Thinking similiar to Statement/Khanspires but ...

[Kornbread]

I don't know what app you are using to measure, but try to change the vertical scale for the distortion measurements.
And if you measure at too low of SPL, you will pick up room noise that messes with things.
Shoot for 80-85 dB on axis at 1 meter level. That's quite loud, but it works better than a low level.


Wolf and filmslayer, I find that measuring HD higher at 90-95dB at a meter is better than lower.

Been measuring with REW (room eq wizard and one of PE's calibrated? mics) quite low, doesn't bother the AWA as much. I'll remeasure at a higher level ... when she's not around.


rpb, It's about four feet into a somewhat irregular ~22'x26' room with sloping ceiling; think upstairs loft.

---k---Your impulse response definitely doesn't look good. I'm guessing this is related to measurement gating. Bad measurements makes everything difficult.

I'd be interested in seeing the individual drivers frequency response before/after crossover.

Did you unibox the Paper drivers box sizes? I never have, so don't know if the paper need bigger boxes. Your boxes look so darn similar to mine, I don't know if you built differently from the original to accommodate the paper versions.

It be interesting to see off axis response and a overall power response to better understand the treble hot issue. Toeing in or out or room may alter your opinion.

Played around with channel delay and never gained much with the impulse response. Key word ... played.

Read Bagby's white paper on measurement technique, sorry guys, it's above my head. Learning guys, maybe someday ... Did figure out how to gate the response and how to find where the first reflections appear in the impulse response. Hopefully, the next graphs should be gated correctly.

Would the individual driver response be ok from the same @1m, on tweeter position, or explain what you'd like and how, I'll give it a shot.

Yes, I did unibox the paper and metal woofers. IIRC, they were fairly close. Will try and find that, and post later.

The outer dimensions are the same as the Statement II, but the mid's enclosure changes the internal woofer volume. To me, the paper woofers looked fine in the larger box, and the cut sheet sealed the deal. No matter how many times I do the math, measure, and cut, I always screw some part of that up. Sooner or later, I'll get the dimensions up.

[Kornbread]

Placed the mic as close as possible; just off center of the phase plug on all drivers, and came up with this. Then let REW 'average the responses.'

Also, there were some peaks laying ~25db down on the woofer and mid. Changed some xover slopes/types/points and now there ~45db down.



And then got [email protected] meter. [email protected] (floor bounce?) but looks like the first reflections was closer to 3.3?



What I should do is blend the gated response with the nearfield woofer response plus port? How do you get an accurate measurement at the port when it's on bottom of the cabinet ... and it's bloody heavy?

After some playing around, here's @1 meter gated @3.3ms. It still sounds hot in the lower treble ... ▼see note below.▼



Learned something today. The cheap android tablet was adding hash to the lower treble. The laptop with normal quality Spotify sounds better than the cheap android tablet set to extreme quality. Listening to cd's using an older Marantz setup, and flac files from Foobar>external dac>aune preamp I 'think' the speakers sound more like what the graph looks like. Not awesome, but getting better.

[---k---]

Placed the mic as close as possible; just off center of the phase plug on all drivers, and came up with this. Then let REW 'average the responses.'

I'm not the measurement / crossover guru in Khanspire team. CJD is.
But, that measurement method doesn't sound right to me. I believe you measure @1m. Measure the tweeter, then the pair of mids together, and then the pair of the woofers. This way you're capturing the baffle effects and the summation of the drivers. You should be gating off a lot of the base, so you'll have to bring that in. I'm guessing Jeff's paper goes through all this, but it is the hardest part of the process - which is why I've never tried.
Bad measurements are going to give you bad design. Active doesn't make it easy.

[donradick]

Kornbread

If the measured frequency response is pretty flat, but (as you said "still sounds hot in the lower treble") sometimes you need to measure off axis.
Simple enough, leave your mike in the same position, and rotate the speaker 10-20-30-45 degrees off axis.

But that only works if you got your gating right and have good clean results. Sorry, I can't help you with REW.

[Kornbread]

Finally got around to taking measurements @0,10, 20, 30, and 45 degrees off center. All gated at 3.3ms.
Purple=0* Maroon=10* Light Blue=20* Dark Blue=30* Orange=45*

[Kornbread]

Had a bit of free time today. Blue trace should be the same as earlier measurement, Red trace from today. Went from sounding hot in the lower treble to heavy, slow, and distant. Using the minidsp, I'm able to make changes on the fly, I don't see how you guys do it passively. I'd play around more but the wife is taking her beauty nap, and we all know what happens when we wake the wife from her nap.

Measurements taken same as before; @1 meters on tweeter axis, gated 3.3ms

[Kornbread]

The wife finished her beauty rest ... didn't help much.

Green is the same measurement as the earlier red trace, blue is new settings @1meter, gated 3.3ms, which should give a fairly accurate measurement to ~300hz?





Anyhow, cleaning up the area from ~400hz to 1500hz and balancing out the tweeter really helped, but something was still wrong in the bass.

So I opened up the gate to 100ms. Right way to do it or not, it showed some fairly large peaks centered around 190hz., and 88hz., each peak spanned about 80hz. Squashed them. This is the result gated at 100ms. When this measurement is gated to 3.3ms. it gives the Blue graph above.

There starting to sound better. Roger Waters sounds and images good, other good recordings sound pretty nice, and the pile of overly compressed 80's hair band cd's laying around sound just as harsh and compressed as ever, so this leaves me thinking, there getting closer to accurate. I'll get a better handle on their sound over the next couple weeks.

Something to note, at my .... cough ... age, I didn't think the area at 10k and up would be noticeable, so I just left it alone. Now I'm not so sure, there seems to be a bit to much shimmer on top.

Overall, this has been a very productive day.

jhollander, one of these days I'll make it to your neck of the woods. Want a crack at them?

[---k---]

So what active filters are you currently using?

I think you're finding that it is very easy to get decent results with an active crossover. But to get truly amazing results takes just as much effort and experience as a passive.

[Kornbread]

Woofer crossed at 250hz., BW24db/oct, (it took this slope to clean up a rather large resonance? well into the midrange)
Mids at 250hz BW24db/oct and 1700hz BW36db/oct
Tweeter 1900 BW42db/oct. (steep slope cleaned up a peak below the cutoff freq.)

Played around with the xover points, and slopes, but so far, these settings measured better. I know, some of you guys are going to look at this and scratch your heads thinking, we've got a blind man driving down the freeway! And ... basically ... your correct. Anyhow, feel free to comment. I'm trying to learn, and I learn best by doing, even if a few (bunch) of wrong turns are made in the process of getting there.





Overall eq. effecting all channels.



Woofer eq.



Mid eq.




Tweeter eq.



Way more correction than I wanted, and some may even be contradicting others. All I know is it sounds better than it was. You can see where a high shelf was used to raise the overall output, and a low shelf was used to remove some materiel below ~20hz to ease some of the strain on the tiny tk2050 chip amps.

[rpb]

Here's where I would start. It's hard to describe the full process, but the first thing I'd look at is the mids measured alone with no eq. See how flat, or lumpy the response is on axis, and off axis to the sides.

Next, I'd add a highpass and a lowpass filter. Maybe 4th order at 225 and 3k. Import the response as an frd into PCD, and compare it to a target curve. Next use a little EQ, or change filter types to get closer to the target.

Switch your displayed results to 5 dB scales, and show us some pictures of just the mid response.

[Kornbread]

What has it been now, a couple months since slapping them together for some testing. IMO, they now sound significantly better than my long time reference nht2.5i's. I'm liking them. Wish there was someone nearby to give their opinion.

Had a few vacation days that needed used before April and boy is home for Spring break, so, I have a little time and a helper for moving these behemoths (still managed to hurt my back) . With boy's help, all spots have been filled, BIN on for the last time and sanded, it looks and feels smooth. One speaker is ready to paint. But there seems to be a problem, what kind of paint can I use over BIN (shellac)? No one around town mixes lacquer (some kind of regulation) and these things are just too big for a rattle can. Besides, these things need to shine like a new penny, what are my options?

[ontariomaximus]

I would consider a small foam roller with a Tremclad Rust Paint. Done it a few times - finish turns out pretty decent.

[Kornbread]

Thanks for the suggestion.

I had forgotten to visit O'Rielly's auto parts store. They have a Dupli-Color brand lacquer ($25 quart) that comes premixed and ready to spray as a four part process; prime, base, candy coat, and clear. And they have Candy Apple Red. That just happens to be the color I was wanting.

[Kornbread]

Been a couple months.

The speakers have been sitting in the bedroom waiting for sanding, a couple more coats of primer, and paint. Note to anyone that might use the premixed ready-to-spray, Dupli-Color primer from O'Rielly's, it doesn't go very far. One 32oz. can wouldn't cover both speakers once. It sprayed like it needed to be cut further. They did have some other 'normal' Dupli-Color primer that cut 1:1 with their 'specialty' reducer. Total cost for 32oz. primer and 32oz. 'specialty' reducer was about the same as one 32oz. can of premixed ready-to-spray, Dupli-Color primer. Coverage is much better with the normal primer; got two coats on both speakers with maybe ~1/4 can of primer/reducer left over.

Wet sanded with 600 after the first and third coat of primer.

For some reason, the local O'Rielly's had an issue getting the premixed candy apple lacquer paint, settled on molten red. If you ask me, molten red looks just like barn red from the local farm supply store. One 32oz. can of the Dupli-Color premixed ready-to-spray paint covered both speakers with two coats, and enough to go back over and hit any light spots I'd missed.

Anyhow, here they are in two coats of barn ... I mean molten red. They have not been cleared, or sanded yet, so not much shine.

Not really a fan.

My thoughts, start from just behind the roundover, and paint the baffle flat black. If that doesn't sound good, they have 'most' of these colors in stock; http://www.oreillyauto.com/site/c/se....00+-+%2425.99

Need your advice. Which of those colors is going to look nice with this barn red? What else can I do to make them look better?

Oh, and ideas for the feet? They need to be about 2" off the floor, and not something that's going to let these heavy behemoths pierce my floor with holes. Trailer hitch balls?

[jhollander]

I assume you have a spray gun and this is not rattle cans? If so find a good automotive paint store and do a candy color on top. A darker tint clear will add depth. You could also add a metallic clear.

I'd not add anything to what you already have for the base. You could paint what you have black and add a rubber feet.

IMO 1 color speakers look the best unless you have a removable baffle. Even then black is meh. For MWAF I'm doing chocolate spray latex, seal coat, semi gloss lacquer, with a amber baffle, gloss.