Using JBL 4333 and 2402 (Crossing Over, etc. Sanity Check)

[kevinkr]

I've been a long time user of a pair of JBL C37 Rhodes with D130 woofer and 075 ar horn tweeters. I generally enjoy the sound quality this set up gives me with a variety of small (tube) SET amplifiers I have designed and built over a number of years.

I am upgrading at this point, have a set of Onken boxes being built which when finished will be equipped with Iconic 165-8G 16" woofers.

I have acquired a pair of JBL 4333 extended bandwidth mid horn drivers. (2420 driver, 2312 horns, 2308 diffusers) These are 16 ohm drivers with alnico magnets.

I have also acquired a pair of 2402 which are very similar to the 075, these also have alnico magnets but are 8 ohm drivers.

I have heard it alleged that the vc in the 2402 is the same for 8 and 16 ohm versions..

I plan to parallel the 4333 with 16 ohms, will probably use 8 ohm LPADS with both mid and tweeter for initial level matching to the woofer and each other.

Eventually I plan on active x-overs and bi or tri-amplification, but for now I am going to design a passive x-over for the system.

This driver combination was common in some pro sound applications, but I have not been able to find a schematic on Lansing Heritage or anywhere else.

The 2420 is usable from 500Hz on up, but the 2312 and 2308 diffuser are designed for cross-over at 800Hz, and I was planning to cross-over to the 2402 somewhere between 7 - 8kHz in case I decide to change to the 2405. (Otherwise I could cross-over somewhat lower if I am only going to use the 2402.)

I thought 2nd order cascaded butterworth, high pass, low pass for an 800Hz - 8kHz bandpass for the mid horn, and a butterworth 2nd order high pass for the 2402 as I am crossing it over well above its natural roll-off somewhat below 2kHz.
(JBL 075 use 1st order at 2.5kHz in their 2 way systems)

The woofers will be crossed over using 2nd order butterworth filters -3dB @800Hz.

Should I (eventually) consider Linkwitz-Riley and given the additional complexity, why? I know they offer much greater attenuation in the stop band and a steeper slope in the transition region + drivers are 180 degrees out of phase at the cross-over point.. I am concerned about in band phase response and group delay characteristics though..

Also the current crop of x-over designers require frd and xma files that are not available for these drivers so it seems that I can't use the most advanced x-over designer software available.

Does anyone here have any experience using these drivers in a high efficiency system such as mine?

I need a sanity check. Any thoughts at all are welcome even if not entirely germaine to my questions.

[kevinkr]

Completed passive cross-over design..

Went with standard butterworth LPF for the woofer, -3dB, -90 degrees @ 800Hz, midrange is -3dB, +90 degrees at 800Hz and -3dB, -90 degrees @ 8kHz, and tweeter is -3dB, +90 degrees @ 8kHz.

If I am not mistaken this means I should invert the polarity of the midrange connection, but not the woofer and tweeter?

I will be using 15W, 8 ohm Lpads to control the tweeter and midrange driver levels, they are considerably more efficient than my chosen woofer with the box tuning I am using. (about 8dB or so difference)

Ordered x-over parts today from Madisound. Solen film caps and madisound 19 gauge air core inductors. (I don't need that much power handling.)

Any thoughts anyone?

[kevinkr]

No one seems to have had any thoughts.. For anyone who might be thinking about doing anything similar the results so far are quite encouraging.

Anyway an update seems in order. My 2402H tweeters arrived yesterday, one of them unfortunately was missing the diaphragm, the other one works fine. A new JBL OEM diaphragm is on the way from the seller who fortunately seems like a decent sort..

The 2402 (075) horn and phase plug polish up really well with MAS (metal polish from the local ACE hardware store) and look almost as good as new. Very easy to do with the driver apart, and not too difficult even on an assembled driver.

I built the last iteration of the 3 way cross-over I asked for advice about. It is a butterworth implementation with -3 dB points & +/-90 degrees phase shift at 800Hz and 8kHz. I used LPADs to allow me to set the 4333 midrange horn driver (2420/2312/2308) and 2402H tweeter output levels.

I built the 2 x-overs on little breadboards I made from a plywood base and maple planks for the front and rear panels. The level control pots are on the front and an excessively fancy set of 8 "high end" binding posts are on the rear. I managed to shoe horn the 4 inductors, and 6 capacitors into about 24sq in of space - it's a bit tight. Everything is accessible for easy modification.

It is obvious from the level pot setting that the 4333 is insanely efficient, at least 3dB higher than the 2402H. Interestingly it is quite beamy without the diffuser installed, even more so than I had expected. In this respect I assume it is quite similar to the LE75/85 horns used in early JBL consumer speaker systems.

Phasing of the drivers is important in the cross-over region, and I spent several hours determining whether I had them phased correctly.

Transition from the midrange to the tweeter is very smooth, I am not aware of it happening. Likewise as long as the level is correct the transition from the woofer to midrange is also unobtrusive.

I am going to acquire a Behringer ECM8000 instrumentation microphone in the near future and short of recommendations for a cheap/flat mic pre with phantom power that I don't have to design and build myself I am just going to buy a Behringer UBS802. Any suggestions? Please.. This will be used in conjunction with AudioTester software and an M-Audio 2496 that lives in my digital media server.

Later once I have some actual room measurements I will apply whatever correction is required to the passive cross-overs. SO far I am not aware of any serious deficits, but I am sure I will find something.

I designed all of this stuff blind, (and also my first speaker project in 30yrs) and probably just a bit ambitious for my "first" project. I think to some degree I am getting away without preliminarily measuring the individual drivers because the JBL drivers are relatively flat over the narrow range of frequencies they have to cover, and these ancient drivers were used with simple x-overs and still performed pretty well.

These are going to be used with the Onken bass cabinets I am working on. The efficiency of the top section is sufficiently high that I am not going to have a problem matching these to the Onkens which will be in the vicinity of 97dB excluding half space effects. (They won't be installed quite in the room's corners.)

Edits to fix typos, etc.

[ocaukrell]

Sorry cant be of any help.
Although the timing is interesting.
My father has speakers he built many years ago as he could afford them.
2215b 16ohm
2420 16ohm with horn and lens
2405h 8ohm
Now after all this time he's realized the 3115 xovers are not that good.
Now we both need to sit down and try to design a xover for these speakers.
Only info I found was schematics of the original xovers from jbl.
And spec sheets for the drivers.
Not what I'd call inspiring. Very crude and basic. (xovers that is)

[kevinkr]

Eventually I'll have the schematic of my x-over in pdf form and would be happy to share it with you. I would love to see that jbl x-over schematic at some point as well.

[ocaukrell]

That would be great (pdf), send me your email and I'll foward on what info I have.
chomper7 at tpg dot com dot au

Obviously replace the at with @ ,the dot with . and remove spaces.

Have asked Troels if he can help, but as I suspected unless he's got actual drivers to measure, he can't help.
Pity, have absolute faith in his work.

Nigel.......

[kevinkr]

Funny how nobody caught the fact that my cross-over points were -3dB which apparently is appropriate for odd order networks and should actually be -6dB as they are even order networks.

Frequency response is not quite as flat as it could be, and so I am going to recalculate the x-overs for -6dB and 90 degree intercept points and see if that flattens things out a bit more..

I think the fact that the levels are adjustable on both mids and highs has to some extent masked this issue, but I think it is audible because the flattest overall measured response is a bit dead sounding, and my preference results in peaks of several dB above mean in the x-over regions which as they say should be a clue...

Any comments would be welcome..

This cross-over design is used in my Onken based DIY speaker system.

[kevinkr]

Here is the newest version of my x-over. This is quite a simple design, and I have not used zobels on either of the horn drivers, based on preliminary room response measurements this doesn't appear to be necessary and in vintage JBL applications would not have been used.

I deliberately reduced the coil dcr in the bass cross-over in order to slightly increase the woofer electrical damping and this seems to have had the desired effect of reducing a moderate midbass hump in the response - subjectively tightening up the response.

I still need to do the full set of room response measurements, but the preliminary findings show an improvement of several dB in flatness from 100Hz - 12kHz and beyond. Not insignificant, and the system does actually sound significantly better overall, less romantic, and more detailed. The speed and imaging I liked previously have been retained.

I designed this using LTspice and based on listening and limited measurements the results are what I would expect. I will try to post some graphs once I do the actual response measurements.

Some assumptions based on my amplifier preferences were accounted for in this design - typically my SE amplifiers do not use negative feedback and have a source impedance of 2 - 3 ohms and the cross-overs have been voiced to amplifiers in this range. (As has the Onken bass alignment itself.) In addition I have used the dcr of the inductors to help control the Q of the individual networks, choosing ones based on their dcr at a specific inductance.. All the inductors used are air core types as I was not certain that using cored inductors might not detract from the sound in some way - the original was all air core and performed well enough to remain in use for well over a year. I wanted to evaluate the revised x-over tuning, not the potential performance differences between inductors.

Caps in the new x-o are all Clarity SA types (UK made) based on my experience with previous substitutions of the original Solens in the old x-o design with Clarity SA and Erse Pulse types which to my surprise yielded significant improvements in detail, and audibly reduced colorations.

[dogbert]

If you want to use a JBL design for a crossover you could start with the L300 network. The schematic is posted at the Lansing Heritage site using L-pads rather than autoformers. This modern version gives the same response as the original version using readily available modern components.

The L300 used a 136A 15" woofer, the equivalent home version of the pro model 2420 and 2312 that you have, and an 077 slot tweeter. The Altec woofer you are using is more sensitive than the 136A woofer and the super tweeter is different, although it uses the same diaphragm, but this is as close as you will get with a pre-engineered network.

Here is the exact thread you want to look at http://www.audioheritage.org/vbullet...highlight=3133

[kevinkr]

Hi Dogbert,
Thanks for the information. I'm quite happy with my network as currently implemented. It's long since built - the posting occurred after its completion. I was just posting so that people could see what I did.

At some point I'm going to see about getting some variable autoformers to try instead of the lpads I am currently using.