Hi,

Diagnosing your tweeter, if it's even the tweeter that's the issue, is going to be difficult how you've done this as it represents the room by a large margin. Your non-bass frequency measurements need to be done with a lot of setup so that you can properly gate the impulse response. If you haven't done this, then your measurement is just the room and placement dependent and will be different every time you measure it from another place, etc, so not representative of your actual tweeter. If you want to try to isolate this, take the speaker into a room and have it as close to equidistant from the walls as you can and have the tweeter as close to equidistant from the ceiling to the floor. Measure those distances so you can work out how much time it takes for sound to reflect and you can then do the measurement and use the impulse response tab to gate the measurement to reject anything from a time longer than your minimum reflection time. If done properly you'll get something closer to anechoic and a better representation of your actual tweeter without the room.

This link has some simplified info on how to do this with screen shots:

Quasi anechoic loudspeaker measurements Part 2 - Audio Judgement

In the second part of our loudspeaker measurements article, we show you, step by step, how to make frequency readings using ARTA.

audiojudgement.com

Another option is to measure the tweeter via impedance sweep and check it's T&S parameters, if you have something that can do this (like a DATS V3).

Very best,

 

... or drag a known-good and the suspected-bad speakers outside and measure there. Just the floor bounce to worry about.

@vlad2601 - You can borrow my DATSv3 any time. DATS sweeps will tell you a LOT about what might be going on with each speaker. That can be done indoors; you just need a windows machine. Compare the results for each speaker against the published JBL specs here (last page). That will tell you which ones are good/bad with a pretty high degree of certainty. The upside of pro speakers is they almost always have good published specs like this.

 

I don’t know that the problem was even defined. However, a tweeter is going to work like it’s supposed to, or it’s not. You should be able to play white noise very quietly and run your ear up against each one. If one of them isn’t working right, remove it and measure it if you want, but in the end, you’ll need to replace it.

 

If you look at the image of the freq responses including the bad speakers, the expectation would be that they’d all be essentially the same. 5 have very significant variations from the other 7. 4 more have a ~3-4db variations from the 3 speakers which I would say are the most like the JBL reference curve that @fattire linked above and to the responses @Archaea posted in his thread here.

 

These speakers will be a little difficult to troubleshoot. Could be individual tweeters are bad. But as their response varies by level and probably something like an all pass filter or delay line for timing/phase adjustment (just guessing here, JBL didn't publish AFAIK the details on how they achieved an approximation of Legendre shading and appropriate delay for this design) it could also be issues in the filter board affecting some drivers. The response posted could be the result of comb filtering of some drivers are functioning properly and others are decreased in level or otherwise not correct.

 

I think I’ll try @fattire ’s DATS and also try switching individual tweeters that seem to be good with apparently bad tweeters to see if that changes anything.

 

Call JBL Pro and ask them how much they'd charge you for a brand new tweeter array. Their pricing isn't as bad as you might expect. They don't seem to gouge you, I mean. You could buy a new board and know exactly what you are working with, and fix one of your speakers at the same time.

All these tweeters are the exact same, but they are shaded in groups of four (I think). So if you have some bad tweeters, it would be a matter of somehow detecting which ones are bad and swapping them out. Much easier said than done. I'm glad to hear you don't think it's your crossover boards - - if you've for sure ruled that out.

One of the things I noticed in my adventure with these was for some of my BAD speakers, if i switched the crossover from Music to Speech on the little dial on the outside the problem went away. So I then knew it was the crossover board. The Frequency Response is different between music and speech settings, but it just is a little less flat and a little bit more boosted in the human vocal mid-frequencies. It's another thing to try.

I don't know how to read your SPL excel chart. Can you give me a legend on that? What are we looking at in each column and row?

 

I met @vlad2601 at his place and showed him how to run DATS against his speakers. Below are the results comparing one of the good speakers (#2) with one of the bad speakers (#5). Impedance sweeps were taken in DATS then exported to be used in REW.

The magnitude response is in-room, on-axis at the mid-point of the tweeter array, from 1M. Ignore everything below 500 Hz due to room effects. It's pretty obvious that there's an issue somewhere in the high end tweeter array. All bad speakers shows huge deviation in both the magnitude and impedance plots. We didn't measure every speaker with the mic. We just confirmed one known good and bad on my system using REW to ensure there weren't measurement errors. Vlad had a great rig built to hold the speaker so his testing of each speaker is quite repeatable and reliable.

Additionally, vlad walked me through his SPL testing of each individual tweeter. I didn't think much of it, but his results correspond directly to the speakers having issues. I.E., the speakers with the largest deviations in the largest number of tweeters are the worst performing with the biggest issues in the impedance response as well.

We pulled one of the tweeter arrays out and played with it a bit in DATS. It appears that it's possible to individually sweep each tweeter in DATS to get an impedance response. We were worried that the array PCB had some funky wiring in it, but that doesn't appear to be the case. If they're wired in banks as expected, worst case is that all tweeters in each bank should measure the same or very similarly.

Vlad's next step is going to be to impedance sweep each tweeter in a known good speaker. Then impedance sweep each tweeter in a known bad speaker to see how things compare. He's probably going to have to end up sweeping all 5 of the definitely bad speakers so that's 96 tweeters total ((1 good + 5 bad)*16 tweeters each)!! Needless to say, I left the DATS for him to borrow for a while .

If the tweeters all match, then it's either the tweeter PCB or something internal to the speaker itself.

 

I met @vlad2601 at his place and showed him how to run DATS against his speakers. Below are the results comparing one of the good speakers (#2) with one of the bad speakers (#5). Impedance sweeps were taken in DATS then exported to be used in REW.

The magnitude response is in-room, on-axis at the mid-point of the tweeter array, from 1M. Ignore everything below 500 Hz due to room effects. It's pretty obvious that there's an issue somewhere in the high end tweeter array. All bad speakers shows huge deviation in both the magnitude and impedance plots. We didn't measure every speaker with the mic. We just confirmed one known good and bad on my system using REW to ensure there weren't measurement errors. Vlad had a great rig built to hold the speaker so his testing of each speaker is quite repeatable and reliable.

Additionally, vlad walked me through his SPL testing of each individual tweeter. I didn't think much of it, but his results correspond directly to the speakers having issues. I.E., the speakers with the largest deviations in the largest number of tweeters are the worst performing with the biggest issues in the impedance response as well.

View attachment 3557094

We pulled one of the tweeter arrays out and played with it a bit in DATS. It appears that it's possible to individually sweep each tweeter in DATS to get an impedance response. We were worried that the array PCB had some funky wiring in it, but that doesn't appear to be the case. If they're wired in banks as expected, worst case is that all tweeters in each bank should measure the same or very similarly.

Vlad's next step is going to be to impedance sweep each tweeter in a known good speaker. Then impedance sweep each tweeter in a known bad speaker to see how things compare. He's probably going to have to end up sweeping all 5 of the definitely bad speakers so that's 96 tweeters total ((1 good + 5 bad)*16 tweeters each)!! Needless to say, I left the DATS for him to borrow for a while .

If the tweeters all match, then it's either the tweeter PCB or something internal to the speaker itself.

Thanks to @fattire for letting me borrow the DATS. I think I’ve found a potential smoking gun with testing the impedance sweep on the individual tweeters. On what I call the “bad” speakers, several of the tweeters have bad impedance response; i.e. instead of an avg impedance around 10-15 ohms, it’s in the kohms!

Test setup:

1. Removed tweeter arrays from the speaker.
2. Using the DATS clips attached to the relevant +/- for each individual tweeter, run an Impedance Sweep in the DATS software.

Good speaker (#3) impedance sweep results for tweeter position 10 in the array (the other 2 good speakers I’ve tested thus far have similar responses to this one)

Bad speaker (#5) impedance sweep result for the same tweeter position 10.

Speaker 3 - impedance sweeps for all tweeters in array. The 4 tweeters which are 4 ohms lower are the center 4 of the tweeters in the array and likely related to the CBT magic. All 3 of the good speakers look the same as this.

Possible next step would be to swap in individual tweeters from a known good array to a bad array and see if that fixes the issue. Need to get with a friend who is comfortable soldering as it’s not something I’ve really done much. I could get there, but first want to see if that’s where the issue lies.

Derek

 

Since those two speakers perform so differently, I would also check the crossovers. Those impedance curves above 1khz are two different speakers on paper!

 

Try flipping a pair of speakers compared both to speech and narrow to see if they behave identically. Speech vs Music made some of my misbehaving units perform the same — letting me know it was the crossover for sure.

 

Without cutting the tweeters out of the array (disconnecting them individually and measuring them individually, then resoldering them individually). I don’t understand how you will know which is bad. The DATs would be measuring all the connected tweeters (through the green PCB traces) at once. Which is perhaps four at a time?

or are you thinking all four may be bad?

 

Let me back up and give you all the raw data and let you all help me interpret what's going on.

What I can note from the testing I did:
1. Only the tweeter to which the leads are connected is energized and generating sound.
2. Each tweeter has a slightly different measurement.
3. Some of the tweeters measure slightly off and others are orders of magnitude off.
4. For 4 of the tweeters in the good speakers, something is affecting their resistance.

Here's the results in spreadsheet form:

Note that when there is a reading out of the "normal" range for the 4 lower impedance positions, the others have readings more like the other tweeter positions.

What are possible causes for the 7 very out of range measurements? Would those causes also pertain to the 2 slightly out of range values?

 

I'm not disagreeing with you that there is more interaction going on with the rest of the array. I'm just not smart enough, nor do I have the full electrical diagram to understand the full list of possible interactions.

My one thought on why only one speaker is being energized is that this is the path of least resistance and thus the circuit is completing across the speaker.

Could it be that the interaction with the rest of the array is only contributing a small amount to the measurement as compared to the amount contributed by the tweeter under test?

 

I de-soldered a 2 tweeters, 1 “good” and 1 “bad”.
Re-ran impedance sweeps on both.
Each had almost the exact same measurement before.
I don’t understand why the test when connected isn’t affected by the other tweeters in the array but that appears to be the case.

Next steps, continue testing the remaining speakers to identify the components not within the tolerances of the others.

 

Super interesting. And makes no sense.

Appears that will make your testing and identification of blown drivers so much easier with the dats tool. I was thinking you might have to unsolder all 16 tweeters to test a single bad speaker.

 

Yeah, I went over to @Montekay ‘s house to help with some of his home theater build out and picked his brain.
He had a super simple test with a AAA battery and touching leads from the tweeter terminals showed the same results my DATS testing did.

The main reason I will go ahead and use DATS on the rest is that I had a few that were just a little out of spec that wasn’t as obvious with the battery test.

@Archaea any hints on removing the mids from the cabinet? The one I tried so far didn’t seem to want to budge after removing all the screws I could find. And also on removing the tweeters and pcb without de-soldering everything?

 

Update:
I finished testing all of the tweeter arrays with DATS. Short of it is that 6 measured fairly consistently and 6 had at least 2 or 3 which measured out of spec or the impedance curve looked significantly different than the other related tweeters.

Using the worst speaker in the bunch as a parts source, I'm going to start repairing the speakers. Once I've disconnected the marginal tweeters, I'm going to remeasure them to make sure they weren't impacted by the other bad tweeters in the array.

Here's the spreadsheet of tweeters
Red = bad
Yellow = out of spec or bad curve
Green = mostly good

 

I am going to use Speaker 9 for spare parts and removed all the tweeters from the array. Once disconnected, I ran impedance sweeps on all the tweeters and were essentially the same as when in the array at a around 14.5 ohm, including the 3 good ones from the middle which had measured 10ish ohm in the array.

As @Archaea found, it's a slow and tedious process to disassemble, de-solder and test all the tweeters. Fortunately, tonight's Youthman podcast was a good listen while working.

I took 3 of the tweeters and swapped them for the bad tweeters in Speaker 8. I got everything soldered and re-assembled. Soooo many screws....I'll do an impedance sweep and frequency test next, hopefully tomorrow.

Thanks for the help and encouragement thus far.

- Derek

 

Success!!

Black line is obviously the repaired speaker. This is with 1/6th smoothing on both.

1 speaker down, 7 to go.... 🤪

 

Looks much better. I have speakers 5, 6 and 7 on the way and they are new/open box. I'm hoping they will serve as good references to measure the hodpodge of others against.

 

I had been looking for another used one to use as another parts speaker but 16 tweeters from JBL are $850 or so shipped, and haven’t seen any cheap ones lately. Sad thing is they really should be like $20, not $50/ea but you’ve gotta pay the JBL tax. 😉

 

JBL PRO has them. Sounds like you have already contacted them. Unfortunately, the price you see is the price you pay.
One of the major negatives to this type of speaker. Refurbishment isn't really a thing with them without DIY, coin, and a bit of luck. Many of the parts are not available for repair purposes, and almost none of the plastic cabinet pieces are orderable. This is why I always recommend buying brand new with the included 5yr warranty and PRO level service. You may be surprised by the actual price you would pay for new.
This is the part you are looking for. Been OOS for a couple of months from this vendor.

Original JBL CBT 70J Tweeter, 444093-001 - The Speaker Exchange

Brand new OEM JBL 444093-001 tweeter 25 mm (1 in) driver, dual magnet neodymium For models: CBT70J, CBT 70J-1, CBT 70JE-1 There are 16 tweeters per cabinet, this listing is for 1 pc

reconingspeakers.com

 

I wish I could have afford 13 of these new or even b stock. That said, even including replacing a lot of tweeters, I’m going to be well south of $500/speaker.

It’s good to know there are other sources for components. JBL Pro prices seem to fluctuate. The first time I called, they wanted a little over $50/ea and this time was around $48. Not complaining on prices dropping, that’s for sure.

- Derek

 

I picked up a used pair of these from epray for $1000 a couple months back, and one of them sounds fine but the other one may have one or two non working tweets. Unfortunately I can't return them. At the end of the day I just don't care for the sound of these so I am back to using My Modulas by Jon Marsh.

 

There are a couple of ways to easily test if you have a blown tweeter, like the AA battery test. A few tweeters being out can make a significant impact. If you decide to get rid of them, let me know. I need a couple more.

 

I am going to get rid of them. When I get home I will check to see if I still have the shipping box.