PDA

View Full Version : Tricky mic position create random response mesure



B&KMan
06-20-2005, 07:28 PM
Hi everybody,

After small controverse in regards of position mic for evaluation calibration SPL, I perform for members, a small experience of danger of mic in near field for coherent response and amplitude value.

I put my microphone in near field : exactly same space mic-driver of all 3 mesures but just keep mesure in arroud top edge of driver center of driver and low edge of driver...

Ok if you look pict the change position is really small maybe 2 inch for all 3 position in vertical and no in deep and no in lateral...

:D

B&KMan
06-20-2005, 07:31 PM
In according of picts the result of 3 mesures is catastrophic random response in frequences and amplitude...

conclusion at this closeup all positions is too critical for good evaluation and reproductible datas. each 1/16 inch back, angle, down or side affect strongly mesure... do not put your SLM or mic too close... inside of 2 feets the close-up is too critical and push a complexe procedure for extraction of fairs stable datas.

I hope this help to better calibration..

:D




Jean.

B&KMan
06-20-2005, 07:39 PM
So At frq 16K and at exactly same power output, the difference value is more than 12DB depend just 1 inch vertical position!! at 1/8 inch higher position; differente curve and power response....

So if you put your big slm into near field for evaluate SPL add the reflextion of instrument + the type of mic, plus if you load in weight "C" or "Lin"...

Indianas jones jungles sound heres baby....

If you consider any this type of response for calibrate high or med or Low Frq, what your chance to drop notch right on the good value ??


No surprise many sceptical is claim : hey my ear is the best (youafafafaf).

:cheers:

Jean.

Zilch
06-20-2005, 07:55 PM
Oh, Jean, Jean...

I BELIEVE Ian was measuring INTEGRATED near field SPL, not FR, and he advised using a tripod such that the microphone could be precisely positioned at the desired location for most reliable determinations.

In any case, as he subsequently reiterated for clarity, the measurement and adjustment is RELATIVE, NOT absolute.

[I'm thinking OTHER things, too, but I'll just stifle them for now....] :p

johnaec
06-20-2005, 08:04 PM
I think a much more meaningful test would be at the standard 1 meter distance, which allows for normal wave development and propagation, especially with a slot type of radiator. Up so close, you're probably dealing with speaker "quantum mechanics" effects. And standard 0, 15, 30 and 45 degree offsets would seem to relate much closer to what you're actually hearing and listening for.

John

B&KMan
06-20-2005, 08:48 PM
Oh, Jean, Jean...

I BELIEVE Ian was measuring INTEGRATED near field SPL, not FR, and he advised using a tripod such that the microphone could be precisely positioned at the desired location for most reliable determinations.

In any case, as he subsequently reiterated for clarity, the measurement and adjustment is RELATIVE, NOT absolute.

[I'm thinking OTHER things, too, but I'll just stifle them for now....] :p


hey hey Zilch... talk your all your point of view with no reserve ( :D )


for spl with sinewave:

mmmmm :blink: what is interest to keep relative SLP of sinewave of 16K is any 1/16 inch change on couple Db the SPL ??? any mesure in interesting if reproductible... this is a basis of any datas...

the interest to expose the many differents fr response in DB and expose the yoyo SPL if you choose generate one frq or other...

choose any fr and check the difference in reagrds at same fr but just little difference keeping...

Look the Global spl line in white and the total of thi difference is gange more than 4 db in pink noise !! (this is result of mesure exposed )...

so the measuring INTEGRATED near field SPL in this method is very very tricky...

So maybe I have completely in field but I do not understand what is interest record arbitral value with very tricky reproduction value for the same driver and for other...

so the result is
not conform to power spectral energy of driver at all fr and at specific fr
and not according to environnemental area listening
Low reproductible value

So open my eyes and explain what good aspect is possible in this set-up :)

This is my 2+2 cents point....

Ian Mackenzie
06-21-2005, 04:43 AM
Okay,

I am now Back, the Daleks nearly got me..lucky for my sonic screwdriver I escaped!

Zilch is correct.

My reference is in effect to the design of the system with respect to relative Spl pressure response. The drivers respond to a specific spl based on sensitivity and the voltage drive applied via the filter sections. Therefore for each driver to operate at a matched spl of acoustic output from flat amplitude the voltage drive to each driver must be adjusted via the L PADs.

In order to understand this it is important to realise we are refering to net spl of each driver at a specifc descrete point. The overall flatness and integration is a function of the filter slopes, phase response and location of the drivers on the baffle. JBL have already done this design work so our JOB is quite simply to adjust the voltage drive.

After some trial measurements about 18 months ago I realised using FFT, MLS and Pulsed Analysers lacked the instringic precison for performing a measurement of net SPL based on differential of the voltage drive.

So I pulled out my trusty Tandy meter and mounted it on my tripod and fed a sine wave via a PC contolled function generator to the amplifier. By placing the meter a specific distance from the baffle in front of each driver I was able to measure with great precision the net SPL via fine adjustment of the L pads.

I used frequencies of 600, 5000 and 150000 for the midrange, horn and slot respectively.

I started with the slot and positioned the meter directly on axis about 2 inches from the baffle and wound down the pad on the horn fully and adjusted the level with the Slot Pad fully open to +4 db on the meter. I then carefully adjusted the L pad till I got 0 db, after repeating this several times I marked the postion on the foil Cal.

I then place a masking tape over the Slot mouth and repeated the same test with the horn with the meter again on axis with a reference of +4db (and wound down the midrange L pad fully) and then adjusted the horn L pad to 0 db.

The same technique was applied to the mid come using several measurements to assure precison of the +3 attenuated back 0 db level.


The location of the meter should be directly on axis but exact location is not overly critical. The reason is simple in the that we are only concerned with the net relative spl after adjustment from fully open L pad back to a desired point for 0 dba. As JBL advise this on their foil cals in Dba and the voltage drives from the crossover filters are already pre determined it is fairly straight forward to adjust the exact levels of each driver.

I think it is reasonable to assume that JBL with all their vast resources would have found the 0 dba reference point.

In fact I have found using this technique the only reliable and repeatable method of adjusting the relative balance of the drivers on a 4 way system.

Mid and far field measurements using even gated analysers do not offer the same precision for this kind of measurement.

When I then played the system it has portrays the smoothest and most remarkable coherent sound and the imaging is amazing.

I find the need for further adjustment completely un necessary

Needless to say my tongue in cheek remark (since edited) in the crossover modification thread is probably apt. It's simply a case of interpreting the available information and applying it with some simple techniques to solve a problem.

But if we come back here in few years time we will probably find people scratching their heads with the same problem.

The Doctor

B&KMan
06-21-2005, 08:34 AM
Okay,

In order to understand this it is important to realise we are refering to net spl of each driver at a specifc descrete point. The overall flatness and integration is a function of the filter slopes, phase response and location of the drivers on the baffle. JBL have already done this design work so our JOB is quite simply to adjust the voltage drive.

As JBL advise this on their foil cals in Dba and the voltage drives from the crossover filters are already pre determined it is fairly straight forward to adjust the exact levels of each driver.
I think it is reasonable to assume that JBL with all their vast resources would have found the 0 dba reference point.

In fact I have found using this technique the only reliable and repeatable method of adjusting the relative balance of the drivers on a 4 way system.
The Doctor

Hi Ian,


Ah I feeling this great member put light on this phenomenon... :applaud:

Now I understand and I realize the reliability and theorical approach...





but I'm not understand again.... :D

1--- If It Is the interest is put 3 dB (or 4 for horn) down spl why JBL put Lpad ??? just put resistance and voilą !!!

2--- Well any driver is a specific variation on regards of reference driver (this is true for the etwork too )...
But I Assume the tolerance factory is tight but not identical... So if your driver is drop 0.75 dB than reference... I assume you have necessity to upper correct with L-pad for boost current for compensate low efficience...

Assuming 2 same test of same type driver in aternativlly connected in same network:

If you put L-pad in max for driver A and for Driver B but .75 Db difference power on driver response and if you work your method you fix L-pad at same place for each and your report the exact 3 Db down response electric and acoustic but the start level is difference for A to B ??? your result is same differencial power and arrival... No ?? normally you arrival at same L-pad position but absolute SPL result is different right ???
If you add the network imprecision tolerance: the differencial result is more no ??

On other hand if this result is good in theorically and perfect in anechoic chamber ?? but it is good in any room... :blink:

:jawdrop:

Okay

3--- your method push a big question Why JBL put L-pad ?? If just for fix exactly -3 dB (4) electric power for driver... I loose probably another aspect... :blink:

4--- My litterature expose the interest of L-pad is for opportunity to place SPL and ideally spectrum fr resp in flatness possible response on room and reagards of sweet spot listening...

Because any room is different and many materials absord , diffract, diffuse, and reflexe, the necessity to tune speaker on regards of this is important...

the final result is flat spectrum as possible arround your ear , in regards of all environnemental variables...

5--- I'm not shure the perfect down 3dB acoustic response is corrolar to ideal electric balance on network... and I'm not shure if the network is better " happy" in this set-up and produce best transfert...


6--- I'm feeling the interest if this method is fix the unprecision of scale arround L-pad only...


aouch !! my skin is iritate by too scratching head...

Please doctor dont drop me :D ...


Jean.

:cheers:

Ian Mackenzie
06-21-2005, 04:06 PM
Okay,

The L Pads are there at the end of the day toi do what ever you like with them.

I don't understand your other assumptions. Assuming you follow JBL standard test rig and measure the voltage drives there should be no issues.

If you are hearing impaired, have a very dead room or a very live room you might want to adjust for a variation or like deafening skreachy sound turn the horn up full by all means. Some owners resort to using a 1/3 octave equaliser in an attempt to make the response as smooth as possible at the listening position. In small moderation this can be useful.

However if the latter is true the problem is the room and not the speaker.

In my 25 years experience with 4343's the single biggest problem with listening satisfaction is setting the L pads of all the drivers to attain a acoustically correct setting. You can spend months or even years and not get it right.

After performing the above procedure I have also measured the on axis response using MLS and obtained the smoothest overall response.

Where the response of individual drivers is uneven this is extremely difficult. Even with smoothing and you have to account for room interference and micophone location.

The method above is simple and accessable for most owners.

Go here for further information.

http://www.audioheritage.org/vbulletin/showpost.php?p=25742&postcount=141

rek50
06-21-2005, 05:17 PM
Ian, Jean, between this thread and the 4343 mod thread:blink: would you both consider getting serious about sound :D ...You both make a lot of sense in your methods. Jean, with the multiple mic positions and Ian with the Tandy. I can relate to adjusting for MORE than 25 years, as well as using an EQ. Just about the time I thought I had that "Magic" setting, I would most always get out of the chair and diddle the knobs once again. It seemed that every song or new music source would beg for a "Minor" adjustment. Finally, I boxed the EQ and let the knobs alone. NOW, I have to have a Tandy. Thanks guys....

Ian Mackenzie
06-21-2005, 05:53 PM
Rec,

That is the most sensible post about this topic I've read for a while and I appreciate your thoughts.

Asbolutely, even in the vinyl days I recall only one or two albums sounded correct, Billy Joel's 52 Street was one of them. This is another dilema of owning accurate monitors. Everything that's wrong tends to show up and only complicates the issue of tonal balance.

I do find less insanity however in knowing that the pads at least are set at a true reference point to start with.....bass and treble knobs on your amps anyone. Were'nt they a thing of the past?

The Doctor

Edit.Sorry I missed you first point.

I suppose I got serious about JBL's sounding right at the time when I heard my 4343 at my brothers place. At that point a changed out the 2344A bi radials, sold them off to Cyclotonguy and spent 12 month rebuilding a set of 4345's.

However my quest for serious sound started really when I switched to Passlabs Class A amplifiers about 4-5 years after indulging Cyclotonguy, Karen and Neslon Pass . But getting the L Pads set correctly is a good place to start.

boputnam
06-21-2005, 07:48 PM
I guess I don't get it.

Measurements at those proximities will have little bearing, IMO, on what is heard at the listener position - that is, unless things are REALLY outa phase / time alignment. There is so much sonic/acoustic reaction in the near-field, that to measure things well inside that field would seem to be ignoring real affects outside that domain. But, I'm confused...

:help:

Assuming the engineers/makers had things fairly close, FFT will reveal what is occuring at the listener position and guide your fine tuning.

B&KMan
06-21-2005, 09:33 PM
:help:



Hi,
first I dont try the Ian Method so I'm not put judgement in now...

actually the question is SPL in regards of spectral density, spectral power, power vs energy and corellation with SPL. a FFT and DFT approach and quality of each in regards of time capture ... problem of each approach, the precision of one record to average, room response, cepstrum, convolution and deconvolution, hilbert transform, Coherence, zwicker's loudness, zoom window, etc, etc, my acoustician pratice is push me work with many standart internationnal ISO and ANSI and create a path of relatively same repetitive methodologie... :blah:

But is getting too serious ;) so I equalizing my response a little bit by exemple to come of JBL Engeener...

If you buy a beautyful high level grade synthesis sytem ( 100K USD) a Jbl gentlemen representation is go at your home with biggest box....

Is extract 4 mic (1/2) pressures field (so the mic is vertical position) for best random response in quasi difuse field and connect this 4 mics at console to integrate sum signal response. The mic is different positions in height and distance of wall center, for better average... the pseudo-pulse train is produce the lattop deliver the result and calculate the compensation curve. In same time you look curve the operator decide if apply correctif , if yes the curve correction is sending at the processor sound include in system synthesis... after this another pulse train is emitted by system for look the result... this loop test is run 5 minutes, hours or days, in regards of precision and problem appear in curve response...

Many professionnal RTA (DFT approach) offert a extension module for connect 8 or 16 mic... but a millionnaire set-up !! so I keep one and call function average... so each record is sum calculation... I'M decide if I try 3, 5, 6, 9, or 100 record... of course at certain point the average difference is too small and irrelevant...

this is ISO and ansi method for calculate echo, room response and mony other parameter difficult to catch with precision...

conclusion : For non ideal environnement: the average is the key... because the position mic is all the time affect by 3 types of mode occuring in the room, axial, tangencial, and oblique.

this is the professionnal approach and recommendation, Of course now the sound intensity, and mapping sound device is exist and fix relatively this problem, but one sound intensity, analyser and calibrator is same price of synthesis ( ... )

I hope this cue is not really help many guy for fix problem but is put light on why and ... Help consideration...

Best regards.

Jean.

B&KMan
06-21-2005, 09:41 PM
Jean, would you both consider getting serious about sound .

thanks for reply,


Well, well well, the shock of idea create light and sound !!! :applaud:

I'm consider this forum with serious and many member help seriously with my network and other tips so I just return the elevator... but it is possible to joke a few time :bouncy: no ???

manys guy read this and other tecnical tips so if start on my recommendations and finish in dead end : is it seriously not respect... ;)

jean...

Ian Mackenzie
06-21-2005, 10:42 PM
I guess I don't get it.

Measurements at those proximities will have little bearing, IMO, on what is heard at the listener position - that is, unless things are REALLY outa phase / time alignment. There is so much sonic/acoustic reaction in the near-field, that to measure things well inside that field would seem to be ignoring real affects outside that domain. But, I'm confused...

:help:

Assuming the engineers/makers had things fairly close, FFT will reveal what is occuring at the listener position and guide your fine tuning.

I guess I might be able to imagine why you are confused.

Its a case of discipline of the mind to analyse specific facts out for some's desire for a dust cloud of confusion.

We are not talking about a whole Lab case study here on designing a system from stratch nor are we searching for micro changes in air density to locate zenomorph's on the edge stratasphere.

Think of this a striping down the paint layers of colour on a car till you get down to the bare metal. The top layer is the visible image but as you go deeper and deeper ...Perhaps the reconing and clean down to the bare metal is a better analogy. We know how important that is!

Anyway, as I posted originally one needs to understand the relationship b/n voltage drive and net spl. When a speaker is designed one key aspect of the design is matching driver sensitivities...we know that from Project May. But we are not talking about a two way where you can itteratively compare low and high and eventually get it right. This is s bloody 4 way!

One needs to know that the net spl of each driver is matched between all four drivers BEFORE one can start making claims to the effects of near field reflections, room reverberation and all this other crap. Phase anomallies and what happens latter on is not the point.

Attempting to look at the entire response curve of a whole system will send you cross eyed trying to work out which aspect of the curve is right., what's up, down etc and then you say okay I better boost that with equ and then you wind up with a complete mess.

This is why I am say to those particularly who are building their own systems and networks to consider the approach I outlined.

If you engage yourselves in the practicallity of such an exercise the benefits of what appears to be an over simplification become very obvious. As you turn the L Pad and the needle moves to zero 0 dba mark you will understand why you are doing this.

Studies have been done that support evidence that small amplitude broad band deviations of 2 - 3 octaves over the entire spectrum are accutely more audible as shifts in tonal balance than irregular or random peaks and dips. This is why RIAA curve accuracy is so important, a +- 0.75 db variation or less is quite audible. So obviously the better the match of the 4 driver spls the more true the system will perform, doing that at 3 metres in your room is lunacy.

And so if we apply ourselves to the task of finding a way to determine the correct level match of driver sensitivity in a multiway system it is more likely to sound correct.


The Doctor.

Mr. Widget
06-21-2005, 11:10 PM
"Studies have been done that support evidence that small amplitude broad band deviations of 2 - 3 octaves over the entire spectrum are accutely more audible as shifts in tonal balance than irregular or random peaks and dips. This is why RIAA curve accuracy is so important, a +- 0.75 db variation or less is quite audible. So obviously the better the match of the 4 driver spls the more true the system will perform, doing that at 3 metres in your room is lunacy."

I fully agree. Here is a quasi anechoic plot of a stock 4343 taken at 1.5m. The system was raised off the floor in a room with 15' ceilings. I limited the response below ~450Hz as my measurement was no longer meaningful at those frequencies. Due to severe comb filtering between the 2420 and the 2405, the response will measure differently as the mic is moved horizontally or vertically even at the distance of 1.5m.

Widget

Ian Mackenzie
06-22-2005, 12:37 AM
A sigh of relief,

Thankyou Widget for posting this response curve.

I will be on the planet of the apes (work) tomorrow so please carry on.

The Doctor:drive:

B&KMan
06-22-2005, 06:35 AM
Hi


Well I understand the relationship of net spl and electrical signal but reason of L-pad is not for this....

the reason of Lpad is for flatness response for sweet spot mixing room studio.

of course is play suberly in room house but this is a studio monitor, so asuming the nature of the thing.

It is really easy to fix relation of driver spl and electrical with high degree of precision in laboratory and determine exact resistance to put in place of l-pad and voilą home solution...

The professionnal speaker is for professionnal enviroment tool, exterior, interior, bar, disco, studio... the studio X refresh driver at 2 years and scrap speakers fast because non accurate, better new drivers performance....

and the studio is not hesitate to pay a professionnal for calibration system or buy sofisticated tool...

Of course I accourage any to do it yourself set-up, but it is important to repositionning each method and the final precision value of each...


your method is exellent but too limite at final response of the room is not evaluated... and is effect dracticly the final balance tonal...

Anyway, if you try to found a magic set-up with SLM only or sofisticated tool but one record average, your loose to much more precision... I have a big tool and I experienced this... My point is just benefit the forum to the limitation...

for members who found more tips in SLM, go in cie of electronics network (rane is realy generous in this) and dowload manual instruction : is full of tips for tune with SLM... but after each tips, all is explain the best is sofisticated tools. Is not my fault...

finnaly , this controverse have exposed the complexity of this question and I hope members is understand more in this domain, the pleasure to play with L-pad...

:cheers:

Jean.

boputnam
06-22-2005, 06:51 AM
One needs to know that the net spl of each driver is matched between all four drivers BEFORE one can start making claims to the effects of near field reflections, room reverberation and all this other crap. Phase anomallies and what happens latter on is not the point. Yea, I get that. But you missed my closing...


Assuming the engineers/makers had things fairly close, FFT will reveal what is occuring at the listener position and guide your fine tuning.I was implying this need not be necessary for a properly designed speaker, is all. But this does make hella sense for a ground-up DIY.

B&KMan
06-22-2005, 07:51 AM
Yea, I get that. But you missed my closing...

I was implying this need not be necessary for a properly designed speaker, is all. But this does make hella sense for a ground-up DIY.


EXACT !! The designer have necessity to keep response without room effect or other, but in listening approach, you deal with. 2 approachs, 2 different nature of problem, 2 goals, ...

the quasi anechoic approach is try to simulate close the anechoic room. The designer have specialized tool to fix with precision the relation electric and real SPL output. So why L-pad ?? For corrected response in regard of room... the L-Pad is for corrected response in regards, angle, height, and other parameters. for compensation of non ideal anachoic room !!!

:cheers:


Jean.

Ian Mackenzie
06-22-2005, 05:18 PM
Yea, I get that. But you missed my closing...

I was implying this need not be necessary for a properly designed speaker, is all. But this does make hella sense for a ground-up DIY.

Bo,

One of the problems of this type of discussion is relevance, and dealing with such a specialised topic as measurement (there are multitudes of volumes covering measuring) it is easy to become lost I agree.

I was proposing to cover the complete topic in a thread covering the upgrade of the 4343 to the 4344 specification.

But I am not sure that this will happen now as concerns have been raised that such imodifications should have documented subjective improvements so that those attempting such modifications can be assured of value for money.

While the issue of L pad adjustment is perhaps less ponderous if you altready own a 4345 monitor, as I understand it this thread was linked to 4343 crossover modifications to upgrade to the 3145 equivalent crossover.

Therefore anyone considering such an upgrade will need to address the issue of setting the L pads as I note in this schematic the L Pads are 8 ohms and the built in attenuation in the crossover is different to the 3143 network.

I also read the Rane operating instructions for the AC 22 last night covering level settings. From this reading they make it clear level testing and measurement is a tricky business. They suggest either test tones or pink noise at a distance of at least 15 feet in reference to a PA system using additional equipment. They also imply that the crossover levels can be used to moderate room and other system related issues which can be further addressed with specialised EQ.

For the benefit of those reading this thread I entertained the idea of the Tandy meter (in the earlier prescribed manner) so that A) you would not be frigthened away by the thought buying expensive test equipment over and above the cost of the upgrade from a 4343 to 4344 specs and then spending 6 months figuring out how to to use it properly and B) Having to languish the frustrations of such adjustments and the dissapointment of less than anticipated performance after the effort and the expense.

This is after all supposed to be a fun hobby where at least if you are following a guide written by someone who has done it before there is a good probability it will work out well.

So For those that are interested in following up on an upgrade from the 4343 to the 4344 ( or improvements to the 4343) send me a PM so I can put you on a list for a "detailed how to guide".

Best regards

Ian Mackenzie

boputnam
06-22-2005, 07:33 PM
One of the problems of this type of discussion is relevance... Exactly. :yes:

My reason for posting was to remind the average reader that this level of detailed measurement is not needed for a properly designed JBL Monitor Series multi-element configuration. However, if one does go about their own design, it is quite important that the elements are acoustically balanced, otherwise no degree of "room EQ" can remedy the symptoms.

And believe me I know this from working the FOH with some complete crap mains and wedges that are far from "acoustically balanced". In that scenario, the "un-optimised" build results in pathetically low GBF. And there is little that EQ'ing can do to surmount it. :banghead:

B&KMan
06-23-2005, 02:02 AM
Bo,

One of the problems of this type of discussion is relevance, and dealing with such a specialised topic as measurement (there are multitudes of volumes covering measuring) it is easy to become lost I agree.

I agree too


I was proposing to cover the complete topic in a thread covering the upgrade of the 4343 to the 4344 specification.

yes and I agree this tye of thread is dirth and extend too mouch the principal thread (this is reason I started this)

But I am not sure that this will happen now as concerns have been raised that such imodifications should have documented subjective improvements so that those attempting such modifications can be assured of value for money.

While the issue of L pad adjustment is perhaps less ponderous if you altready own a 4345 monitor, as I understand it this thread was linked to 4343 crossover modifications to upgrade to the 3145 equivalent crossover.

Therefore anyone considering such an upgrade will need to address the issue of setting the L pads as I note in this schematic the L Pads are 8 ohms and the built in attenuation in the crossover is different to the 3143 network.

so this methode it is good for all speakers ???


I also read the Rane operating instructions for the AC 22 last night covering level settings. From this reading they make it clear level testing and measurement is a tricky business. They suggest either test tones or pink noise at a distance of at least 15 feet in reference to a PA system using additional equipment. They also imply that the crossover levels can be used to moderate room and other system related issues which can be further addressed with specialised EQ.

yiah acoustic is realy tricky (see title ) :D

For the benefit of those reading this thread I entertained the idea of the Tandy meter (in the earlier prescribed manner) so that A) you would not be frigthened away by the thought buying expensive test equipment over and above the cost of the upgrade from a 4343 to 4344 specs and then spending 6 months figuring out how to to use it properly and B) Having to languish the frustrations of such adjustments and the dissapointment of less than anticipated performance after the effort and the expense.

Well your right but in certains point if you play in mecanical it is normal you buy certains specialize tool, more your hand is go more specific tool... this is choice of each... buy it is important to expose the complexity of jobs and manys occurs surprise is come with it...



This is after all supposed to be a fun hobby where at least if you are following a guide written by someone who has done it before there is a good probability it will work out well.

So For those that are interested in following up on an upgrade from the 4343 to the 4344 ( or improvements to the 4343) send me a PM so I can put you on a list for a "detailed how to guide".



Well Ian, your a master and I respect you idea, maybe the language is push my expression or comprehension in wrong way...

Your experiences is deep, complex and relevant in manys parameters arround JBL... I respect your point of view... Many small divergeance appear but with more talking, probably just communication problem...

And because I respect you, i run your test on my old 4343 (unfortunately )
little modified by fresh caps on hf and uhf (but in same value)...

And because the eratum methode is easy I resume the test here...

1--- put SLM at A weighted for less low frecquency interference in mesure...
2--- I test noise floor (below 40 dB)
3--- I perform sinwave at 15K, 5K, 500hz for respectively ech driver...
4--- I put SLM in 2 inch of flush verticale plane. full axis and full 0° all side
5--- I put pot in full range
6--- I perform sinewave at average 85 dB.
7--- I adjust SLM and I positionning in same time my body in exact placement for altering L-Pad.
a--- the uhf is not sensible of body in field but hf andparticulary med is big influence of you presence close to slm...
b--- because I have calibrated tone pulse for SLM I keep on my and to uncalibrate my SLM for this test==> If you have not this tool do not touch adjustement of yous SLM...

8--- I put 4 DB down response for UHF and HF and 3 for Med...

9--- Same procedure in other channel...

================================================== =====
10--- result:

(see first pict of L-pad position)

Listening test...

I perfom a SACD Fim 029 for the tesT listening. (no wine or other drink before or after... :D )

before read it is extremely subjectif aspect and it is not a judgement just my feeling on the spot...

First impression is more details and more informations, the med is more foward and the high is sweep and liquid, no harsh,etc...

but the bass guitar, timbal, contrebasse, is here but no body, power loose.. the drum is snap but you feel less the body of casing drum...

I replace my L-pad on regard on complex average mesurement set-up.
(see second pict of L-pad placement)
look the difference set-up ( :blink: )

Of course, if you look the pict, you undertand easy the sound is more bloom so I put little more power for same clarity level of high... but the detail of female voise is little desapear but more natural resonnace body: is feeling you put out mic on the mouse to put at 1 or 2 feet...

anyway, the major difference is bass presentation... in SLM set-up, I listen bass note but not power of bass and is relatively back presentation... now the integration is more coherence... all instrument have body and feel the body of xylophone, piano, sax, bass, drum, voice... Better ??? question of point of view... but yes the bass create the feeling to loose high fr details...

But attention surprise : 2 methods create a good integration driver and feeling of transparence :applaud:

The major tips in this comparable is : where if your fine tune your speaker, position and pad is linked, so you change pad maybe you have obligation to change speaker position... more complexe comparison... and repetability

Anyway I expose this not for try to foward one method or other but just expose each aspect of tricky and complex problem and question of speaker tune... I hope member is more understand the tricky aspect of mic)


Now I keep a glass wine

:cheers:

BTW Other members have tried this SLM method ??? your feeling ???

Jean

Ian Mackenzie
06-23-2005, 04:33 AM
Jean,

Are those tests with the new crossovers installed?

Also I am afraid the markings on the 4343 foil cal appear arbitrary and appear not to be calibrated in the manner of the 4344 & 4345 foilcals therefore nominal pink noise equidistant from each driver may be the only way to set the levels on the 4343....perhaps this is why people have so much trouble setting the levels of the 4343?


Ian

Ian Mackenzie
06-23-2005, 06:19 AM
Well Ian, your a master and I respect you idea, maybe the language is push my expression or comprehension in wrong way...

Your experiences is deep, complex and relevant in manys parameters arround JBL... I respect your point of view...


Some time back I think it was in the 1940's I landed my Tardis in a young man's workshop in Los Angeles and found him winding round wire on a former for a loudspeaker voice coil. It took some convincing but he eventually accepted the idea that winding edgewound copper on the former was much better.

I forget his name .....it was Jim something........:hmm:

The Doctor.

Ps Time to go .....I can hear the hoards stampedi

B&KMan
06-23-2005, 08:14 AM
i run your test on my old 4343 (unfortunately )
little modified by fresh caps on hf and uhf (but in same value)...

Jean



I not perform with new network because it is not ready...

:(

I create as schematic with conjuction of your concept cascade caps with the concept of Dc charge but I insert many erratum by cut & paste and other mistake electrical interpretation... (I'm null in electro)
And I run many tech test, (this thread and 4343 mod thread who expose the mesure of different transfert vibration of depend of the support fixation...)


so I limit time in a days : the BigNet is not ready...


If this set of speaers is able to talk long ride nightmare is expose...

I found this pair in morom cie of instalation pro audio... the 15" is reconed multiple time with with dirst third stuff... the guy is save a thrird part coil (standart circle type wirewound, is fix a other paper cone cutted in other driver and glued all in this ... the 10" is tilt inside... after verification the diagram of Hf is original but chaged and bad inverted signal path... so driver is out of phase... originally this spearker is works many years in great studio record... I hope this state descrition is not revalue in this time ...
before cancel the swich of post speakers and change selected caps section, the put is more downbecause the energy of the 15" is more low power... orignally the tweeter is really close to off med -3 and Hf is -4 or 5...

Maybe your method is work better in all correct components...
(remember, It is my first impression reserve on this approarch... )

My room correction is not perfect because it is not dedicated room but substential correctif is here, echo control on regard of sabine and kuttruf curve. position speaker in regards of phenomenon in 3 axis and phase respect, hight level components electronic, control, noise electrical and vibration analyse in each path. Yiah Yiah, I loose the count of couple hundred hours heres...

For just SLM I perform the best close performance to average by Rane Instruction... 1.5 meter is appear for my room the relativement better response in regard of reference test. the critical point is your position in room during test altered big the signature response so anybody is work test please put yourselt exactly same place and outside of 1.5 meter of SLM area...

and for average approach I repost my ask of any member...

Anybody please try SLM method and post your impression or mesure system...
maybe I great idea here but too other parameter slash the result...


:cheers:

I put picts for this thread my old network, modified, and the BigNet in work

Ian Mackenzie
06-25-2005, 03:51 PM
As proof in the pudding here is a response run of the 4345 after adjustment with the meter as described earlier using the equivalent 3145 network. The system is unequalised.

I defy anyone to attain a flatter un-smoothed in room response.

For comparison the JBL 4345 tech sheets.

The Doctor

B&KMan
06-25-2005, 10:50 PM
As proof in the pudding here is a response run of the 4345 after adjustment with the meter as described earlier using the equivalent 3145 network. The system is unequalised.
The Doctor

Bravisimo !!!

This is a excellent proof...

I have never doubth your method is works in yours stock...
:cheers:

I try to understand what is what and if many experience is repeatable in other set-up... with other member


is not work in my room + my systems + my old 4343...
But maybe it is not a hasard to rebuild my crossover... :D





BTW
I'm curious for couple point mesure...

1--- It is possible to explain your method mesure and expose your set-up software parameters and placement mic for more comprehension ???

2--- why your cut low mesure at around 75Hz ??
3-- why you have big drop at 10K ???


thanks for all

Jean.

Ian Mackenzie
06-26-2005, 01:45 AM
Bravisimo !!!

This is a excellent proof...

I have never doubth your method is works in yours stock...
:cheers:

I try to understand what is what and if many experience is repeatable in other set-up... with other member


is not work in my room + my systems + my old 4343...
But maybe it is not a hasard to rebuild my crossover... :D





BTW
I'm curious for couple point mesure...

1--- It is possible to explain your method mesure and expose your set-up software parameters and placement mic for more comprehension ???

2--- why your cut low mesure at around 75Hz ??
3-- why you have big drop at 10K ???


thanks for all

Jean.

I learn't my technique from watching a my tutor at tech college.

I used the highest sampling rate and sample length my computor will allow which is 44.1 kherts and 8192. The system is self calibrating.

You have to understand some issues of measuring in a home environment and have a strategy to obtain what you want. The software is a very basic package called Winairr. More elaborate software may produce pretty graphs but for the purpose of proving an existing design where most technical facts have already been determined it is not justified. (If I was designing a system professionally from scratch I might consider Leap with LMS, Clio or Praxis.)

The microphone and preamp I built myself using a long thin aliminium tube with a hi quality Panasonic electret insert mounted on a tripod.

As you no doubt appreciate muti way speaker it is not possible to have a single mic on axis with all drivers within 1 -3 metres and at longer distance room effects cause the measurement to be polluted. Gated measurements at longer distance to remove room effect also smooth out important details. It is also impossible to use small divisions at longer distance to obtain meaningful results.

For this reason I chose about 1.5 metres (as a compromise) on axis vertically with horn lense and mid way between horn and slot radiator. The mid range for this reason is slightly lower in level, but when measured directly on axis is level.

My objective was to assess overall flatness between the midrange driver, horn/lens and slot and I think this result is obtained.

The loss of details at below 100 hz is due to some time gating to minimise room effects from floor and ceiling. The sound lounge has only a 9 foot ceiling despite the Tardis being Dimensionally Transendential. This is so if the auto gravity fails, when I fall of the ceiling I don't break any bones...LOL

The sharp dip at 10 khertz is the comb effect of the horn and slot and is normal for this kind of measurement with this design. In practise it is imho not audible.

The Doctor

Ian Mackenzie
06-26-2005, 06:10 AM
Some other curves of the individual drivers with crossover filters.

I was really just interested to see the function of the crossover filters here.

The Doctor

B&KMan
06-26-2005, 11:44 PM
Ok this is my curve in according of your set-up mic at 1.5 m

impulse tone, rectangle weignt window... etc, the pict contain all indication mesure...

the notch of 8k is the result of bad cutting network...

normally is cut at 9.5K but at 7k the hf is full here and UHF is full here too soo the cummulative is little over 4.7 DB... but impossible to redure this notch with-out loos the rest of sprectrum of UHF... sic !!! this is another reason I build a new network...

:cheers:

Jean.

Ian Mackenzie
06-27-2005, 01:50 AM
Hi,

Well that is good, it will be interesting to see what your new network does!

The Doctor

jfine
06-27-2018, 08:03 PM
Okay,

I am now Back, the Daleks nearly got me..lucky for my sonic screwdriver I escaped!

Zilch is correct.

My reference is in effect to the design of the system with respect to relative Spl pressure response. The drivers respond to a specific spl based on sensitivity and the voltage drive applied via the filter sections. Therefore for each driver to operate at a matched spl of acoustic output from flat amplitude the voltage drive to each driver must be adjusted via the L PADs.

In order to understand this it is important to realise we are refering to net spl of each driver at a specifc descrete point. The overall flatness and integration is a function of the filter slopes, phase response and location of the drivers on the baffle. JBL have already done this design work so our JOB is quite simply to adjust the voltage drive.

After some trial measurements about 18 months ago I realised using FFT, MLS and Pulsed Analysers lacked the instringic precison for performing a measurement of net SPL based on differential of the voltage drive.

So I pulled out my trusty Tandy meter and mounted it on my tripod and fed a sine wave via a PC contolled function generator to the amplifier. By placing the meter a specific distance from the baffle in front of each driver I was able to measure with great precision the net SPL via fine adjustment of the L pads.

I used frequencies of 600, 5000 and 150000 for the midrange, horn and slot respectively.

I started with the slot and positioned the meter directly on axis about 2 inches from the baffle and wound down the pad on the horn fully and adjusted the level with the Slot Pad fully open to +4 db on the meter. I then carefully adjusted the L pad till I got 0 db, after repeating this several times I marked the postion on the foil Cal.

I then place a masking tape over the Slot mouth and repeated the same test with the horn with the meter again on axis with a reference of +4db (and wound down the midrange L pad fully) and then adjusted the horn L pad to 0 db.

The same technique was applied to the mid come using several measurements to assure precison of the +3 attenuated back 0 db level.


The location of the meter should be directly on axis but exact location is not overly critical. The reason is simple in the that we are only concerned with the net relative spl after adjustment from fully open L pad back to a desired point for 0 dba. As JBL advise this on their foil cals in Dba and the voltage drives from the crossover filters are already pre determined it is fairly straight forward to adjust the exact levels of each driver.

I think it is reasonable to assume that JBL with all their vast resources would have found the 0 dba reference point.

In fact I have found using this technique the only reliable and repeatable method of adjusting the relative balance of the drivers on a 4 way system.

Mid and far field measurements using even gated analysers do not offer the same precision for this kind of measurement.

When I then played the system it has portrays the smoothest and most remarkable coherent sound and the imaging is amazing.

I find the need for further adjustment completely un necessary

Needless to say my tongue in cheek remark (since edited) in the crossover modification thread is probably apt. It's simply a case of interpreting the available information and applying it with some simple techniques to solve a problem.

But if we come back here in few years time we will probably find people scratching their heads with the same problem.

The Doctor

I know this is an old thread,

Wouldn't you get *about* the same result if you played some pink noise, medium low volume, then measure the db of the woofer (or driver that is not adjustable via lpad), then measure horn, tweeter, and adjust those to the same db level as the woofer?

B&KMan
06-27-2018, 11:20 PM
I know this is an old thread,

Wouldn't you get *about* the same result if you played some pink noise, medium low volume, then measure the db of the woofer (or driver that is not adjustable via lpad), then measure horn, tweeter, and adjust those to the same db level as the woofer?


Hello


Well, if you know the exact frequency cut off and only play it on the 15 '', note the level in dB or dBA. Then you increase the power of 8 '' until you have + 3 dB or dBA that goes. Then you redo the same procedure with compression driver and then the tweeter.


the problem is obtain exact cut off frecency dividing network...

If you take a pink noise and measure globally in dB or dBA, it will be impossible to find the right level of adjustment because the acoustic energy is based on the frequency range covered.


ps.s. in fft it is necessary to use white noise rather than pink noise

----------------

for this thread, the fundamental question is what is the ideal position or angle to arrive at having an answer that "ear sound" balanced. In other words where should we place the microphone to have a calibration that corresponds to a good sound balance listening.

Two approaches: on axis at 1 to 1.5m in the axis at the common height between the 8 '' and the flute or squarely at the sweet spot of listening


That said, several microphones offer an approximation of balance significantly more performance than a single microphone.




But that does not take into account the work on phon in a diffuse field.


The tonal balance changes according to the fact that it is diffuse (no free field) and according to the volume.


Several approaches give a downward slope as the frequency increases, but by exactly how much?


And what is the answer of the room itself? In short, not as simple as it seems ...


my 2 cents