Which Components Should I Use?

[Wonderboy]

Hmmm, well I just spent at least four hours trying different set ups, tweeters, amps, passive or active etc. I have come to the conclusion I would like to keep it as simple as possible (well, except for the fact I'm stuck on a four way system and using the 2012). I know what I plan to do will not go over to well with guys that are really into measurements and doing it by the book but if it works for me and I am happy what can I say. I have thrown my Radio Shack SPL meter out the window, it was modified years ago to be more sensitive to bass freqs. when I made my insane subs and now find it is not accurate for the higher ranges. So really now I have no test equipment, except my ears. (and I'm going deaf from testing, lol!)

I have decided to go with the 2370a horn like I mentioned earlier, I was having second thoughts and was thinking of switching over to the 2380a but after doing some serious listening and comparisons I'll stick with the 2370a's. Mind you, I was using the 2380a's with the 2482 driver and a phenolic diaphragm they did not sound good at all for music. So, my plan is to bi-amp the system, a separate power amp for the 2226's and another amp for the other three drivers, the 2012, 2425 and the 2405. I already have crossovers for the 2405 and the 2425's so really the only thing left I really need is a crossover design for the 2012 mid-bass driver. Baron030 suggested a parallel trap circuit for an active crossover to level out the response and get rid of the 1kHz spike but is there a way to do it passsive? I was thinking of crossing the 2012 with a high pass filter around 300Hz, 12dB per octave and a low pass around 1.2kHz, 12dB per octave, how does that sound?

Kevin.

[Ian Mackenzie]

Hi Kevin,

I recall this was mentioned in a post recently. Something like 1.5 mh will flatten the rising response of the 2012.

It is very difficult to make recomendations without actual hard data from measurements. Stuff like baffle step compensation and more to the point how the 2012 mates with your horn of choice is critical.

You cannot design a crossover with an RTA or a sound level meter unless you have a lot of experience with a particular drivers you plan to use.

If it were my with such a fantastic driver I would invest in the Woofer tester (speaker version) that measures proper gated impulse responses, assists with crossover calculatiions and a host of other great stuff. Its around $600-700.

This may sound a lot but you will never look back. Trust me..

Ian

[Baron030]

Hi Kevin

I will say this again. I really don't think that my parallel trap circuit can be added to the low pass section of a passive crossover network. It is really only a solution for active crossover systems.

And I think Ian Mackenzie might be remembering one of first attempts at flattening the 2012H with a 1.5mH coil. The 1.5mh coil trick did work, but it was only good for a very narrow frequency range and I used it with my Ashly. I have found by switching to a parallel trap circuit. That it works a whole lot better, over a wider frequency range, and with less insertion losses as well.

Now, having said all that, maybe I can throw out some ideas that might be helpful. If the low pass section of the 1.2 KHz crossover network had a low Q. Like a Linkwitz-Riley curve for example, then maybe some of the 2012s rising response could be compensated for. We are talking about some kind of an asymmetric crossover network here. And then, hopefully the network has enough series impedance to be able to really limit the current going to the 2012 at above 500Hz. In which case a series notch filter could be added to the circuit across the 2012 voice coil. This could take a really big bite out of that nasty 1,000Hz peak that the 2012 has. And I would think that maybe a third order network might be better then a second order network. There are some large peaks in the 2-3 KHz range. And a higher order filter would do a better job of kill them off.

Unfortunately, I don't have any crossover design software. So, I really can't help much in the crossover design. You see all of my attempts at making passive 2012 EQ circuits are all based on the trial and error method. And that is a very slow process. I can’t tell you how many hours I have put into building and mostly testing that very simple 2012 corrector circuit.

But, I could generate some response measurements of both my left and right 2012H drivers using CLIO 7.10 LITE. And maybe, that could come in handy…

Baron030

[johnaec]

I've also got a new pair of 2012H I'm debating crossing over to 2447J 1.5" drivers w/2381 horns, so I've been watching this with great interest. I need to see if I can move the crossover below 1K with this combination. BTW - crossover is Ashly XR4001.

John

[Zilch]

Get the 8024 running and learn how to use it.....

[4313B]

But, I could generate some response measurements of both my left and right 2012H drivers using CLIO 7.10 LITE. And maybe, that could come in handy…

Post the response curve data file and the impedance curve data file so others here can pick them up and run with them if so inclined. Save the data with maximum points and set the phase to minimum as I showed in another thread.

There are other considerations such as what baffle is the driver on and what is the voice coil x, y, z orientation as well as the mic distance and drive voltage but that's out of scope in this particular instance.

[Wonderboy]

Hey Guys,

Thanks for the advice. Maybe I should just take Johnaec's route and cross the 2012 under the 1kHz with a third order? Actually at the moment the 2425's are crossed over at 800Hz, second order, but that might be cutting it a little close eh? I'm not sure how much overlap is needed at crossover points etc. Also I read somewhere that with a second order crossover in a three way or more system the polarity should be switched on one of the drivers to compensate for 180 degree phase shift? (something like that, not sure if I remembered it correctly). I was just reading in The Loudspeaker Design Cookbook, the author suggests possibly using a "T" bandpass topology that had been presented in Speaker Builder 2/87 for passive three way networks, is anyone familiar with this?

Thanks for the advice on the woofer tester Ian Mackenzie, at the moment I don't have that much money to spend. I think I saw a woofer tester for sale at Parts Express.com, not sure of the price or if that is what you are talking about? A week or two ago I read a whole bunch of your posts on your "Building the 4345/4344" found it very interesting, well done!

I think Zilch had some good advice for me, get the DSP 8024 up and running, I mentioned I had one earlier and was so hung up on my Radio Shack meter the DSP slipped my mind. I have had the unit since about 2003, think it's time to open it and get into it, not sure if I will be able to fit it in to my system though. My system is pretty domestic rather than pro audio but as soon as I post this thread, I will head down to the basement and dig it out and open the bloody box and get reading!

Baron030 go ahead and generate some response measurements for the 2012 if you feel up to it, but do not go to a lot of trouble for me but some of the other guys might like to see them, as mentioned by 4313B.

Another idea is, if I was to replace the phenolic diaphragms in the 2482's with aluminum or titanium or what ever, along with the 2380a horns, that might make them sound better and enable me to cross the horns lower (compared to the 2425s) as well as the 2012, further from the 1kHz region. Or is it the phenolic diaphragms that allow that driver to go that low?

What I don't understand is why would JBL construct a midrange/low-frequency speaker with all these peaks and problems, or did they design it for another application. In the Technical Notes for the 2012H, Volume 1, Number 24 it describes the "Desired" Performance Features of this driver, '#1' on the list it states: "1. Extended power response;rising axial response" they say it is "desired", or is that not the same as 'freq. response'? (please excuse my ignorance)

Kevin.

[Baron030]

Hello to All
Well, it looks like this is kicking up some 2012 driver interest around here. I will be happy to post the response and impedance curve data files, so that anyone here can pick them up and run with it. But, with Thanksgiving coming up soon, it might take me few days to compose a comprehensive posting with as accurate data as possible in it. So, just be patient here.

Hi Johnaec
I am using a 1.2 KHz. crossover point with my 2382a/2446h combination. And the reason why has a lot more to do with the 2382a low frequency roll-off then anything with the 2012. In comparing the 2381 to the 2382, it looks like you can go lower then I sure can. But, just lowering the crossover point is not going to be the quick fix to the rising response of the 2012H. I still think you need some 2012 EQ in the mix to make it all work right. The peak in the response really does gain +6db from about 500Hz to a 1 KHz. So, even a 4th order crossover will not fix this problem alone.

For anyone planning on using a passive crossover, I really do think that there is a relatively simple solution out there. It's just coming up with the correct component values that will be the hard part. Be it a 2nd or 3rd order crossover, I still think a series notch filter might be a good place to start. Maybe the schematic pictured will help explain what I am thinking about. And here are a few of my series notch filter assumptions. The series notch filter will need to resonate at 1 KHz. And if it is set to resonate below 1 KHz, then it will create a dip in the response at about 500Hz. And if it set too high, then it will miss the peak and not do its job. And “R1” value will most certainly need to be less then 8 ohms to provide the necessary amount of attenuation at 1 KHz. But, what I don't know is the Q value of the notch. These are the “C3” and “L3” values. If the Q is set too low, then there will be a big dip in the response at about 500Hz. And if the Q is set too high, then the single big peak will be replaced with two smaller ones and the response will not be as smooth as it could be. And I would also think that the “C3” and “L3” values will also vary depending on the selected crossover frequency. So, picking the correct series notch filter values will involve some trade offs.

Baron030

[johnaec]

But, just lowering the crossover point is not going to be the quick fix to the rising response of the 2012H. I still think you need some 2012 EQ in the mix to make it all work right.

I've already determined I'll be using the Ashly XR4001 4-way active crossover in my case, and resigned myself to the fact I'll also hopefully just need a parametric EQ to smooth out the response, but may also have to include a 1/3 octave graphic EQ. The EQ's will be digital, while the Ashly is analog. Thank heaven for the low noise/distortion of most modern components, though the phase issues I'll be dealing with might become interesting.

What I'd really like to sew things together is an all in one 2-in 8-out digital signal processor that didn't cost an arm and a leg...

Of course - getting everything to work passive would be fantastic - 'too bad it's not practical to design passive with all the components instantly adjustable, like in the digital domain...

Either way, I'll be following this discussion with great interest.

John

[Ian Mackenzie]

Perhaps something more simple.

I messed with for an hour using a model of the 2012H and edited the response reasonable closely.

The upper response is purely resistive.

That is good for the driver but hard for crossover design in this case. The peaks from JBL data are at 3 and 5 khertz. The cone exhibits beaming at the onset of 1000 hertz.

Therefore a target crossover point of 1000-1500 hertz for domestic use is desireable.

Juggling a 2nd a order L / C network I got a 18 db slope at 1100 hertz. This assumes a given seal volume of around 10 litres.

In the end I found 1.38 Mh and 49uf have a nice curve that matches a 18 db low pass slope closely

[Baron030]

Hi Ian

This is very interesting and your timing could not be better. Since, I was thinking of running some more CLIO tests on my 2012H drivers anyways and posting the results here. I could also test something that is very close to your crossover suggestion. I have a pair of 1.50mH 0.23 ohm Jantzen foil air coils just sitting in a box with a big bunch solen caps. For testing purposes, I would not want to unwind these expensive coils down to 1.38mh. But, I could very quickly test a crossover circuit that would be very close to your design. It will be interesting to see the results. Assuming a 1.50mH coil, what cap value do you think I should try this test with?


Baron030

[Ian Mackenzie]

A few things,

This is only a model a made while munching my favourite pasta.

Second, the actual values are very critical to within 2 %.

A bit of this way and that and it turns to shit.

Its was just an indication of what you can do by manipulating the filter values in relation to the characteristics of a given driver.

The people at JBL are jab hands at this if you look at their networks.

But is not trivial and you might have several attempts following actual auditions.

Before going any further we need data of the actual measured driver

You really need to spend time getting this right with Clio and fed the data into a package like Soundeasy or LEAP Crossoevr Shop

Its not a Quick and Dirty RTA job on such a nice driver.

[Zilch]

Slander!

[And gratuitous, to boot.... ]

[Baron030]

Hi John
As an owner of an Ashly XR4001, I have to say that you have great taste in electronic crossovers networks. I love my Ashly. But, don't resign yourself just yet into thinking that a parametric EQ is required to smooth out the 2012H. Go back and take another look at my simple LRC circuit solution for use with active crossover networks. Ok, there are some here that will be quick to point out that a wavy 3 to 4 db peak to valley response isn't all that great. But, actually it does sound pretty good in my apartment. And I have been toying around with some ideas that could improve upon that simple circuit. In other words, this tread could start to get really interesting.

Hi Ian
I fully understand exactly what you are saying. Probably a hell of a lot more then most people around here. And yes, I will be spending a lot of time making sure that my CLIO measurements as accurate as possible before posting them here.

Maybe, I should look into buying a software package like Soundeasy. Wait a minute, I have been here before.
This is one of those red pill, blue pill decisions. As in, "You take the blue pill, the story ends, you wake up in your bed, and believe whatever you want to believe. You take the red pill, you stay in Wonderland, and I show you just how deep the rabbit hole goes."

You see I took that red pill, bought CLIO, and I now I am finding out just how deep the rabbit hole goes.

Baron030
There is no spoon....

[Ian Mackenzie]

I have come to the conclusion I would like to keep it as simple as possible (well, except for the fact I'm stuck on a four way system and using the 2012
I was thinking of crossing the 2012 with a high pass filter around 300Hz, 12dB per octave and a low pass around 1.2kHz, 12dB per octave, how does that sound?

Kevin.

Hi Kevin,

Sounds like a plan,

The peaky response is not such an issue if you intend to crossover from 1.2 - 1.6 khertz, I recommend a careful examination of the JBL data sheet regards loading the 2012 and setup. (the operating range is up to 1.6 khertz).

You may also like to consider the 2397 horns with the 2445 driver.
If you have $1100 to spare or want to trade other drivers I can steer you in the direction of some nice S/H 2 inch Tad 4001's (pair) available in the USA

The 2012 has very low power compression and very low distortion so it should be a nice match with the 2226 woofers. You can add some mild LF boost to the 2226 to help balance the system overall.