PDA

View Full Version : time delay problem JBL 4343 (3143 network)



B&KMan
10-31-2004, 12:19 PM
Time delay problem JBL 4343 (3143 network)

Hello fantastik group !!

(sorry for my poor english)

Recently, I run test calibration with my analyser for check spectrum response on driver.

Big surprise result the time delay integration is little false on cross-over 3143.
The time delay is 0,671 MS more faster for the 2405 for th rest of the drivers.

Well this effect is not change pretty well the sprectrum response but ruin the phase match...

Do you have any information for modification, repair or upgrade cross over for fix this effect ??

(I attache too small pict where the phenomenon is produce.)
thanks

boputnam
10-31-2004, 01:09 PM
Originally posted by B&KMan
Do you have any information for modification, repair or upgrade cross over for fix this effect ?? Well, I ain't no expert in this, but one solution would be to take the 3143 out of the signal path, and go with an external four-way active crossover and time-delay the four signal components until you get the result you need.

Or, just have another glass of fine red wine, and sit back and enjoy the pleasures of vintage JBL engineering! :cheers:

B&KMan
10-31-2004, 01:18 PM
I hope found the trick for sending better pict in regards with the limitation forum.
:)

B&KMan
10-31-2004, 01:24 PM
Originally posted by boputnam
Or, just have another glass of fine red wine, and sit back and enjoy the pleasures of vintage JBL engineering! :cheers:

Yes this second option is particulary work : the acohol have a hight fading effect in high frecuency perception...

:cheers: and :cheers: again...

:)

GordonW
10-31-2004, 08:17 PM
I'm just wanting to be clear here- so, the bigger of those two spikes on the time response curve, is ALL THREE of the other drivers in the 4343 stacked up on top of each other, ie, the midrange horn, the midbass and the woofer? If so, that's pretty remarkable... that would very consistent response from drivers of different mechanical acoustic centers.

My suggestion, might be to seperate JUST the 2405 from the other three drivers- run a tri-amp setup- with the 2231/2235 woofer on one amp, the 10" midbass and mid horn on one amp, and the 2405 on a third amp. Then, use a digital time delay, to delay the 2405 .67ms. That shouldn't be too terribly hard- all you'd have to do, is seperate the 2405's crossover section from the rest, and just run another amp from a time delay unit for just that tweeter and crossover...

B&KMan
10-31-2004, 11:21 PM
Originally posted by GordonW
I'm just wanting to be clear here- so, the bigger of those two spikes on the time response curve, is ALL THREE of the other drivers in the 4343 stacked up on top of each other, ie, the midrange horn, the midbass and the woofer? If so, that's pretty remarkable... that would very consistent response from drivers of different mechanical acoustic centers.


:yes:
YUP, the rest of the driver is very consistent time delay integration. But other hand the lower frequency is low = long in time so, the impulse test response on lower frequency is too long for test as the same rate speed.

But my many test with corrolaire speed vs frequency is expose excellent integration for the rest of driver... Superbe job. :rockon2:

but if the engeener is macht delay of the HF why is not ability to match UHF ??? the rest of picture is too clean for I accept this delay... :hmm:
---
If I calculate the distance to backward UHF driver for delay alignment: it is +- equal to position of the HF diaphram position (back 22 cm) (sic)

Unfortunately I'm not very good in electronic theorie and I have difficult to understand if my cross-over (3143) is busted in one component or is a error design... or mismatch connection inside cross-over...

but this problem is same at two monitor in exact same time delay result.

this time delay it is consider small or realy problematic ???

Ian Mackenzie
11-01-2004, 03:51 AM
JBL has previously published a grou delay for this and other monitors. They claimed its group delay falls within the accepted limits of human perception (ref Sound Engineering John Eargle)

Ian;)

B&KMan
11-01-2004, 07:51 AM
Originally posted by Ian Mackenzie
JBL has previously published a grou delay for this and other monitors. They claimed its group delay falls within the accepted limits of human perception (ref Sound Engineering John Eargle)

Ian;)


WOW !!! you have a big and precious conncection :spin:

It is possible to give the link or exact reference of paper or copie paper

please please please

I'M shure anybody is real interestd by this paper issue.
;)

Guido
11-01-2004, 10:53 AM
Originally posted by boputnam
Or, just have another glass of fine red wine, and sit back and enjoy the pleasures of vintage JBL engineering! :cheers:

Bo, :cheers:

Ian Mackenzie
11-01-2004, 11:59 AM
Originally posted by B&KMan
WOW !!! you have a big and precious conncection :spin:

It is possible to give the link or exact reference of paper or copie paper

please please please

I'M shure anybody is real interestd by this paper issue.
;)

Are you kidding.

Go buy the book on Amazon, millions of copies sold

ISBN: 0442222211

My copy is currently on loan to Kent English Passlabs

Ian

boputnam
11-01-2004, 12:36 PM
Originally posted by Ian Mackenzie
(ref Sound Engineering John Eargle)
Sound Recording by John Eargle - on Amazon:

http://www.amazon.com/exec/obidos/tg/detail/-/0442222211/qid=1099341256/sr=1-9/ref=sr_1_9/104-8350722-3283143?v=glance&s=books

My copy is way too valuable to loan-out to any hi-falootin' audio wizard in the mystical Sierra... :rotfl:

Ian Mackenzie
11-01-2004, 01:53 PM
My copy is way too valuable to loan-out to any hi-falootin' audio wizard in the mystical Sierra...

I find sharing knowledge the most valuable aspect of this interest.

Those who don't stand dormant.

The question is Bo do you understand and have learnt from the John Eargle text or does it sit gathering dust in the bookcase.

"For those unaware Kent is a physicist formerly of the nuclear energy commission and one of the nicer people I met in my travels and the least I could offer for his hosting & friendship was a permanent loaner of my in-flight reading. Besides I plan to order the current edition soon enough."


Ian:D

B&KMan
11-01-2004, 02:03 PM
thanks for your references, I attemp to buy soon...

for my side the B & K info relativ of standart question for average impulse mesure and typical response of ear is expose other face ...

Show pict...

In third face of problem the problem of builder amp is build a more fast possible response in circuitry and speaker is normally the reflect of the same problem, impulse and transient information is very very fast...

the ear is catch the difference....

boputnam
11-01-2004, 02:49 PM
Originally posted by Ian Mackenzie
The question is Bo do you understand and have learnt from the John Eargle text or does it sit gathering dust in the bookcase.
Actually no, friend - it's stuffed into my gig bags along with that excellent "Live Sound Mixing" by Australian Duncan Fry (from ARX) you gifted me on your visit! When the band (and you, too...! ;) ) wander off chasing the babes, I try and get in a little reading... :yes:

Ian Mackenzie
11-01-2004, 05:09 PM
Bo,

Well with so many talented people around you they can afford to wonder and enjoy darlings of pleasure and soil the drapes...all work and no play makes a dull boy after all.

Any way your married to a sweetie, keep your nose in the book and behave yourself.

B&K,

About the whole group delay thing, its got more ins and out than a duck's arse to measure and make an sense of, then there's the subjective debate among researchers. A number of notable and damn fine brands like Duntech has exploited this I think for marketing reasons but there are equally as many other alternatives.

I suppose it would be nice to have a perfect point source but as was explained by guru's in another thread, this can only exist for one sweet spot in time and space.

With the 43xx 4 ways the advantages far outway the minuses and I think the focus was on flat amplitude response on axis rather than absolute phase at the time, flatness of response certainly more audible than phase shift.

Ian

Tom Loizeaux
11-01-2004, 05:29 PM
Why not pull the 2405 out of the cabinet (temporarily blocking the hole), and put the 2405 on top of the cabinet. Then move it back in small increments to see if a better phase alignmemt is found? That, with bi-amping the woofers and setting any required delay on the active crossover to bring the lows in line with the rest, might help with your phase concerns.

Let us know.

Tom

B&KMan
11-01-2004, 06:20 PM
Originally posted by Ian Mackenzie

I suppose it would be nice to have a perfect point source but as was explained by guru's in another thread, this can only exist for one sweet spot in time and space.

Ian

Bonjour,

well, yes and no this test is performed in couple of placement on axis and 30 axis with different lenght.

all result expose the same problem....

the point source of off time...


Originally posted by Ian Mackenzie

I think the focus was on flat amplitude response on axis rather than absolute phase at the time, flatness of response certainly more audible than phase shift.
Ian

the problem is not the spectrum response because in this question the response is not problem. The problem is the transient response and harmonic response. Floue, blur, fuuzy response. And the spacious response afected.

the first attack is drop over 20 DB before the second is start. the integration of HF & UHF is false but each peak analysing separately expose good phase response...

My litterature is explain the human ear is particulary sensible to phase shift... but many university recheard trace the line exact the variation of one is more perceive by variation of other...

I understand this monitor is a great. (I buy fo this reason)
But over 25 years of research if run foward many improvement. theS9800 is a astronoshing exemple of that...

Anybody have a complete spec of the cross-over 3143: slope, response driver, real cut frequency ???
Maybe I have a batch modified or wrong build...

B&KMan
11-01-2004, 06:24 PM
Originally posted by Tom Loizeaux
Why not pull the 2405 out of the cabinet (temporarily blocking the hole), and put the 2405 on top of the cabinet. Then move it back in small increments to see if a better phase alignmemt is found? That, with bi-amping the woofers and setting any required delay on the active crossover to bring the lows in line with the rest, might help with your phase concerns.

Let us know.

Tom

Yeah this is the simple solution : put out driver and go back with in time integration is perfect but I start other big problem, blend with HF, stable fixing, space for good dispertion., cable, and E.T. look...

If you are any more tips in this way let me know...

:cool:

Ian Mackenzie
11-01-2004, 07:28 PM
B&K,

Can you post the time impulse to include scales and mark the UHF peak? so easier to read.

I have previously done near field time and response of individual drivers in 4345 4 way. I recall when only one driver being tested the time response was much cleaner.

It comes down to interpreting what the test results are telling us.
What are the peaks, what is causing them? What time gating are you using?

I might be able to repost my results if you are interested?
But I must have glass of Red first.

Ian

B&KMan
11-01-2004, 08:25 PM
Originally posted by Ian Mackenzie
B&K,

Can you post the time impulse to include scales and mark the UHF peak? so easier to read.

I have previously done near field time and response of individual drivers in 4345 4 way. I recall when only one driver being tested the time response was much cleaner.

It comes down to interpreting what the test results are telling us.
What are the peaks, what is causing them? What time gating are you using?

I might be able to repost my results if you are interested?
But I must have glass of Red first.

Ian

Well, eastern time is 23:10 and I tried to give info soon.

For little explain the dual channel analyser is syncronised with internal generator... the gate is simple max peak detector - 10% is applied on channel A
The generator is connected with T bnc of channel A and channel where test interest. the channel B is receive the response via preamp an mic.

The rectangle weighting is applied on this impulse signal at channel A and B

Because time delay is automatiuely non syncro the analyser permit to delay response on each channel for perfect syncronisation.

Finally the original response is compared by the second with completely automated taste and response H1 is produce. well in this case this syncronisation is not critical for expose the 2 peaks. But for complete analyse it is important...

But for determine what is what the analyser accept to weighting transient window. In this any time and size of time is manual calibrated by user and reveal by analysing function.

The first step is the response of UHF.

My second pict expose the delta x is intervale and delta y is the difference in Db of the peak comparison.


euh it is too hermeneutic explanation?? :moon:

I have no problem to send more pict but long process in this forum restriction... let me know.

and I open to send pict for your test.

the exchange build concrete of this forum :bouncy:

B&KMan
11-01-2004, 09:34 PM
I hope this more picture is help more your mind to realize the phenomenon.

this pict is ruff analyse of the fisrt peak.

B&KMan
11-01-2004, 09:36 PM
this pict is ruff analyse of the second peak.

B&KMan
11-01-2004, 09:47 PM
Finally this pict explain mesure set -up
each Line of time analyse is value of 15.3 US time.

Sorry for poor pict.

It is better comprehensive ??? :hmm:

oups it is time for me to replenish the second solution (red wine)

:cheers:


p.s. 1 --- the signal generator is one pure positive square pulse.
the spectrum response is a line !! If you have more curious I send pict of this...

P.S. 2--- it is easy to buid spectrum in the same way with near field mic.
I have no problem to buid this analyse too if you interest. And Yes the UF is go up to over the cut off frequency cross-over.

Ian Mackenzie
11-01-2004, 11:48 PM
Very interesting and clever.

Here is my PC analyser reference pulse.

Ian

Ian Mackenzie
11-01-2004, 11:51 PM
This one the UHF only.

Near field .

There is some blurr on the tail edge, this could be the system, noise from the amps ect or possible ringing from the crossover.

Ian

Ian Mackenzie
11-02-2004, 12:05 AM
The UHF and Horn .

The combined effect of the delays and other crap shown here.

I have no doubt with better equipment I could evaluate this better. Third order filters do suffer poorer transient performance than say first order so it is likely there is some transient distortion. But whether that's audible over other compromises with these designs is another issue.

In some respects the 4430/35 has a better crossover characteristic as it has miminal group delay according to the white paper. But that ain't the whole story.

B&K, elsewhere our member 4343mod has used a modifed 1st order passive network for the 4343 and claims startling results although this would not solve the time domain issues of the UHF driver. No doubting it would sound different but such a dramatic change to the network may create other compromises such as dispersion lobes and limited power handling in some circumstances.

Ian

B&KMan
11-02-2004, 07:06 AM
Thanks very mch for info too.

cool stuff.

What exactly rate intervale ??
the little inconsistence is origine of the type of weighting filter for this type of input signal. Normally the hanning is overal best filter but for pulse it is preferable to run in rectangular filter for rectangular signal ... (according to many tech note on Bruel & kjaer. but maby is different for this software...)

Do you have possibillity to expose the imaginary part of signal ??

This is my pict of my UHF

Ian Mackenzie
11-02-2004, 11:49 AM
What exactly rate intervale ?? The sample length is defined by the control panel as is the sample length.

I will check the authors web page for more details, this is a very simple package that used a soundcard to generate the pulse and do the FFT functions. Tim G has a better package , perhaps he is following this thread he can run a similar test.

Ian

Bill Shenefelt
03-14-2007, 10:34 AM
Might another approach be to run the horn and slot radiator on a single amp and the midbass on a second one and the bass on a third? The way the physical configuration is right now is with the 15 inch in a 6 cu ft enclosure like an L300 laying on its side (It was the same box I had used in the home built L300) A second box about 15 inches square on the front face contains the 10 inch midrange. It just lays on the bass box. The face of the box for the 10 inch is extended vertically upward and free air mounted to that face are the horn and slot radiator. These could be moved rather than using an electronic delay.
My main problem right now is that after I changed from the 3 way design of the L300 to the 4 way design incorporating the 10 inch midbass the system sounded "warmer" but lost the great punch it once had. I thought a mismatch of the crossover from 15 inch to 10 inch was the culpret. The 15 could no longer go up to 800 cps and possibly provide the leading edge of the waveform. This is just what I imagine to be happening.
I do have coming to me a 3 way crossover from Marchand currently configured for either a 300 or 400 cps 24/ocatve for the cone speakers and provisions for adding a 1200 cps crossover for the cone to horn interface. Looking for the best way to set it up right now. I want to recover the nice punch I had with the L300 setup, yet keep the warmth I get with the 10 inch. All drivers and passive crossover design and parts are from the 4343 but I have no 4343 15 to 10 inch cone passive and wanted to use active there if beneficial. With my age and ears I cant hear much from the slot readiator anyway.


I'm just wanting to be clear here- so, the bigger of those two spikes on the time response curve, is ALL THREE of the other drivers in the 4343 stacked up on top of each other, ie, the midrange horn, the midbass and the woofer? If so, that's pretty remarkable... that would very consistent response from drivers of different mechanical acoustic centers.

My suggestion, might be to seperate JUST the 2405 from the other three drivers- run a tri-amp setup- with the 2231/2235 woofer on one amp, the 10" midbass and mid horn on one amp, and the 2405 on a third amp. Then, use a digital time delay, to delay the 2405 .67ms. That shouldn't be too terribly hard- all you'd have to do, is seperate the 2405's crossover section from the rest, and just run another amp from a time delay unit for just that tweeter and crossover...

Chas
03-14-2007, 11:23 AM
Bill, this should not be a problem. Per my PM earlier today, I moved from three ways to four way 4345's and certainly found nothing missing in this department.

If I recall correctly, the 2122H's in my 4345's need to be installed in 0.5 cu.ft. enclosures for optimal transient response. Mine have a light fibreglass fill, covering the rear and all sides. I think it's about 0.5 to 1" thick. What box volume and fill are you using?

Bill Shenefelt
03-20-2007, 10:09 AM
My suggestion, might be to seperate JUST the 2405 from the other three drivers- run a tri-amp setup- with the 2231/2235 woofer on one amp, the 10" midbass and mid horn on one amp, and the 2405 on a third amp. Then, use a digital time delay, to delay the 2405 .67ms. That shouldn't be too terribly hard- all you'd have to do, is seperate the 2405's crossover section from the rest, and just run another amp from a time delay unit for just that tweeter and crossover...

With the active substituted for the passive 077(2405) using a 24/octave at 300 and the 077 at 9000(?) with 24/0ctave is that the delay needed or is it different. I would think the phase would shift and the delay would be quite different compared tog using 24/octave. I can use whatever is best in tthe active 3 way and get is set for the needed delay. Physical location of the drivers would suggest to delay the top and bottom bands to move them back to the location of the horn output, not the other way around or is the target to get summed outputs not time alignment? . Ellectrical stuff always did not make sense to me like mechanical stuff did. Even looking at the schematic on the 3143, what are the driver crossover slopes. I have a tough time discerning zobels from hi from low in the bandpass network for the horn. The 10 inch and slot radiator crossovers look like 12/octave but the horn (12 at 1200 and 6 at 9000????)