Quick & Dirty 4430-Inspired Two-Ways Part II

[Ian Mackenzie]

Quote: "And Ian, too, who's reputed to be a closet two-way fan from WAY back."


For what it worth Zilch I have never built a two way design so I would prefer you delete your inaccurate reference or ask a moderator to do so. Actually I have been heading in the opposite direction for years and years.

It would take me only one nonsecond to confirm just why a two way as you have gestated over here is just not up to the task of hi end home audio as we know it today but an infinite time span to explain to you just why this is so. Non the less a two system is a fun diy project.

enjoy

Ian

[Zilch]

The baseline for reference is here, though they've got the wrong PT waveguide listed:

http://www.jblpro.com/ae/pdf/spec_shts/AM4212_00.pdf

2023H is 95 dB, according to that. LE14H-3 is about 92.5, I think, so I'd be expecting to have to pad down the HF 2.5 dB.

Instead it's 20 log (Vr) = 9.5424 dB, Vr being 6/(1/(1/6+1/3)). I've lost 7 dB somewhere.

First suspect was the LF filter. RTA says it's consuming about 2 dB. Since it was built to the AM4212/00 schematic (link in post #785, above,) I'd assume there's nothing untoward occurring there. 2 dB seems about right for this 4-pole filter, no?

My post above reflects my frustration with Spice, actually. It generates "Incomplete Matrix" error; the version I'm using prohibits series capacitors.

I did develop a workaround, and the HF curve appears below, clearly illustrating the requisite compensation for this new family of UHF drivers on these waveguides. They're certainly not "Plug 'n' Play." These are compression drivers, not the direct ring radiators we've known.

Spice says it's down 20.89 dB at 1.3 kHz. 2407H is spec'd at 110 dB, so it should be in the range without my additional 9.5 dB L-Pad.

Inductor DCR's are:

L1 = 0.46 Ohms
L2 = 0.55 Ohms
L3 = 1.28 Ohms
L4 = 0.28 Ohms
L5 = 0.36 Ohms

We don't have the response curve of 2023H, that I can find, in which case, the best I can do is operate on assumptions.

Somewhat of the response is shown here, on a different crossover:

http://www.jblpro.com/mpro/PDF/MP412.pdf

Crossover frequency there is 1.8 kHz:

http://www.jblproservice.com/pdf/MPr...ries/MP412.pdf

I'm aware that the LF filter is not likely optimum for the best response from LE14H-3. The purpose here is primarily to implement a 24 dB/octave crossover to work with the 2407H HF driver. JBL's crossover for that would seem to be a good starting point.

After getting Spice working, I was able to adjust the filter to produce the RTA curve at the bottom on the JBL OASR horn. It looks like I've boosted 2.5 kHz a bit too much, but I think it's close enough for listening until CLIO is up and running.

I still have no clue as to why I'm having to pad down 7 dB more than anticipated to achieve it.

I'm sorry you're irritated, Giskard. I am, too, at not having a better answer to this. As you see, I have certainly NOT ignored what advice you've generously provided along the way here. Your substantial contribution to the endeavor has always been very much appreciated.

For the record, I would be greatly pleased if this thread were moved over to become the first entry in the new DIY forum.

[4313B]

The baseline for reference

Should be done by you using the original drivers and network.

though they've got the wrong PT waveguide listed

Oops!

My post above reflects my frustration with Spice, actually. It generates "Incomplete Matrix" error; the version I'm using prohibits series capacitors.

Ah! Use a 1E12 ohm resistor to ground whenever you get that error.

I did develop a workaround, and the HF curve appears below, clearly illustrating the requisite compensation for this new family of UHF drivers on these waveguides. They're certainly not "Plug 'n' Play." These are compression drivers, not the direct ring radiators we've known.

Ok.

Inductor DCR's are:

L1 = 0.46 Ohms
L2 = 0.55 Ohms
L3 = 1.28 Ohms
L4 = 0.28 Ohms
L5 = 0.36 Ohms

Nice information! It irritates the hell out of me that JBL doesn't do every schematic like this one - http://www.jblproservice.com/pdf/Net...%20Network.pdf

We don't have the response curve of 2023H, that I can find, in which case, the best I can do is operate on assumptions.

That's a bad thing.

I'm aware that the LF filter is not likely optimum for the best response from LE14H-3.

Let me crank up spice and post the voltage drive of the network. You compare that voltage drive to what you know about the LE14H-3.

After getting Spice working, I was able to adjust the filter to produce the RTA curve at the bottom on the JBL OASR horn. It looks like I've boosted 2.5 kHz a bit too much, but I think it's close enough for listening until CLIO is up and running.

Unfortunately you need to post a schematic showing your changes so we know what you are talking about. Worst case example below.

[Hamilton]

It would take me only one nonsecond to confirm just why a two way as you have gestated over here is just not up to the task of hi end home audio as we know it today.....

I'm listening, you have my undivided attention.

[4313B]

I'm listening, you have my undivided attention.

Limitations in bandwidth.

Non the less a two system is a fun diy project.

Indeed. I've still got a 2-way project on the bench as well as several 3-ways and a 4-way. I'm under no illusions as to the outcome of each. They will all have their pros and cons.

[4313B]

They sound good, too. I think this matters. These'll be the only available new JBL UHF drivers soon, maybe....

Ok..... here's some info for you to look over then.

[4313B]

Make us proud Zilch....

[4313B]

I'm sorry you're irritated, Giskard. I am, too, at not having a better answer to this.

Yeah, you piss me off. Whatever! I'll get over it... Do you see what it takes to come up with an answer? You are posting with less information than you really need. There is absolutely nothing wrong with trying all this stuff out. People just have to know that you are trying all this stuff out and if they follow suit and blow stuff up, or it sounds like crap to them, it's their problem. I would imagine your desire to see it moved over to the DIY forum will be fulfilled. It's a great piece of work and we'll see if anyone else will come forward and try these kinds of things as well.

I apologize for ragging on you.

[Zilch]

Do you see what it takes to come up with an answer? You are posting with less information than you really need.

Yup. The LF filter compensates for the rising response of 2023H. LE14H-3 driver response is closer to flat, so this network is pulling it down as it approaches the crossover frequency. Setting the levels empirically, I've had to pull the HF down that extra 7 dB to compensate.

In the course of doing that, I've inadvertently generated an overall downward slope "Voice" to the entire system, which may be a good thing, from examining the S9800 system response, for example. I'd show that, but I forget where it was published here. From a review, if I recall.

EDIT: Found it, bottom. Stereophile listening room "Exemplary" K2 performance, 1/3 octave, spatially averaged, 5 dB per major division, baseline 73 dB. Compare to RTA curve in post #796, above.

Below is the adjusted HF circuit, bumping up 2.5 kHz. It's what you suggest, with the capacitance adjusted, as well. I've removed the inductor impedance resistors for clarity, but left in my 100m Spice "workaround." Don't know why putting it down there worked, but it did.

Thank you for posting the scoop on 2407H. Other members may be interested in trying these. Perhaps from this we'll be able to develop a simple compensation circuit to permit "drop-in" use with particular horns/waveguides.

People just have to know that you are trying all this stuff out and if they follow suit and blow stuff up, or it sounds like crap to them, it's their problem.

Yup. Hamilton's L200t3 clones were initally "Piercing," I believe he described it, using 2426H on 2342 horns. He had to do considerable experimental adjustment with L-Pads to tune them, including comparative listening to various sources (music vs. HT) using his 4430's as standard. Same with his DIY center channel.

As others of us well appreciate, he'll be tweaking for years to come. It's an ongoing rite of passage here!

[Zilch]

For those who may be interested, Spice is circuit schematic capture and emulation software. It greatly facilitates filter circuit understanding, design, and adjustment.

I'm using a limited functionality version, and have another free one I haven't tried out yet. It was developed at Berkeley.

What's a good source for the "real deal" P-Spice? How much does it cost?

[Hamilton]

Limitations in bandwidth.

Giskard, so a bottom feeder like me might understand, what would be another word for "bandwidth"?

Thanks.

[speakerdave]

. . . . what would be another word for "bandwidth"?

bandwidth Abbr. BW 1. Electronic filters The numerical difference between the upper and lower -3 dB points of a band of audio frequencies. Used to figure the Q, or quality factor, for a filter. See Figure 1 of RaneNote Constant-Q Graphic Equalizers; also download "Bandwidth vs. Q Calculator" as a zipped Microsoft Excel spreadsheet in the Rane Library. 2. Telecommunications The size of the communications channel. In analog communications, bandwidth is measured in Hertz (Hz), while digital communications measures bandwidth (data transfer rate) in bits per second. For example, an analog telephone channel has a bandwidth of 4,000 Hz, while a digitally coded telephone channel has a bandwidth of 64 kilobits/second. See the RaneNote Audio Specifications.

This is from the Rane Glossory website:
http://www.rane.com/digi-dic.html

Here's an online dictionary;
http://www.m-w.com/

It's also in my hard copy Webster's Third New International Dictionary

David

[johnaec]

The LF filter compensates for the rising response of 2023H. LE14H-3 driver response is closer to flat

Zilch - what I'm curious about is why you're even experimenting with the 2032H? It's such a different speaker than any of the LE14 series that it's almost like comparing apples and oranges. AFAIK, it's pretty much an entry level speaker for sound reinforcment, and hardly worth the trouble trying to match filters for it, (unless working on a SR application). Or did I miss something?

'Just curious...

John

[stevem]

Limitations in bandwidth.

Why did JBL design their most recent large format monitors (44xx, DMS-1) as two-way designs, do you suppose? It seems to me that these need to be wide bandwidth systems in order to fulfill their design goals.

I'm on the fence about the best approach myself. I'm thinking of going back to a two-way for simplicity's sake, but as you also said, they both have their pros and cons.

[Mr. Widget]

Now this is a damned nice two-way... anything else is just a compromise.


Widget