Uneven response on 2440 with new diaphragm

[Pellets]

Hi all,

recently I fitted new Radian 1245 diaphragms to my old 2440 drivers. The old diaphragms were battered and worn so I expected a change for the better. Well I can't say the new ones have made a big change and I tried to measure the response with may AudioTools app. The response looks quite ragged to me. Is this what to be expected?

I don´t have the standard H93 horn, instead I use the Selenium HL14-50N which some say is a better horn for this driver. Measured with pink noise through a 800/8K bandpass filter.

Best regards

[edgewound]

Hi all,

recently I fitted new Radian 1245 diaphragms to my old 2440 drivers. The old diaphragms were battered and worn so I expected a change for the better. Well I can't say the new ones have made a big change and I tried to measure the response with may AudioTools app. The response looks quite ragged to me. Is this what to be expected?

I don´t have the standard H93 horn, instead I use the Selenium HL14-50N which some say is a better horn for this driver. Measured with pink noise through a 800/8K bandpass filter.

Best regards

Couple things could be going on.

Did you install and test with a sweep generator and listen for anomalies? Did you use any of the included shims? Have you ever had the magnets recharged? Freshly charged magnets can make a huge difference.

[Earl K]

The jaggedness might just be caused by the app's resolution when using Pink Noise.

Maybe you need a better TEST setup ( like REW and a UMIK1 USB mic ) ?

REW works on both Windows and Mac ( and is share-ware >. essentially free ).

Also, try your 2440/Radian combo on a different horn ( if you have one ) to get a sense of changed perspective ( iow; which ripples are horn induced and which are from the driver/diaphragm combo ).

Here's a pretty ancient trace ( made using a Sine Sweep in ARTA ) of one of my 2482 drivers loaded with a Radian R1282 diaphragm and mounted onto an H5038 ( Paragon ) horn.
- Note#1; there's is an 8 ohm resistor with a 3uF cap paralleled to it , both in series with the driver ( this RC combo offers a bit of HF compensation ).
- Note#2; no mic calibration was loaded back then > so the HF is exaggerated by about 3db too much above 10K .


88831

I think my hi-rez curve is a bit smoother than yours.

:)

[Pellets]

Thanks Edgewound and Earl K for your responses. I´ll try to explain better:

The new diaphragms came with shims. I contacted Robert at Radian who said that if the old diaphragms didn´t have any shims then don't use them on the new ones either, so I didn't.
I ran a frequency sweep from 350-3000 Hz through the coil during the assembly and everything sounded fine. The drivers have not been recharged and they seem to have been around for some time. I spent a long time cleaning gap and phase plug.

Test setup: I used pink noise through an electronic L/R crossover with 24 dB slopes and a passband 800-8 kHz. Then through a Rotel power amp to the unit. I picked up the signal with a Pearl studio condenser mic at 1 meter and fed the signal to the SMAART TOOLS app in the Audio/Sound Toolbox on my computer. This is basically a fast RTA which I set at 1/12 oct smoothing with a 8192 FFT size.

What concerns me is the almost 10 dB variation in the passband.

I will try another horn when i can find one. The Selenium horn is not foreshortened and should be able to deliver down to at least 800 Hz, but I understand what you mean that the horn itself can introduce ripples.

And I will certainly try out the REW software to see if the results get better (or my understanding of it)!

Best regards

[withTarragon]

Before you get too caught up with REW, I would go back and try what was mentioned earlier using tone sweeps and listening.

My limited experience has been with slightly newer drivers (JBL 2445 & 2446) and using JBL diaphragms. What I found was 1) even there were no shims previously, I had to fabricate one using thin paper (to prevent it from sitting too deeply). 2) using repeated tone sweeps, I had to torque the screws very evenly to see if the response was even. I then had to go back and try torquing with a bit more force in one quadrant vs another. You can actually turn the screws for a loose fit and then simply push on the diaphragm while the tones are sweeping. This gave me the ballpark positioning.

Two observations 1) when it was sitting too deeply, there was a harshness (harmonic distortion) at the lowest frequencies (quite audible). 2) when the diaphragm was not centered, there was low pass filtering effect. The output was decreased overall, but even more decreased at the higher frequencies. In your case, a 10 dB difference would be "audible" as about "half as loud", Ultimately REW will verify this.

Two comments, 1) In the past, some said these are self-centering diaphragms and you do not need any adjustments using tone sweeps and ultimately REW. That is not the case for the 2445 / 2446 units. The 2440 is older so my guess, and it is only a guess, you may be having the same problem. 2) this is a tedious back-and-forth process and will take some ingenuity and dexterity (I assume the pros have some sort of jig to help out). It may not go quickly, so it is much faster to do it with tone-sweeps and your ear for the initial positioning. Stopping to do REW measure every time would take forever. Save REW for the last stage.

Also, I know you mentioned that you cleaned out any debris (and I assume you did not use composed air). Are you confident you did a thorough job?

good luck,
-Tom

[grumpy]

I don't know... if I eyeball average, it looks like a quite expected first cut at a phone measurement.
Definitely room for both kinds of tools for the task at hand.

[Pellets]

Hi again,
I´ll try to wrench the monster out of its box this weekend and have a go at the alignment and the shims issue. Radian said that the shim is mostly used by those who need a greater clearance between diaphragm and phase plug which might be necessary at higher power levels. The high frequency rolloff will then be lower.

When you fit the diaphragm there are two locating pins securing it into a centered position. Maybe there is some play there anyway. But you should be able to hear some unwanted buzz when you run the frequency sweep and it's not properly centered, otherwise I don't now how to proceed with this.

I took my time to clean the driver before installing the new diaphragm. The gap is wide enough to be inspected with a light and magnifying glass. I was mildly surprised to find all that debris in the gap though!

So hopefully I´ll be back with some improved response curves soon.

Best regards

[toddalin]

I assume that your measurement was conducted in an actual habitable room with furniture as opposed to an anechoic chamber and this can make a big difference.

[Pellets]

Yes absolutely. The 2440:s sit in Olympus cabinets in my living room. They are not easily moved about. I measured at one meter distance from the midrange horn without the grille. The 2440:s are a devil to disassemble but I will do it.

On the other hand there is this school of measuring real close to the radiator thus reducing the room effects. It's easy to check whether this is of any importance or not. I´ll come back to that.

Best regards

[Pellets]

Hi again.
I got the driver and horn out of the cabinet today and I opened it up for inspection. The gap was clean and after replacing the diaphragm I ran a frequency sweep which was OK. There is no way to move the diaphragm assembly once it is set onto the locating pins.

I made a first test with pink noise at one meter distance and one close up to the horn mouth. See attached graphs. After that I fitted the shim that came with the Radian diaphragm and ran the test again. To me the response looks much worse with the shim. I have adjusted the output level between the readings in case you wonder.

It seems now that there is a variation of about 5 dB or so in the passband. Maybe that's as far as it goes with this driver/diaphragm. What do you think?

Best regards

[Mr. Widget]

If the diaphragm is snug on the pins and there is no buzzing, you are likely properly centered.

As for the the apparent choppy response, have you tried other horns? This type of response can easily be caused by the horn itself. Most horns are far from linear.


Widget

[grumpy]

what are you using for measurement settings? averaging? (yes? or 4-8s delay) "Advanced FFT settings"(Hamming window, 50% overlap)?
There's an unusual very low-resolution look to the plot that has me wondering about the settings...
The FFT plots from the StudioSix AudioTools app on my phone looks quite similar, but much better resolution (including when noise is used as stimulus).
I expect your driver is fine, and there is something going on with the measurement.

[Odd]

As recommended earlier here download REW (https://www.roomeqwizard.com/) (free)
Support Norwegian here (https://www.hifisentralen.no/forumet/threads/rew-m%C3%A5linger.79532/)
Support Forum (https://www.avnirvana.com/forums/official-rew-room-eq-wizard-support-forum.10/)

[Pellets]

Hi again.
Sadly I don't have a 2" horn available right now. I can guess that neither the H93 nor the Selenium are perfect, and as soon as I can I´ll buy - yes, what? Would a horn like the 2380 fit? I read somewhere that the throat geometry is different between old and newer JBL drivers.

The difference in response between closeup and one meter distance are maybe due to room effects. It actually looks better at one meter though.

As for the SMAART Tools setup - I don't have the FFT windowing selection feature and I selected a FFT speed of 8192 because it was recommended as a good tradeoff between accuracy and speed. Nor do I have an Average decay setting - I have used 6 seconds delay but maybe I should have tried up to 10 secs (Infinites seems to save the top readings). For readout I used the 1/12 octave RTA line setting. Perhaps it's better to go for the solution that Earl K and Odd recommend with the REW software.

It's interesting though that once I got the driver/horn out and onto the top of the cabinet the response became notably smoother. But I can't have it like that, people here already think the speakers look a bit overwhelming.

Best regards

[Pellets]

Hi again. It took a while to sort out the REW software but now it seems I have a grasp of the functionality. After some calibration I ran a curve of the 2440 through my electronic crossover (800Hz-8k@24 dB/oct).

The settings in the REW RTA were 1/12 oct 8 average 50% overlap. I used the PN pink noise as recommended.
Here is a picture of the response. This time though the driver/horn was mounted in the cabinet and not free standing as in the previous pictures.

The curve looks quite similar to the previous one apart from the very deep notch at 2.2KHz or so - this could be seen on the previous curves but not that deep. Maybe it's exaggerated by the baffle. Anyhow it seems that the response is quite ragged with at least 5dB variation in the passband. What can be done? Can I measure better?

I'm on the hunt for 2311 horns but they are very scarce. The 2380 is too big for the cabinet. Any ideas?

Best regards

[Mr. Widget]

Here is a link to a thread I posted long, long ago. This was my first pass at measuring speakers. I was using a top flight measuring system with calibrated mic, but I was just beginning to learn how to use it. I have subsequently learned a lot and have been able to make much more meaningful measurements.

That said, the 2441s were essentially brand new. I had sent them back to JBL who at that time would still strip down a driver and rebuild it at their factory.

Take a look at these posts and view them realizing that the person running the measurements was really pretty clueless at that time.

http://www.audioheritage.org/vbulletin/showthread.php?3849-JBL-2397-Curves



Widget

[Pellets]

Thank you Widget for passing on this information! I have read it with great interest and will try some of the tricks used for measuring the frequency response. I'm not too concerned by the ragged graphics in the measurement but more the shouty nature of the driver/horn combo, a bit like what you could expect from older Lowther drivers. I will renew my efforts to find the original 2311 horns just to have something to compare with.

I have probably seen pictures of the 2397 horn or similar mounted on top of Olympus cabs, they seem too deep to fit inside the grille.

I´ll come back with a better curve if it also sounds better... it takes some time and experience to make the best of REW (or for that matter the Sound Tools app).

Best regards

[Earl K]

@Pellets

You should be able to achieve smoother response than what you are currently showing.

( Using REW ) run a Full-Range Sweep measurement ( using the normal REW sine stimulus ) that starts at 200hz and then goes out to ( at least 22,000hz ).

Then open & look at your Distortion window , it should look something like this ( this is my 2482 with a Radian R1282 diaphragm mounted on the Selenium HL14-50 conical ).


89014

Pay attention to the profile/tracking of the Harmonic Distortion ( HD ) traces > 3HD should always stay below 2HD ( from 1K to 10K ).

Your diaphragm might not be properly loaded/seated ( that'll certainly be the case if you see a spike in 3HD distortion anywhere from1K to 10K that juts//rises above the 2HD trace ).

FYI, the curve that I show is very sweet sounding with Bach Violin Concerto's playing ( even without a woofer working below it ) // so I think you have a couple of problems to sort out ( due to your stated discontent of your setup).

Also, I very much doubt that 2311's will solve your present discontent ( I sold off my 4 2311's ) since owning the Selenium made them acoustically redundant ( my Seleniums were purchased at least 20 years ago ).

- Personally, I would run your 2440/R1245's full-out, past 10K, and then simply cross-in a tweeter above that point.

:)

[Riley Casey]

If you have significant differences in response when the horn / driver combo is measured stand alone versus mounted in the box behind a grille I suspect it's time to start looking at mounting problems creating resonances or bad grille design.


... This time though the driver/horn was mounted in the cabinet and not free standing as in the previous pictures.

The curve looks quite similar to the previous one apart from the very deep notch at 2.2KHz or so - this could be seen on the previous curves but not that deep. Maybe it's exaggerated by the baffle. Anyhow it seems that the response is quite ragged with at least 5dB variation in the passband. ...

Best regards

[Pellets]

Thanks Earl K, that is potentially very interesting. I´ll try to emulate your measurements - were they done at 1 metre distance? I had a look at the test measurement I did when trying to understand REW and it doesn't look like yours at all... OK, I will take the driver and horn out again and do a new more controlled measurement. This weekend hopefully if my back ache disappears. I´ll come back as soon as I have something meaningful to show.

Riley: I measured with the Olympus grilles off. But the horn is mounted at the top of the baffle with many bits and pieces around it possibly disturbing the frequency curve. We´ll see if this will be bettered when the unit is out of the box again.

Best regards,

[Earl K]

Thanks Earl K, that is potentially very interesting. I´ll try to emulate your measurements - were they done at 1 metre distance? I had a look at the test measurement I did when trying to understand REW and it doesn't look like yours at all... OK, I will take the driver and horn out again and do a new more controlled measurement. This weekend hopefully if my back ache disappears. I´ll come back as soon as I have something meaningful to show.

Riley: I measured with the Olympus grilles off. But the horn is mounted at the top of the baffle with many bits and pieces around it possibly disturbing the frequency curve. We´ll see if this will be bettered when the unit is out of the box again.

Best regards,

My measurement was done at 27" ( not 1 Meter // 40 inches ) .

Distance won't matter that much when checking for 3HD distortion ( assuming 1 meter or less ) .

From a subjective ( quality ) listening perspective > you ought to try a 10db ( resistive pad on those drivers ) to see if they mesh better with the woofer below them .
- This ( pad usage ) assumes you can rebalance the system parts after knocking down the 2440 driver level by that much.
- A 16R resistor paralleled across the driver terminals and then another 16R resistor in series to the Hot terminal of the driver gives this approximate 10db pad.

The idea behind the pad is to better align the dynamic qualities of the disparate parts .
- IOW; knock a 110db CD horn driver combo down to 100db ( dynamic potential ) to better track ( dynamically ) with a 95db woofer.

IanMac has reported ( quite recently ) that this is something that Greg T. does ( and from my own experience is entirely correct ) .
- This treatment makes multiple drivers behave just a bit more like a full-range speaker ( so, the common resistor can be your best buddy here > 5 or 10 watt wirewounds would do for now ) .


Here's a look-see ( of my 2482/Radian on the Selenium HL14-50 conical ) with the 10db pad in place ( it's the green trace with 1/24 smoothing done in XSim // while the red trace is the raw response that I captured before manipulation ).
-

89040

:)

[Earl K]

One other thing to try Pellets ( in the search for better SQ > sound-quality ) is to do a passive/active hybrid when dealing with the 2440 Horn combo.

- While still Bi-Amping ( "Y-Split" the drive signal going to the Horn amp before your electronic crossover ) and then use a passive filter on the Horn/Driver combo by implementing something like in the following pic.
- ( balance levels by using the volume knobs on your horn amp )

With the following values you'll get a response curve very much like this ( this is highly listenable IME ).

89041

This offers you an approximate 650hz crossover point ( low I know > but these big 4" diaphragms can handle this "abuse" at home listening levels ) .
- Just adjust your existing electronic crossover ( for the woofer's low-pass ) for this new point.

:)

PS; "R4" + "L2" are out-of-circuit > IOW > they are just unused placeholders equating to straight pieces of wire .

[Riley Casey]

One big short coming of REW ( or perhaps a get what you pay for element ) is the lack of any coherence data. A narrow notch like the one one in your curve at around 2.3 kHz is often a result of cancellations from a variety of possible sources like reflections or vibrations. A coherence curve like that in Smaart will alert you to the reliability ( or lack there of ) of the measurement. I would be suspicious of validity of that notch particularly if it's not duplicated in measurements taken with the horn outside the box.


Hi again. It took a while to sort out the REW software but now it seems I have a grasp of the functionality. After some calibration I ran a curve of the 2440 through my electronic crossover (800Hz-8k@24 dB/oct).

....
Best regards

[Pellets]

Hi all helpful people,
I just want to share some graphs from the latest measurement on my 2440+HL14-50 combo. The unit is now out of the cabinet and resting on a table. Measuring distance 27". Sorry about the levels, I didn't have the possibility to calibrate levels but they are reasonable (not 150 dB...). Frequency sweep 200-22 000 Hz.

First I measured the unit as is. See the graph, the bad notch at 2.3 kHz is gone but there is quite a huge variation across the passband. Also the 3d harmonic distortion is quite high.

I then took out the diaphragm, cleaned both the gap and the coil carefully, inspected everything and put the diaphragm back on its pins. I tried to move it sideways but there is no play. Tightened down the screws gently and put back the lid. Now look at the second graph. Different isn't it. I have no explanation other than sheer luck. So maybe this is as far as I can get with this particular unit or what do you think?

I´ll come back on the other advices you kindly sent me.

Best regards

[Pellets]

Hi. I admit the graphs from yesterday didn't look too good. Here is one with calibrated levels and correct frequency span and also with 5th harmonic distortion. /P

[Tim Rinkerman]

Have you tried changing the tightness of the screws while passing sine waves through the driver? Just because the driver has alignment pins and they are snug in the holes, doesn't necessarily mean the coil is centered in YOUR driver. Electro-Voice used to be notorious for having to ream the locating holes to get the diaphragm to seat properly, no reason not to suspect Radian of being off a few thousandth's here or there.
People tend to take replacing a diaphragm as a "remove and replace" operation, but there is far more to it to do it correctly. You want clean, smooth response at every frequency..but sometimes it takes hours of tightening, retightening, reaming the alignment holes and "tuning" the diaphragm until it's really right. There is definately an audible difference between one that has been done right and one that was just put in and not clearanced and torqued down correctly.

[Mr. Widget]

Are we sure there is a problem here?

I am not familiar with this horn, but these curves don't scream there is a problem to me just looking at them. This horn and driver will never give you a beautiful textbook case curve.


Widget

[Pellets]

Hi again.
Thank you Widget, maybe it doesn't get better than this with a passband variation of around 5 dB. I compared my latest curve to the one Earl K showed in one of his answers and it seems we are pretty close. His driver is about the same as mine but I guess it has a phenolic diaphragm.

The worry I have is that I don't really know why the response improved so much just by me taking out the diaphragm, cleaning gap and coil and putting it back again. I tighten the screws as you would on a cylinder head, crosswise, and I try to tighten evenly. A more scientific approach would perhaps be to borrow a torque screwdriver and see if the trick is to get the screws down as evenly as possible with a reasonable end force.
To randomly adjust the screws and measure the result would probably take a long time and one problem would be how to go back to a previous better setting.

I read in the brochures (or somewhere) on the 2440 that the diaphragms have the advantage of being field replaceable and if this is so I think there has to be a more definitive way to go when doing the job.

I´ll pull out the second 2440 soon to see if it behaves in the same way as the first one. Next step is to try to implement some of Earl K:s suggestions. And finding a torque screwdriver...

Best regards

[Mr. Widget]

His driver is about the same as mine but I guess it has a phenolic diaphragm.His 2482 driver and aftermarket diaphragm are different from your driver/diaphragm, but also according to the name of the png file the plot is of a driver with HF compensation... it would take quite a lot of compensation to achieve that HF performance from a phenolic driver.


The worry I have is that I don't really know why the response improved so much just by me taking out the diaphragm, cleaning gap and coil and putting it back again. I tighten the screws as you would on a cylinder head, crosswise, and I try to tighten evenly. A more scientific approach would perhaps be to borrow a torque screwdriver and see if the trick is to get the screws down as evenly as possible with a reasonable end force.
To randomly adjust the screws and measure the result would probably take a long time and one problem would be how to go back to a previous better setting.If you open up a bunch of these, you will see notations with + and - numbers that had to do with shimming and adjusting the diaphragms. Also, sometimes you need to ream out the guide holes to let you wiggle the diaphragm side to side just a bit to get the best alignment... then there are the inherent inaccuracies of the diaphragms themselves. Not all brand new factory built diaphragms are perfect. :banghead:

After measuring a few drivers and realizing how difficult it is to get them really well matched, I lost interest in repairs and shipped mine back to the factory to rebuild (no longer a possibility) or sought new drivers. Of course that is no longer possible either for so many of these drivers.

At the factory they had a test jig set up to verify performance as they installed the diaphragms... few of us can match that.


Widget

[Earl K]

Yes, my 2482 is loaded with an aluminum domed, Radian diaphragm ( R1282 ). That's why it has the HF response that you see.
- The 2482 with the original phenolic diaphragm drops like a stone above 4.5K ( and no HF compensation can alter that reality ).

http://www.audioheritage.org/vbulletin/attachment.php?attachmentid=89074&stc=1&d=1623165914

Your curve ( seen immediately above ) shows a natural setup where one could run the driver/horn combo full out ( HF wise ) > and then cross in a tweeter just above 10K-12K ( should sound nice ).

:)

[toddalin]

I would not doubt that removing/cleaning/retorquing the screws could be beneficial, or detremental.

Anyone who has ever wiggled/loosened/tightened the screw/phase plug on a 2402 and it's family, knows that if the diaphragm is allowed to twist or flex at all, it changes the response.