JBL 4365 Active Crossover settings

[bubbleboy76]

I have done a very basic setup with bi-amp and active crossover on my 4365s. In my ears the sound-improvement is big in my system compared to stock crossover and single-amp. The sound is subjectively more clear and dynamic, and much more bass the way I like it.
It is very fun to test different settings, and hear and measure what happens. This is to me the best way to learn.

Equipment:
JBL 4365 with temporarily bypassed internal crossover.
DBX Driverack 260, internal output gain jumper lowered to its lowest setting.
Oppo BDP-105, xlr output 4.2 vrms at full volume. I use this as preamp with the digital volume control in the ESS Sabre dac.
DBX RTA-M microphone.
Focusrite Scarlett 2i2 sound-interface.
REW analysis program, with DBX RTA-M generic calibration file
Hypex nCore nc400 6-channels (DIY). 3 channels with about 12Db lower input sensitivity/gain by removing resistor r141. 26 Db, respectively about 13 Db, gain. I use the high-gain channels to drive the LF, and the low-gain to drive the HF.

I know there are a LOT to work on and improve, but my initial settings sounds surprisingly good!

Measurements done only in one position, the listening position. I plan to sum more positions later on.

2-way at the moment. Hence, the UHF is not in use. I keep it simple for now.

Work on delay in the crossover-region is my next planned step, and off course more work on the crossover region overall. I plan to do it with close microphone, and as far away from the walls as possible. The speakers are huge, so I can not do it outdoors.

48Db slope in the crossover sounds slightly better than 24Db at the moment. I think it is best to keep the crossover narrow until it is finetuned/more correct.

The system is dead silent at idle, even when putting the ear to the horn. I am quite impressed by this. No noise at all.

Comments and tips are welcome, I am a beginner at this. But try to keep it positive, I do this for the fun of it!

Settings, measurements and voltage drive of the original network will follow.

Other owners of 4365, please post your settings if you go this route! I know goophy did active crossover as well.

[bubbleboy76]

My current DBX 260 settings, that roughly gives a JBL/Harman house-curve in my listening-postion.

INPUT:
-6Db

(I want to use as much voltage out from my consumer-level source as possible)

PRE-PEQ:
FC: 78.8Hz
Q: 12.364
Level: -6Db

(My room standing-wave roof/floor reduced)

CROSSOVER HF:
High-pass: 793,7Hz
High-slope: LR 48Db/oct
Low-pass: 20.159Khz
Low-slope: BW 24Db/oct
Gain: +3.2Db

CROSSOVER LF:
High-pass: 20.3Hz
High-slope: BW 24Db/oct
Low-pass: 793,7Hz
Low-slope: LR 48Db/oct

(
790Hz inspired by the M2 filter. Initially, I used 750Hz like the original network. 790Hz feels safer to not stress the 476mg.
LF high-pass to protect the 1501fe, I do not know if it is needed? It sounds better without this filter, more "body" in the bass.
HF low-pass to not make the 476mg go out-of-bounds. I do not know if it is needed? I hear no difference, so it do no harm in that way.)
)

POST-PEQ:
FC: 1.46KHz
Q: 14.065
Level: -6Db

High-shelf FC: 4.76KHz
High-shelf slope: 12Db/oct
High-shelf level: +6Db

Low-shelf FC: 136Hz
Low-shelf slope: 6Db/oct
Low-shelf level: +9Db

(Low-shelf inspired by the M2 filter and the JBL/Harman target curve)

[bubbleboy76]

The JBL/Harman target curve I am referring to attached.

[bubbleboy76]

Voltage drive and frequency response for the JBL 4365. Source: Italian magazine Audio Review.

[bubbleboy76]

Ok, a dummy question!
Why are crossovers done in this overlapping and summing way? Why not cut straight off? Digital dsp must be able to do it?

[bubbleboy76]

Can someone give me advice what protective caps to use on the horns?
The 476mg is 8ohm, and I use 800Hz crossover frequency.

[fpitas]

Ok, a dummy question!
Why are crossovers done in this overlapping and summing way? Why not cut straight off? Digital dsp must be able to do it?

It's not possible to cut straight off, but the slope can be quite steep with DSP. Steep slopes have mathematical consequences, though. For a "conventional" filter topology, a steep slope will introduce a large peak in the group delay and ringing in the time response. DSP can achieve another result using an FIR filter, but the price is pre-ringing.

[ivica]

Ok, a dummy question!
Why are crossovers done in this overlapping and summing way? Why not cut straight off? Digital dsp must be able to do it?

Hi bubbleboy76,

Today DSP driven networks can provide sharp (usually called "brick wall") filters using different type of filters while the price can be "acceptable" (see 'miniDSP' OpenDRC products), but using such type of filter, as drivers are separated each to other (due to its size, or horn size) the listener can get a feeling (in the smaller rooms especially) that a part of the listened music from the instruments is coming from one driver and the other part from the other driver as the reproduced frequency spectrum is divided sharply between the drivers, so a kind of overlapping would give much more smoother passage from one driver to the other.
Drivers directivity is different too.
On the other side we have to be aware, that whenever the same part of spectrum is reproduced by two (ore more) sources, that are some distance apart, a kind of "comb filter effect" would be present when off-axis listening is done (fortunately, the reflection waves have 'soften' that effect).

I believe that 68uF to 100uF capacitor would be good protection of the compression driver with less influence on the sound over 700-800Hz.

Regards
Ivica

[fpitas]

ivica raises a good point about spectral overlap. Some overlap is usually necessary to get good driver integration. I suppose if you sat hundreds of feet from the speakers overlap would not be necessary, but in a typical listening situation, it is.

[bubbleboy76]

I believe that 68uF to 100uF capacitor would be good protection of the compression driver with less influence on the sound over 700-800Hz.

Regards
Ivica

Thanx.

I found some old JBL-guidlines stating 52uF is recommended for 8ohm and 750Hz crossover. Minimum 250v.

Any recommendations of brand/type?

[bubbleboy76]

My next planned thing to play with is time-alignement (with delay of the LF) by measuring max cancellation at the crossover point with the drivers out of phase.

Should I do the measurement in the listening position(s) or close to speaker placed in the middle of the room? Any thought about this would be welcome.

[1audiohack]

Confound you Bubbleboy!!!

I am trying to ignore this thread since I am perfectly happy with my 4365's, however I do have a couple of 260's that are not in use, and there's that pile of amps just waiting,,,

Grrrrr!

[bubbleboy76]

Confound you Bubbleboy!!!

I am trying to ignore this thread since I am perfectly happy with my 4365's, however I do have a couple of 260's that are not in use, and there's that pile of amps just waiting,,,

Grrrrr!

It is your destiny. Join me, and together we can rule the galaxy as father and son!

:)

Edit clarification: You will be the father in that case! I have read a lot of your posts on this forum, and I have great respect for you.

[bubbleboy76]

Take this with a grain of salt, because I am a bit excited just now...

I tried some different delay-settings without measuring, just listening.

OH MY GOOOD! The sound went to the next level.

Done some other changes as well. Just listening, not measuring. I am on the verge to too much bass now, but I am a bass junkie.

My current DBX 260 settings:

INPUT:
-4Db left
-5Db right

PRE-PEQ:
FC: 78.8Hz
Q: 12.364
Level: -6Db

CROSSOVER HF:
High-pass: 771.1Hz
High-slope: LR 24Db/oct
Low-pass: 20.159Khz
Low-slope: BW 24Db/oct
Gain: +3.2Db

CROSSOVER LF:
High-pass: 19.7Hz
High-slope: BW 12Db/oct
Low-pass: 771.1Hz
Low-slope: LR 24Db/oct
Delay: 0.10m

POST-PEQ:
FC: 1.46KHz
Q: 14.065
Level: -6Db

High-shelf FC: 4.76KHz
High-shelf slope: 12Db/oct
High-shelf level: +6Db

Low-shelf FC: 136Hz
Low-shelf slope: 4.5Db/oct
Low-shelf level: +9.5

[bubbleboy76]

One reflection, my measurement system and my driverack is not in sync. For example, if I put a notch-filter at 80Hz in the driverack, then it takes effect at 70Hz in the measurements.
This makes it a bit tedious. I can not just measure and then know which frequency to attack in the dsp. I have to trial and error, until the measurement-curve looks like I want.
I am trusting the driverack more than the measurement-system, so probably something is wrong in the setup.

[bubbleboy76]

Regarding dsp, is there a performance difference in doing + or - when it comes to gain?
For example, I have +3.2Db gain on the HF. Is it better to do -3.2Db on the LF instead?
Same with my low-shelf +9.5dDb in the bass. Is it better to make it an high-shelf with -9.5Db?

[Mctwins]

Regarding dsp, is there a performance difference in doing + or - when it comes to gain?
For example, I have +3.2Db gain on the HF. Is it better to do -3.2Db on the LF instead?
Same with my low-shelf +9.5dDb in the bass. Is it better to make it an high-shelf with -9.5Db?

Hallo!

You have to do a proper Gain Structure, don't guess...:D

[bubbleboy76]

Hallo!You have to do a proper Gain Structure, don't guess...:D
You clearly did not understand my question. Read it again, slowly.
My gain structure is optimized (to my knowledge) as much as it can be in my system and my conditions.

[1audiohack]

It is your destiny. Join me, and together we can rule the galaxy as father and son!

:)

That's hilarious!


Edit clarification: You will be the father in that case! I have read a lot of your posts on this forum, and I have great respect for you.

I don't know about that but it's very kind of you to say. As hard as I try, I am not a Jedi yet. :)

All the best,
Barry.

[ivica]

One reflection, my measurement system and my driverack is not in sync. For example, if I put a notch-filter at 80Hz in the driverack, then it takes effect at 70Hz in the measurements.
This makes it a bit tedious. I can not just measure and then know which frequency to attack in the dsp. I have to trial and error, until the measurement-curve looks like I want.
I am trusting the driverack more than the measurement-system, so probably something is wrong in the setup.

Hi,

Is it correct sampling rate 44.1kHz or 48kHz.
I think that it would be better to reduce gain vs to enhance the gain.

regards
ivica

[bubbleboy76]

Hi,

Is it correct sampling rate 44.1kHz or 48kHz.
I think that it would be better to reduce gain vs to enhance the gain.

regards
ivica

do not know about the sampling rate, will check that out, thanx!
I do not trust microsoft windows and sound at all...
I know in certain cases/devices increasing digital gain is bad, can introduce digital clipping.

[bubbleboy76]

This will make some guys happy :)
Yesterday, I reconnected the passive crossover to the horn to see what would happen to the sound. The sound memory is short!
The sound lost a bit of dynamics and clarity, but it gained coherence and sounded more "correct". So the stock passive filter won over my rudimentary dsp-setup (for the HF).
I have learned that getting a crossover right for the HF, and a CD-horn, is over my head (at the moment), but you do not know until you try :)

The LF on the other hand...

It is a so simple crossover function / voltage drive in the passive filter that I think "what could possibly go wrong" :)
The bass was so profoundly much better in my system when I went to active crossover, so I will continue with that (for now), and try to get it to match the passive HF-crossover the best I can. Will play with that today.

[bubbleboy76]

The added delay to the LF-driver made a significant improvement in my system. The sound was "released" from the speakers. Before, I could hear the sound coming from the speakers. After, the sound comes from the room, if you get what I mean.
I am now a strong believer of the importance of time-alignement of speaker drivers.

[4313B]

I have learned that getting a crossover right for the HF, and a CD-horn, is over my head (at the moment), but you do not know until you try :)
Yes. I think I have mentioned several times that it isn't trivial. ;)


The added delay to the LF-driver made a significant improvement in my system. The sound was "released" from the speakers. Before, I could hear the sound coming from the speakers. After, the sound comes from the room, if you get what I mean.
I am now a strong believer of the importance of time-alignement of speaker drivers.G.T. mentioned the same thing on several occasions. ;)

[bubbleboy76]

Tuned it in now, and it resembles very much the stock sound that I did not fully like. It is most likeley due to my bad room accoustics causing this. It is my livingroom and unfortunately I can not do room treatment.
Will try to post my measurements later on.
I found out today that the 4365 uses 12Db/oct LR crossover, not a fancy 24DB/oct like k2S9900.
I think I will have another go at fully active.

[4313B]

I found out today that the 4365 uses 12Db/oct LR crossover, not a fancy 24DB/oct like k2S9900.I don't remember that being the case. I posted the voltage drives to both networks years ago, I don't remember where though. For example, the low pass filters on the 1500AL-1 and 1500Fe are nearly identical. 6.2 mH series, 27 uF parallel versus 39 uF parallel, and 1.0 mH series.

[bubbleboy76]

I don't remember that being the case. I posted the voltage drives to both networks years ago, I don't remember where though. The low pass filters on the 1500AL-1 and 1500Fe are nearly identical. 6.2 mH series, 27 uF parallel versus 39 uF parallel, and 1.0 mH series.

I might be wrong, but it looked like that when I was measuring.

[bubbleboy76]

Measurement with the stock crossover (LF, HF and UHF) in my listening position.
Looks very good, but in some way the sound is to bright for my taste.

[bubbleboy76]

Measurements of crossover with active on LF and passive on HF/UHF.
Mic 1m from speaker.

12Db/oct Linkwitz-Riley low-pass at 600Hz of the LF in the DBX 260.
-8Db high-shelf at 136Hz, 6db/oct.
Notch 6Db reduction at 80Hz
LF delay 0.10m

The summing graph looks good to me.

[bubbleboy76]

Measurement in listening position right now, with active LF and passive stock HF and UHF.

[bubbleboy76]

I don't remember that being the case. I posted the voltage drives to both networks years ago, I don't remember where though. For example, the low pass filters on the 1500AL-1 and 1500Fe are nearly identical. 6.2 mH series, 27 uF parallel versus 39 uF parallel, and 1.0 mH series.

Can you find/get the voltage drive? I would be very grateful. I only have the italian magazine one.

[4313B]

Can you find/get the voltage drive? I would be very grateful. I only have the italian magazine one.At some point I need to load LEAP onto a PC and then I can get them for you. I have a laptop in mind I just haven't gotten around to loading any software onto it. It needs an OS, AutoCAD, LEAP, HiQnet, and DEQX loaded onto it.

[bubbleboy76]

At some point I need to load LEAP onto a PC and then I can get them for you. I have a laptop in mind I just haven't gotten around to loading any software onto it. It needs an OS, AutoCAD, LEAP, HiQnet, and DEQX loaded onto it.

Please help a brother in need! ;)

[bubbleboy76]

After some more listening tonight, I think I prefer how my fully active setup sounded.
I will probably make another go at trying to improve those dsp settings.

[bubbleboy76]

At a first glance, looking at the voltage drive for the K2S9900 posted at this site, the curve looks pretty "simple" to translate into dsp-settings. But then I realise it is dbV on the y-axle and not Db. Are those equivalents? Or do I have to convert dbV to Db in some way to know what to enter into the dsp?

Can I use S9900 voltage drive for my speaker you think? The compression driver is the same, but maybe not the horn?

[bubbleboy76]

K2 S9900 voltage drive (source: user 4313B) next to 4365 voltage drive (source italian magazine Audio review)
I note that S9900 has this shelfing in the bass that I also added to my own settings.

[pos]

I think your best option is to push 1audiohack to do the measurements and crossover settings on his, and share them on the forum :D

What measurement tools do you have at hand?

[pos]

OK, I just reread your first post.

If you could take precise measurements and share the impulses then we could probably be able to try some crossover settings.

[bubbleboy76]

If you could take precise measurements and share the impulses then we could probably be able to try some crossover settings.

Will do. But I am still waiting for the protection caps. Jantzen Crosscap 68uF.

[bubbleboy76]

I think your best option is to push 1audiohack to do the measurements and crossover settings on his, and share them on the forum :D


:)

[bubbleboy76]

I tried to change to Linkwitz-Riley 24Db/oct and, whoooah, the sound I want!
I must measure to see what happened, I guess the cancellations/suckouts after the crossover point dissapeared.

Active LF, Passive HF/UHF settings:
24Db/oct Linkwitz-Riley slope at 600Hz.
+8Db low-shelf at 136Hz, 6db/oct slope.
LF delay 0.10m.

[bubbleboy76]

It sounds insaneley good now :)
If I only could replicate this sound fully active, with that extra percentage clarity and dynamics, then I would be in heaven.
Who will convert the 4365 HF voltage drive into dsp-settings for me?! I promise to send some Swedish delicacies to you! (we have some scary stuff as well, fermented herrings, snus, cray fish etc)

[spkrman57]

From a city-billy audiophile..

I found the 1501Fe to sound best for me in a 2 - way with 2445/Be/500hz Edgarhorn when crossed over @ 375hz 24db LR and the Horn @ 750Hz 18db oct BW and passive EQ (Pi spkr X-over type).

I'm looking at 2206/E-145/2226 passive 2-way run with tubed amp bi-amped between 100hz to 150hz with SS amp on 1501Fe for LF duty.

All the cabs can be small and still provide all the LF reproduction I need,

Just my 2 cents...

Regards, Ron

[bubbleboy76]

From a city-billy audiophile..

I found the 1501Fe to sound best for me in a 2 - way with 2445/Be/500hz Edgarhorn when crossed over @ 375hz 24db LR and the Horn @ 750Hz 18db oct BW and passive EQ (Pi spkr X-over type).

I'm looking at 2206/E-145/2226 passive 2-way run with tubed amp bi-amped between 100hz to 150hz with SS amp on 1501Fe for LF duty.

All the cabs can be small and still provide all the LF reproduction I need,

Just my 2 cents...

Regards, Ron

Interesting, thanks! The M2 uses that kind of different slopes as well, if I remember correctly.

[4313B]

K2 S9900 voltage drive (source: user 4313B) next to 4365 voltage drive (source italian magazine Audio review)
I note that S9900 has this shelfing in the bass that I also added to my own settings.That is the voltage drive G.T. recommended for bi-amp operation. A person plugs that voltage drive into their active/DSP filter. For example, a Crown can emulate that. Those two voltage drives are an invalid comparison. I wish I could find the thread all that information was posted in.

[bubbleboy76]

There is an post in the "system information" part of this forum for 4365, but no file is attached. Must been forgotten by that Giskard-guy :)

http://www.audioheritage.org/vbulletin/showthread.php?29286-4365

[pos]

Who will convert the 4365 HF voltage drive into dsp-settings for me?!

If you manage to get your hands on high resolution voltage drive curves (or better still a txt files with frequency and amplitude columns) I can try to emulate them with rephase using settings that would be doable on your DSP crossover.
That being said, the passive network also tries to address the geometrical misalignment of the drivers withint the slopes, whereas it would typically be best addressed using delays and symmetrical slopes with a DSP crossover. This is exactly what GT did with the S3900 and S4700 "DSPification". But for that you need a good set of measurements.

[bubbleboy76]

If you manage to get your hands on high resolution voltage drive curves (or better still a txt files with frequency and amplitude columns) I can try to emulate them with rephase using settings that would be doable on your DSP crossover.
That being said, the passive network also tries to address the geometrical misalignment of the drivers withint the slopes, whereas it would typically be best addressed using delays and symmetrical slopes with a DSP crossover. This is exactly what GT did with the S3900 and S4700 "DSPification". But for that you need a good set of measurements.

Maybe plan a vacation to Sweden this summer POS?! I have a boat and live by the sea, I recommend it! :)

[4313B]

There is an post in the "system information" part of this forum for 4365, but no file is attached. Must been forgotten by that Giskard-guy :)No, it was a thread wherein I posted the voltage drives of the K2-S9900, 4365, DD65000, DD66000, and DD67000 showing the similarities and differences. Anyway, I'll reload LEAP at some point and look at the files again.

[bubbleboy76]

But for that you need a good set of measurements.

Can you define this a bit more, please.
Should I measure each driver very close, like 5cm, or should I measure in the middle between them 1m away? Or some other way?

[pos]

Maybe plan a vacation to Sweden this summer POS?! I have a boat and live by the sea, I recommend it! :)
Hej, why not, and you could help me remove that pesky R141 on my nc400 modules :D
Seriously though, traveling is not easy with a 2 years old at home...


Can you define this a bit more, please.
Should I measure each driver very close, like 5cm, or should I measure in the middle between them 1m away? Or some other way?
You can follow GT's method as exposed in the S3900/S4700 DSP thread.

You should either not use any filter (outside of the protection caps of course, that will stay inline and can be considered part of the system) and filter/restrict your measurement signal. If you include a filter you should note it so that is can be taken into account (compensated) during the correction.
You will need a low noise floor in order to get an accurate view of the stop band.

You should take several measurements for each driver, at different distances and angles. No need to match levels or try to calculate delays between drivers at this point.
Then just send me these impulse with comment for each measurement (filters, distance, etc.).

[bubbleboy76]

I know what you mean, I have two little sons, a 3-year and a 10-month.
Please send me your email in a PM.
Can I send REW measurement files?

[Mctwins]

Hallo!

Here is something for you to read. I am using ARTA program and to me it is a better tool than REW.

http://www.artalabs.hr/papers/Mateljan-ELA-02.pdf

http://www.artalabs.hr/papers/Mateljan-InRoomResponse.pdf

http://www.artalabs.hr/AppNotes/AP4_FreeField-Rev03eng.pdf

on the other hand, I really don't understand what you are trying to achive with your speakers. :confused:

To me, the 4365 performs in a outstanding way and I trust what the designer has come up with in terms of aligning the x-over settings, voltage vs freq, constant directivity, power response and so on.....my room response shows this very well.

Good Luck and Happy Reading :)

[bubbleboy76]

on the other hand, I really don't understand what you are trying to achive with your speakers. :confused:

To me, the 4365 performs in a outstanding way and I trust what the designer has come up with in terms of aligning the x-over settings, voltage vs freq, constant directivity, power response and so on.....my room response shows this very well.


You have said this now a 1000 times. What I try to achieve is very clearly stated in this thread (if you take the time to read it), so the question is really what you are trying to achieve?
I know what you think about my little experiment, do not worry about that!

[bubbleboy76]

Hallo!

Here is something for you to read. I am using ARTA program and to me it is a better tool than REW.

http://www.artalabs.hr/papers/Mateljan-ELA-02.pdf

http://www.artalabs.hr/papers/Mateljan-InRoomResponse.pdf

http://www.artalabs.hr/AppNotes/AP4_FreeField-Rev03eng.pdf



Thanks for the links.
What is ARTA doing better? Worth the extra 79 euros?

[ivica]

It sounds insaneley good now :)
If I only could replicate this sound fully active, with that extra percentage clarity and dynamics, then I would be in heaven.
Who will convert the 4365 HF voltage drive into dsp-settings for me?! I promise to send some Swedish delicacies to you! (we have some scary stuff as well, fermented herrings, snus, cray fish etc)

Hi bubbleboy76,

It is very nice to have such interesting toys as JBL 4365 supported by DSP and several amplifier around. As You have said using active network and separated amp for driving bass driver would produce more realistic bass response as there would be no driver to network mutual interaction, especially in the about 100Hz ~ 200Hz region. I have done some calculation and it can be seen the driver to network interference. As 4313B has told such behavior has been discussed by G.T. while talking about 4345 speakers and his suggestion to drive its bass (2245) active.
From the figure it can be seen, too, why, when active driven, a kind of shelving from about 130Hz has to be done in order to emulate passive network driver driven.

The problem how to active drive Horn-driver combo is much more complex if You want to eliminate passive network. It would not be enough to emulate voltage network response only, as there are driver&horn and passive network mutual interaction. As You can see from the Italian 4365 presentation, off-axis response of the driver&horn combo has shown a kind of rise in the response from 3kHz to 9kHz relative ot on-axis measurements, so I would suggest You to use off-axis response about 15~20 degs off-axis, as a response You would equalize, I believe that after that less "aggressive" sound would be get.

While equalize the response, I would suggest You ONLY to use filters that would REDUCE part of spectrum, that means NEVER ENHANCE a part do spectrum, and try to use AS LESS as possible number of EQ filters.

From the mentioned 4365 presentation it can be seen that VHF (476Mg) driver and 045 are not time aligned. It seems to me that about 0.75ms ( about 25cm) delay has to be included int to 045 driver drive (if separated driven amp for UHF would be used). Bass to VHF driver (476) time alignment You have done jet.

regards
Ivica

[bubbleboy76]

Hi bubbleboy76,

It is very nice to have such interesting toys as JBL 4365 supported by DSP and several amplifier around. As You have said using active network and separated amp for driving bass driver would produce more realistic bass response as there would be no driver to network mutual interaction, especially in the about 100Hz ~ 200Hz region. I have done some calculation and it can be seen the driver to network interference. As 4313B has told such behavior has been discussed by G.T. while talking about 4345 speakers and his suggestion to drive its bass (2245) active.
From the figure it can be seen, too, why, when active driven, a kind of shelving from about 130Hz has to be done in order to emulate passive network driver driven.

The problem how to active drive Horn-driver combo is much more complex if You want to eliminate passive network. It would not be enough to emulate voltage network response only, as there are driver&horn and passive network mutual interaction. As You can see from the Italian 4365 presentation, off-axis response of the driver&horn combo has shown a kind of rise in the response from 3kHz to 9kHz relative ot on-axis measurements, so I would suggest You to use off-axis response about 15~20 degs off-axis, as a response You would equalize, I believe that after that less "aggressive" sound would be get.

While equalize the response, I would suggest You ONLY to use filters that would REDUCE part of spectrum, that means NEVER ENHANCE a part do spectrum, and try to use AS LESS as possible number of EQ filters.

From the mentioned 4365 presentation it can be seen that VHF (476Mg) driver and 045 are not time aligned. It seems to me that about 0.75ms ( about 25cm) delay has to be included int to 045 driver drive (if separated driven amp for UHF would be used). Bass to VHF driver (476) time alignment You have done jet.

regards
Ivica

Thank you very much for this helpful information!

My time-alignement delay of the woofer is not done the properly way yet, I only used my ears so far. Can you roughly calculate what is should be?

[ivica]

Thank you very much for this helpful information!

My time-alignment delay of the woofer is not done the properly way yet, I only used my ears so far. Can you roughly calculate what is should be?

Hi bubbleboy76,

It would be difficult to say what is time delay of the bass driver, in the box as it varies from about 15ms at 32Hz, to about 0.4ms ate 470Hz (with passive network applied), but using active network, delay would be preset too. Mentioned VHF driver&horn has about 25cm mechanical offset (I guess) , but applied VHF network ( passive or active) would introduce its time delay around crossover too, so time delay compensation can be get , may be experimentally, applying "square wave" signal, and then looking at the (in time domain) reproduced waveform. Changing time delay try to get the best reproduced signal.
A software like TRUERTA ( https://www.trueaudio.com/rta_abt1.htm ) can be very usefull as it can produce square wave signal, and can show mic received signal.

http://www.musicanddesign.com/Speaker_Transient_Analysis.html
http://sound.westhost.com/ptd.htm
http://www.bksv.com/doc/17-198.pdf

http://www.kirchner-elektronik.de/~kirchner/AutoDSPeng.pdf

Have a Good Luck,

regards
ivica

[bubbleboy76]

thanks again ivica.
Yes, the physical distance is about 25cm.
I received the protection caps today. I am shocked by the size of them, I thought it should be a little small component :)

[ivica]

thanks again ivica.
Yes, the physical distance is about 25cm.
I received the protection caps today. I am shocked by the size of them, I thought it should be a little small component :)

Hi
As known: "there is no better enhancement .... then displacement (volume)":p

regards
ivica

[hjames]

Hi
As known: "there is no better enhancement .... then displacement (volume)":p

regards
ivica
Badly quoted ... among my brother and his car guy friends, the quote has always been:

"There is no substitute for cubic inches ..." (displacement)

[bubbleboy76]

"There is no replacement for displacement" is the one I have heard :)

[bubbleboy76]

Some more measurements today, with my current settings.

Stock passive filter on the horn.

Active dsp filter on the woofer:
24Db/oct Linkwitz-riley at 595.6Hz
-6Db notch 80Hz (the largest room mode reduced)
+8Db shelf at 136Hz with 6Db/oct slope
Delay 10cm

I measured 5cm and 1m from speaker elements, on axis and 60 degree off axis. Only one element at the time (the graphs are overlayed).

Sleeping baby at home, so I had to measure on lower SPL. I am on parental leave :)

[bubbleboy76]

I updated to the latest version of REW, and the latest usb soundcard drivers for my Oppo BDP-105.
Now, the measurements are in sync with my dbx. If I add the notch at 80Hz in the driverack, it takes effect at 80Hz in the measurements.

[bubbleboy76]

I should have even lower crossover point of the LF, should I not, judging from the measurements above?

[ivica]

"There is no replacement for displacement" is the one I have heard :)

Hi ,

You are right,
sorry for my bad English

regards
ivica

[ivica]

I should have even lower crossover point of the LF, should I not, judging from the measurements above?

Hi bubbleboy76,

1.LF
Looking at 1m distance measurements, the graph looks like (to me) very similar as on the "theoretical' shown on the:
http://www.audioheritage.org/vbulletin/showthread.php?36940-JBL-4365-Active-Crossover-settings&p=375799&viewfull=1#post375799

2.VHF
Looking at the VHF response, I think that You have to reduce "window time" in order to eliminate the reflection signal from the measurements, or to "lay down the box on the fore" with the face up, and then do the measurements, in order to reduce large surface reflections. Shown data are too bouncing, for me.

regards
ivica

[bubbleboy76]

Hi bubbleboy76,

1.LF
Looking at 1m distance measurements, the graph looks like (to me) very similar as on the "theoretical' shown on the:
http://www.audioheritage.org/vbulletin/showthread.php?36940-JBL-4365-Active-Crossover-settings&p=375799&viewfull=1#post375799

2.VHF
Looking at the VHF response, I think that You have to reduce "window time" in order to eliminate the reflection signal from the measurements, or to "lay down the box on the fore" with the face up, and then do the measurements, in order to reduce large surface reflections. Shown data are too bouncing, for me.

regards
ivica

Will do! I found this, describing how to do it in REW:

http://www.minidsp.com/applications/acoustic-measurements/loudspeaker-measurements

[Mctwins]

Hallo!

Your measurement looks strange at least for the UHF. Strong Comb filtering even with mic close to the speaker. It looks the same at 5cm or 1meter.

You have take into account for your room modes when looking from 20Hz-500Hz.

Could you present a mdat file? I have REW also. I would like to see your waterfall graph?

In the link provided and the headline "Low-frequency measurement" is interesting to read.

[hjames]

Hi ,

You are right,
sorry for my bad English

regards
ivica

Hopefully that did not come across harshly,
it was not meant to be harsh ...
Growing up, my brother and some of his friends drag raced
and it was pretty well known in the hot rod crowd
(my brother raced a '64 with a Chevy big block V8 overbored .030)

I was more into british and italian sports cars

[ivica]

Hopefully that did not come across harshly,
it was not meant to be harsh ...
Growing up, my brother and some of his friends drag raced
and it was pretty well known in the hot rod crowd
(my brother raced a '64 with a Chevy big block V8 overbored .030)

I was more into british and italian sports cars

Hi hjames,

No, problem at all.
Many years ago, I have heard, the phrase, but can not remember at the moment,
but I think this one is very good: "There is no replacement for displacement"

regards
ivica

[bubbleboy76]

Hallo!

Your measurement looks strange at least for the UHF. Strong Comb filtering even with mic close to the speaker. It looks the same at 5cm or 1meter.

You have take into account for your room modes when looking from 20Hz-500Hz.

Could you present a mdat file? I have REW also. I would like to see your waterfall graph?

In the link provided and the headline "Low-frequency measurement" is interesting to read.

Yes it is strange for the high frequencies. I measured with the speaker very close to the concrete sidewall, I think that is why. I did not have time to move around the speaker this time. I played a little bit with time window now, but it did not help. The same hilly curve even with very low time-window.
Next time I will measure with the speaker more far away from the walls.

I have the .mdat files, but that is not a valid file-extension to attach to this forum. I can attach a full range waterfall from listening position next time I measure.

[bubbleboy76]

I also had the microphone taped to the top of chair because I have no tripod for it, yet.
I realise now this must have screwed it up :)

[bubbleboy76]

MiniDSP has some good instructions on my level, not only mathematical mumbo-jumbo:

http://www.minidsp.com/applications/digital-crossovers/stereo-2-way-xover

http://www.minidsp.com/applications/dsp-basics/time-alignment

[Jakob]

Reading the information in the link provided by bubbleboy76 it says:

"A good approach is to use the output channel Parametric EQ to correct for the response of the individual drivers, and the input channel Parametric EQ for overall response shaping and taming room issues"

Is this true in general or only apply to miniDSP units and if general, why is it better to use the EQ on the input for overall shaping and taming room issues? Is it to reduce work if some drivers are exchanged and you then only need to rework the output EQ?

[bubbleboy76]

Reading the information in the link provided by bubbleboy76 it says:

"A good approach is to use the output channel Parametric EQ to correct for the response of the individual drivers, and the input channel Parametric EQ for overall response shaping and taming room issues"

Is this true in general or only apply to miniDSP units and if general, why is it better to use the EQ on the input for overall shaping and taming room issues? Is it to reduce work if some drivers are exchanged and you then only need to rework the output EQ?

It is like that in my DBX 260 as well. Divided into Pre-crossover-PEQ and Post-crossover-PEQ. The thinking is to separate the correction of the room and the correction of the speaker. So when you change room/location/venue, you do not have to mess with the speaker setup again.

[bubbleboy76]

This one you meant, 4313B?

http://www.audioheritage.org/vbulletin/showthread.php?19493-Compression-Driver-Compensation&p=215951&viewfull=1#post215951

or maybe this one:

http://www.audioheritage.org/vbulletin/showthread.php?29853-h66000-h9900-h4365&highlight=voltage+drive+s9900

[bubbleboy76]

Found this one in one of the threads above. Posted by 4313B, so it is the real stuff...

[bubbleboy76]

No, it was a thread wherein I posted the voltage drives of the K2-S9900, 4365, DD65000, DD66000, and DD67000 showing the similarities and differences. Anyway, I'll reload LEAP at some point and look at the files again.

Would it be possible (and easier) for you to just send me the file?
I would be eternally grateful.

[bubbleboy76]

Installed this little guys yesterday.

[4313B]

Would it be possible (and easier) for you to just send me the file?
I would be eternally grateful.You have a way to open and view LEAP files?

[bubbleboy76]

You have a way to open and view LEAP files?
Pos is helping me, I think he can do it, he is a resourceful guy :)
I wonder if the LEAP free demo software can open files? I can not try it without a file.
My email is [email protected] if you want to send me something, remember those swedish delicacies I promised?!

[bubbleboy76]

Found this one in one of the threads above. Posted by 4313B, so it is the real stuff...

Anyone who knows (read 4313B) if, and what, high-pass filter was used on this anechoic measurement by JBL?

[4313B]

The 4365 network...

[pos]

But what about the black curve, the unfiltered one?
Does it have some sort of protection network, or maybe a limited sweet frequency range?
Looking at the way the slope seem to get shallower down low it might just be some sort of windowing on the measurement impulse?

[4313B]

But what about the black curve, the unfiltered one?It is unfiltered.

Does it have some sort of protection network, or maybe a limited sweet frequency range?He doesn't use protection caps during measurements.

Looking at the way the slope seem to get shallower down low it might just be some sort of windowing on the measurement impulse?Yes. You can apply tail correction in LEAP.

[pos]

Thanks
Would it be possible to extract the impulse response (as a wav or txt file) from the LEAP file?

[bubbleboy76]

try to use AS LESS as possible number of EQ filters.


Would you like to elaborate on this? What are your experiences?

[ivica]

Would you like to elaborate on this? What are your experiences?

Hi bubbleboy76,

Any the influence of each EQ (parametric EQ) can be understand reading 4313B posts around
http://www.audioheritage.org/vbulletin/showthread.php?29853-h66000-h9900-h4365&p=301665&viewfull=1#post301665

Enhancing short FR region is less welcome, then reducing. Trying to compensate comb filer effect, I would never suggest too.
High Q factor ( Q = Bandwidth/fc ) to enhance some region I would not suggest to use.
Usually JBL drover are very linear FR in the FR range where it is designed to be used, saying that bass not over 800~1000Hz,
4-inch compression driver to be used used from about 800Hz up to about 10kHz, UHF drivers to be used over 8kHz ~10kHz.

regards
ivica

[pos]

I don't think it is as much a matter of positive or negative EQ as it is a matter of where the defect you wants to correct comes from.
For example you don't want to correct comb filtering or diffraction effect, as these will vary from one position to another.
On the other you do want to correct any defect that comes from the source itself, be it high or low Q, negative or positive (within reason, that is taking into account the capabilities of the driver...), because those corrections will likely be valid across a wide listening area.

[bubbleboy76]

A little update. I have been waiting for some XLR to 1/4" TRS adapters, so I could use the same soundcard (focusrite 2i2) for both input and output.
Previously, I have had the focusrite for input and the oppo for output, and POS advised me strongly against this.
Hopefully, this was the cause of the comb filtering in my measurements.
Now the adapters have arrived, so I will have better quality measurements as soon as I get the time to do it.
I also used 4 sweeps for every measurement before, and that can under some circumstances cause spikes in the curve according to the REW helpfile. So I will use 1 sweep from now on.
I will also use ASIO drivers instead of Java. It was not possible in REW when choosing different input/output soundcard.

[bubbleboy76]

New measurements done today and sent to POS for analysis.
No more comb filtering :)

[bubbleboy76]

1Audiohack, I have the anechoic 4365 horn/driver measurement curve in REW-format if you want it. POS did some magic and extracted it from the picture.

[bubbleboy76]

Measurements ongoing.
110cm to side walls. The longest my cables allows.

[Ed Zeppeli]

Funky looking room! Nice Les Paul too.

Cheers,

Warren

[Jonas_h]

Interesting thread! I fiddled around with active crossover (without known presets) with my last speakers and the process I ended up with which gave the best result:

1. Create two target curves in REW with 24db/oct LR filters at the desired crossover frequency. I.e. one target curve with a high-pass filter and one target curve with a low-pass filter.
2. Measure the drivers and assign filters so each driver follows this slope.
3. Do time-alignment on both drivers. (Either use RTA and adjust delay which is the easiest or compare the phase trace for both drivers and adjust accordingly).

The above procedure gave me really good results.

But the lack of free-field measurements and the constant thought about "Have I done it correctly" lead me to sell the speakers and buy speakers with passive crossover. Then I am doing "passive bi-amping" with known JBL presets to time-align the drivers.

[bubbleboy76]

Interesting thread! I fiddled around with active crossover (without known presets) with my last speakers and the process I ended up with which gave the best result:

1. Create two target curves in REW with 24db/oct LR filters at the desired crossover frequency. I.e. one target curve with a high-pass filter and one target curve with a low-pass filter.
2. Measure the drivers and assign filters so each driver follows this slope.
3. Do time-alignment on both drivers. (Either use RTA and adjust delay which is the easiest or compare the phase trace for both drivers and adjust accordingly).

The above procedure gave me really good results.

But the lack of free-field measurements and the constant thought about "Have I done it correctly" lead me to sell the speakers and buy speakers with passive crossover. Then I am doing "passive bi-amping" with known JBL presets to time-align the drivers.

Another way of achieving this, POS told me, is to first EQ each driver acoustically flat one octave past crossoverpoint, and then apply the electrical crossover slope of choice. But my dsp is to limited in number of PEQs to be able to do this.

Today, we measured delay properly, and 16-17cm delay of the woofer it should be. This also is consistent with the physical distance of the drivers, that I have estimated to be about 20cm.

We now have a pretty decent acoustical linkwitz-riley 36Db/oct at 800Hz (using electrical BW24db/oct). But the work is ongoing.

I would like to have a properly calibrated mic. The soundcard is now calibrated with loopback.

[bubbleboy76]

I just ordered a calibrated microphone from Cross-Spectrum Acoustics.
It feels pointless to have a meaurement system that can not be fully trusted.

[bubbleboy76]

I once voted for 4313B as president, but in the next election I have to put my vote on POS!

[Jonas_h]

Another way of achieving this, POS told me, is to first EQ each driver acoustically flat one octave past crossoverpoint, and then apply the electrical crossover slope of choice. But my dsp is to limited in number of PEQs to be able to do this.

Today, we measured delay properly, and 16-17cm delay of the woofer it should be. This also is consistent with the physical distance of the drivers, that I have estimated to be about 20cm.

We now have a pretty decent acoustical linkwitz-riley 36Db/oct at 800Hz (using electrical BW24db/oct). But the work is ongoing.

I would like to have a properly calibrated mic. The soundcard is now calibrated with loopback.

That is a solution as well. But I would feel that I added way too many filters and "reverting" them afterwards without utilizing the natual response of the driver. I am no expert though, so maybe my worries are not significant.

Congrats with your microphone. I bought an EarthWorks M30 for the same purpose to be sure that I had reliable measurements - it measured identical to a Behringer microphone in the 100-2000hz area and slight variations below and above...

[bubbleboy76]

This is marketing, but still shows how it can look.

[Jonas_h]

This is marketing, but still shows how it can look.
I bought 8 Behringer mics which measured identical. I could have been lucky though. But no matter what, I understand the need to "be sure". Thats why I bought the Earthworks myself.

[bubbleboy76]

I will calibrate my old mic with the new one as reference, then I can duct tape them to my head by the ears and do the perfect listening position measurement!
;)

[Mctwins]

I will calibrate my old mic with the new one as reference, then I can duct tape them to my head by the ears and do the perfect listening position measurement!
;)

Why do you want to do this?

[bubbleboy76]

Why do you want to do this?

It was a joke.

[jarrods]

then I can duct tape them to my head by the ears and do the perfect listening position measurement!

Not as funny as it may seem. When I worked at Deakin University building up electronic/computing rigs for research academics one Senior Lecturer in Psychology was doing basically that. Dr Simon Oldfield was using a dummy head (not some cheap model used to display hats etc but physical mimic of real human head) and he had put tiny condenser microphone elements inside where the eardrums would be. He would record things in anechoic chamber and play back with a rig that just had two simple speakers above your head. One day he recorded jingling his keys around the dummy head and then played it back to me while I was sitting in the subject chair. You could hear the keys go around your head, even fully behind you. I have never heard such accurate pinpointed 3D sound as good as this in my life and this was some 30 years ago. He went on to do research for Australian Dept of Defence on advanced auditory cockpit displays for fighter planes.

[ivica]

Not as funny as it may seem. When I worked at Deakin University building up electronic/computing rigs for research academics one Senior Lecturer in Psychology was doing basically that. Dr Simon Oldfield was using a dummy head (not some cheap model used to display hats etc but physical mimic of real human head) and he had put tiny condenser microphone elements inside where the eardrums would be. He would record things in anechoic chamber and play back with a rig that just had two simple speakers above your head. One day he recorded jingling his keys around the dummy head and then played it back to me while I was sitting in the subject chair. You could hear the keys go around your head, even fully behind you. I have never heard such accurate pinpointed 3D sound as good as this in my life and this was some 30 years ago. He went on to do research for Australian Dept of Defense on advanced auditory cockpit displays for fighter planes.

Hi jarrods,

Such experience can be expected, but the problem in real sitations, is more complicated, as a listener, while listening live music, rotates his head (left-right, up-down,...), depending to his interest or towards the part of the orchestra he is mostly interested in, and I believe that is the main reason that usual recordings are not made using "using a dummy head", but may be there are much more other reasons (I believe)

reagrds
ivica

[macaroonie]

65889


https://youtu.be/QQzS3yReIx0

[bubbleboy76]

Now I have received calibrated mic.
I will do new mesurements as soon as I get the house free from kids :)
I also did a new soundcard calibration today, including the dbx 260. See the attached picture.
The yellow line is including the dbx (on full passtrough), the other line show only the Focusrite 2i2 soundcard.

[ivica]

Now I have received calibrated mic.
I will do new mesurements as soon as I get the house free from kids :)
I also did a new soundcard calibration today, including the dbx 260. See the attached picture.
The yellow line is including the dbx (on full passtrough), the other line show only the Focusrite 2i2 soundcard.

Hi,
so almost no error.

regrads
ivica

[bubbleboy76]

Hi,
so almost no error.

regrads
ivica

Yes, that was anticipated, but it is still nice to know.

I will include the dbx in the timing reference channel for the next measurements, previously I only had the soundcard loop-backed.

[bubbleboy76]

Looking in the mic calibration report...
As worst-case examples, the mic was 15Db wrong a 5Hz, and 3Db wrong at 12500Hz.
That is pretty significant.

[bubbleboy76]

Here is also phase showed in the calibrations of the soundcard, and the soundcard+dbx260.

dbx260 clearly affects phase a bit.

[Mctwins]

Hallo!

Here is my soundcard calibration with Tascam US-122MKII

With REW of course.....

Dbx260 is not a soundcard. Why do you calibrate with the dbx260?

[Mctwins]

Here is also phase showed in the calibrations of the soundcard, and the soundcard+dbx260.

dbx260 clearly affects phase a bit.

You must be doing something wrong here, or your soundcard is corrupt.

[bubbleboy76]

Dbx260 is not a soundcard. Why do you calibrate with the dbx260?

This is a very interesting question to me, but kind of philosophical with the small changes the calibration applies!

POS advised me to not use the sound card calibration at all, due to the insignificant errors.

For me, one reason for having the dbx in the calibration would be to take that device out of the equation, and get more generic dsp settings in the end. The dbx260 is not my end state, it has to few PEQs per separate channel/driver.

[bubbleboy76]

You must be doing something wrong here, or your soundcard is corrupt.

Why do you say this? My calibration of only the soundcard looks very much like your calibration. But it is hard to see the details with the dB scale you show.

It is kind of expected for a non dc-coupled device to introduce some phase shifts and high-pass filtering by the protective caps, don´t you agree? Just like yours and mine soundcard does as well shown in the diagrams (I am just assuming dbx260 is not dc-coupled, I do not know that for sure).

One way to find out would be if you try to do a calibration including your dear dbx260 in the loop :)

[bubbleboy76]

The settings I have now sounds absoluteley amazing. The best sound system I have heard :)

I do not want to post the settings here because it is POS work. It is up to him to post them here, when he finds it fit.

I can say, the current settings are no way near the ones I myself tried in the beginning of this thread.

I am still waiting to get a chance to do new measurements with my calibated mic.

I will also measure the stock passive filter again to compare.

I know now that there is nothing wrong with the stock 4365 passive filter. It is me prefering +9dB in the low frequencies. I have been ruined by having subs "too high for too long" :)

[Jonas_h]

So the upgrade you are experiencing is more bass when going from passive to active? To me it sounds like you could have settled with just adding some boost down low and use the speaker with the passive filter.

[bubbleboy76]

So the upgrade you are experiencing is more bass when going from passive to active? To me it sounds like you could have settled with just adding some boost down low and use the speaker with the passive filter.

Yes, I think so too. It would have solved my "problem".

But now, when I have heard the clarity and micro-dynamics of active crossover, and the difference time-alignement did, there is absolutely no way back.

[Mctwins]

Hmm... I have plenty off bass.....:D

[pos]

I do not want to post the settings here because it is POS work. It is up to him to post them here, when he finds it fit.I need to confirm our settings with a new set of measurement with your new calibrated mic.
The <500Hz settings will potentially not translate very well to other 4365 owners because of the influence of the room that cannot be taken out of the equation with "in-house" measurements... (that said we did not try to address room modes and only used low Q EQS in that range, so it should be okay...).
Last but not least, our settings are partly constrained by the DSP unit you are using in term of Q and frequency setting precision (step), as well as the sheer number of EQ points available.
There are also a few EQ points that will likely only be valid for your particular drivers, because they are done in frequency ranges where unit-to-unit consistency is not (and cannot be) guarantied with enough precision. This is the case for the two high Q breakup peaks in the stop band of the woofer, that were different in your woofer (you only measured one) and the ones in the curves from the 1501Fe EDS.
By the way it is interesting to note that the breakup peaks in the 2216nd are not addressed in the M2 settings either, probably for the exact same reason, whereas some other minor (but more consistent) things were actually addressed.


I know now that there is nothing wrong with the stock 4365 passive filter. It is me prefering +9dB in the low frequencies. I have been ruined by having subs "too high for too long" :)
I would have to look that up but IIRC the setting with a +9dB EQ point down low results in a ~+6dB "gentle" bump when all the other EQs are taken into account.
This is similar to what you typically see in house curves.

[1audiohack]

I am keenly interested in this as I have just powered up the 4365's with a new Crown I-TECH 5000 HD amp. I will also now have this flexability. :D

Thanks guys!

Barry.

[bubbleboy76]

P in POS stands for Phase, right?!

[bubbleboy76]

I am keenly interested in this as I have just powered up the 4365's with a new Crown I-TECH 5000 HD amp. I will also now have this flexability. :D

Thanks guys!

Barry.

Wow, that is one nice amp (and dsp/dac). Do you have 2 of them?

[bubbleboy76]

By the way it is interesting to note that the breakup peaks in the 2216nd are not addressed in the M2 settings either, probably for the exact same reason, whereas some other minor (but more consistent) things were actually addressed.


I suspect your new project will address them, for your drivers :)

Feels like I am the only one here not ordering the M2 stuff. But I will wait with regretting that until I see you get any parts! ;)

[bubbleboy76]

Quality measurements done today, with calibrated mic and 8 sweeps per measurement.
Measured in a lot of different positions of the mic. Exact same position in middle of room for both speakers.
No soundcard calibration used.
Now we just have to wait for the pos magic :)

[bubbleboy76]

Ground plane measurements around listening position done today for the LF drivers.

[Jonas_h]

Ground plane measurements around listening position done today for the LF drivers.
Why are they not attached to your post? ;)

[bubbleboy76]

Why are they not attached to your post? ;)

It is very much data. If you want the raw REW data with impulse responses it is on my dropbox:
https://www.dropbox.com/sh/al5n8gqa98b5yvb/AAD5937WR-x8upWePGuhEVTAa?dl=0

[bubbleboy76]

Pics from latest measuring session.

[bubbleboy76]

I have an dbx4820 incoming from Ebay. Will be interesting to see if it sounds different than the 260.
It will give me 2 more PEQ per individual driver. But most importantly for me, digital inputs.

Anyone tried using 4800/4820 for volume control? There are wired external volume knobs from dbx:
http://dbxpro.com/en-US/product_families/zone-controllers-us

But I would like to do the volume control with an IR-remote, any suggestion/tips of how to do that?

[1audiohack]

....Anyone tried using 4800/4820 for volume control? There are wired external volume knobs from dbx:
http://dbxpro.com/en-US/product_families/zone-controllers-us

But I would like to do the volume control with an IR-remote, any suggestion/tips of how to do that?

I am keenly interested in this as well.

Barry.

[bubbleboy76]

Voltage drive and frequency response for the JBL 4365. Source: Italian magazine Audio Review.

When I measure the stock passive filter of 4365 I also get this comb filtering (as Audio Review magazine) where the UHF kicks in.
This seems to be because of the HF and UHF drivers are very out of time/distance alignement, right pos?

Note, I do not get this comb filtering with active setup with exact same measurement, so it is not a fault in the measuring or induced by the room.

[Mctwins]

When I measure the stock passive filter of 4365 I also get this comb filtering (as Audio Review magazine) where the UHF kicks in.
This seems to be because of the HF and UHF drivers are very out of time/distance alignement, right pos?

Note, I do not get this comb filtering with active setup with exact same measurement, so it is not a fault in the measuring or induced by the room.

I have not notice this when I did the measurments close to the HF and UHF with my 4365. Some happy day I will remeasure and see if I have some anomalies. Right now it is more fun to play music.

Remember, you must intstall AudioArchitecht to be able to work with dbx4820. The 4820 don't have a front panel display. This is how I understand it.

[pos]

When I measure the stock passive filter of 4365 I also get this comb filtering (as Audio Review magazine) where the UHF kicks in.
This seems to be because of the HF and UHF drivers are very out of time/distance alignement, right pos?

Note, I do not get this comb filtering with active setup with exact same measurement, so it is not a fault in the measuring or induced by the room.

Even if you perfectly time-ailgn the HF and UHF for a given measurement position, it will only be valid at certain points in space, typically one axis...
The wavelength are so small at these frequencies that even a tiny geometrical difference will cause phase errors and comb filtering.
There is nothing that can be done about it, but the good news is that at these frequencies it is not really a problem either :D
In a 3-way FIR active setup you could use brickwall filters to get the comb filtering range smaller than the corresponding ERB band, but...

Do you feel you are missing something up top in the active 2-way setup compared to the passive 3-way one?

[bubbleboy76]

Do you feel you are missing something up top in the active 2-way setup compared to the passive 3-way one?

I have not actively listened for it, but no, I have not felt that.

I will stick to 2-way.

[bubbleboy76]

Remember, you must intstall AudioArchitecht to be able to work with dbx4820. The 4820 don't have a front panel display. This is how I understand it.

Yes, I wanted that one without knobs, so my children can not change settings and destroy the speakers by accident.

[Mctwins]

Yes, I wanted that one without knobs, so my children can not change settings and destroy the speakers by accident.

I know, children love knobs:)

If you would go for the display then one can do a "panel front lock", is the same as for 260.

[bubbleboy76]

I know, children love knobs:)

If you would go for the display then one can do a "panel front lock", is the same as for 260.

I never got that to work on the 260, not from the pc-software anyways.

[Mctwins]

I never got that to work on the 260, not from the pc-software anyways.

Hallo!

I do belive that you can do it from GUI. If you open the GUI and go to "EDIT" and go to "Utilities". I am not using it myself or tested it. :dont-know:

[1audiohack]

I locked my 4800 so no one could destroy the components in my 4350's, and then forgot and locked my self out of it. It took me two hours to figure my way back in. I foget how I did it though. :o:

Barry.

[bubbleboy76]

Amateurs using pro-gear... ;)

[bubbleboy76]

Anyone tried using 4800/4820 for volume control? There are wired external volume knobs from dbx:
http://dbxpro.com/en-US/product_families/zone-controllers-us

But I would like to do the volume control with an IR-remote, any suggestion/tips of how to do that?

The HiQNet Motion Control iOS app could be the solution for this. I will dig into that some day.

[bubbleboy76]

Playing with the 4820...
It sounds better than the 260, for sure.
But I am not really friends with the HiQnet Audio Architect software. Not as simple as the Driveware for the 260.
I just want to be able to go online with the device and change a little settings in it, but I can not figure out how to do that. The thing I manage to do is to setup the whole device offline, and then sync everything. But that takes long time.
There is no feedback showing if the program is doing anything, like a progress bar. That is really annoying. I missed to understand several times that the program was actually working syncing all 50 presets to the device.
I have 32-bit windows 7, so I can not use latest version of Audio Architect. Maybe it is better in later versions.
I am also using serial cable. Maybe it works better with ethernet network connection?

[Mctwins]

Playing with the 4820...
It sounds better than the 260, for sure.
But I am not really friends with the HiQnet Audio Architect software. Not as simple as the Driveware for the 260.
I just want to be able to go online with the device and change a little settings in it, but I can not figure out how to do that. The thing I manage to do is to setup the whole device offline, and then sync everything. But that takes long time.
There is no feedback showing if the program is doing anything, like a progress bar. That is really annoying. I missed to understand several times that the program was actually working syncing all 50 presets to the device.
I have 32-bit windows 7, so I can not use latest version of Audio Architect. Maybe it is better in later versions.
I am also using serial cable. Maybe it works better with ethernet network connection?

Hallo!

You wrote very interesting in post 143;)

If you are using the serial adapter for the 260 into 4820 then I think it is wrong. You schould use ethernet cable if Audio Architect is being use. Download the latest version. You must go online and see if you have the latest firmware update for the 4820.

I see in the manual that if serial cable is use then it is called System Architect GUI, not sure if it the same as Audio Architect.

There is some turtorial videos for the 4820 on dbx website.

You have to try both ways, install the System Architect GUI or Audio Architect and see whats works best.

Not so easy when I don't have the unit in front of me so I can test it.

[Mctwins]

Hallo!

If you go to dbx website and go to support and then to software there you will see that for the 4820 there is only HQnet Audio Architect avalible.

Maybe the System Architect is an old GUI version.

[bubbleboy76]

I am using Audio Architect 1.30.2, the latest version that was supporting 32-bit windows (that I have on my laptop).
I have updated to latest firmware of the 4820 using Audio Architect, that was the first I did.
I will try ethernet connection, but it feels like a long shot that it would change anything.
The only thing that will help is me learning how to use the program :)

[bubbleboy76]

I got a hang of it now.
The Audio Architect adds some layers of complexity compared to the obsolete System Architect (that the dbx website tutorials shows).

[bubbleboy76]

Close to be ready now. Measurement of the release candidate settings done today, for delay, phase, and HF/LF relative levels.
I think soon pos can publish his masterpiece here!

An MiniDSP OpenDRC-di is also added to the mix, for phase linearization :)

[bubbleboy76]

At some point I need to load LEAP onto a PC and then I can get them for you. I have a laptop in mind I just haven't gotten around to loading any software onto it. It needs an OS, AutoCAD, LEAP, HiQnet, and DEQX loaded onto it.

Rumours say you have loaded LEAP now...

If you find the 4365 data, I would be very grateful.

[bubbleboy76]

Current state of affairs...I have measured the relative gain levels for LF/HF, and found the one giving the smoothest transition.I measured 3 listening positions and averaged them.Then I tried to do the same measurements to find the correct delay in the averaged listening positions, but it was hard for me to get a clear picture of the cancellation that should occur. From 1m it was much clearer in the measurements. The room playing tricks on me. But POS has a trick up his sleeve for this...The acoustical crossover is now LR48dB/oct at 650Hz.FIR phase corrections is added for the crossover region and the bass reflex port. I hear positive sound enhancements for this, but I will blindtest it on my friend with the golden ears when all settings are final.

[bubbleboy76]

We have also had some hassle with my measurements.
REW does not include/apply the mic-calibration when exporting impulse responses to text-file or wav-file.
So it was fortunate that we discovered this. We both asumed it included mic-calibration when I was sending files to POS.

[4313B]

Rumours say you have loaded LEAP now...

If you find the 4365 data, I would be very grateful.Still looking... going through various hard drives looking for the files as I have time. My data drives "look" even worse than my garages...

[bubbleboy76]

Still looking... going through various hard drives looking for the files as I have time. My data drives "look" even worse than my garages...

Try to search for "GT_4365.lcd". That was the filename in one of your graphs.

[4313B]

I found it.

[4313B]

Data imports

Yes, I know the label says 1500Fe-8 rather than 1501Fe

I'm not sure if that is a typo or if the original woofer was indeed a 1500Fe, don't know, don't care.

Both of those ferrite transducers are awesome, easily my favorites amongst the fifteen-inch ferrites. :yes:

[4313B]

045Ti "removed", which is how the data file was sent to me. A 3.3k ohm resistor was in place of the 3.3 ohm resistor.

[bubbleboy76]

Data imports

Yes, I know the label says 1500Fe-8 rather than 1501Fe

I'm not sure if that is a typo or if the original woofer was indeed a 1500Fe, don't know, don't care.

Both of those ferrite transducers are awesome, easily my favorites amongst the fifteen-inch ferrites. :yes:

Thanks!
Those SPL vs Freq imports, do you have those separate raw data files?
1500Fe-8.txt
476Mg,txt
045Ti.1.txt

Possible for you to upload the raw data high-resolution impulse-response for those?

[bubbleboy76]

Do you have spinorama measurements as well, 4313b?
Or any kind of off-axis data?

[bubbleboy76]

Data imports

Yes, I know the label says 1500Fe-8 rather than 1501Fe

I'm not sure if that is a typo or if the original woofer was indeed a 1500Fe, don't know, don't care.

Both of those ferrite transducers are awesome, easily my favorites amongst the fifteen-inch ferrites. :yes:

As an owner of 4365, that is kind of frightening, if the network was designed for another woofer. The 1501fe was a kind of substantial change to the 1500fe. It got the cloth surround, if I am correctly informed.

[4313B]

These are the imported data files.

[bubbleboy76]

These are the imported data files.

WOW! Thank you so much!

Just to make sure, this is on-axis measurements from the JBL anechoic chamber taken with the elements inside the 4365 horn/box?
And no protection caps or limited frequency range?

[4313B]

As an owner of 4365, that is kind of frightening, if the network was designed for another woofer. The 1501fe was a kind of substantial change to the 1500fe. It got the cloth surround, if I am correctly informed.You have to take it for what it is and not jump to any conclusions based on one LEAP file. The only reason I mentioned it was because I knew someone would bring it up anyway. This LEAP file is just one tool in their tool box. You might want to look at the technical sheet I have included compared to the LEAP schematic. I did years ago but can't remember the differences, if any.

There is another SPL curve in the file labeled "1500Fe M" but there is no associated impedance file so there is no real point in including it. It isn't used in the simulation either as can be seen here:

[4313B]

WOW! Thank you so much!

Just to make sure, this is measurements from the JBL anechoic chamber taken with the elements inside the 4365 horn/box?
And no protection caps or limited frequency range?I was told that was the case. And protection caps would wreck the data so none are used during measurements. It does look like G.T. might have applied some tail correction to the VLF response of the LF below ~ 35 Hz or so. If I remember correctly the large chamber is "only" good to around 25 Hz and the smaller chamber to something like 30 or 35 Hz.

[Mctwins]

Thanks 4313B

What smoothing is it in those graphs?

Why is there difference in SPL vs Freq graphs in post 156 and 158. There seems to be more combing or peaks and deeps between 12kHz-20kHz(Sum) in post 156 compared too without the UHF in post 158.

[bubbleboy76]

I also had a lot of comb filtering up high when measuring the stock filter, compared to when I measured without the UHF (in exact same mic position, 1m from speaker).
If I remember correctly.
Have you tried that on yours mctwins?

[Mctwins]

I also had a lot of comb filtering up high when measuring the stock filter, compared to when I measured without the UHF (in exact same mic position, 1m from speaker).
If I remember correctly.
Have you tried that on yours mctwins?

Ok,

No, I have not done this. I have only measured at my LP.

[Mctwins]

In post 156 SPL vs Freq graph, it would be nice to see higher up in freq response, to 50kHz. Is this possible to see?

[bubbleboy76]

I did a quick comparison of one of my 1m measurements with this JBL anechoic data.
I am amazed how correct my measurements seems to be for the HF.
For the LF the room messes more with me. And also maybe not the same sort of driver compared, 1501fe vs 1500fe.
No gating.
1/6 smoothing.
Calibrated mic.
68uf protection cap on the 476mg.
600Hz BW 12dB/oct dsp highpass filter applied on the 476mg to protect it when measuring.

[bubbleboy76]

It would be fun to EQ the LF flat up to 20KHz and see how it sounds. Anyone tried that?!

[ivica]

It would be fun to EQ the LF flat up to 20KHz and see how it sounds. Anyone tried that?!

Hi,
May be try E140 too.


Regards
Ivica

[4313B]

In post 156 SPL vs Freq graph, it would be nice to see higher up in freq response, to 50kHz. Is this possible to see?Ok

[bubbleboy76]

So here we have the explanation why I thought my stock 4365 sounded too bright.
No, I am not a dog!
This is the anechoic thruth mctwins. Maybe now you can stop haunting and mocking me? I think even an apology would be in order.

[4313B]

I left the 045's turned off... even then the 476Mg's ended up on the lowest setting. High frequencies have always bothered me though so nothing new here. Most systems I listen to are too bright out of the box or people have cranked up the top end. I never needed slots added to my 4430's.

DSP will allow you to shelve the whole thing however you want. ;)

[bubbleboy76]

"There is nothing wrong with the 4365" GT said to you 4313b? What am I missing? The measurements looks like shit up high.
(give me some slack, I bought this ones new for a heck of a lot money)

[4313B]

"There is nothing wrong with the 4365" GT said to you 4313b? What am I missing? The measurements looks like shit up high.
(give me some slack, I bought this ones new for a heck of a lot money)They look normal to me with all the comb filtering going on. You simply might have more sensitivity to high frequencies than most people? Your room might be too hot? Just turn the top end down. Heck, I ran my slots in every single 43xx system on one quarter turn for all the many years I had them. That is certainly a lot lower than they were calibrated for at JBL.

I just pulled up the LEAP file for the K2-S9900 and that won't help to compare with. Greg didn't even bother putting the 045 data into the LEAP file as can be seen in the screen shot below. However, as both Greg Timbers and Doug Button pointed out, the 045's offer substantial improvement in dispersion above ~10k, that can arguably be considered their primary purpose.

So... just shelve the top end response to your liking and enjoy! :)

[bubbleboy76]

You simply might have more sensitivity to high frequencies than most people?.

Woof woof!

Ok, so this is a good anechoic curve?? Really??
If that is the case, I can apologize to whatever I have offended!
But I do not fully beleive in it, because now as a 2-way active with flat eq up to 20khz, they sound perfect to me.

[4313B]

DD66000 for comparison. (remember - 045Be vs 045Ti) JBL takes tons of measurements with various microphone and system (enclosure) alignments, not just the single measurement you see here for the purpose of generating a potentially reasonable network solution in LEAP. This is the summed response generated by LEAP given a specific data set. The result is then evaluated for validity.

[4313B]

If that is the case, I can apologize to whatever I have offended!People freak out about curves all the time...

Some people only listen to curves ;)

That's why all those armchair speaker designers were freaking out about the DD66000 when they saw the published curves and specs from JBL.

Remember the guy who bought a pair of S4700's and thought they had "issues"? Well, there is nothing wrong with the S4700 either. Maybe some people just don't like it but can't come to grips with the fact that they just don't like it? I don't know. Not everything is plug and play. Some things take a little bit of tweaking in specific environments to work as desired.

One of worst systems I ever heard included a pair of JBL 4343's in a funky (but very nice looking) living room. The living room was open on the back end to a kitchen and a hallway with staircase. It just sounded like absolute ****. But the guy loved them. I cannot fathom how. And quite frankly, it really isn't any of my business. :p My 4343's sounded killer in my room.

[4313B]

I thought you guys had all this figured out via DSP?

[bubbleboy76]

I thought you guys had all this figured out via DSP?

Yes we have, it sounds wonderful now in my setuo :)

And I am very grateful for your data, so I could verify my own measurements.

[bubbleboy76]

However, as both Greg Timbers and Doug Button pointed out, the 045's offer substantial improvement in dispersion above ~10k, that can arguably be considered their primary purpose.


Hmmmm...
Worth pursuing it?
Worth the extra comb-filtering?
I am very happy how it sound up high now as a 2-way.
And I think pos is growing a little bit tired of me :)

[4313B]

The 045 is merely "tacked on" to these systems. There is no cutoff (low pass) from the 476Mg/Be to the 045Mg/Be/Ti in any of them, simply a high pass filter on the 045Mg/Be/Ti. Given that, you can play around with the leading resistor in the 045 network and change it to a larger value to decrease output. I would not worry about the comb filtering, it's all a trade off, you might prefer the increased dispersion of the 045.

If I remember correctly, the only systems that had an actual low pass filter on the smaller 435Be compression driver were the 1400 Array, K2-S9800 and SAM1HF and they crossed over to the 045 at something like 10 kHz.


Yes we have, it sounds wonderful now in my setuo :)

And I am very grateful for your data, so I could verify my own measurements.Sure. Anyone should be able to get pretty close to anechoic in their home above ~ 1 kHz. It is the information below ~ 1 kHz that can be the problem. That's probably why alot of guys like designing those smaller two-way systems with smaller woofers crossing over higher (coupled with the fact that their math tells them fifteens "don't work" to 1 kHz.) ;)

[Mctwins]

Hallo!

Thanks again 4313B for the graphs.

I am not using any EQ or DSP in my system, nor do I use shelving to reduce the high freq. I have set the knobs for HF and UHF to 0dB. I don't have any harshness in high freq in my room.

Those combfiltering that occur in the SPL vs Freq graphs is inaudible to human ear becasue it is only +/-2,5dB, give or take and from 15kHz and beyond. That dip around 21khz, Well, this is definitly not audible. I don't know what smoothing it is in those graphs, I asume it is 1/6oct band.

There are many factor that can cause harshness in a system, room acoustics, bad recording, no control in early reflections, room resonances, little bass contra high freq content and the list can go on.

I am using Wing diffusors and this helps alot in my room that controls the early reflections and for the bass control the Helmholtz resonators Varitunes V4 and V6 makes this room super duper sound.

:)

[4313B]

:)

[Robh3606]

Hello 4313B


Thanks for posting this

Rob:)

[Mctwins]

So here we have the explanation why I thought my stock 4365 sounded too bright.
No, I am not a dog!
This is the anechoic thruth mctwins. Maybe now you can stop haunting and mocking me? I think even an apology would be in order.

Hmm...I really don't understand you here. Where have I mocked you.:confused:

When putting a loudspeaker in a room the room acoustics will have it's own charateristic behavior compared to the anechoic data presented here.

I am gonna be totally honest with you here. It seems that you are not satisfied with the performance with the 4365 as it is. I think you have gone in the wrong direction to bi-amp these speakers as you have done, namly gone actively. Even if you would have M2 speakers the setting in those Crowns DSP shelving, PrePEQ, PostPEQ, x-over settings and so on has already been done and there is no point to adjust or change this settings due to that it sounds harsh or the bass is not good in the room you are putting it in. If changes is done in these settings, M2, you will destroy the good performance in those loudspeakers.

The end result will be the same, in your case, if you leave it with the stock network or going activly and then adjust to have the same result as in the stock network. I really don't see any point in doing this. The only thing you have done is removing the UHF.

Regarding your measuerments in post 170, I don't see the same result compare to the graphs 4313B is showing. For example, in SPL vs Freq graphs there is a peak at 150Hz and in your .txt file you don't have this peak at all. There is no comparison. In your graph "4365 LF. JBL anechoic and Rajapruk mic1m", both are green and I don't see this peak at 150hz in neither of them. And this goes for the 476Mg graph as well, no comparison.

:)

[pos]

And I think pos is growing a little bit tired of me :)Not at all ;)

I think The HF measurements are pretty accurate and follow the anechoic ones quite well, but as 4313B pointed out the LF measurement in a room cannot be as accurate as in an anechoic room.
But that is not really a problem as this part of the range has to be EQed to fit the room and placement anyway.

[bubbleboy76]

Hmm...I really don't understand you here. Where have I mocked you.:confused:

When putting a loudspeaker in a room the room acoustics will have it's own charateristic behavior compared to the anechoic data presented here.

I am gonna be totally honest with you here. It seems that you are not satisfied with the performance with the 4365 as it is. I think you have gone in the wrong direction to bi-amp these speakers as you have done, namly gone actively. Even if you would have M2 speakers the setting in those Crowns DSP shelving, PrePEQ, PostPEQ, x-over settings and so on has already been done and there is no point to adjust or change this settings due to that it sounds harsh or the bass is not good in the room you are putting it in. If changes is done in these settings, M2, you will destroy the good performance in those loudspeakers.

The end result will be the same, in your case, if you leave it with the stock network or going activly and then adjust to have the same result as in the stock network. I really don't see any point in doing this. The only thing you have done is removing the UHF.

Regarding your measuerments in post 170, I don't see the same result compare to the graphs 4313B is showing. For example, in SPL vs Freq graphs there is a peak at 150Hz and in your .txt file you don't have this peak at all. There is no comparison. In your graph "4365 LF. JBL anechoic and Rajapruk mic1m", both are green and I don't see this peak at 150hz in neither of them. And this goes for the 476Mg graph as well, no comparison.

:)

I have severe problems understanding you, and you seem to have severe problems understanding me.
Let's leave it at that.

[4313B]

I don't see any problems with either view point.

The best part is where the 4365's can be returned to stock simply by hooking up the passive filters and viola!

If it were me I would try and duplicate the passive voltage drive using active DSP and see how that worked out. Then I would adjust for the room using the remaining PEQ points.

[bubbleboy76]

Working on setting the correct delay LF/HF for the listening positions now.

It was 17cm (0.5ms) from 1m.
POS calculated that it should be approx 14cm in the listening position, based on a lot of parameters. And it seems to be correct. I get the largest cancellation with 14cm delay on LF (3 listening positions / seats in my sofa averaged), compared to 15, 16 and 17 cm.
But the cancellation is not as pronounced and clear as from 1m.

I attach some curves from latest measurement. 1/6 smoothing. No gating.
Sleeping baby in the apartement, so I had to measure on a fairly low volume.
I never did generate waterfall before. I do not know how to get more colors and stuff. I have to play more with that.

[bubbleboy76]

They are not quite fullrange-speakers the 4365:s, looking at these measurements. The 4365 box is tuned to 27-28Hz, if I remember correctly.
Some 2269h:s, in closed boxes, would add that extra missing in the 10-30Hz region, I guess? ;)

[4313B]

.

[bubbleboy76]

Thanks. Consistent with my measurements then, where it starts going down at around 40Hz.

[bubbleboy76]

I have no experience analyzing waterfall diagrams, or room correction.
Anyone having comments or reflections on my waterfall?
Does it look good, like sh..., or something in between? This is a normal livingroom, without any acoustic treatment. Speakers are located in corners (wife requirement), and the sofa is placed againts the back wall.
I interpret the diagram that my Schroeder-frequency is about 500Hz, correct?

[Mctwins]

Working on setting the correct delay LF/HF for the listening positions now.

It was 17cm (0.5ms) from 1m.
POS calculated that it should be approx 14cm in the listening position, based on a lot of parameters. And it seems to be correct. I get the largest cancellation with 14cm delay on LF (3 listening positions / seats in my sofa averaged), compared to 15, 16 and 17 cm.
But the cancellation is not as pronounced and clear as from 1m.

I attach some curves from latest measurement. 1/6 smoothing. No gating.
Sleeping baby in the apartement, so I had to measure on a fairly low volume.
I never did generate waterfall before. I do not know how to get more colors and stuff. I have to play more with that.

Hallo!

Set the waterfallgraph to 300ms in the "Time Range" In the "Controls" button and then set it to "Apply default setting". This way it is more easily to see. Have the scaling on dB axis to where you have your highest peak and 40 dB down.

Try this first.

[Mctwins]

Hallo!

This is my response at LP with 4365. I have around 12dB down from 20-30Hz. You have 30dB from 20-40Hz. And also I have more even slope from 100-700Hz.

Maybe this is why you hear the harshness more compare to mine. As I see it, you have little or less midbass compare to beyond 1kHz.

Remember, this is WITHOUT any Pre or Post PEQ!!!

[bubbleboy76]

Better waterfalls attached, with other scales. 1/48 smoothing.

[bubbleboy76]

Hallo!

Set the waterfallgraph to 300ms in the "Time Range" In the "Controls" button and then set it to "Apply default setting". This way it is more easily to see. Have the scaling on dB axis to where you have your highest peak and 40 dB down.

Try this first.

Here it is.

[bubbleboy76]

I do not like to look at discrete mic postions like this, but I have not figured out how to do waterfall from averaged measurements in REW yet.
Move the mic 1cm and it is another curve.

[Mctwins]

Ok, thanks

I see that you have lack of bass power response and a huge suckout between 80-400Hz. Due to lack of bass the reveling in resonances will also be very hard to see.

For comparison, please set the scaling to 20Hz-20kHz and to 500ms. And keep the default settings.

Here is the upstairs rooms freq and waterfall graphs with the Cinema Screen Arrays. You see how dense the waterfallgraph is!

[bubbleboy76]

Hallo!

This is my response at LP with 4365. I have around 12dB down from 20-30Hz. You have 30dB from 20-40Hz. And also I have more even slope from 100-700Hz.

Maybe this is why you hear the harshness more compare to mine. As I see it, you have little or less midbass compare to beyond 1kHz.

Remember, this is WITHOUT any Pre or Post PEQ!!!

1/3 smoothing here? How does it look with 1/6 and 1/48?

[Mctwins]

1/3 smoothing here? How does it look with 1/6 and 1/48?

I was waiting for you to ask this question.

We don't hear above 1/6 octave band.

The Yellow=1/3
The Red =1/6
The Green =1/24

And the scaling can be changed to look better.:)

[Mctwins]

In ARTA there is no 1/48 octave band.

[bubbleboy76]

Here is my curve with 1/3 smooting.

[bubbleboy76]

Average over 3 listening positions.
1/3 smoothing.

[bubbleboy76]

As always, I disagree with almost every word you write mctwins :)
I am not sure if I am Superman or lex Luthor here, and I do not care :)
And I will not argue with you, there is nothing in it.

[Ed Zeppeli]

We don't hear above 1/6 octave band.



What does this mean exactly?

Surely there are 12 audible notes in an octave.

[ivica]

What does this mean exactly? Surely there are 12 audible notes in an octave.

Hi Ed Zeppeli,

May something more can be find in the:

http://www.diyaudio.com/forums/multi-way/103872-geddes-waveguides-745.html#post4463702

http://www2.ling.su.se/staff/hartmut/bark.htm#conv
and
http://www2.ling.su.se/staff/hartmut/cb_tbl.htm#bark
so can be seen that Q about 6 corresponds to critical bandwidth

regards
ivica

[bubbleboy76]

So 1/6 smoothing corresponds to human hearing granularity?

If that is the case, we can all enjoy nicer looking curves then :)

[bubbleboy76]

A graph from my measurements displaying the vertical off-axis response of 476mg on H4365.
1/6 smoothing.
Mic 1m from horn.
Calibrated mic.
No filter except protection cap, and BW 12db/oct highpass at 600Hz for protection.

At 8KHz it starts to beam a bit. Maybe more than expected?

[ivica]

A graph from my measurements displaying the vertical off-axis response of 476mg on H4365.
1/6 smoothing.
Mic 1m from horn.
Calibrated mic.
No filter except protection cap, and BW 12db/oct highpass at 600Hz for protection.

At 8KHz it starts to beam a bit. Maybe more than expected?

Hi bubbleboy76,

It seems to me that the response +/- 40 deg is very good, but using from the 600Hz is too low, I believe.

regards
ivica

[bubbleboy76]

Hi bubbleboy76,

It seems to me that the response +/- 40 deg is very good, but using from the 600Hz is too low, I believe.

regards
ivica

We crossover at 650Hz. Linkwitz-riley 48dB/oct acoustical. Do you think it is too low?
In my tests it sounds better with low crossover than high. I think 476mg/horn is better than 1501fe in that region.
The diagram was without any EQ or crossover (except protection).

[bubbleboy76]

This is my woofer measured from 2cm distance.
Amazingly flat to me.
No smoothing at all.
No EQ or passive filter.

[bubbleboy76]

This is the same vertical off-axis measurements for the woofer, as comparison to the horn I posted previously.
I do not know if it is useful for anything though!
1/6 smoothing.

[ivica]

We crossover at 650Hz. Linkwitz-riley 48dB/oct acoustical. Do you think it is too low?
In my tests it sounds better with low crossover than high. I think 476mg/horn is better than 1501fe in that region.
The diagram was without any EQ or crossover (except protection).

Hi bubbleboy76,

As I have remembered diaphragm displacement (Xd) is proportional to Xd ~ 1/f2
so the displacement for 800Hz would be about 1.5 TIMES SMALLER then at 650Hz as, (800/650)2 = 1.515,
and the 'problem' become larger as music material usually has larger amplitude in the 650Hz region then at 800Hz,
and if it is expected that the voltage amplitude become smaller with the frequency rise about 1/f then, the difference
would be proportional ~1/f3, what would give us that it can be expected that the diaphragm displacement relation would be
(800/650)3=1,86 times larger at 650Hz network, then at 800Hz network, almost TWICE.

Fortunately, at usual home listening levels, using high efficiency driver, the mentioned calculation is not so emphasized.
One think has to be 'in mind' that JBL 'diamond shape diaphragm suspension' is more sensitive to the large diaphragm excursion.
I have no idea about the used horn driver loading is applied , but usually short horn (the distance from the driver "phase plug mouth" to the "horn mouth")
does not produce good 'driver loading' under 1000Hz (what is "the problem of the OSWG" horn wall surface flare,

http://www.audioheritage.org/vbulletin/showthread.php?36154-OSWG-Acoustic-Impedance&p=366503&viewfull=1#post366503
http://www.audioheritage.org/vbulletin/showthread.php?36154-OSWG-Acoustic-Impedance&p=366614&viewfull=1#post366614


and that is why Mr.G.T.- I believe, in his design has applied a "short horn" part at his horn design at the horn throat,
as can be seen on 4365,design DD66000, 9800,.....even that ca be seen on M2 horn ).
On Your measurements such not so emphasized FR response of the driver&horn combo under 1000Hz can be seen too,
reaching maximum at around 1.5kHz ~ 2kHz.

regards
ivica

[bubbleboy76]

pos told me that our steep 48dB/oct crossover slope at 650Hz would be safer, and able to handle more power, than the stock network (that is 24dB/oct slope at 750Hz I think).

[ivica]

pos told me that our steep 48dB/oct crossover slope at 650Hz would be safer, and able to handle more power, than the stock network (that is 24dB/oct slope at 750Hz I think).

Hi bubbleboy76,

As POS told You, he is very, very good in that field, so I believe that it would be OK.
I have just said what has to be aware of, when using compression drivers under lower frequency range....
as You have really nice and worthy 476 Mg driver.

regards
ivica

[Ian Mackenzie]

If you go lower than the prior system crossover point the verticel directivity of the horn may well increase due to loss of pattern control.

This can be a good or a bad thing depending on the acoustic/ electrical phase of the drivers at the crossover point.

One thing is for sure the vertical lobes will change and the phase compensation in the crossover point will require optimisation. That adds up to less/more deviation in the power response in the listening window around the crossover point.

I am not sure is you can do with your active crossover it but if the horn has an acoustic roll off of 12 do you could use a LR 24 network and optimise the acoustic response for 6th order LR.

Then apply a 6th order low pass filter on the woofer. This will facility less deviation in the power response of the listening window. That equates to better sound reproduction according to Floyd.

Applying it would be interesting but you would need to test the system extensively after getting the basic crossover slopes to ensure correct phase response by reversing the drivers phase.

This will produce a notch in the crossover point on axis and around the power response and I'd best done with several measurements.

The trick is to get the on axis forward lobe centred directly at the listening window so as to provide a smooth power response around the listening window.

You can't do this by just doing a attempting to get a smooth crossover response in phase so try reversing the driver phase.

I have simulated 40 db notch in phase reversal in LEAP will careful adjustment of the driver offset.

Then you might gain some of the pseudo performance of the M2 in the midrange.

[bubbleboy76]

Thanks for the input Ian.
I beleive pos is on top of this, and more... :)
I hope he will answer to this when he is out of jetlag!
In post 192 you can see we did the phase reversal thingy:
http://www.audioheritage.org/vbulletin/showthread.php?36940-JBL-4365-Active-Crossover-settings&p=381777&viewfull=1#post381777

[bubbleboy76]

Currently, reading the Floyd E. Toole book about reproduction... :)
I am half-way through it, and I am more and more interested in trying to add the 045ti-1 to the mix, for wider dispersion 8-20KHz.
Let's see what 2016 brings :)

[4313B]

Well the nice thing is that you can add it in at any time if you simply do it like G.T. did in any of those systems. You should be able to find the various schematics for the high pass passive filters and they are easy enough to build. I have to admit that when I walk into a room without them the dispersion pattern from the large horns is apparent. Not an issue when one sits down in the general field of the listening position but certainly noticeable when standing up in the periphery. The M2 waveguide isn't nearly as noticeable but the high end "pinching" is still there. The newest waveguides I am told don't require additional UHF transducers at all.

[bubbleboy76]

Well the nice thing is that you can add it in at any time if you simply do it like G.T. did in any of those systems. You should be able to find the various schematics for the high pass passive filters and they are easy enough to build. I have to admit that when I walk into a room without them the dispersion pattern from the large horns is apparent. Not an issue when one sits down in the general field of the listening position but certainly noticeable when standing up in the periphery. The M2 waveguide isn't nearly as noticeable but the high end "pinching" is still there. The newest waveguides I am told don't require additional UHF transducers at all.

That is a simple way to do it. But... I remember the severe UHF comb-filtering in the anechoic measurement of the 4365.
The stock 4365 was painful to the ears at high volume, in my setup.
I do not know if you remember my little conspiracy theory about the "detuned" 4365 filter? ;)
If I do it, I will do it the way pos advices me to do it.

[Mctwins]

Well the nice thing is that you can add it in at any time if you simply do it like G.T. did in any of those systems. You should be able to find the various schematics for the high pass passive filters and they are easy enough to build. I have to admit that when I walk into a room without them the dispersion pattern from the large horns is apparent. Not an issue when one sits down in the general field of the listening position but certainly noticeable when standing up in the periphery. The M2 waveguide isn't nearly as noticeable but the high end "pinching" is still there. The newest waveguides I am told don't require additional UHF transducers at all.

The newest waveguide.....are you talking about 4367 or M2? One can adjust the UHF in 4367.

[Mctwins]

That is a simple way to do it. But... I remember the severe UHF comb-filtering in the anechoic measurement of the 4365.
The stock 4365 was painful to the ears at high volume, in my setup.
I do not know if you remember my little conspiracy theory about the "detuned" 4365 filter? ;)
If I do it, I will do it the way pos advices me to do it.

I don't have any pain listening to high volume in my 4365..:confused:

[4313B]

The newest waveguide.....are you talking about 4367 or M2? One can adjust the UHF in 4367.Newer. Not released yet.

[bubbleboy76]

I don't have any pain listening to high volume in my 4365..:confused:

I do not doubt that. I have seen very rolled-off UHF measurements on gearslutz-forum from your system, if I remember correctly.
Maybe your room takes the edge off?

[Mctwins]

Newer. Not released yet.

Aha...cool:)

[Mctwins]

I do not doubt that. I have seen very rolled-off UHF measurements on gearslutz-forum from your system, if I remember correctly.
Maybe your room takes the edge off?

What measurement are you refering to, 4365 or Cinema Screen Array?

[bubbleboy76]

What measurement are you refering to, 4365 or Cinema Screen Array?

Don't know. It was your thread on gearslutz.

[bubbleboy76]

What is the appropriate protection-cap value for the 045ti-1?
It is 4ohm, and crossover will be somewhere 8-15KHz.
I have 6.8uf and 0.68uf caps laying around, can they be used?

[pos]

If it is like the 045Be it has its fs around 4.5kHz, so you need to aim for a corner frequency lower than that to avoid having that big impedance spike messing up the response. The other solution is to use an lpad to somewhat linearize the phase beforehand.

http://www.audioheritage.org/vbulletin/showthread.php?33558-045Mg-045Be&p=339108&viewfull=1#post339108
(any pointer to the curves for the 045Ti ?)

Aiming for a 2kHz corner frequency looks to be a reasonable solution here, and that would translate into a 20uF cap.

[Mctwins]

Don't know. It was your thread on gearslutz.

Hallo!

All the measurements in Gearslutz is at the listening position, around 5meters and both speakers fireing. So, of course the HF is rolling off. To me, I can be wrong here, UHF is beyond 20000Hz.

We are not testing how the loudspeakers is performing, we trust fully in the JBL engineering department, but more how the room is performing.:)

[bubbleboy76]

The newest waveguides I am told don't require additional UHF transducers at all.

Is this the new waveguides?

http://www.jbl.com/arena-series/

http://www.jbl.com/loudspeakers/CONTROL+X.html?cgid=loudspeakers&dwvar_CONTROL%20X_color=Black#start=1

http://www.jbl.com/loudspeakers/STUDIO+220.html?cgid=loudspeakers&dwvar_STUDIO%20220_color=Black#start=1

[bubbleboy76]

"If they sounded like this stock, JBL would sell miliions!"

Quote from my friend after hearing my system with this active setup for the first time.
He is an old JBL fan, and tricked me into being one, once upon a time.

[1audiohack]

"If they sounded like this stock, JBL would sell miliions!"...

You're killing me! Grrrrr.

Barry.

[bubbleboy76]

You're killing me! Grrrrr.

Barry.

It's the extra bass that got him going, I think.
And you have subwoofer with yours, right?

[1audiohack]

I do run a pair of Array 1500's with them.

I just know what you can get with high quality DSP and getting some of the passive stuff out of the way. It's a fight to leave anything alone.

Barry.

[bubbleboy76]

My current 4365 active 2-way dsp settings with a 650Hz 48dB/oct‏ Linkwitz-Riley acoustical crossover.

HF :
68uf Jantzen cross-cap
high-pass filter Butterworth 24dB/oct 810Hz
-4.3dB Q=5.0 520Hz
+2.3dB Q=4.0 740Hz
-12dB Q=1.0 1600Hz
-2dB Q=3.0 4200Hz
-1dB Q=2.5 6100Hz
+3dB Q=1.0 11kHz
-3.3dB Q=2.0 16000Hz

LF :
low-pass filter Butterworth 36dB/oct 600Hz
Delay 0.5ms (valid at 1m)

This is the settings I have gotten from POS, based on my measurements.

Then, I have added a shelf in the bass to my taste. Low-shelf starting at 135Hz, +8dB, 4.5dB/oct slope.
I also have low-shelf starting at 30Hz, 6dB/oct slope, down to -12dB, to not put unnecessary power below the port tuning.

Barry, try it you might like it... :)
But, you have to measure and set HF/LF relative gain levels, phase, and delay, correct according to your own system.

Attached are also POS "rephase" phase linearization files that can be used with MiniDSP OpenDRC. One correcting only the crossover, and one correcting crossover+bass reflex port.

[Ed Zeppeli]

Thanks for this thread.

I would appreciate if, in a nutshell, it could be explained to me how to replicate a given voltage drive plot via DSP. I am using the dbx venu360 and would like to get as close as possible to the given curve using what I have at hand.

Is it all done within the HF/LP settings of the crossover or are there also eq points to consider?

Thanks much,

Warren

[grumpy]

Almost always both. I don't have a specific recipe for you though.

It really helps to have a way to measure the output of the crossover/EQ
though (often the plots pictured in a DSP setting GUI are not of
sufficient resolution or don't show the crossover and EQ functions summed).

I'd start with crossovers that were close to the target voltage drive curves
and try fairly low Q (not narrow band) corrections where needed, and avoid
large gains/cuts. Compression driver + horn eq being possible exceptions
(e.g., constant directivity horns have an expected correction)

[Ruediger]

Thanks for this thread.

I would appreciate if, in a nutshell, it could be explained to me how to replicate a given voltage drive plot via DSP. I am using the dbx venu360 and would like to get as close as possible to the given curve using what I have at hand.

Is it all done within the HF/LP settings of the crossover or are there also eq points to consider?

Thanks much,

Warren

A voltage drive plot calculated with constant loads (e.g. 8 Ohms resistors) won't help you much to get started. You should have the real drivers in place and should measure the effective voltage at the driver terminals.

I don't know the dbx venue 360. The Behringer DEQ 2496 can help you to find an equalizing curve in a few minutes. Maybe dbx can do something similar, no idea. The Behringer doesn't cost a fortune.

Ruediger

[Ed Zeppeli]

It really helps to have a way to measure the output of the crossover/EQ
though (often the plots pictured in a DSP setting GUI are not of
sufficient resolution or don't show the crossover and EQ functions summed).

I'd start with crossovers that were close to the target voltage drive curves
and try fairly low Q (not narrow band) corrections where needed, and avoid
large gains/cuts. Compression driver + horn eq being possible exceptions
(e.g., constant directivity horns have an expected correction)

Thanks grumpy. So I can create the general slopes with the hi and low pass filters but then I add the dips and moguls via EQ used as broadly as possible. This is making more sense to me now.

So, having a way to verify these tweaks electrically and not acoustically is truly the only way to confirm that I have duplicated the voltage drives as intended by the designer. It would be a helluva lot of work with a multimeter and sine waves. :)


A voltage drive plot calculated with constant loads (e.g. 8 Ohms resistors) won't help you much to get started. You should have the real drivers in place and should measure the effective voltage at the driver terminals.

I don't know the dbx venue 360. The Behringer DEQ 2496 can help you to find an equalizing curve in a few minutes. Maybe dbx can do something similar, no idea. The Behringer doesn't cost a fortune.

Ruediger

Yes, I have the drivers in place and am now just seeing how close my crossover is to the designer's. I have lots of capabilities to get things relatively flat within the dbx and REW but is flat the only concern here?



This is what I'm trying to replicate.

75313

I've hunted through various posts and literature and discovered that Mr. Timbers used a linkwitz crossover centred at 750hz. There appears to be a 24db/octave drop on the woofer but there is a much gentler slope on the horn module. It looks to be about 6db/octave. I should definitely be playing with that slope within the venu360 as I had just done both with a 24db/octave and then tweaked things to flat with various parametrics.

Also to throw a wrench into the works is the fact that I'm using a 2234 woofer as opposed to the LE14H-3.

Thanks for the information.

Warren

[grumpy]

Also, keep in mind that some of these kindly supplied system bi-amp voltage-drive curves assume that the passive HF crossover elements in the cabinets remain in-circuit. In this case, additional rolloff and eq. With the driver horn acoustic rolloff also contributing to the final HF output curve. Guessing the result would give a rolloff complementary to the LF.

I haven't looked at it hard, but you might also approximate the LF, dual-knee curve with two series Low-pass crossovers (set for different frequencies) e.g. one 6dB and one 18.

[Earl K]

<<<SNIP>>>Yes, I have the drivers in place and am now just seeing how close my crossover is to the designer's. I have lots of capabilities to get things relatively flat within the dbx and REW but is flat the only concern here?

It's actually quite simplistic to run a line level (REW) signal into your dbx unit and then back to REW for study.

That will allow you to see what your finagled & finessed curves look like.

If you get hip to using a graph tracer ( & then saving the traced results to either a .txt or .frd file ) you can import your target curve into REW ( eg; the one you just posted ).

Then you can overlay ( within REW ) your measured dbx filter efforts over your target curve ( to see how you are doing ).

:)

PS; Apparently this tracer http://labs.fprawn.com/wp-content/uploads/2015/01/Screen-Shot-2015-01-01-at-4.27.05-PM-300x198.png (http://labs.fprawn.com/index.php/fpgraphtracer/) is quite good.

[Ed Zeppeli]

Cool info Earl. Thanks. That will be very useful.

[Ian Mackenzie]

Hi Warren

The 24 db slope is the final acoustic slope.

However the final tuning of the filter slope will require you reverse the phase of the horn and adjust the anti phase notch in the response manually for the deepest uniform notch at the design point ie on axis with the horn.

Attempting to adjust the filter while the drivers are in phase will not result in the drivers being in phase at the crossover point. Ie it won't perform correctly.

There are references to this by respected people on the internet.

[Ed Zeppeli]

Had some time today.

It looks like I may be fighting the internal SAM1HF crossover in my attempts to get a gentler slope. Though I have yet to measure the electrical output of the dbx as per Earl K's suggestion I did import the voltage curve using the trace program.

Measured from 1 metre. Blue Line is LW24db/octave high pass at 750 hz and Green line is LW12db/octave at 1.82khz. As you can see, both dbx crossover settings result in a steep drop at around 600Hz which I can only attribute to the internal SAM network. The purple curve is the imported voltage drive.

75329

So, either I dig in hard and modify the low pass of the internal SAM network, live with the system as is and just tune the crossover and room (which I've been quite happy with so far) or go ahead and purchase the HF section of the 1400 Array networks from Synthesis. Last I checked these were in stock for around $95 each.

I suppose there is also the charge coupled option as well which is fairly well-documented here.

:hmm::hmm:

Having fun experimenting, so it's all good here!

[JeffW]

Came across the JBL voltage drive jig sheet

http://www.jblproservice.com/pdf/Network%20Schematics/Standard%20Test%20Fixture.pdf