Adding yet another variable (current vs. voltage driven) sure makes for a lot more measurements.

I survived figuring out Soundeasy's quirks once again with minimal bursts of cursing.

The following are the 3 driver's frequency response at varying levels of current drive vs. voltage.

For the woofer I'll clearly have to limit myself to more voltage drive given the peak at resonance, but for the mid tweet I can go current drive easily.

I'm not sure where I got the idea that current drive altered phase. The phase for the mid and tweet are identical for current vs. voltage and the woofer is close enough to measurement error.

Tony asked about the B1 active XO's. There's really not much to say. Any active filter that uses an opamp as a buffer (I.e. where output is directly tied to the '-' input) AND a cap blocks DC from the output to the feedback loop should be able to use a B1 like buffer. The circuits in sections 1, 2, 3 and 7 of SL's filter page all qualify, but the rest may be sketchy in my mind since the B1 can have some DC offset.

Attached is a rough drawing I scribbled last night for my own benefit, so it ain't pretty. An input buffer is something I'm still debating needing or not, but I'm leaving some room for it.

I'm going to let Soundeasy help me define the values, but the drivers are pretty well behaved in their desired bandwidth such that nearly textbook filters can be used.

I don't have enough JFETs of any one type for everything, so I'm going to have to mix types, probably LSK170s for the woofer and Toshi 2SK170s for the mid/tweet.

...

A main reason I'm active on this project is the event coming up on the 5th. For me there's nothing like a deadline to get me out in the garage.

A lot left to do:

• MLS measurements - on/off axis at varying output impedances for 3 drivers will make for quite a few measurements
• Figure out how to enclose the mid - I'm thinking open back at the moment
• Design the active crossovers in Soundeasy
• Build the active XOs - point to point likely using FirstWatt B1 type buffers (two JFETs isn't much more work than an opamp)
• Paint

Two weeks left!

Unlike the mid and woofer - the tweeter doesn't seem to show any gains under current drive.

Some decent reductions in 3rd harmonic.

So - I setup Soundeasy to take some distortion measurements. I don't claim to be a wizard at taking these measurements and balancing the various factors (overloading the mic, etc.) so take 'em with some grains of salt but they are interesting.

Taken with a Behringer ECM8000 mic (uncalibrated) via a UB802 to a M-audio 410 PCI card.

You can pretty clearly see my noise floor. Taken in a living room with computers/fridges turned off, noisy computer in another room.

Mic was nearfield.

There's some clear differences between the plots when driven with current vs. voltage. Ignore the nasty bits around resonance for the current driven one - I clipped the soundcard input with the resonance peak.

A Wilson type setup - the mid/tweet box is separate from the woofer box. The upper box is open front and back - replaceable baffles on both sides in case I care to change something, either drivers or sealed vs. open back.

The bottom cab is MDF. I had some of that concrete board left over from an outdoor kitchen project and figred I'd laminate it to the front baffle under the woofer. I don't think it was worth the effort. A knuckle rap test on the sides vs. front showed no huge difference.

The top cab is MDF laminated with baltic birch.

The drivers chosen were somewhat dictated by what I had lying around.

Tweeter: Fountek NeoCD3.0 - I've been wanting to use a ribbon for a while. I have fond memories of hearing old Apogees in the 80's. I bought the prior version (JP1?) quite a while ago. (I'm good at buying drivers for projects and then not building them when I get some other bright idea) When I opened them to use them one of the unit's magnets had delaminated/disintegrated. Big bummer. Madisound came through with great service though and gave me the current model at no cost and no hassle.

Mid: Tangband W4-1337S - The Statement guys seemed to have success with this mid with the Fountek, so I figured why not.

Woofer: Peerless 850146 - I bought these a while ago. Supposedly one of the few woofers appropriate for sealed enclosures. One reason to go sealed is the double hump in the impedance for a vented enclosure would cause the frequency response to rise at both...

Here's the power supply brick for all amps. A toroid with a CLC. -35V-0V-+35V

Note the color coded terminals on the right so I don't roach another LM4780! (see last post)

Here are two of the assembled boards. A mono one for the woofer on the left and a stereo on the right for mid and tweet. I designed the PCB to do both.

Originally I did two in point-2-point, which I found tedious, especially for the LM4780 which has tiny pins with little room for error. And after I blew up the LM4780 board (I hooked up the power supply backwards - poof!) I decided to go PCB, since there was no way to replace the chip given how close the pins were to each other.

The ExpressPCB software is pretty easy to use. Eagle, for how often I attempt PCB's, had too many annoyances for me when creating new objects, etc. The service is pretty quick too, ordered on a Tues and received them on a Fri. They give you 3 boards in their low cost service - so I'll use the one P2P LM3886 and 3 PCB boards.