzed for different horn patterns is a bit of a ??? for me. I would have assumed a change in crossover point was necessary ( or a tinkering of the slope ) / so, I guess I can't help you much more on the subject .

:)

[toddalin]

Hi Todd,
- I saw the pic of your RTA showing the net results. It did have better midrange blending .
- It's still somewhat a mystery to me how you got "Outstanding" results with a crossover that was designed around an 8 ohm woofer . If I remember correctly, you substituteded in a 3 ohm woofer load ( or 12 ohm depending on how you wired the dual voice coils ) . Why it all worked out, is now anyones' guess . :)

:)


It's wired for 12 ohms.

A mystery to me too, but the analyzer shows what the analyzer shows. :o:

BTW, without the crossover, the W10GTI measured horrid! There is a broad valley starting ~250 Hz to over 1,000 Hz, then a massive peak at about 2K Hz that doesn't show in any of the JBL literature.

What was amazing to me was that the crossover actually seems to increase the volume in this broad valley making it relatively flat.

I've seen where some crossovers increase the volume of the frequency just below the crossover point, and that appears to be what is happening here to the benefit of the speaker.

[Earl K]

- What was amazing to me was that the crossover actually seems to increase the volume in this broad valley making it relatively flat.

- I've seen where some crossovers increase the volume of the frequency just below the crossover point, and that appears to be what is happening here to the benefit of the speaker.

- Some of these descriptions imply that the woofer load is causing the LC components to "ring" a bit / with an attendant increase in energy . Usually the "ringing" won't be over 1 octave wide & typically it's much smaller in bandwidth .

- Or / you could just be seeing the results of the horn circuit adding it's energy into the mix .

- Or some of each scenario .


:)

[toddalin]

- Some of these descriptions imply that the woofer load is causing the LC components to "ring" a bit / with an attendant increase in energy . Usually the "ringing" won't be over 1 octave wide & typically it's much smaller in bandwidth .

- Or / you could just be seeing the results of the horn circuit adding it's energy into the mix .

- Or some of each scenario .


:)

This is with the horn disconnected. :blink:

[Robh3606]

Hello Toddalin

I can't see the links at work so I am shooting from the hip. If all else is equal as far as driver impedances and crossover slopes it could simply be that they are just changing the amount of high frequency compensation. The wider the pattern typically the more HF compensation you need. The narrower horns tend to do a bit of self EQ. The easiest way to check is to look at the directivity plots of the 2 horns and you can usually tell from there what crossover mods they would need between the two horns. It's not exact but it gives you a good idea. Comparing the two horns the more the directivity changes with frequency the less compensation you need for flat on axis response. The key to this is too find the horn plots or take a crack at the schematics.

Rob:)

[Earl K]

This ( mid boost ) is with the horn disconnected.

Ahhh,, Okay .

- Then the most likely scenario is that you've created a " resonant, ringing midrange filter" which fills in the response dip ( just before F3 on the low pass section of the crossover ) .
- FWIW; back in the 70s', JBL used do this quite a lot in network design ( the 4343 is a prime example ) .
- Nowadays, this approach is generally frowned upon / though I notice that the DD66000 used something very similar in its' horn circuit.

- Anyhow, since you enjoy the sound of this induced resonant energy ( transmitting into a quite heavily coned 10" subwoof ), I wouldn't recommend that you go chasing a higher crossover point, just to satisfy the theory of "matching" the horizontal horn dispersion to the woofers axial dispersion .
- Undoubtably, raising the crossover point ( and redesigning for a more normal LC transform ) will eliminate this midrange "ringing" . And then, more than likely, your satisfaction ( that you presently enjoy for the sound of this woofer ) wil disappear .

- As RobH has pointed out, the 2370 needs some CD horn boost to flatten out it's response / so you'll need to study ( & borrow ) some 2426J circuits ( for the 2370 ) to marry onto the existing LF portion of your network .

:)

[Zilch]

Todd - what is the resolution of your RTA?

If you haven't done so yet, let me recommend you upgrade to 1/6 octave.... :yes:

[toddalin]

Todd - what is the resolution of your RTA?

If you haven't done so yet, let me recommend you upgrade to 1/6 octave.... :yes:

Obviously from the photo, 1 octave.

I agree, and day before yesterday had a Behringer 2496 Ultra Curve for $102.05 on evil-bay with 1 second to go when two scalpers moved in and one ultimately got it for $112. :banghead:

[toddalin]

Hi Todd...
busy this Saturday morning?
-grumpy

Saturday would be great! Linda's going to a yoga retreat all day and Glen wants to pop by and pick up an Altec 421A. Maybe this could happen at the same time.

Glen?

[grumpy]

The earlier, the better for me, but I could probably rotate the day and do some
work in the morning & early afternoon. Let me know. :)

[toddalin]

The earlier, the better for me, but I could probably rotate the day and do some
work in the morning & early afternoon. Let me know. :)

Let's see if Glen chimes in. Anytime after 9:00 a.m. is good for me.

[glen]

Let's see if Glen chimes in. Anytime after 9:00 a.m. is good for me.

I generally like to sleep in a bit on Saturdays :snore:
so 10:00AM or so would be more my speed,
afternoon would be O.K. too.
If it needs to be earlier not mess up people's schedules
I can make it whenever you say.

Sounds like fun! :D

[toddalin]

Who doesn't :D? If OK, I'll drop by ~9am and can stay up to a few
hours & overlap w/Glen. Measurement setup and social niceties
will take a bit of time anyway:). Looking forward to the visit. -grumpy

Actually, Linda also expressed a desire to sleep in and would prefer that we don't start until after 10:00 a.m. :biting: Her Yoga retreat starts at noon.

[glen]

I'll see you around 10:00AM then!

[grumpy]

Thanks Todd and Linda for having both Glen and me over.
Very nice place and a very entertaining system. :)

Context for FR plots to follow:

"L200" is L200 cab with 2235, 175/HL91, 075 and custom crossover.

WT10GTi is in 1.4ft3 CV cab with tweeter left out (leaving a 3/4" long,
2" dia port)... stock port blocked off, woofer coils wired in series.

(Todd, please let me know if my descriptions are off, and I'll correct)

Measuring equipment (conveniently portable):
Intel-Mac/FuzzMeasure v2.0.11 (swept sine technique)
MOTU Firewire audio I/F
Earthworks M30 (actually S30 from when smaart sold equipment)
JBL 6215 amp
Short boom stand

[grumpy]

1) 2235 in L200 cab, playing solo (1m)
2) All drivers, HF L-pad adjustment range (1m)
3) All drivers, MF L-pad adjustment range (1m)
4) Measurements at couch (~3.3m)

I believe the red plot lines were where the L-pads were
adjusted for nominal listening (plots 3 and 4)

Measurement height was approx 075 location.

[grumpy]

1) Orange - WT10GTi in CV cab, no crossover (1m)
Green - WT10GTi in CV cab, with crossover (1m)

(crossover = "keeper" implementation)

2) Blue - WT10GTi in CV cab, no crossover (5mm, between dustcap and cone)

3) CV cab with WT10GTi as bass portion of L200 system (2235 not connected)
three snapshots of minor L-pad adjustments.

[grumpy]

Commentary: Although it appeared that some crossover changes might be
helpful to flatten the system response in areas that are a bit rough, the
follow on listen and A/V demo were very entertaining. Adjustments made
after the initial plots were viewed did result in improved sound quality and
were not far off from the settings Todd had determined by ear.

Fun times folks :applaud:

-grumpy

[toddalin]

Thanks Todd and Linda for having both Glen and me over.
Very nice place and a very entertaining system. :)

Context for FR plots to follow:

"L200" is L200 cab with 2235, 175/HL91, 075 and custom crossover.

WT10GTi is in 1.4ft3 CV cab with tweeter left out (leaving a 3/4" long,
2" dia port)... stock port blocked off, woofer coils wired in series.

(Todd, please let me know if my descriptions are off, and I'll correct)

Measuring equipment (conveniently portable):
Intel-Mac/FuzzMeasure v2.0.11 (swept sine technique)
MOTU Firewire audio I/F
Earthworks M30 (actually S30 from when smaart sold equipment)
JBL 6215 amp
Short boom stand

Thanks for coming by. Was a ball! We'll all have to do it again sometime. Maybe when you have time to sit through a move and listen to some SACD.

Glen, it was great! Still don't know how I let you talk me "down" on the 421A. Come on by sometime and lunch/dinner is on me. :o:

Only thing I might add is that the L200 cabinet with the 2235 had one of the ports plugged.

Might also add that no portion of the home theater processor system was used in the testing. Grumpy brought by a dedicated JBL power amp that was connected directly from the computer sound card to the speaker under test so no "inadvertant" eq was added in the tests.

[toddalin]

1) 2235 in L200 cab, playing solo (1m)
2) All drivers, HF L-pad adjustment range (1m)
3) All drivers, MF L-pad adjustment range (1m)
4) Measurements at couch (~3.3m)

I believe the red plot lines were where the L-pads were
adjusted for nominal listening (plots 3 and 4)

Measurement height was approx 075 location.

The dip that centers at 650 Hz is my biggest peeve with the sound of the 2235s. To me, the dip reduces intellegability of the male voice and drives me nuts.
I initially though it was from the crossovers, and was even prepared to take some windings off the inductors. But, they certainly seem to work with the W10GTIs, so... maybe woofer/crossover interactions with 2235s?

[grumpy]

FWIW, I've attached two more plots.

One is a partial, which shows fair agreement with the JBL Tech Sheet plot
showing the 2235H in a 10ft3 box. In this frequency range, it would seem
to validate the measurement system.

http://audioheritage.org/vbulletin/showthread.php?t=10651

This first plot shows both a 2234H (newly reconed/red) and a
2235H (aged,old-refoam/blue) in the same cabinet (4430)... 1dB/minor
division. 1/24oct smoothing.

Second plot shows same 2235H in 4430 cab trace (blue) vs the 2235 in the 1-port
L200 cabinet (red), approximately the same measurement distance (I've shifted
levels for visualization). I'm a bit surprised at the differences above 300Hz.
Without further testing, I don't have a good explanation. -grumpy

[toddalin]

FWIW, I've attached two more plots.

One is a partial, which shows fair agreement with the JBL Tech Sheet plot
showing the 2235H in a 10ft3 box. In this frequency range, it would seem
to validate the measurement system.

http://audioheritage.org/vbulletin/showthread.php?t=10651

This first plot shows both a 2234H (newly reconed/red) and a
2235H (aged,old-refoam/blue) in the same cabinet (4430)... 1dB/minor
division. 1/24oct smoothing.

Second plot shows same 2235H in 4430 cab trace (blue) vs the 2235 in the 1-port
L200 cabinet (red), approximately the same measurement distance (I've shifted
levels for visualization). I'm a bit surprised at the differences above 300Hz.
Without further testing, I don't have a good explanation. -grumpy

FWIW, the "2235s" in my L200 cabinets were alnico 2205s recharged and reconed as 2235s by OCS in Dec 2005. The 2235 is a valid replacement cone for the 2205 basket.

[Robh3606]

FWIW, the "2235s" in my L200 cabinets were alnico 2205s recharged and reconed as 2235s by OCS in Dec 2005. The 2235 is a valid replacement cone for the 2205 basket.

If I read Grumpys post correctly that could be the answer. Maybe they didn't recharge them properly. If the magnets are partially demagged you would see a roll off up top. They used JBL kits??

Rob:)

[grumpy]

FWIW, the "2235s" in my L200 cabinets were alnico 2205s...

We can re-measure and I can bring a fresh 2235H to compare in/ same box, same env.


If I read Grumpys post correctly...

I think you did :). 2nd plot: red=2205/2235 in L200 cab, blue=2235H in 4430 cab

[Earl K]

Nice Work Grumpy !

Very Intriguing ( there's quite a mystery happening here ) ;

http://audioheritage.csdco.com/vbulletin/attachment.php?attachmentid=28574&stc=1&d=1190608245


If I read Grumpys post correctly snip,,,,,,,,snip

- I agree that some of the posts are a tad ambiguous and invite assumptions .


Grumpy, some housekeeping notes ;


(i) It seems that the above picture shows 2 different woofers ( a stock 2235H as the blue trace & a hybrid 2205a/2235 in red ) , measured in 2 different cabinets ( stock 4430 & stock L200 with one port blocked ) and / also measured in different acoustic environments ( your place & Todds' place ) .
- Are those observations all correct ?

- For the sake of clarity ( now & for future reference back to this interesting file ) / this sort of info ought to be stressed or included within the pic-files' info header ( or better yet, IMO , inserted into the pic as an overlay via some image editor ). Anyways ,,,

(ii) I'm really surprised ( shocked somewhat ) at the smoothness of the blue curve . Perhaps this is partially due to the offset location of the woofer on the 4430s' baffle board . That will reduce ( somewhat ) destructive diffraction effects .

- It would be quite illuminating to see if the jagged "Hills & Dales" ( on Todds 2205/2235 hybrid ) smooth out some with his woofs placed ( & remeasured ) in your 4430 cabinets ( & measured in your acoustic environment ). It would be nice to know the relative contribution of your acoustic environment .

- Todds' "refurbished" 2205/2235 woofs should have a full set of TS parameters run on them, to see what's up with them. ( Woofer Tester 3 is now out & is only $100. at P.E. , TODD ??? ) .

- As RobH has suggested, that OCS "remag" on Todds' hybrid, looks very suspect ( a partially demagged alnico would explain quite a lot here ) .
- Checking the TS parameters ( including the BL figure, with Woofer Tester 2 or 3 ) would be a decent QC "reality-check" of OSCs' handiwork .

http://audioheritage.csdco.com/vbulletin/attachment.php?attachmentid=28573&stc=1&d=1190608224

- The midrange peaks centered at around 680 hz ( in stock 2234s and 2235s, seen immediately above ) , are also pretty typical of the 2225h-style, ribbed cone (FWIW ) .
- It's pretty bizarre, that Todds woofs ( seen below ) show almost the complete opposite characteristic in this critical FR area .



<> :)

[grumpy]

Thanks Earl,


- Are those observations all correct ?Yes... and I could make the explanation both longer and more obvious.
Points taken on annotating plots better :) I got lazy.


I'm really surprised ( shocked somewhat ) at the smoothness of the blue curve Well, this was after a bit of measuring and moving in my living room.
It's certainly possible to make things worse.


It would be nice to know the relative contribution of your acoustic environment .Indeed. Also do have WT2, but we didn't have time to get that far into it.... next time:)
Anyone want a discount deal on a "MINTY" WT2 ? (works great, new one's faster) ;)

[grumpy]

Nothing really new, but a refresh, given the recent in-situ measurements... and
Todd was interested :)

So here I've plotted another model of Todd's latest "keeper" crossover.
I used loads closer to actual/current setup with WT10GTi,
and spec'd (not measured) inductor DCRs.

Zilch posted schematic:
http://audioheritage.org/vbulletin/showpost.php?p=93836&postcount=62

HF for this "cascaded" version (per schematic) is shown in orange.
Todd wired his HF section in parallel, and is shown in green.
L-pads are set to maximum.

I have not repeated the plots using 8/16/8ohm loads for LF/MF/HF... see referenced
thread/post if interested:
http://audioheritage.org/vbulletin/showpost.php?p=94443&postcount=94

Note. Todd's 2205/2235 is getting hi-passed even lower than plot below.
Adding that info to the 2205/2235 FR plot shown earlier in this thread gives a good
hint about why there's a hole in Todd's L200/2235 response.

[Earl K]

Folks, sorry for the blog-like turn of this thread. My intent in adding to it was to share
a history of a "get together" and to provide some measurements for Todd to use as he
sees fit. It certainly is not intended as a how-to for crossover/system design. ...

Well,, I'm not the blog police but ,,,

- I see no need for an apology. :)


AFAIK ( or remember ) ,,, a working defination of "blogging" ( as it applies to participating here at this forum ) has never been fully worked through .

AFAIC ,,, a thread becomes a blog once it has become nothing more than a "one-way" dictation of personal information or experience .

- IOW, as long as there remains in place ( on some temporal basis ) a 2-way exchange of ideas and information / no blog exists .

<> :)

ps ; Things here maybe off-topic / but that's a different concern .

[Robh3606]

I see no need for an apology. :)

Ditto

Rob:)

[toddalin]

Server was down all day yesterday.

When I had the woofers done, I had three done at the same time (all 2205s to 2235s). To my ear, and crude equipment, all three exhibit the same tendancies.

I would love to get to the bottom of this as it's been bumming me out since I replaced the 130As back in Dec 2005. :(

If it is a recharge ($35/each) that would certainly beat $190/each for recones. (Still, as I just had them done... it's :bs: ) And yes, as I recall, the cones did look to be JBL (with #s).

[4313B]

AFAIC ,,, a thread becomes a blog once it has become nothing more than a "one-way" dictation of personal information or experience .Perfect. :) And we've had a few of them.

I've personally deleted posts when I perceived that I was talking to myself.

However, we also have a few bloggers who are constantly encouraged by at least one other member to continue to do so.

- I see no need for an apology. :)Agreed!

Refreshing! :yes: Thanks Grumpy!

[grumpy]

Well, thanks. :)

Hope to make it back over to Todd's sometime soon... now that there are some specific
things identified that really -need- measuring and I know what to bring. -grumpy

[toddalin]

Well, thanks. :)

Hope to make it back over to Todd's sometime soon... now that there are some specific
things identified that really -need- measuring and I know what to bring. -grumpy

Thanks. I bought the woofers on ebay as baskets and they went straight in to OCS for recharge/recone to replace the 130As. As I noted, these woofers have been a thorn in my side (ear) since I put them in.

Zilch can attest to how frustrated I was at the time. I went to the point of building the new "keeper" crossovers, thinking that the old N1200s may have been the trouble (i.e., horn not reaching low enough to meet the new woofer). And while the crossovers are sweet (and were costly to build), evidently, that wasn't the cause and they weren't the cure. :blink:

[glen]

Thanks for coming by. Was a ball! We'll all have to do it again sometime. Maybe when you have time to sit through a move and listen to some SACD.

Glen, it was great! Still don't know how I let you talk me "down" on the 421A. Come on by sometime and lunch/dinner is on me. :o:

Todd and Linda were great hosts, opening up their fun and beautiful home to show off the great sound system, GIANT train layout and cool cars.

Grumpy had a really nice and compact setup for measuring the speaker performance and his proficiency with it was quite impressive as he could focus in on nuances of the speaker response.

You know you're having fun when your butt is shaking to the Star Wars pod race! :bouncy:

[grumpy]

Follow up: Tested 2205/2235 using WT2...

BL was down ~17-18 vs a new 2235H (and an older/used one) that
came in around 23. Qes (and hence Qts) was also much higher than spec,
presumably due to the weak magnet... someone tell me if they know this
is cause and effect is true or false.

All other parameters, whether using Delta-Mass or In-Box
measurement techniques, indicated that the cone/suspension was correct
for a 2235H. Time to talk to OCS about the recharge?

Replacing Alnico-framed 2235 with a new one did alter the response some
but after many measurements and mic placement changes, we decided to
move the speakers relative to the walls. Sidewall reflections appear to
be the main suckout culprit in the 500-900Hz octave.

Scooting the systems closer (inches) to the rear walls helped smooth the
response a bit elsewhere.

http://www.acoustimac.com/products.asp , some light cloth, and a bit
of molding :)

Good news: when the speakers are wired with one cabinet out of phase,
it's now -really- obvious :p (a brief mix-up which was sorted out quickly).

Doing a L-R comparison FR plot also brought out that there was a polarity
issue with the 075 in one channel (Todd, I looked at the impulse plots
later, and this was indeed an issue, which we corrected near end of day).

+/- 3dB from ~40-14KHz (albeit 1/3 octave) is still nothing to sneeze at
and some by-ear tweaking of levels for the listening position made the
system pretty smooth to listen to (the point, after all). L-R matching (FR
tracking is within 1/2dB almost throughout the useful bandwidth... in
no small part due to Todd's carefully measured speaker placement and
fairly symmetrical room.

Looking at a few more options w/Todd.

-grumpy

[Mr. Widget]

I know that the W-series were intended as automotive "subwoofers", but here is the measured response of the W10GTI in a 1.4 cu ft cabinet with just a 2" hole in the cabinet front to serve as a port. This is not bad for a very simple project! The keeper crossovers actually seem to flatten out the downslope trend making the curve even flatter.:blink:

That is a rather odd looking plot... could you tell us more about it? i.e. is it a nearfield measurement?, a groundplane measurement?, one taken at a meter or more? ...also, is it of just the woofer, or woofer and network?


Widget

PS: I reposted the plot so that it could be viewed without the need to download it.

[toddalin]

:blink:

That is a rather odd looking plot... could you tell us more about it? i.e. is it a nearfield measurement?, a groundplane measurement?, one taken at a meter or more? ...also, is it of just the woofer, or woofer and network?


Widget

PS: I reposted the plot so that it could be viewed without the need to download it.

This one is the very nearfield (almost touching the woofer cone where the dustcap is glued to the cone) with just the woofer in the 1.4 cu ft box with a 2" diameter hole (where tweeter was) with no crossover at all.

[grumpy]

That is a rather odd looking plot... could you tell us more about it? i.e. is it a nearfield measurement?, a groundplane measurement?, one taken at a meter or more? ...also, is it of just the woofer, or woofer and network?

"2) Blue - WT10GTi in CV cab, no crossover (5mm, between dustcap and cone)"

a very nearfield measurement. I agree it's an interesting response, but I wouldn't fuss
over it too much.

BTW, the 3110 referred to earlier may very well be ... impaired. I haven't had time to
run tests with it solo, so I'd toss out any "data" related to that. Just because we could
shift levels around to make it look flatter, doesn't make it a good test.:)

I would expect that Todd's surround speaker update idea could benefit from a
bit more thought before putting a lot of money into specific passive crossover parts,
but the general idea of using something like the WT sub when there are space
limitations doesn't bother me if it's used within it's limits.

-grumpy

[Earl K]

Follow up: Tested 2205/2235 using WT2...

BL was down ~17-18 vs a new 2235H (and an older/used one) that
came in around 23. Qes (and hence Qts) was also much higher than spec,
presumably due to the weak magnet... someone tell me if they know this
is cause and effect is true or false.

- Since BL is the "product" ; of the length of wire ( in the gap ) times the gauss strength ( as measured in the gap ) , it stands to reason that reducing either of these factors ( or both ) , will result in a lower value ( BL ) .
- Right now it's only a working assumption that the gauss strength in the gap is down in value. ( gap strength needs to be measured with the cone removed )
- A similarly smaller ( than stock ) BL product could be achieved by a reconer using a voice coil that has less turns than the stock JBL . A dead giveaway of a coil with fewer turns is an "Le" that is lower than JBL spec .
- Anyways, a higher Qts woofer will have more bass / though at a price of reduced motional control over the cone ( ie; resulting in, less defined musical "notes" ) .


All other parameters, whether using Delta-Mass or In-Box
measurement techniques, indicated that the cone/suspension was correct
for a 2235H. Time to talk to OCS about the recharge?

- If you're confident those TS parameters ( that you've recently measured ) are accurate / it's time to let OCS know that their rebuilds don't measure up ( literally ) . I'd ask them if they have a recommendation to rectify the situation .


Replacing Alnico-framed 2235 with a new one did alter the response some
but after many measurements and mic placement changes, we decided to
move the speakers relative to the walls. Sidewall reflections appear to
be the main suckout culprit in the 500-900Hz octave.

- That's interesting, I thought Todds walls were too far away to be the "obvious" culprit . But measurements rarely lie. Nice sleuthing, Grumpy .

<> :)

[grumpy]

If you're confident those TS parameters ( that you've recently measured ) are accurate...

Never confident with new toys... that's why I brought an old and new 2235H with me
to Todd's. Consistent, relative measurements I have more faith in. Both ferrite 2235's
were in the right ballpark... The Alnico version was consistenly not. Le's were all spot
on. Concur on the query to OCS.

note: WT2 users, make sure you re-measure Fs, Q for each driver before performing
Vas test (accidentally skipped this once & had to think and scratch for a minute)...:o:
Vas test (includes BL calc) uses currently loaded Fs, Q info as part of basis.


I thought Todds walls were too far away to be the "obvious" culprit
Me too. We tried that late, for grins. Lo & behold, a FR dip moved about the same
delta-F as you would expect for the two-way air-travel distance change. I'd really like
to nail this as truth or bogus (spend time looking for this specifically... e.g.,
turn speaker to directly reflect off side wall to see major interference frequencies,
etc...) Next time :) ... at least the pay is good ;) (mmm... hamburgers)

[grumpy]

OK, Todd sent the WT2 .woo files and I've had a refreshed look.

Todd's Alnico 2235 in L200 cab, network and other drivers connected.

When 20ohm resistor was added, it appears that it was in series with
more than just the 2235 (plot is uniformly 20 ohms more, across the
entire frequency range). Red is ohms. I'm going to call this test invalid
for now. -grumpy

Sorry about the resolution and inconsistent window scaling.

[toddalin]

OK, Todd sent the WT2 .woo files and I've had a refreshed look.

Todd's Alnico 2235 in L200 cab, network and other drivers connected.

When 20ohm resistor was added, it appears that it was in series with
more than just the 2235 (plot is uniformly 20 ohms more, across the
entire frequency range). Red is ohms. I'm going to call this test invalid
for now. -grumpy

Sorry about the resolution and inconsistent window scaling.

Resistor was simply hung between the crossover and the woofer.

The phasing goes weird in the problem area. The dip seems to corrospond with the hump created by the Zobel network. The resistor actually reduces the phasing effects to some extent. Maybe this is what I hear? How could this area be flattened?

[grumpy]

Thinking we should run separate LF/MF measurements and look at the
overlap in the crossover area, and run a higher resolution impedance plot
(keep in mind this is electrical phase of the network and drivers, not
audio phase).

Again, overall this isn't "bad", it's learning what might be done better
and more suited to the owner. See first attached plot.

Then "no more Mr. 1/3 octave nice guy" :D... see second plot along
with my theory... should be able to "mask" the sidewall reflection
at the higher frequencies by artificially making the omni mic less omni...
if the nulls are reduced, ... well, there you go. -grumpy

[Mr. Widget]

This one is the very nearfield (almost touching the woofer cone where the dustcap is glued to the cone) with just the woofer in the 1.4 cu ft box with a 2" diameter hole (where tweeter was) with no crossover at all.Based on the apparent notch around 47Hz, I thought that the measurement might be a nearfield measurement... nearfield measurements of woofers do tend to jack up the bottom a bit, but not this much. A lot of what we are seeing is the actual performance of this woofer in this enclosure. I'd guess that grump's measurements are reasonably accurate and useful.

Your two inch port is tuning the cabinet to ~47Hz... not ideal for a subwoofer. I haven't studied those woofers, but I am sure you would be happier with a lower tuning frequency or perhaps a sealed box. Either modification will likely extend the bass and make it a bit truer sounding.

As for the extended midrange performance, I wouldn't count on a heavy coned sub to provide "quality" reproduction above a few hundred hertz... sure it will make noise at 700Hz and above, but it will also very likely sound pretty colored and wooly.

Another interesting thing about ported boxes that your mentioning tweeter holes has reminded me. I discovered that simply having four unfilled 1/4" holes for tweeter mounting can measurably change the tuning frequency.


Widget

[toddalin]

Your two inch port is tuning the cabinet to ~47Hz... not ideal for a subwoofer. I haven't studied those woofers, but I am sure you would be happier with a lower tuning frequency or perhaps a sealed box. Either modification will likely extend the bass and make it a bit truer sounding.

As for the extended midrange performance, I wouldn't count on a heavy coned sub to provide "quality" reproduction above a few hundred hertz... sure it will make noise at 700Hz and above, but it will also very likely sound pretty colored and wooly.

Another interesting thing about ported boxes that your mentioning tweeter holes has reminded me. I discovered that simply having four unfilled 1/4" holes for tweeter mounting can measurably change the tuning frequency.


Widget

Agreed that a 2" hole in the box is far from ideal, but was an easy trial. If I unblock the 3-3/4" hole on the back (that used to house the port), the tuning goes way up, the upper bass gets much louder, and the spectral balance really goes to he--.

Unforetunately, the woofer now partially blocks this port so the port tube (3-1/2" x 7") had to be removed from the cabinet and this port closed off.

"As for the extended midrange performance, I wouldn't count on a heavy coned sub to provide "quality" reproduction above a few hundred hertz... sure it will make noise at 700Hz and above, but it will also very likely sound pretty colored and wooly."

This is the 2205/2235 in the L200 cabinet (although at 1 meter) with one port blocked and no crossover in the system. Do you think this is any less wooley?

http://audioheritage.org/vbulletin/images/attach/pdf.gifL200_2235_only.pdf (http://audioheritage.org/vbulletin/attachment.php?attachmentid=28551&d=1190592719) (39.6 KB, 21 views)

[grumpy]

Do you think this is any less wooley? :)
I think that the WT10GTi is not a typical car subwoofer...
No real point in running it all the way up to 1KHz, but in a
sealed box, it will probably work OK. "Keeper"-type crossovers
might be overkill.

[Zilch]

"Keeper" was lifted from L200B and modified for three-way:

http://manuals.harman.com/JBL/HOM/Technical%20Sheet/L200B%20ts.pdf

I don't believe we changed anything in the LF section, but Todd can verify.

JBL conjugates are rarely textbook Zobels, rather, custom and integral to the filters.... :yes:

Edit: Here's the schematic. Todd runs the UHF directly rather than cascaded, as I recall:

http://audioheritage.org/vbulletin/showpost.php?p=93836&postcount=62

The ones I built are here:

http://audioheritage.org/vbulletin/showthread.php?p=105370#post105370

Footnote: A year and a half later, that system is still set up here. They were "Keepers," indeed.... :thmbsup:

[toddalin]

"Keeper" was lifted from L200B and modified for three-way:

http://manuals.harman.com/JBL/HOM/Technical%20Sheet/L200B%20ts.pdf

I don't believe we changed anything in the LF section, but Todd can verify.

JBL conjugates are rarely textbook Zobels, rather, custom and integral to the filters.... :yes:

Zilch is correct with minor exception:

The 7.5 ohm resistor in the Zobal network was reduced to 7 ohms due to lack of availability of the 7.5 ohm piece. I think that at the time I reviewed some of the other JBL crossovers and found a 6-7 ohm resistor used in the Zobal for a 2235 or maybe 136A.

The HF boost circuit (1 uF, 0.16 mH, 5 ohm) was omitted.

A 20 ohm 12 watt resistor was placed in parallel with the mid and high.

Other changes in the circuit would have no bearing on anything but include:

Addition of Theta Audio-Cap 0.1 uF by-pass caps on all caps

A 33 uF to replace the two 16.5 uFs in the Zobal network.

A 24 uF to replace the two 12 uFs in the LF

A 24 uF to replace the two 12 uFs in the HF

Two 8.2 uF to replace the 16.5 uF in the HF

The noted addition of the N7000 components using 0.5 mH inductors.

[Zilch]

Compare to Grumpy's results on Todd's driver(s) here:

http://audioheritage.org/vbulletin/attachment.php?attachmentid=28704&stc=1&d=1191219641

[grumpy]

Cool. With all of Todd's CV cabs semi-soffit mounted and two-wall
reinforcement, I doubt the port would be needed. L40 cab = 1.8ft3?

[Zilch]

L40 cab = 1.8ft3?Seems I calculated it at 1.6 cuft some time ago.

http://audioheritage.org/vbulletin/showthread.php?p=132152#post132152

In any case, this driver is eating up a substantial chunk.... ;)

[Zilch]

1) These are lease-breakers. If I crank them much more, it seems like the cabs will self-destruct. No sub required, but some better bracing certainly is.

2) Depth-wise, the drivers barely fit in L40, which is just under the recommended vented box size, 1.6 vs. 1.75 cuft. I might have to remove a patch of fiberglass from the rear in the area of the vent, as there's not much clearance there.

3) L40 tuning is too high; its 3" port is 7.5" long, whereas W10GTI specs say it should be more like 12". See the port response, where it's clear the wrong tuning is generating a boomy bump at the bottom end. I'll fix that with a 3" elbow and a short length of duct, as required.

4) High frequencies above 2 kHz sound nasty. It looks like crossing at 800 Hz is a good choice, and 1.2 kHz may be workable. I have some concerns about IM distortion, as the cone is really pumping to make the extended bass, but I also read somewhere that IM is fiction. Can't say I know what to listen for, actually.

5) It's 3110A and 3120A shown, the latter lowpass having too high "Q". AutoEQ immediately yanked the bump out, but the HF is apparently well handled for now:

[4313B]

1) These are lease-breakers. If I crank them much more, it seems like the cabs will self-destruct. No sub required, but some better bracing certainly is. Yeah, they're too high Q for my taste, the old LE10H is about as high a Q as I want in a driver, but these W series are loads of "fun" in confined car spaces (kid's bedrooms too).

I'll fix that with a 3" elbow and a short length of duct, as required.What kind of group delay are you going to get out of that?

[grumpy]

I don't really understand why sealed wouldn't be just fine...

[Zilch]

What kind of group delay are you going to get out of that?Nasty, but still O.K., maybe. Red = vented, Blk = sealed:

[Zilch]

I don't really understand why sealed wouldn't be just fine...EBP (Fs/Qes) says it's basically a closed-box driver.

Sealed is just fine with boundary reinforcement, as you suggested above, -10 dB @ ~24 Hz without it. :yes:

Vented is better for freefield, or plain ol' driving parents crazy, -10 dB @ 20 Hz. It's L100 boom moved down an octave. :p

These are not very efficient, 84 dB @ 1W, and are not going to take rated power at those frequencies, bottom, but vented (perhaps either,) will play mighty damn loud for a 10" driver down there.... :thmbsup:

[toddalin]

These are not very efficient, 84 dB @ 1W, and are not going to take rated power at those frequencies, bottom, but vented (perhaps either,) will play damn loud for a 10" driver down there.... :thmbsup:

Granted they are inefficient. Does 84 dB @ 1 watt consist of the standard 2.83 volts (into an 8 ohm load), or are you basing it on 3.46 volts into a 12 ohm load for a real 1 watt rating? I don't imagine you're running it at 3 ohms. :biting: Can you imagine two in a cabinet at 6 ohms with a 2425? Talk about mini goliaths. :bouncy:

[Zilch]

The 2.83 V spec over 3 Ohms is 2.67 Watts, if thats how they're spec'ing it. 1 Watt would be 4.26 dB lower, so I don't know who's zoomin' what with that.

I can never make sense of car audio numbers.

Where'd I get 84 dB? I clicked the "Calc" button for 1-W SPL in BB6P driver parameters, and it says 83.93 dB... :dont-know

[toddalin]

The 2.83 V spec over 3 Ohms is 2.67 Watts, if thats how they're spec'ing it. 1 Watt would be 4.26 dB lower, so I don't know who's zoomin' what with that.

I can never make sense of car audio numbers.... :dont-know

I thought that maybe the rating was a result of your personal measurements where you knew th input voltage and this would shed some light on it.

Spec sheet notes a sensitivity (2.83 V @ 1 M) of 92 dB. It doesn't say, but I assume that is into the 2.95 ohm load parallel wiring as some of the other spects are also based on the parallel wiring. That would equate to 2.71 watts and 86 dB at 11.8 ohm load series wiring.

[grumpy]

For this particular application:

These are HT surround speakers and will not be in a freefield space.

Todd has a Sunfire sub that will loosen your fillings. I don't think
pushing near 100dB/1m (+ boundary reinforcement and x4) at 30Hz is
going to be much of a limitation, loudness-wise :) It will be interesting
to see what the midband sensitivity is (the 92dB-SPL value might be
max value vs averaged over a more typical use band).

[grumpy]

got bored during lunch today... v(5) and v(17) are "outputs"

R4P can be lowered to 64ohms ... has effect of changing
LF Q (small bump, to rolled off early)

RN1 can be adjusted some to play with the dip depth (compensates
for a peak in horn+driver response)... speaking of LCR networks :)

RLP1 and RLP2 form a 16ohm L-pad, maxed out.

Intended to work with acoustic roll off and FR of 242[5|6]J on 2470,
as well as WT10GTi with voice coils in series (~12ohms nominal),
roughly matching sensitivities.

Might want to raise the top end more (tilt, 5KHz+) if listening off axis.

Impedance roughly stays within 6-12 ohms (simulated)... RL1P is
included for simulation only.

I have no idea if this would actually work in a real-life case, but would be
entertaining to build up with existing/inexpensive parts and compare (to improve
model/assumptions).

WT10GTi and 2425 on 2470
Vin 1 0 AC
L1 1 2 2.8mH
RL1 2 3 .3
RL1P 1 3 10MEG
C1 3 0 21uF
L2 3 4 1.0mH
RL2 4 5 .2
Cz 5 6 11uF
Rz 6 0 12
Le 5 7 1.6mH
R4 7 0 12
R4P 5 0 64MEG

CH1 1 12 14.0uF
RH1 12 0 15
LH1 12 202 1.2mH
RLH1 202 0 .2
RH2 12 13 20
RH3 13 14 20
CN1 13 15 10uF
LN1 15 16 .8mH
RN1 16 0 20
RLP1 14 17 .1
RLP2 17 0 72
RHP 17 0 20
RHF 17 0 16

[toddalin]

got bored during lunch today... v(5) and v(17) are "outputs"

R4P can be lowered to 64ohms ... has effect of changing
LF Q (small bump, to rolled off early)

RLP1 and RLP2 form a 16ohm L-pad, maxed out.

Intended to work with acoustic roll off and FR of 242[5|6]J on 2470,
as well as WT10GTi with voice coils in series (~12ohms nominal),
roughly matching sensitivities.

Might want to raise the top end more (tilt, 5KHz+) if listening off axis.

Impedance roughly stays within 6-12 ohms (simulated)... RL1P is
included for simulation only.

I have no idea if this would actually work in a real-life case, but would be
entertaining to build up with existing/inexpensive parts and compare (to improve
model/assumptions).

WT10GTi and 2425 on 2470
Vin 1 0 AC
L1 1 2 2.8mH
RL1 2 3 .3
RL1P 1 3 10MEG
C1 3 0 21uF
L2 3 4 1.0mH
RL2 4 5 .2
Cz 5 6 11uF
Rz 6 0 12
Le 5 7 1.6mH
R4 7 0 12
R4P 5 0 64MEG

CH1 1 12 14.0uF
RH1 12 0 15
LH1 12 202 1.2mH
RLH1 202 0 .2
RH2 12 13 20
RH3 13 14 20
CN1 13 15 10uF
LN1 15 16 .8mH
RN1 16 0 20
RLP1 14 17 .1
RLP2 17 0 72
RHP 17 0 20
RHF 17 0 16

Guess I need to see the schematic to understand the part placement, circuit nuances, and their implications. :o:

[grumpy]

2nd and 3rd columns are network "node" numbers.
Parts connect the nodes. Node 1 is the voltage input, Node 0 is ground.

(Inductor DCR's are included in the sim)

[toddalin]

2nd and 3rd columns are network "node" numbers.
Parts connect the nodes. Node 1 is the voltage input, Node 0 is ground.

(Inductor DCR's are included in the sim)

OK, that makes sense now and I can visualize it looking at a typical network. Guess I need to draw it out to fully understand it. (I've never expressed a schematic that way nor seen it done that I can recall.)

You're proposing 10 and 60 MEG resistors in a crossover network?

Also, how many watts would the various value resistors need to be rated for a 200 WRMS crossover?

BTW, All Electronics has a good price on nonpolarized caps that could be used for experimental purposes.

http://www.allelectronics.com/cgi-bin/category/140900/Non-Polar.html

[grumpy]

old SPICE netlist format... free versions abound. Requires a fair bit of
background knowledge to be useful and understand limitations. I haven't
fooled with it much for many, many, years but it's familiar ... so, for me,
it's fast.

I'm sure most crossover design tools that folks discuss here are much more
suited to the task, in terms of practical topologies and the ability to
import/export data to/from measurement programs.


You're proposing 10 and 60 MEG resistors in a crossover network?I don't think so :) :

"R4P can be lowered to 64ohms ... has effect of changing
LF Q (small bump, to rolled off early)"

"RL1P is included for simulation only."

I left R4P in as a placeholder... putting RL1P in allowed SPICE
to converge on a simpler representation of the circuit.



Also, how many watts would the various value resistors need to be rated for a 200 WRMS crossover?

If I have time, I'll do this more rigorously.... guesses for now:

As the 12 ohm resistor in the LPF leg doesn't see too much under 1KHz, I'd think 20-40W, depending on
use would be OK.

In the HPF leg, 10W for the 20ohm resistors would probably be fine. The 15 ohm unit on the input should be
larger ... 40W...

Most home use environments won't run anywhere near full-steam for very long, so specifying 100% sine-wave
capability would be a waste. Also a 200W/8ohm amp will usually run out of gas at ~130W/12 ohms, ~100W, 16ohms
(power supply voltage limited).

[grumpy]

Using a tunable sine-wave input: for full "200W" input, you'd be pushing a bit
over 100W into the 15 ohm resistor above 1KHz... while the driver would be seeing
about 3W (roughly 113dB @ 1m). I'll presume that most listening isn't done at 3dB
below clipping, (now down to ~50W) and that most of the energy is below 1KHz
(so make it smaller). In hindsight, I probably would weigh individual resistor
power dissipation more heavily in the design considerations.

The 20ohm resistors on either side of the LCR notch are dissipating about
30W at full sine-wave power (over fairly small bandwidths), resistor in LCR is less.

In the low-pass leg, making the 12 ohm resistor 30W, would be overkill... even
at "full power" (peak dissipation frequencies in this leg are ~550 and 890Hz,
less than that everywhere else)... but then resistors that would work for this duty
are cheap (three 33ohm 10 watters in parallel).

So, for now, I'll stand by my guesses for real-world use :)

One could certainly design this differently to keep the HF impedance higher and
not waste so much heat. Keep in mind this was just an exercise and I am not
even close to being an experienced crossover designer. -grumpy

[Zilch]

Also a 200W/8ohm amp will usually run out of gas at ~130W/12 ohms, ~100W, 16ohms (power supply voltage limited).I was asking myself that very question just yesterday. At 12 Ohms, any amp is gonna start clipping at a lower SPL, no, 'cause it's outta voltage headroom?

I'll measure the actual impedance with the voice coils in parallel. While there are amps spec'd below 4 Ohms, these are likely to smoke most conventional amps at 3 Ohms if cranked, wouldn't we think?

Running passive two-way, which sounds great, BTW, the HF is attenuated ~20 dB to balance with the woof voice coils series connected. :biting:

[grumpy]

At 12 Ohms, any amp is gonna start clipping at a lower SPL, no, 'cause it's outta voltage headroom? unless you have an output
transformer, or rail switcher, or ... but generally, yes.


...wouldn't we think? mostly, yeah... I'd think you'd want
to think twice about hooking a potentially sub-3ohm load to (-two- channels here)
a typical HT receiver, for example.



BTW, the HF is attenuated ~20 dB to balance with the woof voice coils series connected. :yes: Yep. If you look at the FR plot, in the legend you'll
note I added 17dB to the HF portion for visualization... I figure you can dial back the rest
with the L-Pad.

-grumpy

[toddalin]

Running passive two-way, which sounds great, BTW, the HF is attenuated ~20 dB to balance with the woof voice coils series connected. :biting:

So what crossover circuit are you using and do you have a FR plot (yet)?

[Zilch]

BMS 4552 with P-Audio PH-35 throat adapter on $10 90° x 50° PT waveguide.

3120A Mid = Min, HF boost = Med. Outboard L-Pad on HF driver.

HF is black curve in second image, post #50, +/- 1.5 dB, approx.

I haven't "fixed" the 3120A LP filter to match the W10 driver. AutoEQ's still doing that, FR-wise.

I'll take the impedance measurements, then I'm done with it 'til I want to build some L100 killers.... :p

[Zilch]

WT2 impedance curves.

Series connected voice coils, in vented L40 cab, Zmin = 14.46 Ohms @ 137 Hz. Z @ 1 Hz = 13.18 Ohms.

Parallel connected, free air, Zmin = 3.66 Ohms @136 Hz. Z @ 1Hz = 3.34 Ohms.