Jump to content

Measurements from BIG CD's and Horns


Recommended Posts

As some of you probably know I switched out the top end of my big warehouse system a long time ago. I went from JBL 2445H's on 2360A's to Celestion AXI2050's on Klipsch K-402's. I took a bunch of measurements but never got around to posting any of it. 

I'm not going to devote too much time writing up a big break down of all of this. 

The short version is the K-402 horn is an overall improvement over the old JBL 2360A in most ways. The JBL does offer better loading down to deeper frequencies, but has more acoustic issues overall. Resonances etc...Physically the K-402 is slightly, smaller, lighter and less deep too. 

The old 2445H CD is not a cutting edge design any more but it is surprisingly viable still. 

The AXI2050 is considered cutting edge and it does offer some intriguing benefits, but it isn't all wins against the relatively ancient 2245H (or other more conventional 4" CD's).

My short take on the AXI2050 is this: It offers excellent lower midrange performance with the possibility of really using it XO'd down to 350-400Hz (with less than full out pro use SPL and duration), or 500-600Hz at full on pro / arena use levels and depending on the horn/ waveguide used with it of course. Coupled with large horns such as these, the SPL and distortion results in the lower midrange are impressive. This driver does technically make it up to 20kHz but it requires a really healthy dose of EQ, even more than with most large CD's. The old 2445H kicks the AXI2050's ass in the top octave. The AXI2050 sensitivity and response above 10kHz is serviceable, but not great. Where it shines is in the 400-8kHz bandwidth. It measures well and sounds stellar in this range. Other options cover >10kHz much better with greater sensitivity. Even after EQing the top octave into shape it only sounded adequate. It wasn't bad but it wasn't anything to write home about either. The midrange on the other hand sounded great. Also this is a really robust CD. It will handle border line abuse that I'm not sure some other CD's would. 

In summary. AXI2050 = Killer midrange / so so top octave. I ended up switching to and keeping the AXI2050 because the midrange is way more important to me than top octave treble. 

  • Like 1
Link to comment
Share on other sites

Impedance measurements of both JBL 2445H CD's. Overall the matching of the drivers isn't bad considering their age. Unfortunately Mounting the 2445H to the K-402 would've required cutting into the mounting bracket to clear the magnet diameter. I didn't feel like taking that step. 



Impedance of both 2445H CD's loaded on 2360A horns. 



Impedance of one of the 2445H CD's without a lens and loaded to the 2360A.





Impedance comparison of AXI2050's. 



AXI2050 impedance with no lens vs loaded onto 2360A



AXI2050 impedance with no lens vs loaded onto K402




AXI2050 impedance on K402 vs on 2360A



2445H vs AXI2050 raw impedance





2445H vs AXI2050 both loaded onto 2360A horn



  • Like 1
Link to comment
Share on other sites

This is all taken in a LARGE warehouse room with a fairly high background noise level and I wasn't super strict about the minimizing mic reflections,  so take that for what it is. The measurements are taken at 1m and gated below 150Hz. Ideally the measurements would be made at a further distance from these huge horns and in an outdoor setting to get into the far field. This is the best space I could get and what I had to work with at the time. I figured I'd note this lest someone thinks this was an outdoor test, or a lab grade endeavor. 



JBL 2445H on 2360A frequency response. Comparison of both drive units. 





AXI2050 on 2360A frequency response. Comparison of both drive units. 





AXI2050 vs 2445H...Both on 2360A Horn






AXI2050 on K402 frequency response. Comparison of both drive units. 





AXI2050 on K402 vs 2360A comparison












  • Like 1
Link to comment
Share on other sites

I wanted to see just what these big horn / CD combos are capable of so a bunch of increasingly loud measurements were taken to capture output compression and distortion. I used 512m length sweeps. Even with earplugs and shooting phones this got to be really annoying. I stopped with the JBL's at 16 volts. I quit at 22.5 volts with the AXI2050, before I got to 28.3 volts (100W nominal). They would've survived at 28.3 volts, if the sweep was limited to above 500Hz I'm sure, but 130dB in the 500-2.5kHz region is brutal!  Burst testing to look at dynamic capability, would've gotten really ridiculous. I decided to forego that. 


First up the 2445H on 2360A.



Increasing drive levels. 



2445H on 2360A. The same data as above presented normalized as output compression. Note the expansion near 10 and 12kHz likely due to shifting of resonances. 



JBL 2445H on 2360A. THD from the same measurement set. 




Distortion makeup at 2.83V



Distortion makeup at 15.9 volts.




Horizontal response. 



Vertical Response.


  • Like 1
Link to comment
Share on other sites

Measurements of the AXI2050 on the JBL 2360A horn


Sweeps at increasing voltage.

Note the 22.5 volt sweep goes down to 200Hz!



Normalized data showing only the amount of compression occurring as the voltage is increased. 



THD captured during the measurements. 



Distortion harmonic makeup at 2.83 volts



Distortion harmonic makeup at 22.5 volts



Horizontal response. Note the 75deg measurement is missing.



Vertical Response.



  • Like 1
Link to comment
Share on other sites

Measurements of the AXI2050 on the Klipsch K402 horn.


Measurement sweeps at increasing voltage.

Note the 22.5 volt measurement cuts off at 300Hz. 



The same data normalized to show only the compression of the output as the level is increased. 



THD captured during the measurements. 



Distortion harmonic makeup at 2.83 volts




Distortion harmonic makeup at 22.5 volts



Horizontal response. 



Vertical response



  • Like 1
Link to comment
Share on other sites

To wrap up here are some comparisons. 

AXI2050 vs 2445H both on the 2360A horn. 


Both are nominally 8ohm drivers but the AXI2050 has a much higher avg impedance.



Response and voltage sensitivity.

The 2445H shows higher sensitivity, smoother response and much more extended high frequency response on the 2360A horn.



Maximum level reached during testing. 

2445H is with 15.9 volts input and limited to 400Hz to prevent damage. AXI2050 measurement is at 22.5 volts and starts an octave lower at 200Hz. Despite the higher voltage during the loudest measurement the output of the AXI2050 still lags behind the 2445H above 5Khz.



THD during the above measurements. NOTE: The JBL was receiving 15.9 volts. The AXI2050 was receiving 22.5 volts. In general the AXI2050 has lower distortion. Keep in mind it is producing higher output below 3.5kHz and less output above 5kHz. Reference the graph above. 



2.83v THD comparison. The AXI2050 is showing lower distortion but, note that at this input level the 2445H is producing more output above 500Hz. Reference the 2nd graph in this post. 



Output compression observed at the maximum output level reached. 15.9 volts for the JBL 2445H and 22.5 volts for the AXI2050. Again keep in mind that the AXI2050 is taking more voltage and it is producing higher output below 3.5kHz and less output above 5kHz. Reference the graph above. The AXI2050 is exhibiting much better and more uniform behavior here. The 2445H driver  and 2360A combo exhibit some strange behavior in the 8-16kHz octave in this test. Most likely related to resonances and response shifting. 



  • Like 1
Link to comment
Share on other sites

Comparison of the 2360A and K402 horns using the AXI2050 driver. 

In general the K402 seems to be a better more "hifi" design. The 2360A loads obscenely low in frequency and has good pattern control but it has a lot more pronounced issues and these can be seen in the measurements. Still it is a useful and listenable horn decades after being introduced. I used them for a very long time with the 2445H drivers tested here and enjoyed them. There was always a bit of extra "character" to the sound of them that I never could eliminate fully, but they had some positives also.   The K402's just seem to be easier to work with and get out of the way a little more, despite also being a giant horn. 


Impedance. This measurement alone shows about just how different these 2 horn designs are and how they match up to the driver. 



Frequency response. The K402 appears to be a better match for the AXI2050 driver. The response is more sensitive and smoother overall. The 2360A has the advantage in LF extension and output <500Hz. 



THD at 2.83 volts. Reference the graph above. The 2360A exhibits lower THD below 400Hz but much higher THD throughout the rest of the bandwidth.



22.5 Volt measurement comparison. 



22.5 volt THD comparison. Similar to the much lower volume 2.83 volts measurement, the same driver loaded onto the 2360A exhibits much higher distortion in the output. 10-20% THD vs only 5-10% or roughly half when loaded on the K402 horn. Note this is a brutally loud playback level. 



A comparison of the output compression occurring during the 22.5 volts measurement doesn't show much difference. There's perhaps a very slight advantage for the 2360A data but the K402 was averaging just slightly higher output at the same voltage. 


  • Like 1
Link to comment
Share on other sites

That's a lot of data! I like that most of the distortion is the relatively inoffensive 2nd harmonic. The near 130db sweeps must've been a pain, I'm already annoyed when I take 100db sweeps with hearing protection in the studio. I have a vacuum cleaner hose which has an incredibly annoying resonance. I suspect it produces something like 110db at around 10khz. It's almost unbearable when I use over ear hearing protection with IEMs beneath, playing music.

I'd be interested how the B&C horn with the DCX would compare. It's a fair bit cheaper too, I suppose. 1k+ for a compression driver is insane. Similar price regions to the Beryllium ones I forgot the name of.

It's good to see that power compression doesn't distort the frequency response too much and is only 1.5db approaching 130db.

Link to comment
Share on other sites

It was most definitely uncomfortable when it got loud. I thought doubled up ear plugs with shooting muffs would be enough. I was noticing a weird sensation of HF tactile feel at frequencies you don't normally feel. Pressure, etc.

No doubt that it's not going to be easy to blow a 2" exit CD outside of legit large scale events. Unless you are some kind of masochist, run it with no HPF, or hit some sort of turn on thump / cable disconnect snafu, I don't see it happening. 

I debated between this driver and the Radian Be driver. I was going to go that way but right when I went to purchase the price went way up. It was near $1200 if I recall. At the time I was able to get the Celestions for about $900. I do wonder how those Radian's would perform. I know they wouldn't have the lower midrange headroom or be able to cross as low, but it would likely have been good enough for a 500-600Hz XO. The top end would've been better I'm sure. 

Too bad it's not easier and cheaper to test this type of devices like this. It's a bigger PITA than subs! Harder to get an adequate environment, more difficult to setup and easier to screw up. Plenty of measurement sources out there but no one really goes past 1w equivalent voltages. 

Link to comment
Share on other sites

I wonder how the 18sound BE driver compares, it sure is expensive too.

Measuring tweeters shouldn't be that hard. Get a room that is big enough for a 5-10ms gate time and you can measure down to 500hz quite well. I bought and compared a bunch of tweeters and mids for current builds. Put them on the intended speaker baffle, EQ flat and compare distortions at varying levels. Of course not the massive IEC baffle :)

In room gated distortion measurement of an inexpensive SBacoustics tweeter on a tiny 4x4" baffle. 8' ceiling height:



Link to comment
Share on other sites

Nice results. Lot's of good tweeter tests out there. Not many that check the behavior at >100dB though.

I'm talking about the big CD and horn combos. There's much less info out there on those and it's more dispersed. That JBL horn + CD combo weighs probably 80lbs / 36kg and is 775mm square. Awkward as hell to move around and do polar measurements. Then there's the loudness factor. 

Link to comment
Share on other sites

Yea, I typically only test up to 100db, since I mostly test HiFi components. I put DE360 into my center speaker, which would've been more annoying to test. Being inside a small room only a few feet away from a 120db HF source.. I don't know if I need that. But I know that my center speaker will also never reach those levels so it's fine for now :)

Link to comment
Share on other sites

Very useful data.  I'm impressed you tested close to 130dB I have tested at ~115dB before and also experienced this tactile HF sensation.  Under 10% HD in the 22.5V test of the AXI2050 on K402 is impressive.  Its a shame the Klipsch charge so much for this horn.  I did consider buying a pair of 2360A until I heard a them in a club in Berlin (more of a listening room).  Initially I the sound was very impressive but after a while I noticed an unreal kind of cupped nature to the midrange which would have annoyed me if I had bought some.



Link to comment
Share on other sites

Lots of nice data here!

I believe teeth may resonate with treble, and then you have bone conduction.  That may be where the pressure sensations are coming from.

Gating is a very useful technique for getting good high frequency data outside of ideal measurement conditions, but keep in mind that the trade-off between window size and "low frequency limit" is actually a *frequency resolution* trade-off.  For a given window, the resolution in *absolute frequency* is the same.  That means that the log-frequency resolution (what we usually care about) drops by half with each successively lower octave until the "low frequency limit" where the data is no longer useful at all.

The high end on the AXI2050 + K402 here may be attenuated, but the dispersion looks pretty good and the response looks EQ-able, at least in theory.

I couldn't find any specs at all on the K402.  How big is that thing?


Link to comment
Share on other sites

The K-402 is listed as 39.5x22.5x16.6". It's also a BIG piece of kit, but not quite as bad as the 2360A which is just huge. 

Shelf EQ is required for 2" CD's on setups such as these. I don't push the top end all of the way flat on axis. I generally try to go for something similar to the Harman or JBL listening curves. A gentle and gradual slope from the lower midrange to the extreme treble. Flat from the mids to the treble sounds harsh to me on this type of setup. It's too much HF energy. Even with not EQing the top end all the way to flat it takes a lot of boost >8kHz. Thankfully there's little avg power in that bandwidth and big CD's are rugged as far as tweets go. 

To me the piece of data that sticks out like a sore thumb are the THD results between the 402 and the 2360A. A lot of prevailing wisdom says we are very insensitive to THD and have trouble reliably hearing differences until they become grossly distorted, but none of the other data seems to show such a large difference. The polar response is different of course but I don't think this is why the K402 sounds smoother/ more "Hifi" to me with both EQ'd and on axis. The 402 exhibits about half of the THD at roughly the same output levels. This elevated THD and cupped hands sound effect at some frequencies, is likely due to the narrow / deep diffraction slot + some amount of increased throat distortion. The 2 to 1 THD ratio tracks with the output level. I wish I had data for the 2445H on the K402 to corroborate, but it wouldn't fit the K402 due to the motor diameter. 

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Create New...