JBL 308P MKII Studio Monitor Review

[amirm]

Got it.

 

[JRAudio]

To get a complete impression of the DUT, a measurement far above 20kHz would certainly be interesting. Beside possible (low-pass) filters (with active speakers) one would also be able to see the break-up modes of the tweeter.

@amirm has to decide if the additional effort for such a measurement is worth the possible findings.



Here I can not follow your arguments any more.

How should the 90ms group delay come about and how should these 90ms (if they are correct) be determined from the frequency response measurement, if the absolute phase is not known and only the minimum phase can be calculated from the frequency response measurement?

In your shown measurement in post#270, a filter seems to limit the frequency response (of the JBL-308P) above 20kHz.
The resulting acoustic order of the filter should be somewhere around 16th order (at 21kHz) if we consider the filter slope of your JBL-308P frequency response measurement (red straight line in the diagram).
View attachment 92919

How should a low-pass 16th order at 21kHz lead to 90ms group delay?
Let's look at an example of an ideal loudspeaker with a typical LR4@2kHz crossover. The FR, phase frequency response and the group delay is as follows (red - woofer, green - tweeter):
View attachment 92942
Everything as expected. The 4th order filter shifts the phase around the allpass filter crossover frequency by 4x90° = 360°. The resulting group delay is about 0.3ms. The higher the crossover frequency, the lower the effect on the group delay.

Now we provide the speaker with an additional low-pass filter 16th order at 21kHz:
View attachment 92946

The new filter results in an additional phase shift of 16x90°. However, the influence on the group delay is negligible. So it is not possible that the additional filter creates a group delay of 90ms.

Hi “CTRL”

Thanks for noting. I was in hurry to send this speakers back, so it looks like I missed the right coma position, it would be 90 uS. And it would have been better, not to 1/12 oct smooth, when showing this response. I have the phase and group delay measurement, but am on my way right now. I can post it later tomorrow, but as far as I remember, I have had 5 to 6-times 360 degree wrappes in the phase measurements.

But anyway, my main interest is / was mainly to wish an additional measurement, showing if the active speaker does have a 48k bottle neck or not. The natural low pass behavior of a tweeter itself is fine, or an 88k2 / 96k DSPs in active speakers or room processors, but I would like to see up front, if an active speaker has a 48k digital bottle neck, if this isn't even mentioned in the manual.

Thanks, Juergen

 

[hyperplanar]

Gotta say, the FR is a lot more linear than what I was expecting considering how people seemed to universally say that the 308s didn't sound as neutral as the 305s (albeit the original versions, not the MkII).

I wonder if it would be advisable to use the -2 dB treble adjustment switch to get the FR more neutral if no other EQ capabilities are available? Or would that be too much?

 

[Deleted member 2944]

But anyway, my main interest is / was mainly to wish an additional measurement, showing if the active speaker does have a 48k bottle neck or not. The natural low pass behavior of a tweeter itself is fine, or an 88k2 / 96k DSPs in active speakers or room processors, but I would like to see up front, if an active speaker has a 48k digital bottle neck, if this isn't even mentioned in the manual.

Thanks, Juergen

Please post your measurement of the JBL tweeter by itself (without the DSP/amplifier in the path.)

Dave.

 

[Robbo99999]

Gotta say, the FR is a lot more linear than what I was expecting considering how people seemed to universally say that the 308s didn't sound as neutral as the 305s (albeit the original versions, not the MkII).

I wonder if it would be advisable to use the -2 dB treble adjustment switch to get the FR more neutral if no other EQ capabilities are available? Or would that be too much?

It's right, the 308p spin is not far off neutral, so if you EQ the Listening Window Flat then you'll get a better tonal balance, 308p is a bit bright, but with an Anechoic EQ you can leave the treble adjustment switch at it's default position - here's my EQ and you can see there's subtle boosting of the bass and subtle but wide reduction of the treble, so those two together change tonality relatively substantially:

Which turns into the following in room measurement & RoomEQ:

Note I brought down a couple of bass peaks (as you can see) but I'd say that's a fairly balanced tonality. So I think the Anechoic EQ will fix the tonality for you.

 

[hyperplanar]

It's right, the 308p spin is not far off neutral, so if you EQ the Listening Window Flat then you'll get a better tonal balance, 308p is a bit bright, but with an Anechoic EQ you can leave the treble adjustment switch at it's default position - here's my EQ and you can see there's subtle boosting of the bass and subtle but wide reduction of the treble, so those two together change tonality relatively substantially:
View attachment 93513
Which turns into the following in room measurement & RoomEQ:
View attachment 93515
Note I brought down a couple of bass peaks (as you can see) but I'd say that's a fairly balanced tonality. So I think the Anechoic EQ will fix the tonality for you.

Thank you for the info but I think there’s a misunderstanding here... I mean hypothetically, if one has no access/ability to EQ the speakers (lets say it’s directly connected to a TV or something), would using the -2 dB treble switch on the JBLs be a net benefit to the frequency response? I’m asking because one of my friends has a set up like this, and it doesn’t make sense to spend the $250 or whatever on a miniDSP for $400 of speaker.

 

[SpaceMonkey]

Thank you for the info but I think there’s a misunderstanding here... I mean hypothetically, if one has no access/ability to EQ the speakers (lets say it’s directly connected to a TV or something), would using the -2 dB treble switch on the JBLs be a net benefit to the frequency response? I’m asking because one of my friends has a set up like this, and it doesn’t make sense to spend the $250 or whatever on a miniDSP for $400 of speaker.

This is such an easy test. Just do it.
This was measured for 305s ( http://noaudiophile.com/JBL_LSR305/ ):

HF switch works as advertised. So if HF is too much just flip it.

 

[hyperplanar]

This is such an easy test. Just do it.
This was measured for 305s ( http://noaudiophile.com/JBL_LSR305/ ): HF switch works as advertised. So if HF is too much just flip it.

I was just asking because I don't have the 308s...
was simply curious if the -2 dB switch would nicely compensate for the base anechoic FR having too much treble, or if it would be too much of a cut, since the shelving frequency/Q of the treble cut aren't documented either.

If it does closely cancel out the overly hot treble and results in the anechoic FR becoming flatter, then wouldn't it be recommendable for anyone to start with the -2 dB treble cut activated as a baseline when setting up the 308s?

 

[SpaceMonkey]

I was just asking because I don't have the 308s...
was simply curious if the -2 dB switch would nicely compensate for the base anechoic FR having too much treble, or if it would be too much of a cut, since the shelving frequency/Q of the treble cut aren't documented either.

If it does closely cancel out the overly hot treble and results in the anechoic FR becoming flatter, then wouldn't it be recommendable for anyone to start with the -2 dB treble cut activated as a baseline when setting up the 308s?

I was gonna write some intelligent explanation. But i will just say "because marketing" and all people are different. Only a small number of people will need hf cut due to hearing them bright.

 

[Robbo99999]

Thank you for the info but I think there’s a misunderstanding here... I mean hypothetically, if one has no access/ability to EQ the speakers (lets say it’s directly connected to a TV or something), would using the -2 dB treble switch on the JBLs be a net benefit to the frequency response? I’m asking because one of my friends has a set up like this, and it doesn’t make sense to spend the $250 or whatever on a miniDSP for $400 of speaker.

Yes, you're right, flipping the -2dB HF Switch will fix it quite well. It's what I did before I had access to Amir's anechoic measurement. JBL 308p is about 2dB hot in the treble as seen in the spinorama, so it's really a pretty good solution flipping that -2dB HF Switch.

If it does closely cancel out the overly hot treble and results in the anechoic FR becoming flatter, then wouldn't it be recommendable for anyone to start with the -2 dB treble cut activated as a baseline when setting up the 308s?

This was something I tried to suggest to Amir (via PM) before he even measured the speaker, ie to do a Klippel measurement with the -2dB HF Switch activated because I was sure from my in-room measurements and from all the reports I'd read on the internet as well as my own experience that the 308p was too hot in the treble and the HF -2dB switch fixes that almost perfectly with just a little work required on the crossover with EQ. I figured most people would be flipping that switch, so I argued that it was most likely to the be the most widely used "default setting" by consumers and therefore the most relevant/useful for Klippel measurement.

 

[Robbo99999]

Hi,

Before talking about EQ, here is a point I tried to make in the Adam T7V thread:
Adam T7V, EQ spinorama.

I just started to add the sub score as I have been asked several times was it was after EQing.
I don't really see the point of it or least the way I interpret it seems different.

Here is the calculation I perform, which seems to match what others are doing:
PPR_LF = 12.69 - 2.49*NBD_ON - 2.99*NBD_PIR - 4.31*log10(14.5) + 2.32*SM_PIR
see there for some details:
https://www.audiosciencereview.com/...urements-community-project.14929/#post-467858

This assumes:
- Frequency response 14.5Hz @-6dB. 14.5Hz is chosen so the theoretical max score is 10 although nothing prevent a system from achieving better performances and therefore the score could exceed 10; this is the first issue.
- Perfect integration whatever that might mean, which is the second issue and where I don't quite adhere to the concept.

The idea, I guess, is to compare speakers if the LF extension is literally taken out of the equation, akin the pound-for-pound rating.
To me it would make more sense to use it this way rather than thinking "if I buy a sub then I'll get the astonishing system predicted by the sub score". That is just not going to happen: sub or not the room will still determine the system LF response.

This being out of the way, here is my take on the EQ.
Score: 5.58, with sub: 7.31
Spinorama no EQ:
Directivity, wide horizontal coverage up to +/-20deg but better stay at tweeter height.
Contour plot:
EQ design:
I have generated two EQs. The APO config files are attached.

• The first one LW is targeted at making the LW flat.
• The second, Score, starts with the first one and adds the score as an optimization variable.
• The EQs are designed in the context of regular stereo use i.e. domestic environment, no warranty is provided for a near field use in a studio environment although the LW might be better suited for this purpose.
• The deviation from flat looks very small and might be related to production tolerance (example tweeter sensitivity is 1dB up) on this very unit rather than "designed-in" therefore the EQ here derived might not translate well on other units.

It seems like the EQs are acting more like tonal controls rather than trying to solve any significant issues.
ABX listening looks critical here.

Score EQ LW: 6.23 with sub: 7.93
Score EQ Score: 6.63 with sub: 8.23

JBL 308p mkII APO EQ LW 96000Hz
November092020-114203

Preamp: -1.3 dB

Filter 1: ON PK Fc 85.5 Hz Gain 1.27 dB Q 2.66
Filter 2: ON PK Fc 1013 Hz Gain -0.59 dB Q 3.79
Filter 3: ON PK Fc 1381 Hz Gain 1.9 dB Q 5.02
Filter 4: ON PK Fc 1599 Hz Gain -1.26 dB Q 8.42
Filter 5: ON PK Fc 2362 Hz Gain 1.21 dB Q 9.67
Filter 6: ON PK Fc 2871 Hz Gain -1.42 dB Q 0.43
Filter 7: ON PK Fc 3005 Hz Gain 0.98 dB Q 4.47
Filter 8: ON PK Fc 16165 Hz Gain -3.06 dB Q 6

JBL 308p mkII APO EQ Score 96000Hz
November092020-114055

Preamp: -1.2 dB

Filter 1: ON PK Fc 87 Hz Gain 1.23 dB Q 2.77
Filter 2: ON PK Fc 953 Hz Gain -0.89 dB Q 2.9
Filter 3: ON PK Fc 1384 Hz Gain 2.41 dB Q 5.79
Filter 4: ON PK Fc 1585 Hz Gain -1.13 dB Q 7.87
Filter 5: ON PK Fc 2395 Hz Gain 0.72 dB Q 6
Filter 6: ON PK Fc 2879 Hz Gain 1.72 dB Q 13.72
Filter 7: ON PK Fc 4189 Hz Gain -2.12 dB Q 0.29
Filter 8: ON PK Fc 16022 Hz Gain -4.23 dB Q 5.45

EQ dZoom PIR-LW-ON
Regression - Tonal
EQ LW spinoramaEQ Score spinorama
Radar No EQ vs EQ score, appreciable improvements
Rest of the analysis is attached

Comparing with the other 8inch monitors
Kali LP8 vs Presonus Eris E8 XT, Spinorama and EQ design inside
Adam T8V
The 308p looks second to the LP8 with no EQ but the EQ versions would need careful listening for accurate evaluation.

Hi, thanks for your excellent work there, I briefly tried your "Score Optimised EQ" vs my own "Listening Window EQ". The biggest differences were in tonality with your "Score Optimised EQ" bringing out the bass a little more (by virtue of the decreased treble) than my "Listening Window EQ", but it did give a slightly more muffled, albeit bit smoother sound than my "Listening Window EQ". I listen at 2 metres within an equidistant triangle (with speakers toed-in and also angled downwards slightly to face me), do you have any thoughts on which EQ might be the most applicable for my listening setup, "Score Optimised EQ" or "Listening Window EQ"? I haven't done any in room measurements with your "Score Optimised EQ", but for reference here are screenshots of my "Listening Window EQ" I did combined with in-room measurement (the highlighted line is showing the predicted effect of my roomEQ filters with the original measurement being the faded line behind it, and to be clear the in-room measurement already has the Listening Window EQ applied before measurment):

As an additional or related thought, I do wonder how close my own JBL 308p Mkii is to the measured sample that Amir measured, I wonder how much variance there could be unit to unit.

(I had to delete the graphs in your post I quoted because otherwise it wouldn't let me attach my own pics - perhaps a max pic limit per post allowed, or was a temporary glitch on my PC)

EDIT: attached are some Thumbnail sized pics of a REW simulation of my room showing speaker positions & room size, if that would come into play with any recommendations you may have for me re whether I should use "Listening Window EQ" or "Score Optimised EQ", note that my room has got a little recess on the back left wall where I have a dining table so I couldn't model my room size & shape totally accurately in REW, it also has 50cm deep "window boxes" recessed into the front wall which again couldn't be modelled in REW (I also didn't know what values to input for Surface Absorption):

 

[hyperplanar]

Yes, you're right, flipping the -2dB HF Switch will fix it quite well. It's what I did before I had access to Amir's anechoic measurement. JBL 308p is about 2dB hot in the treble as seen in the spinorama, so it's really a pretty good solution flipping that -2dB HF Switch.

This was something I tried to suggest to Amir (via PM) before he even measured the speaker, ie to do a Klippel measurement with the -2dB HF Switch activated because I was sure from my in-room measurements and from all the reports I'd read on the internet as well as my own experience that the 308p was too hot in the treble and the HF -2dB switch fixes that almost perfectly with just a little work required on the crossover with EQ. I figured most people would be flipping that switch, so I argued that it was most likely to the be the most widely used "default setting" by consumers and therefore the most relevant/useful for Klippel measurement.

Excellent, thank you for the useful info!

We could simulate a spinorama of the 308 with the -2 dB treble cut activated—someone needs to measure a 308 once with it activated and once with it deactivated, grab the difference which would be the filter's frequency response, and apply that to Amir's spinorama data. Curious to see how the predicted preference score would be affected by that.

 

[Maiky76]

Hi, thanks for your excellent work there, I briefly tried your "Score Optimised EQ" vs my own "Listening Window EQ". The biggest differences were in tonality with your "Score Optimised EQ" bringing out the bass a little more (by virtue of the decreased treble) than my "Listening Window EQ", but it did give a slightly more muffled, albeit bit smoother sound than my "Listening Window EQ". I listen at 2 metres within an equidistant triangle (with speakers toed-in and also angled downwards slightly to face me), do you have any thoughts on which EQ might be the most applicable for my listening setup, "Score Optimised EQ" or "Listening Window EQ"? I haven't done any in room measurements with your "Score Optimised EQ", but for reference here are screenshots of my "Listening Window EQ" I did combined with in-room measurement (the highlighted line is showing the predicted effect of my roomEQ filters with the original measurement being the faded line behind it, and to be clear the in-room measurement already has the Listening Window EQ applied before measurment):
View attachment 93692
View attachment 93694

As an additional or related thought, I do wonder how close my own JBL 308p Mkii is to the measured sample that Amir measured, I wonder how much variance there could be unit to unit.

(I had to delete the graphs in your post I quoted because otherwise it wouldn't let me attach my own pics - perhaps a max pic limit per post allowed, or was a temporary glitch on my PC)

EDIT: attached are some Thumbnail sized pics of a REW simulation of my room showing speaker positions & room size, if that would come into play with any recommendations you may have for me re whether I should use "Listening Window EQ" or "Score Optimised EQ", note that my room has got a little recess on the back left wall where I have a dining table so I couldn't model my room size & shape totally accurately in REW, it also has 50cm deep "window boxes" recessed into the front wall which again couldn't be modelled in REW (I also didn't know what values to input for Surface Absorption):
View attachment 93701View attachment 93702View attachment 93703

Hi,

Thanks for your feedback on the EQ.

Orientation:
Form the data gathered by @amirm it looks better if you listen on tweeter axis, I think this is what you are doing but...
Horizontally pretty much anything between 0 to +/20deg is fine (that's the waveguide strength).
I would experiment with toe in by 10deg (you listen at 10deg off axis) just to see if it improve something with imaging as it may help dosing/delaying some reflections.

With regards to unit to unit variation/tolerance very difficult to say.
I would think +/-2dB 100 - 8000Hz but I have no data to back it up but only practices i am aware off.
Therefore tiny adjustments may not be realistic just based on the NFS data alone.

You seem handy with with REW, I do suggest you measure the LW on both your units yourself and generate the LW EQ this way.
That is the best you can do, then you may use my EQs as reference to see how far/close the unit tested by amirm is.

It'll require attention and dedication to make the measurements between the two sample as similar as possible but is it doable.
You'll then know how well matched your very speakers are and you'll have the LW data for both to perform your EQ.
The caveat is that you'll most probably be using windowing technics which will smooth the data (say you window cut off is 300Hz then you resolution will be that for the whole spectrum, OK of HF may be a bit coarse under 1000Hz).

What EQ should you start with? I published the two EQ, they are very close to each others and not miles away from the the bare speaker.
They are more or less acting this way No EQ / LW EQ = No EQ -1dB form 800Hz/ Score EQ = LW EQ -1dB form 3000Hz.

1 Try to true ABX (you should not know what EQ is being listened to when comparing) no EQ with each of the EQ individually, you may not be able to distinguish them. If you can can distinguish them (blind and statistically often enough) then proceed to the next step. If not leave the speaker alone and just do Room EQ for LF.
2 If you can distinguish them apart repeat with preference in mind, you may be able to tell the different but not prefer one... If not leave the speaker alone and just do Room EQ for LF.
3. If you you prefer one then use that one... and do Room EQ for LF.

 

[Robbo99999]

Excellent, thank you for the useful info!

We could simulate a spinorama of the 308 with the -2 dB treble cut activated—someone needs to measure a 308 once with it activated and once with it deactivated, grab the difference which would be the filter's frequency response, and apply that to Amir's spinorama data. Curious to see how the predicted preference score would be affected by that.

That seems like quite a good idea, and I was thinking about this yesterday....I might have time to do this today. I'd measure at 6 different points around my listening position within a 40cm wide by 25cm deep rectangle at my listening position. Create an average curve from that and then flip the HF Switch to -2dB and measure at the same positions, create an average curve again. Then I could use VirtuixCAD to SPL Trace the HF Trim -2dB Curve, I'd turn that into a Target Curve, import that Curve into REW as a Target and then use manual filters in REW to EQ the HF TRim 0dB Curve to the Target. I'd then take that filter set which is showing the difference and then apply that to Amir's Listening Window measurement and that will be the modified spinorama which I could SPL Trace again using VirtuixCAD which would turn it into a text file for use by other people (for EQ purposes). I think I'd use some kind of relevant smoothing because I'm sure the HF Trim Switch on the back of the speaker creates a quite smooth transition, so we just need to know where the transition starts and by how much. Hmm....that's quite a lot of work now I've outlined it, might be able to do it today, we'll see.

(As a shortcut I might be able to just numerically add & subract the two measurements if I use VirtuixCAD to SPL Trace both curves and then import the results into a spreadsheet, but the problem with that is that x-axis data points aren't likely to be the same (I imagine), so I don't think I'd be able to just line up the 2 columns of data in the spreadsheet & subtract, although that would be wonderfully simple if that were possible!)

Hi,

Thanks for your feedback on the EQ.

Orientation:
Form the data gathered by @amirm it looks better if you listen on tweeter axis, I think this is what you are doing but...
Horizontally pretty much anything between 0 to +/20deg is fine (that's the waveguide strength).
I would experiment with toe in by 10deg (you listen at 10deg off axis) just to see if it improve something with imaging as it may help dosing/delaying some reflections.

With regards to unit to unit variation/tolerance very difficult to say.
I would think +/-2dB 100 - 8000Hz but I have no data to back it up but only practices i am aware off.
Therefore tiny adjustments may not be realistic just based on the NFS data alone.

You seem handy with with REW, I do suggest you measure the LW on both your units yourself and generate the LW EQ this way.
That is the best you can do, then you may use my EQs as reference to see how far/close the unit tested by amirm is.

It'll require attention and dedication to make the measurements between the two sample as similar as possible but is it doable.
You'll then know how well matched your very speakers are and you'll have the LW data for both to perform your EQ.
The caveat is that you'll most probably be using windowing technics which will smooth the data (say you window cut off is 300Hz then you resolution will be that for the whole spectrum, OK of HF may be a bit coarse under 1000Hz).

What EQ should you start with? I published the two EQ, they are very close to each others and not miles away from the the bare speaker.
They are more or less acting this way No EQ / LW EQ = No EQ -1dB form 800Hz/ Score EQ = LW EQ -1dB form 3000Hz.

1 Try to true ABX (you should not know what EQ is being listened to when comparing) no EQ with each of the EQ individually, you may not be able to distinguish them. If you can can distinguish them (blind and statistically often enough) then proceed to the next step. If not leave the speaker alone and just do Room EQ for LF.
2 If you can distinguish them apart repeat with preference in mind, you may be able to tell the different but not prefer one... If not leave the speaker alone and just do Room EQ for LF.
3. If you you prefer one then use that one... and do Room EQ for LF.

Hey, thanks for taking the time for that detailed response. You say I should experiment with toe in, but I've already got the speakers toed in to face me directly so the face of the speakers is flat pointing at me (so I listen at on-axis).......or do you mean I should experiment with toeing in the speakers by an extra 10deg which would mean that the two speakers "on axis lines" will be crossing over in front of me? Is this advice in relation to your "Score EQ" and trying to remove some of the "muffled sound" that I mentioned....by virtue of manipulating the reflections in the room?

About unit to unit variation, I see your point re the small variations in EQ based on spinorama not being that relevant if it's potentially varying +/-2dB. Do you reckon this unit to unit variation would more be in the form of broad shelf-like changes....so retaining the detail of the small peaks and troughs and instead just offsetting the measured SPL by a constant offset in some places....for instance in your original post you talked about the tweeter perhaps varying in hotness from unit to unit.....so I've envisaged this as the same effect as a High Shelf Filter placed just slightly before the crossover dip.....in fact I've thought about experimenting with placing a High Shelf Filter at that point and experimenting with 0.5dB reductions (because my Listening Window EQ is good but I feel the tweeter might be just a tad hot still). Ah, going back to my first point in my previous sentence (I got misguided in that long sentence!), do you think that the fine & small peaks and troughs (the detail) seen in the spinorama would be retained from unit to unit, but just be offset from unit to unit in "Shelf Filter manner"? I suppose I'm thinking about retaining the EQ detail of my Listening Window EQ, but just experimenting with High Shelfs at the crossover point to simulate different tweeter hotness. I say this because I think the fine adjustments of the Listening Window EQ have made positive improvements in the sound, especially at the crossover.....it sounds really very natural apart from perhaps a little too much heat in the tweeter still.

Measuring my own Listening Window in order to create my own detailed Listening Window EQ is an interesting idea. I might struggle in determining the angles at which to measure....even though I can research where exactly the measurement points are for a Listening Window Measurement when I try to apply that in practice by trying to place the mic in the exact positions that is gonna be problematical I think. I'd have to do each speaker on it's own and move furniture (although if I measure at 1m distance I won't need to do that), but I don't have a mic stand that I can accurately place in air.....I currently use the mic stand that came with my UMIK, and I measure at different points on my sofa by using cushions to build up the mic to ear height, as well as placing the mic on the back of the couch for some measurement positions, so I've not needed an extendable mic stand.

Thanks for your advice on how to approach EQ in terms of ABX listening and the thought process behind whether or not you even need to EQ based on that. I can't really do ABX testing, as don't have anyone to flip the switches randomly for me.........or unless as an idea I place my mouse on the EQ On/Off button and then close my eyes and then click the mouse button furiously for a few seconds, press the Alt Tab button (with my eyes closed still) to take me to the music player & press play....listen to speaker & form an impression.....return to laptop and see if EQ is off or on (ie stock or EQ). But, the thing is I listen to one or two tracks that I listen to parts of almost everyday to check my headphones are positioned correctly on my head, but also to check EQ differences on my speakers....so I do sighted tests on these well known tracks where I knowingly flip the EQ on or off or between different EQ's....I tend to pause the music after a particular section of the track....change to the different EQ.....then replay the same section of the track listening for differences.....I'm quite clear in my mind the differences I can hear (if they are there). Tracks I use for this testing is Super Massive Black Hole from the band Muse (very busy track with a very detailed/powerful and busy bass line, but a well recorded track so surprisingly easy to distinguish details in all the competing parts on a well setup system, but will sound terrible with poor EQ's or equipment), I also use a couple of tracks from Massive Attack too (for similar reasons), and I sometimes cross check it's OK on vocals by using a track or two from Florence and the Machine (tracks Various Storms and Saints / St Jude). For instance, it was very easy to notice a difference between my Listening Window EQ and your Score EQ, big differences....mostly tonal, but some clarity differences.

 

[hyperplanar]

That seems like quite a good idea, and I was thinking about this yesterday....I might have time to do this today. I'd measure at 6 different points around my listening position within a 40cm wide by 25cm deep rectangle at my listening position. Create an average curve from that and then flip the HF Switch to -2dB and measure at the same positions, create an average curve again. Then I could use VirtuixCAD to SPL Trace the HF Trim -2dB Curve, I'd turn that into a Target Curve, import that Curve into REW as a Target and then use manual filters in REW to EQ the HF TRim 0dB Curve to the Target. I'd then take that filter set which is showing the difference and then apply that to Amir's Listening Window measurement and that will be the modified spinorama which I could SPL Trace again using VirtuixCAD which would turn it into a text file for use by other people (for EQ purposes). I think I'd use some kind of relevant smoothing because I'm sure the HF Trim Switch on the back of the speaker creates a quite smooth transition, so we just need to know where the transition starts and by how much. Hmm....that's quite a lot of work now I've outlined it, might be able to do it today, we'll see.

(As a shortcut I might be able to just numerically add & subract the two measurements if I use VirtuixCAD to SPL Trace both curves and then import the results into a spreadsheet, but the problem with that is that x-axis data points aren't likely to be the same (I imagine), so I don't think I'd be able to just line up the 2 columns of data in the spreadsheet & subtract, although that would be wonderfully simple if that were possible!)

As long as you have a mic stand or similar mounting arrangement for the measurement mic to keep its position 100% fixed, you should be able to just do two measurements on-axis (treble cut on and off) and derive a pretty good measurement of the treble cut filter response. The room reflections and such are not particularly relevant, since they would affect both measurements in exactly the same way, and cancel out when the difference of the two are taken.

 

[Robbo99999]

@hyperplanar and @Maiky76 , I'm quite excited about this, managed to work out exactly the filter that is used for the HF Trim -2dB switch on the back of the speaker, and I mean exactly! Long story short is that the exact filter used is a 6dB per octave High Shelf Filter at 3000Hz for -2.0dB gain exactly! I'll show my workings in the following, and also find attached a converted SPL Trace for the Listening Window which includes the -2.0dB High Shelf, so this is effectively what the Listening Window measurement would be if Amir flipped the JBL 308p Mkii to HF Trim -2dB position & measured the speaker, I'll attach the text file here:
Effect of Trim Switch -2dB on Listening Window.txt
So the data in the above file can be used for folks who would have wanted an HF Trim -2dB measurement from Amir for EQ purposes - use that file above as the measurement if you want to EQ with the Trim Switch in -2dB position.

OK, so here are my workings on how I derived all this.

Step 1: REW measurements to derive effect of HF Trim Switch at -2.0dB position
I basically did measurements in the exact same 6 places close around my listening position in a 40cm x 25cm rectangle around my listening position, and flipped the HF Trim Switch to either the 0dB or -2dB position for each measurement, so 12 measurements overall. I noticed immediately that the Trim Switch produces a nice smooth offset of -2dB for each measurement I took, so it was very repeatable. However, I still did an average of each of the two sets of measurements (one set being HF Trim switch at 0dB, and the other set at HF Trim Switch -2dB position)....and it is these 2 averages compared that you can see in the pic below. You can see it starts to deviate from 1kHz.

Step 2: Working out the Exact Parametric Filter that is being used by the HF Trim -2dB Switch
See that green line in the above pic, that's the -2dB line obviously, I took that Green Line graph and used VirtuixCAD to SPL Trace that line so I could turn it into a text file which I could then import into REW as a Target File for EQ purposes.....so that's what I did. Imported that text file into REW as a Target Curve and then EQ'd the Trim Switch 0dB line (blue line in above pic) to the Green Line if you like. Here's the pic of that, and this was the nuts & bolts of working out the exact parametric filter that is being used by this switch:

Note in the above graph you can't see the Target Curve line because it is perfectly obscured by the highlighted green line which is the EQ result which proves how accurate this filter is to what is actually happening in the JBL 308p speaker when you flip that HF Trim switch to -2dB. Yes, so you can see that I'm using "HS 6dB at 3000Hz with -2.0 Gain" in the list of filters in that pic.

Step 3: Applying the above Determined Parametric Filter to Amir's Listening Window Anechoic Data
Right, so this is the last step, and was just a means of creating a text file showing the measurement data if Amir had indeed measured this speaker with the HF Trim Switch in the -2dB position. I simply took Amir's Listening Window data from his file and imported that into REW as a measurement, and I then applied that "HS 6dB at 3000Hz with -2.0 Gain" Filter that we determined in Step #2 above:

So above you can see the Effect of the -2dB Trim Switch. I then did an SPL Trace of that EQ'd curve in VirtuixCAD to turn it back into text file data for use by folks for EQ purposes, and following you can see the sucessful SPL Trace I did in VirtuixCAD (zoomed in for better accuracy of data capture):

Yes, so find attached the text file which would be effectively the same as if Amir had measured the speaker with HF Trim Switch in -2dB position, this is for the Listening Window only:
Effect of Trim Switch -2dB on Listening Window.txt

In-Room Measurements shown against Harman Curve - Effect of Trim Switch at -2dB position:
Right, so you can see from the following pics that HF Trim Switch at -2dB position is a better fit for the Harman Curve. First pic is HF Trim at 0dB position, and 2nd pic is with HF Trim Switch at -2dB position:

I hope you folks find this all useful, and I'm certainly gonna revisit my EQ using the Trim Switch at -2dB position. In fact, I'll redo my Listening Window Anechoic EQ using the text file I linked you.....basically I'll be taking that modified measurement and EQ'ing it flat.....although thinking about it I'll just need to take a look at all the graphs & see which makes most sense.....but that's neither here nor there for now. The main point is we now know exactly what the HF Trim Switch does in it's -2dB position, and now you've got the data file upon which you base EQ attempts or comfort for folks who can't or don't EQ and instead want to see what the HF Trim Switch at -2dB position is really doing, which is important because pretty much everyone on the internet says these JBL 308p Mkii speakers are a tad bright.

@amirm , can you envisage in your experience or knowledge any negative issues associated with using the processing inside the speaker (the HF Trim Switch) as a means of influencing the frequency response....any possible degredation of sound fidelity not related to the frequency response change? EDIT: on this point in this paragraph, just did some listening tests comparing a "software HF Trim Switch -2dB" vs using the actual HF Trim Switch -2dB and tonality is exactly the same as expected, and if anything the actual real HF Trim Switch -2dB sounds better, but in reality probably the same.

(@BYRTT , thanks for telling me about VirtuixCAD a few weeks ago re SPL Trace, wouldn't have know about this fantastic program, wouldn't have been able to do without for this little project).

 

[Maiky76]

@hyperplanar and @Maiky76 , I'm quite excited about this, managed to work out exactly the filter that is used for the HF Trim -2dB switch on the back of the speaker, and I mean exactly! Long story short is that the exact filter used is a 6dB per octave High Shelf Filter at 3000Hz for -2.0dB gain exactly! I'll show my workings in the following, and also find attached a converted SPL Trace for the Listening Window which includes the -2.0dB High Shelf, so this is effectively what the Listening Window measurement would be if Amir flipped the JBL 308p Mkii to HF Trim -2dB position & measured the speaker, I'll attach the text file here:
Effect of Trim Switch -2dB on Listening Window.txt
So the data in the above file can be used for folks who would have wanted an HF Trim -2dB measurement from Amir for EQ purposes - use that file above as the measurement if you want to EQ with the Trim Switch in -2dB position.

OK, so here are my workings on how I derived all this.

Step 1: REW measurements to derive effect of HF Trim Switch at -2.0dB position
I basically did measurements in the exact same 6 places close around my listening position in a 40cm x 25cm rectangle around my listening position, and flipped the HF Trim Switch to either the 0dB or -2dB position for each measurement, so 12 measurements overall. I noticed immediately that the Trim Switch produces a nice smooth offset of -2dB for each measurement I took, so it was very repeatable. However, I still did an average of each of the two sets of measurements (one set being HF Trim switch at 0dB, and the other set at HF Trim Switch -2dB position)....and it is these 2 averages compared that you can see in the pic below. You can see it starts to deviate from 1kHz.
View attachment 93899

Step 2: Working out the Exact Parametric Filter that is being used by the HF Trim -2dB Switch
See that green line in the above pic, that's the -2dB line obviously, I took that Green Line graph and used VirtuixCAD to SPL Trace that line so I could turn it into a text file which I could then import into REW as a Target File for EQ purposes.....so that's what I did. Imported that text file into REW as a Target Curve and then EQ'd the Trim Switch 0dB line (blue line in above pic) to the Green Line if you like. Here's the pic of that, and this was the nuts & bolts of working out the exact parametric filter that is being used by this switch:
View attachment 93900
Note in the above graph you can't see the Target Curve line because it is perfectly obscured by the highlighted green line which is the EQ result which proves how accurate this filter is to what is actually happening in the JBL 308p speaker when you flip that HF Trim switch to -2dB. Yes, so you can see that I'm using "HS 6dB at 3000Hz with -2.0 Gain" in the list of filters in that pic.

Step 3: Applying the above Determined Parametric Filter to Amir's Listening Window Anechoic Data
Right, so this is the last step, and was just a means of creating a text file showing the measurement data if Amir had indeed measured this speaker with the HF Trim Switch in the -2dB position. I simply took Amir's Listening Window data from his file and imported that into REW as a measurement, and I then applied that "HS 6dB at 3000Hz with -2.0 Gain" Filter that we determined in Step #2 above:
View attachment 93901
So above you can see the Effect of the -2dB Trim Switch. I then did an SPL Trace of that EQ'd curve in VirtuixCAD to turn it back into text file data for use by folks for EQ purposes, and following you can see the sucessful SPL Trace I did in VirtuixCAD (zoomed in for better accuracy of data capture):
View attachment 93902
Yes, so find attached the text file which would be effectively the same as if Amir had measured the speaker with HF Trim Switch in -2dB position, this is for the Listening Window only:
Effect of Trim Switch -2dB on Listening Window.txt

In-Room Measurements shown against Harman Curve - Effect of Trim Switch at -2dB position:
Right, so you can see from the following pics that HF Trim Switch at -2dB position is a better fit for the Harman Curve. First pic is HF Trim at 0dB position, and 2nd pic is with HF Trim Switch at -2dB position:
View attachment 93903
View attachment 93904


I hope you folks find this all useful, and I'm certainly gonna revisit my EQ using the Trim Switch at -2dB position. In fact, I'll redo my Listening Window Anechoic EQ using the text file I linked you.....basically I'll be taking that modified measurement and EQ'ing it flat.....although thinking about it I'll just need to take a look at all the graphs & see which makes most sense.....but that's neither here nor there for now. The main point is we now know exactly what the HF Trim Switch does in it's -2dB position, and now you've got the data file upon which you base EQ attempts or comfort for folks who can't or don't EQ and instead want to see what the HF Trim Switch at -2dB position is really doing, which is important because pretty much everyone on the internet says these JBL 308p Mkii speakers are a tad bright.

@amirm , can you envisage in your experience or knowledge any negative issues associated with using the processing inside the speaker (the HF Trim Switch) as a means of influencing the frequency response....any possible degredation of sound fidelity not related to the frequency response change? EDIT: on this point in this paragraph, just did some listening tests comparing a "software HF Trim Switch -2dB" vs using the actual HF Trim Switch -2dB and tonality is exactly the same as expected, and if anything the actual real HF Trim Switch -2dB sounds better, but in reality probably the same.

(@BYRTT , thanks for telling me about VirtuixCAD a few weeks ago re SPL Trace, wouldn't have know about this fantastic program, wouldn't have been able to do without for this little project).

Thanks for your efforts.
Model of the trim disregarding the LF, not sure what went on there.

Here are the score and comparisons with the NO EQ and the EQ score EQ.
Score no EQ: 5.58
Score -2dB HF trim: 6.01
Score EQ Score: 6.23

Spinorama with -2dB HF Trim:

Zoom PIR-LW-ON
The trim effect is very similar to the EQ score within the tweeter range

Regression Tonal:
The trimmed version looks a bit flatter ON so it might help with the "Muffled sound"

[...]do you mean I should experiment with toeing in the speakers by an extra 10deg which would mean that the two speakers "on axis lines" will be crossing over in front of me?[...]

Yes this is what I mean. I don’t think it will help much with “muffled sound” but might on imaging, anyways that is a cheap test to perform…

From my original post:

• The deviation from flat looks very small and might be related to production tolerance (example tweeter sensitivity is 1dB up) on this very unit rather than "designed-in" therefore the EQ here derived might not translate well on other units.

It seems like the EQs are acting more like tonal controls rather than trying to solve any significant issues.
ABX listening looks critical here.

Ideally the tolerances would create wide and smooth deviations from the average say for example the tweeter is 1dB hotter than the reference which would be akin to a 1dB high-shelf.
There may be some interferences between the units (around the XOver point) that could exceed as well as or some shifts in resonances that may also exceed this value.
In general, I would not try to correct the “grass” or small/narrow deviations from the curve with high resolution measurements as it may not be quite clear what these are and they may not be the same on your units and/or may not be noticeable on you measurement system.
Applying reasonable smoothing helps seeing the trends rather being obsess with tiny details that’s why I restrict myself to the minimum number of biquads. Additional biquads may yield better results but these improvements become more and more tailor made for the very unit under test and therefore their impact may not necessarily beneficial on ALL units.

For the 308p, the EQ are very mild and might fall within the tolerance range of the manufacturing process.
Even the setting provided to trim the HF may be enough to achieve 96% of the results at zero extra cost.
Of course in room LF EQing is still needed.
Therefore measuring your very pair of speaker might be highly beneficial and the only real improvement you could make.
Tuning through listening is possible but very tedious and difficult especially since you would have to do do it twice, once for each speaker.
Not to mention that repeatability will be low...

Again, you measuring your own units LW will greatly help.

 

[amirm]

@amirm , can you envisage in your experience or knowledge any negative issues associated with using the processing inside the speaker (the HF Trim Switch) as a means of influencing the frequency response....any possible degredation of sound fidelity not related to the frequency response change?

I don't know in the specific. Such filter if done in digital can overshoot for example and do more than they say. I personally never bother with them and use upstream digital EQ.

 

[Robbo99999]

Thanks for your efforts.
Model of the trim disregarding the LF, not sure what went on there.
View attachment 94076
Here are the score and comparisons with the NO EQ and the EQ score EQ.
Score no EQ: 5.58
Score -2dB HF trim: 6.01
Score EQ Score: 6.23
View attachment 94074
Spinorama with -2dB HF Trim:
View attachment 94075
Zoom PIR-LW-ON
The trim effect is very similar to the EQ score within the tweeter range
View attachment 94072
Regression Tonal:
The trimmed version looks a bit flatter ON so it might help with the "Muffled sound"
View attachment 94071

[...]do you mean I should experiment with toeing in the speakers by an extra 10deg which would mean that the two speakers "on axis lines" will be crossing over in front of me?[...]

Yes this is what I mean. I don’t think it will help much with “muffled sound” but might on imaging, anyways that is a cheap test to perform…

From my original post:

• The deviation from flat looks very small and might be related to production tolerance (example tweeter sensitivity is 1dB up) on this very unit rather than "designed-in" therefore the EQ here derived might not translate well on other units.

It seems like the EQs are acting more like tonal controls rather than trying to solve any significant issues.
ABX listening looks critical here.

Ideally the tolerances would create wide and smooth deviations from the average say for example the tweeter is 1dB hotter than the reference which would be akin to a 1dB high-shelf.
There may be some interferences between the units (around the XOver point) that could exceed as well as or some shifts in resonances that may also exceed this value.
In general, I would not try to correct the “grass” or small/narrow deviations from the curve with high resolution measurements as it may not be quite clear what these are and they may not be the same on your units and/or may not be noticeable on you measurement system.
Applying reasonable smoothing helps seeing the trends rather being obsess with tiny details that’s why I restrict myself to the minimum number of biquads. Additional biquads may yield better results but these improvements become more and more tailor made for the very unit under test and therefore their impact may not necessarily beneficial on ALL units.

For the 308p, the EQ are very mild and might fall within the tolerance range of the manufacturing process.
Even the setting provided to trim the HF may be enough to achieve 96% of the results at zero extra cost.
Of course in room LF EQing is still needed.
Therefore measuring your very pair of speaker might be highly beneficial and the only real improvement you could make.
Tuning through listening is possible but very tedious and difficult especially since you would have to do do it twice, once for each speaker.
Not to mention that repeatability will be low...

Again, you measuring your own units LW will greatly help.

Awesome work to incorporate my -2dB Trim revelations into your own work there! I too noticed the strange error below around 40Hz - it happened as part of the SPL Tracing procedure, I zoomed in on the response and only traced from around 40Hz, in order to increase accuracy of the trace - I noticed that VirtuixCAD creates extremes that aren't really there at both ends of the spectrum - for some reason it creates "phantom values" below the low cut off point and also above the high cut off point - it's not a problem in practice because you can just disregard that extra data, the relevant traced data is accurate though, I suppose I could have easily fixed it by just trimming the phantom data out of the text file. I just skimmed your post quickly before work so will EDIT this post with any other thoughts I have later - probably on the specifics of HF Trim -2dB & whether that is better or not from the data you've presented.....but I have a feeling from my in-room measurements that my tweeter is a tad hotter than the one Amir measured, so I'm using the Trim -2dB at the moment.

EDIT: (some extra thoughts)
Yes, so your work is showing the HF Trim -2dB increases the preference score then from 5.58 to 6.01, that's a positive move, even though I know real fine grained decimals in Preference Score are not really that relevant.....I think I remember the guys that created the score say that or was it Amir, but yeah it's a move in the right direction for sure and it definitely changes the character of the speaker, so in no way can it be viewed as insignificant. Bloody hell, having just looked at the Speaker Ratings for all speakers reviewed on this site that score of 6.01 with the HF Trim Switch at -2dB position would take this speaker to the 5th highest rated speaker of all time measured here!

Certainly in my listening tests I like the -2dB option better and also my in-room measurement which I showed in my last post shows it to be a better fit to the Harman Curve. When I was running the speaker at 0dB on the Trim Switch and using a flat Listening Window EQ it was just a tad too bright and I found it was tickling my ears a bit after living with that EQ for a few days, so I know it wasn't quite right.....after looking at your original post and of the graph showing (Trim Switch 0dB) Listening Window EQ (https://www.audiosciencereview.com/...view.17338/page-13#lg=attachment92377&slide=0), it can be seen PIR only slopes by about 7dB over the whole range, whereas the Harman Curve is a 10dB slope over the whole range - I think in conclusion I prefer the full 10dB downslope of the Harman Curve vs a Flat Listening Window EQ, and the HF Trim Switch at -2dB position achieves that perfectly pretty much. I have used some elements of my HF Trim Switch -2dB Listening Window Anechoic Measurement to EQ the crossover and also to EQ down the rise in frequencies at 16kHz (as well as a few other places) as seen in following two pics:

Yes, so I've been enjoying listening to this particular EQ above, but I may re-activate Filter #4 even though it would make the dip worse in the in-room measurement, but that's above transition frequency, so I think I should have faith in the Anechoic Measurement and re-activate Filter #4 because it does result in a smoother Anechoic Listening Window Curve, I'll do some listening tests of this to see if I can notice a difference, it's a small change but it is in the sensitive 1khz area (Filter #4):

I'm thinking most people should run this speaker with the HF Trim Switch at -2dB position, certainly if you're not using any other EQ tools - I think both Maiky76's score analysis and my own in-room measurements & listening tests support that....plus you can then say you've got the fifth highest rated speaker ever measured on this site!

EDIT #2, did a quick test of including Filter #4 and it takes too much of the edge away from vocals and removes some crispness from other elements of the music, I'll be using the filters in my first pic in this post then.

 

[Robbo99999]

I don't know in the specific. Such filter if done in digital can overshoot for example and do more than they say. I personally never bother with them and use upstream digital EQ.

(I'm gonna use the HF Trim Switch -2dB, rather than applying it upstream digitally, because it means using less parametric filters, which fits better within the limited number of filters available in miniDSP 2x4.)