• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

JBL 708i Monitor Review (Passive: Part 1)

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,086
Location
.de, DE, DEU
Toni Mas said:
The sum of port + driver would also be interesting, though both responses cannot be summed directly without taking into account the port and driver size ratio.

You have that on the Spinorama chart. No need to use a calculator. Speaker measurements progressed since Keele published his paper in 1974, almost 50 years ago!
[off-topic] Analysis of near field measurements

The analysis of near-field measurements provides a good measure of the influence of the BR port resonances on the midrange frequency response and the midrange decay.

If the NFS is used, two measurement runs are required, one with the BR port open and one with the BR port closed. The difference of both on-axis frequency responses and the comparison of the decay behavior (via CSD diagrams) should provide in most cases so much information that an evaluation is possible.

If you don't have an NFS at hand, you can achieve similarly (but not as good) results with near-field measurements, without spending a lot of time.
How exactly near-field measurements are processed can be read in detail here ("how-to-make-quasi-anechoic-speaker-measurements-spinoramas...").

Here is a "short" example of how it works and what information the evaluation provides.

1. near field measurements of woofer and BR port. Mic distance is <<1cm (ignore the absolute sound pressure levels, these are not calibrated to 1m distance)
1672605927491.png

If we only look at the BR port measurement, it quickly becomes clear that the port resonances are problematic. The BR port resonance around 1kHz is quite bad, till the resonance is damped by -30dB to the reference level more decay periods (about eight full oscillation periods) are needed than with the Helmholtz resonance (about six periods around 50Hz - BR tuning frequency) and this although the SPL of the port resonance is around -11dB below the Helmholtz resonance.
1672606129655.png
But how severely does this affect the overall sound pressure response of the loudspeaker? (Rhetorical question, for the sake of dramatic ;))

2. the sound pressure level of the BR port must be corrected according to the area ratio of woofer Sd and port area to obtain the sound pressure ratios of woofer and port in the far field - corrected BR port FR is cyan.
1672607549128.png

3. Now both frequency responses (woofer FR + corrected BR port) can be summed up.
This gives us the overall frequency response of woofer and BR port without baffle step correction (BSC). To assess the influence of the BR port resonances we do not need to perform a BSC.

At this point, the correctly determined phase frequency response (I won't list them here, otherwise it would become even more complex) is extremely important during the measurements, since it strongly determines how the FRs will sum up. The frequency response curves alone (or with additional minimum phase) are not sufficient for the correct "summation" of woofer and port measurements - just as a side note.

The higher the resonance frequencies to be investigated are, the more important the difference in "time of flight" from the BR port and woofer to the measurement microphone or listener becomes and may also have to be taken into account - just by the way.
1672607804377.png

We see that the port resonances in the example are always at least 8-12dB below the woofer SPL and the influence in the worst case is just 1dB (500Hz and 900Hz range).

We can also nicely see the influence of the phase frequency responses of BR port and woofer, because adding two coherent signals of 98dB and 110dB would be expected to increase the overall SPL by 2dB to 112dB, but because of different phase angles in the frequency range of the resonance, it is less.

The port resonances around 500Hz and 900Hz play practically no role in the decay. Even if we look at the more difficult to interpret time-resolved CSD (second image).
1672610084663.png 1672610314309.png

So in this example, the BR port resonances barely affect the speaker and are therefore acceptable for a mediocre to good speaker - but it's not SOTA.

A similar analysis would have to be done for the JBL 708i speaker to make more definitive statements.

UPDATE:
Please always keep in mind that the use of near-field measurements is only an approximation of reality.
As the size of the sound source increases, the reliability of the FR decreases to higher frequencies. Also, edge diffraction is likely to affect BR port and driver somewhat differently.... you know what I mean - but in general it works ;)
 
Last edited:

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,423
Likes
7,940
Location
Brussels, Belgium
[off-topic] Analysis of near field measurements

The analysis of near-field measurements provides a good measure of the influence of the BR port resonances on the midrange frequency response and the midrange decay.

If the NFS is used, two measurement runs are required, one with the BR port open and one with the BR port closed. The difference of both on-axis frequency responses and the comparison of the decay behavior (via CSD diagrams) should provide in most cases so much information that an evaluation is possible.

If you don't have an NFS at hand, you can achieve similarly (but not as good) results with near-field measurements, without spending a lot of time.
How exactly near-field measurements are processed can be read in detail here ("how-to-make-quasi-anechoic-speaker-measurements-spinoramas...").

Here is a "short" example of how it works and what information the evaluation provides.

1. near field measurements of woofer and BR port. Mic distance is <<1cm.
View attachment 254055

If we only look at the BR port measurement, it quickly becomes clear that the port resonances are problematic. The BR port resonance around 1kHz is quite bad, till the resonance is damped by -30dB to the reference level more decay periods (about eight full oscillation periods) are needed than with the Helmholtz resonance (about six periods around 50Hz) and this although the SPL of the port resonance is around -11dB below the Helmholtz resonance.
View attachment 254056
But how severely does this affect the overall sound pressure response of the loudspeaker? (Rhetorical question, for the sake of dramatic ;))

2. the sound pressure level of the BR port must be corrected according to the area ratio of woofer Sd and port area to obtain the sound pressure ratios of woofer and port in the far field - corrected BR port FR is cyan.
View attachment 254059

3. Now both frequency responses (woofer FR + corrected BR port) can be summed up.
This gives us the overall frequency response of woofer and BR port without baffle step correction (BSC). To assess the influence of the BR port resonances we do not need to perform a BSC.

At this point, the correctly determined phase frequency response (I won't list them here, otherwise it would become even more complex) is extremely important during the measurements, since it strongly determines how the FRs will sum up. The frequency response curves alone (or with additional minimum phase) are not sufficient for the correct "summation" of woofer and port measurements - just as a side note.

The higher the resonance frequencies to be investigated are, the more important the difference in "time of flight" from the BR port and woofer to the measurement microphone or listener becomes and may also have to be taken into account - just by the way.
View attachment 254060
We see that the port resonances in the example are always at least 8-12dB below the woofer SPL and the influence in the worst case is just 1dB (500Hz and 900Hz range).

We can also nicely see the influence of the phase frequency responses of BR port and woofer, because adding two coherent signals of 98dB and 110dB would be expected to increase the overall SPL by 2dB to 112dB, but because of different phase angles in the frequency range of the resonance, it is less.

The port resonances around 500Hz and 900Hz play practically no role in the decay. Even if we look at the more difficult to interpret time-resolved CSD (second image).
View attachment 254062 View attachment 254064

So in this example, the BR port resonances barely affect the speaker and are therefore acceptable for a mediocre to good speaker - but it's not SOTA.

A similar analysis would have to be done for the JBL 708i speaker to make more definitive statements.

I wonder what you mean with “the port closed” because the impedance never really returns to sealed readings unless the cabinet is actually sealed with rigid material.

Was the port screwed shut with a panel and screws here?

Also an enlightening post as always.

By the way I wanted to ask you for a while now, is there a reason why you don’t share your DIY designs here? Are you sharing them somewhere else? Would love to check them out!
 
  • Like
Reactions: MAB

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,086
Location
.de, DE, DEU
I wonder what you mean with “the port closed” because the impedance never really returns to sealed readings unless the cabinet is actually sealed with rigid material.
Well, as close/tight/sealed as you can get ;)

Was the port screwed shut with a panel and screws here?
For the near field measurements of BR port and woofer (as in the example above) this is not necessary and would even make the result useless (you would sum up a sealed speaker with a BR port)

By the way I wanted to ask you for a while now, is there a reason why you don’t share your DIY designs here? Are you sharing them somewhere else? Would love to check them out!
I use speaker analysis (say bad things about other speakers ;) and their insights) and simulation for my own projects.
But for others to build my own projects, I would have to pay them money. For me, only function matters; aesthetics hardly play a role.
Nobody wants to build something like that.
1672613162646.png

But nevertheless I plan to publish my next project here in the forum (so you can all shi.. over me) if it sounds good and has no serious errors. And if there is really interest I would also make detailed construction plans - but I doubt it ;)
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,385
Location
Seattle Area
To clarify, as correctly stated, "near-field" measurements are ad-hoc with the mic more or less placed in front of the radiating driver. I usually set the tweeter distance for NFS at tweeter and then use that to make a near-field measurement of the tweeter. Then without changing X axis, I move Y until the mic is in front of the woofer and that measurement. Due to highly differing sizes and depth of drivers, the measured SPL is not accurate. For this reason, I routinely adjust by eye the levels for the tweeter to make it more "accurate" but the degree of error is quite high.

The port represents more challenges. In this case, it was easy to move the mic down further to measure the port. In other cases where the port is elsewhere, I have to rotate the mic using NFS. If there is a port on the right, the mic then is at an angle to the port and the distance to it is ad-hoc and set by eye. Ditto for rear ports if they don't fully align with Klippel axis. In this and many other cases, the level of the port measurement then is not exactly comparable to woofer even if you put aside the issues of near-field measurements.

Finally, with the mic very close to tweeter, it routinely creates comb filtering, causing un-even response that is not there if the mic was not so close to the driver.

And to be clear, again as stated, there is "NFS" type measurement here. Al we have is a single axis spot measurement. The measurements are part of the Klippel base system capability and not that of the NFS option. But being near-field impact of reflections is very much reduces or eliminated.

Despite all of this, we can get insight into large "acoustic events" such as port resonance. To interpret the impact on overall frequency response,, we then need to look at the Klippel NFS spin graph.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,385
Location
Seattle Area
Folks, please don't compare this to other finished speakers which use DSP internally to correct their response. I am talking about Genelec, Neumann, etc. If you saw their raw response, may have been disappointed in what you see. If you want to compare, look at the 708p which is likewise finished. I found the subjective performance of that speaker to be superb. No way you can compare what it outputs with a much smaller speaker with flat or even flatter response.

I hesitated posting this interim result hoping the above point is understood but many are still judging the speaker. This is not fair to the company or its product here.
 

sarumbear

Master Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
7,604
Likes
7,313
Location
UK
At this point, the correctly determined phase frequency response (I won't list them here, otherwise it would become even more complex) is extremely important during the measurements, since it strongly determines how the FRs will sum up. The frequency response curves alone (or with additional minimum phase) are not sufficient for the correct "summation" of woofer and port measurements - just as a side note.
In this and many other cases, the level of the port measurement then is not exactly comparable to woofer even if you put aside the issues of near-field measurements.
[...]
To interpret the impact on overall frequency response,, we then need to look at the Klippel NFS spin graph.
As @amirm said, we have the final FR via Klippel. Why would you want to go through a manual method from the last century? You wanted to see the port resonances and there they are too.

What characteristic is missing from @amirm tests, is the question I put to everyone who is not happy with the measurement @amirm conducts?
 

sarumbear

Master Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
7,604
Likes
7,313
Location
UK
I hesitated posting this interim result hoping the above point is understood but many are still judging the speaker. This is not fair to the company or its product here.
As I mentioned on post #11. People are confused.
 

dfuller

Major Contributor
Joined
Apr 26, 2020
Messages
3,335
Likes
5,050
Folks, please don't compare this to other finished speakers which use DSP internally to correct their response. I am talking about Genelec, Neumann, etc. If you saw their raw response, may have been disappointed in what you see. If you want to compare, look at the 708p which is likewise finished. I found the subjective performance of that speaker to be superb. No way you can compare what it outputs with a much smaller speaker with flat or even flatter response.

I hesitated posting this interim result hoping the above point is understood but many are still judging the speaker. This is not fair to the company or its product here.
I'll be frank: Brute forcing a speaker with obvious mechanical issues to measure well with DSP is just not good design. EQ can't fix stored energy problems beyond reducing the energy input - which doesn't fix the problem, it just masks it. Are Genelec et al guilty of this? Maybe, maybe not. I know for a fact that Neumann has done some clever things to knock down the resonances in the KH120 that aren't just electrical notch filters, and I'd bet good money Genelec does something similar.

But either way, it's just sloppy for a company that otherwise does really well - their drivers are excellent and so are their waveguides/horns, and it's not as though they're incapable of designing good ports. The "bones" are good, there's just some... questionable tradeoffs being made.

Besides - we know what it looks like post-DSP, it's the 708P, which still has port issues that you corrected with EQ and reported positive effects.
 

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,086
Location
.de, DE, DEU
As @amirm said, we have the final FR via Klippel. Why would you want to go through a manual method from the last century? You wanted to see the port resonances and there they are too.
What characteristic is missing from @amirm tests, is the question I put to everyone who is not happy with the measurement @amirm conducts?
I don't get, that you don't get it! ;)

Assuming Amir would not make near-field measurements, we would have a NFS measurement showing us that there are resonances in the 600-1200Hz range - nothing more.
What causes these resonances? Cabinet, BR port, cone eigenmodes or surround resonances? We have no idea.

1672617438421.png


But Amir does make near field measurements of woofer and BR port. We can see that there are massive port resonances in the range 400-1200Hz and that these resonances almost reach the level of the Helmholtz resonance around 42Hz (this was already explained in detail in posts#83).
1672617454143.png

As Amir explained, the mic distance from baffle/driver is too far to completely eliminate the influences of the cone and baffle.
Therefore, his measurements provide good indications, but are not quite sufficient for a final evaluation, since phase information is also missing.
But often this combination is already sufficient to narrow down the cause of the resonances.

Theoretically, there could be a combination of surround resonances and BR port resonances for example. In order to be able to break this down in more detail, one needs to estimate the share of the port resonances. So, you would have to proceed as described in the post#141 above (with additional NFS measuring or very close nearfield measurements).

Then you get the share of the port resonances in the overall frequency response of the speaker and whether the port resonances have a severe negative influence on the decay behavior of the speaker. Or whether other causes, such as a surround resonance or others, are dominant.

Hope this helps you understand why even "50 year old methods" are sometimes still useful ;)
 
Last edited:

RobL

Addicted to Fun and Learning
Forum Donor
Joined
Mar 4, 2021
Messages
925
Likes
1,509
But either way, it's just sloppy for a company that otherwise does really well - their drivers are excellent and so are their waveguides/horns, and it's not as though they're incapable of designing good ports. The "bones" are good, there's just some... questionable tradeoffs being made.

Besides - we know what it looks like post-DSP, it's the 708P, which still has port issues that you corrected with EQ and reported positive effects.

So basically you’re inferring that JBL are so incompetent that they screwed up not only the port, but also couldn’t even eq the fix properly? :facepalm:
 

dfuller

Major Contributor
Joined
Apr 26, 2020
Messages
3,335
Likes
5,050
So basically you’re inferring that JBL are so incompetent that they screwed up not only the port, but also couldn’t even eq the fix properly? :facepalm:
No, more that EQ can't fix a stored energy problem. Only make it less objectionable.
 

ROOSKIE

Major Contributor
Joined
Feb 27, 2020
Messages
1,915
Likes
3,394
Location
Minneapolis
I'll be frank: Brute forcing a speaker with obvious mechanical issues to measure well with DSP is just not good design. EQ can't fix stored energy problems beyond reducing the energy input
No, more that EQ can't fix a stored energy problem. Only make it less objectionable.

This thread just gets weirder with the various ideas being posted.
What 'obvious mechanical issues' are you talking about?
What stored energy problems are you talking about?

The port pipe resonance is not a 'stored energy problem' nor is it a 'mechanical issue'. So what exactly are you referring to?
 

ROOSKIE

Major Contributor
Joined
Feb 27, 2020
Messages
1,915
Likes
3,394
Location
Minneapolis
For me, this thread is off the rails.

This is not intended to be a speaker used without the PEQ/DSP portion of the design.
It uses a 2 part crossover, part passive components(inductors, caps, resistors) part active (DSP, PEQ)

This speaker is designed for specific applications.
The passive portion of crossover is in the speaker for three reason as I see it and only these three.
1. To allow 1 channel of amplification per speaker vs 2
2. To protect the drive units in case some one cranks them before properly loading the DSP filters.
3. Is a continuation of #1, to allow the speaker to be connected with speaker wire and not require AC nearby. It can now be amplified from a distance.(&centralized)

Again these are designed to be installed with DSP.

Beyond allowing for 1 channel of amplification and protecting the drivers from accidental overload, the only thing the passive components need to get right is maintaining good directivity. They do that and so now the PEQ/DSP based portion of the 'crossover' can do it's job.

If you think anything else about the passive part of the 'crossover' is important in a design like this be my guest but please don't design my next set of speakers. If DSP/PEQ doesn't make sense to you as a useful part of contemporary audio design, but is rather a hack, be my guest but please don't design my next set of speakers.
I appreciate the hobby of those who choose analog, however I am not a someone who scrapes rocks on PVC so I could care less about preserving whatever DSP/PEQ is supposed to be ruining/hacking up. The final product is what matters to me and I see DSP/PEQ as powerful primary design tool not an afterthought.

The below is from JBL
  • Centralized amplification and processing eliminates the requirement for power source at each speaker location
  • BSS BLU link Networked Audio for ease interface and reduced system noise
  • Harman HiQnet equipped components can be centrally controlled via hardware, software or wireless tablet
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,385
Location
Seattle Area
The other benefit of this design is ability to put as much amplification as you need. Running out of power is one of the top reasons active speakers run out of dynamics. They have to limit cost and heat generation when the amps are built into the speaker.
 

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,086
Location
.de, DE, DEU
Wanted to ask the electronic specialists to what extent the sudden impedance step around 5 kHz to 110 ohms could be problematic for amplifiers (Class D?, tube obviously, distortion in transistor amplifiers?...)?
Should one be particularly attentive when choosing an amplifier or does the impedance step not matter and need not be mentioned?
Would it make sense to use a passive impedance linearization?
1672623812269.png
 

375HP2482

Active Member
Joined
Jan 13, 2020
Messages
165
Likes
176
4 ohms@45 degrees at 10 kHz is unusual, esp. for an 8-ohm nominal driver.
 
Last edited:

uwotm8

Senior Member
Joined
Jul 14, 2020
Messages
401
Likes
457
Folks, please don't compare this to other finished speakers which use DSP internally to correct their response. I am talking about Genelec, Neumann, etc.
But even non-DSP Neumann speakers are still lightyears ahead:rolleyes:
If we compare 708P to their DSP speakers - Neumanns are far better again, ruler-flat response instead of that slightly "smiley face"-like (which may be actually preferred for listening depends on taste or environment).

Surely I'd stick with Dynaudio or Focal Be active speakers for home listening:)
 
Last edited:
Top Bottom