DIY - JBL 4345 Clones - Measurements?

[dkalsi]

Hi All,

I know there are quite a few of us here at the forums that have gone down the DIY 4345 route.

I wanted to know whether anyone has taken the time to measure their diy 4345s or factory build 4345s.

I'm only now learning about REQ wizard and taking measurements in general. Hopefully this evening, I will try to post the frequency response of my DIY 4345 - I'm hoping the experts can shed some light on whether I did something incorrectly (if anything) or whether the response of my speakers can be improved.

To be clear - I built my crossovers based on the JBL 3155 networks (given that I use alternate between the 2425/2307 combo inside the cabinet and 2441/Yuichi 290 outside the cabinet for my mid-rage horns)




In the meantime, if someone has taken actual measurements of their speaker - I would love to see how they measure so I can get a understanding of what can be expected.

Thanks,
Dhar

[Robh3606]

When you measure them go out at least 2 meters. I measured my 4344's I used both an RTA and CLIO way back when. Here I have an ungated CLIO with 1/3 resolution and an RTA where they were tweaked as flat as I could get them along with the JBL Factory Measurement in an anecholic chamber.

The difference between the 4344 and 45 is the woofer bandwidth is the same. Depending on how far and what you use as your reference point you can get quite a bit of variation in the measurements. If it looks close to the factory using 1/6-1/12 resolution above 500hZ- 1K in a gated measurement you are where you would expect to be.



Rob

[JeffW]

I think the factory frequency response curves are posted in the 4345 entry in Systems.

[dkalsi]

When you measure them go out at least 2 meters. I measured my 4344's have it someplace I will post it if I can find it. I used both an RTA and CLIO way back when.

Rob

Will do Rob.

I took a bunch of measurements last night from approximately 30 inches away from the tweeter - on axis.

I've see people take pictures with the speaker laying down on its back - do you think that is necessary? My measurements were taken with the speakers standing.

Additionally, could you please confirm whether all measurements should be taken with the mic on axis with the UHF (2405?). I'm not sure where I should be facing the mic given the asymmetrical placement of the components.

Thanks,
Dhar

[Robh3606]

I took mine on the UHF drivers axis actually more between the horn and the tweeter. The 2405 turns into a flashlight beam in the vertical axis so if you measure say on the midrange you will see some roll-off. All my measurements are with them standing on their stands to get the UHF driver at seated ear height. You can do groundplane measurements but they boost up the lowend on the measurement and you need to correct for it. On their backs?? I have seen it done but never used that method myself.

Rob

[ivica]

Will do Rob.

I took a bunch of measurements last night from approximately 30 inches away from the tweeter - on axis.

I've see people take pictures with the speaker laying down on its back - do you think that is necessary? My measurements were taken with the speakers standing.

Additionally, could you please confirm whether all measurements should be taken with the mic on axis with the UHF (2405?). I'm not sure where I should be facing the mic given the asymmetrical placement of the components.

Thanks,
Dhar

Hello dkalsi,

I can suggest You to do measurements , if possible, outside the room , fare from the walls and other reflecting surfaces, with the "speaker laying down on its back". In such way the measurements would be equivalent , almost as in acoustic laboratory 'ambient', so You can compare the results with JBL specified.
The reflections from nearby objects (surfaces etc) can introduce large peaks and deeps in the FR- response.
Measurements 2m (6ft) or 3m (9ft) apart from such large the box (as 4345) would give you more realistic picture what really can be get from the speakers.
Regards
Ivica

[4313B]

As Rob mentioned above:

http://www.mh-audio.nl/Groundplane.asp

As Ivica mentioned:

That kind of measurement should be ok down to around 450 Hz (approximate width of the baffle). Just gate everything lower.

[dkalsi]

Hi All,

Below is a few measurements of my DIY JBL 4345 clones. These measurements were takes with the speaker standing up - with the mic 28 inches from the tweeter - on axis (no reason - why I chose 28 inches - it was totally random). All L-Pad sets to 0 on the folical.

I will re-take measurements from 2m away with the mic centered between the UHF - HF in the coming days.

My system is a bi-amped. My crossover is Marchand XM-9 setup for 290hz (all settings at 0).

Sweeps from 50hz - 20Khz.

This was me just messing around and trying to learn a little more about taking measurements.

My concerns are:
1) the null at the 290hz active crossover point - which seems to dip almost 7db
2) Additionally, I seems the crossover point is approximately 220hz - but I know I ordered the 290 card from Marchand.
3) the steep roll-off of the 2405 - or actually the entire response after 10Khz - seems to jump up and then rolls-off rather quickly.

Four measurements were taken:
1) Full Respose curve - Purple
2) High Pass @ 290hz - Blue
3) Low Pass @ 290hz - Red
4) With the 2425 disconnected (just to see how the crossover was working at the 10KHz point - not sure whether it even makes sense to be measuring this - but I just wanted to under stand the effect of the crossover and whether I built them correctly).

Any input/comments would be much appreciated.

[Robh3606]

Try flipin polarity on the woofer to see if your notch goes away.

Rob

[dkalsi]

Try flipin polarity on the woofer to see if your notch goes away.

Rob

Thanks Rob - I'll definitely give that a try.

But don't you think the woofers starts to roll off prior to the crossover point? Based on the above measurements (last one in red), seems the woofer starts to roll off prior to the 290hz crossover point installed in the Marchand.

Initially, I thought the woofer measured as it did because I was measuring so far off-axis (i.e., the mic was at the tweeter level). However, when I moved the mic to be on axis with the woofer (2245H), the response was very similar to what has been posted above.


In regards to the active crossover point (290hz), it seems the high-pass is working just fine, where we can clearly see the frequency starting to roll off right at 290hz. However, on the low pass, it appears the frequency starts to roll-off at approximately 200hz.

Nonetheless - I will definitely try switching the polarity on the woofer.

[Robh3606]

Do you have a live set-up like an RTA you can use to check levels as you adjust the trim pots on the driver?? Don't worry about where they are just set them up to get the desired in-room response. Try backing up to 2 meters, you are in close and proximity to one or more of the drivers will effect what you see. If it sounds good now then just remeasure out a 2 meters and don't touch a thing level wise until you record the pot positions so you can go back if needed.

For example the 2425 looks hot but it could be a measurement artifact compared to the woofer because of microphone placement and distance.

You don't say what kind of crossover or the slopes. A L/R is 6db down at the crossover frequency buterworth would 3db down. Also the slope and Q will effect where the knee starts so are the 12,18 or 24db per octave?? Based on the measurements you posted you may want to bump up the 2245 a bit. If you look at the average through the center of the curves it looks like the top end is hotter from say the 2425 and up compared to the woofer. Might want to bump up the 2202 a bit as well to see if the hole at 400-1K might get filled in a bit.

If we were doing this live in realtime it would be realtively quick and painless, doing it over the internet make things a bit more complicated and most importantly you can't hear the speakers. Just try to get them so the sound balanced and go from there. If you have some pink noise give it a try. I use it to set my woofer levels on my Biamp systems and once you get the hang of it you can get extremely close by ear.


Rob

[1audiohack]

Hi Dahr;

Break this down a bit, your looking at several things at once.

Have you taken a "native response" measurement of the woofer yet? If not, get the crossover out of the chain and sweep the woofer an octave or two higher than you plan to use it as a reference.

Will your measurement platform let you enter negative numbers? Some systems will let you sweep through zero Hz and will read the negative side in quadrature yielding better LF information in non anechoic environments. As an example if I were measuring your 2245 I would set the sweep as -200 through 1200 rather than 200 through 1200 Hz.

You may consider a near field woofer measurement eg. the mic less than 1/4 inch from the dust cap. That will eliminate any reflected energy from the room from contaminating your measurement. Stick the mic in the port as well.

I would also take a transfer function measurement of your crossover and take the guesswork out of what it's really doing.

My Rules of Data Collection;

1 What am I trying to measure?
2 Why am I trying to measure it?
3 Is it relevant?
4 Is it audible?


Rules of Interpreting Data;
1 Is it believable?
2 Is it repeatable?

Have fun! It a whole new world your jumping into.

[dkalsi]

Rob & Barry,

Thank you for your comments.

For measurement gear, I am currently using:
1) Room Equalizer Wizard 5.0
2) MobilePre USB External Sound Card (with Phantom Power)
3) Behringer Mic

Not sure whether the above allows for RTA capabilities as I am still learning the software.

I will try to take measurements from 2-meters away keeping the setting as-is. One question: should the lens be removed from the 2425/2307 prior to measurements?

The crossover employed is Marchand XM9, which has the following features:
-"XM9L Electronic Crossover Network is a fourth- order constant-voltage crossover"
-"The slope of each output is 24 dB/octave and, because of the fourth-order design, the high-pass and low-pass outputs of the crossover are always in phase with each other."
-"The crossover network is implemented as a fourth-order state-variable filter"
-"The damping control sets the response at the crossover point. In the center position the response is flat. Set fully counterclockwise there is a 4 dB dip at the crossover point and set fully clockwise there is a 5 dB peak."

Upon making some of the adjustments suggested in post #11, I will take additional measurements.

Barry - I will also take measurements based on the methods you suggested.

1) What am I trying to measure?
- Don't know just yet - but question #2 below is the reasoning
2) Why am I trying to measure?
- I want to know whether I built everything correctly. I didn't know anything about speaker building and decided to built the JBL 4345s without knowing how to read crossovers schematics correctly.
- They sound phenomenal to me - and I absolutely love them - but you can say for me peace of mind I want to know whether I built them correctly.
3) Is it relevant?
- Hopefully, as I read more, I'll get a better understanding of what I should be measuring for purposes of achieving my goal above
4) Is it audible?
- I have noting to compare it to :-)

More measurements to come :-)

[JeffW]

That 4th order slope might be what's causing the dip at lower than 290Hz.

[Mr. Widget]

That 4th order slope might be what's causing the dip at lower than 290Hz.

Absolutely... that said, any measurements below about 1,000 Hz are very difficult to make accurately in a typical room. I'm not saying lower frequency measurements are irrelevant, but look at them realizing they are suspect.


Widget