or Connect
AVS › AVS Forum › Audio › DIY Speakers and Subs › Can you smell what Brando's cooking!?
New Posts  All Forums:Forum Nav:

# Can you smell what Brando's cooking!? - Page 8

If the crossover is 1st order, you don't need to put a load on the outputs when measuring the other drivers. If the crossover is 2nd order or higher, you must put a load on the output that doesn't have a driver connected.

When you are measuring the phase response of the each subsystem (crossover+woofer+cabinet and crossover+tweeter+horn), you are looking at the phase response of each subsystem in the acoustic domain, therefore you need to consider the phase difference at the electroacoustic crossover frequency.

### AVS Top Picks

Thank you!
great to see you here skywave...
Quote:
Originally Posted by Jack Hidley

Russell,

Step 1) Measure the frequency and phase response of both drivers in the system without moving the microphone.

Step 2) Compare the measured phase values of the two drivers at the desired crossover frequency. Whichever driver has more phase shift at that frequency means that driver is more delayed. For example, the tweeter is 136 degrees and the woofer is 276 degrees at 1.8kHz. 276-136=140 degrees. At 1.8kHz a phase shift of 140 degrees is a path length difference of 2.95". ((1,800*140)/(13,680*360)). 2.95" is how far you would need to move the woofer forward to make the phase difference zero at 1.8kHz.

Step 3) Model a crossover with an RLC parallel notch filter that is shunted to ground. Here is an example of one used in the woofer circuit:

http://zaphaudio.com/XDS-crossover.gif

The RLC filter is formed by C7, L8, R9. In the case above you want to put the RLC circuit in the tweeter section, since you need to delay the tweeter output to match that of the woofer. The RLC circuit increases the steepness of the crossover function. This always causes more delay (phase shift) as a result.

Here is the transfer function of the crossover above:

http://zaphaudio.com/XDS-transferfunction.gif

You can see the notch at 8kHz and the very steep slope that adds delay from 3-6kHz. When using this circuit to add phase shift, you don't really care how deep or where the notch is per se. You adjust the notch parameters until you get the desired phase shift at the crossover frequency.

Step 4) How do I figure out how to achieve the 140 degrees of phase shift at 1.8kHz? You must have some modeling software such as LSPCad for this. You import the drivers impedance, frequency and phase responses. Build a Spice model of the crossover in the software. The software will then show you what the resulting phase response of each output will be. I'm not hip to all of the latest freeware/shareware crossover modeling software, so others will have to offer suggestions. Having access to crossover modeling software is absolutely essential to building a loudspeaker. Doing it by trial and error/calculator method will take 10 times as long and will nearly always result in a significantly inferior result.

Wow, Been over this in the loudspeaker cookbook, been reading about it online and never was it clear to me how to work with phase shift.

Jack, you are a great instructor!! You should consider creating a "Continuing Education" online speaker building course. I would buy
So I have a pretty good handle on the active version. Just waiting for a DE160 to arrive so I can listen to the pair. I've never used the digital filter CAD in Soundeasy before so this has been a real education. I backmounted the woofer and this did help align phase better for the passive version. Doesn't really impact the active version of course. Backmounting the woofer introduces some wiggle in the response. Now I haven't made the front nice and pretty with roundover or anything, but from experience doing this sort of thing I expect that any improvements from a roundover will be minimal. You're essentially horn loading the woofer and have to accept the anomolies that go with it. For the active version I could easily stay with front mount and tweak the delay, but for the initial modeling I decided to use the measurements with the backmounted woofer for consistency. I'm re-doing all my measurements, so I only have the 0 degree measurement to show right now. Here is the current result:

So what does it take to get that? Quite a few filters, and many with values you wouldn't expect.

Woofer:
Gain: -1 dB
Delay: .15ms
6 dB Shelf: 100hz, gain 3dB
LP: 1st order Butterworth, Fc 337hz
LP: 4th order LR, Fc 2205hz
PEQ: 925hz, gain 2dB, Q 3
PEQ: 2480hz, gain -7, Q 7

Tweeter:
Gain: -10dB
6dB Shelf: 11550hz, gain 8.7dB
HP: 2nd order LR, Fc 1450hz
PEQ: 2400hz, gain -3dB, Q 4
PEQ: 900hz, gain -5dB, Q 4
PEQ: 4900hz, gain -2dB, Q 2

I also modeled how someone with no measurement or crossover modeling capability might set the DCX. First we are aiming for LR4 @ 1300hz. So we set those filters in the DCX. Now maybe we are savvy enough to know that we should also correct for Baffle Step and compensate for horn gain. So we add a shelf filter centered at 100hz with 6dB gain for BSC, and another at 10000hz with 6dB gain for horn compensation. Here is the result:

Sort of proves my point that off the shelf active crossovers only give the impression of doing things correctly, the reality may be quite different. Obviously if you the ability to measure and model they can be quite powerful though.
"You're essentially horn loading the woofer and have to accept the anomolies that go with it."

Dunno why you'd think that.

IMO it's just diffraction.
Noah in a sense it is the same thing, at least in my mind. But simply saying diffraction is as good an explanation as any. So yeah, I can agree with that.

I'm comparing BW3 versus LR4 for the BBV2 MTM. Here is comparison with rough crossover so the treble isn't real pretty yet, 0,10,20,30,40 degrees vertically (which would be horizontal if a center channel):

BW3

LR4

BW3 is a bit better but not as much I thought it would be. Basically the treble bounces back a bit faster, so the null is narrower.
"Noah in a sense it is the same thing, at least in my mind."

Horn loading acts as an acoustic impedance transformer; diffraction is completely different.
I guess I should have said the anomalies associated with putting anything on a horn, not necessarily associated with horn loading. No coffee today Noah?
Just to be clear, you meant horn anomalies from rear-mounting the woofer, correct?

In which case I maintain it's irrelevant.

> No coffee today Noah?

Why, am I being more of a nitpicking nerd than normal?
Quote:
Originally Posted by noah katz

Why, am I being more of a nitpicking nerd than normal?

Kind of bro Or maybe it's jsut me, this week has been hell at work.
Ok a lot of progress to share! The draft active crossover is done, just waiting on a DE160 to arrive to start listening. This version is using the original cabinet with front mounted woofer. The response is slightly smoother with it front mounted. The following results are with the foam I linked to earlier. Here are teh dimensions, note that the sides are straight in the drawing, but in reality would be slightly curved to follow the horn. An electric knife makes it easy to cut, just gotta make sure you mark everything well so you don't get some goofy lopsided thing.

This is really my first attempt at controlled directivity designs and active crossovers, and so far I REALLY like what I see. SE has a cool matching filter feature implemented in the newest version to measure the frequencies below the gate's cutoff frequency to produce essentially anechoic response from an echoic environment (the typical room). So here is that response, no smoothing:

Looks pretty good eh? But that's just one on axis measure, the real trick is getting the off axis to look good too. So here is the horizontal response, 0 to 40 degrees in 10 degree steps, below 300hz is not real:

SE only lets me plot five responses, so here is the 50 and 60 degree plots with the 0 degree for reference:

I think even Harman could live with that How about the vertical response? Here is the response going up, 0 to 40 degrees:

And down:

Depending on how we define it, I would say +/- 20 degrees is good, with the null get pretty deep between 20 and 30 degrees. Now this has 3 PEQ's on the tweeter, and one on the woofer, although the woofer one is for the breakup and down in level pretty good. I don't know if all the EQ is a good thing or not. I am going to make a version with one mild PEQ on the tweeter for the 2.5khz hump too, and see if they sound that different. The filters used are:

Woofer
Gain: 0dB
Delay: .20 ms
LP: 1st order Butterworth, Fc 353hz
LP: 4th order LR, Fc 2415hz
6dB Shelving filter: Fo 100hz (84hz if using a DCX), 3 dB gain

Tweeter
Gain: -4 dB
HP: 1st order Butterworth, Fc 5660hz
HP: 2nd order LR, Fc 555hz
PEQ: 2400hz, Q 3.6, gain -2.8dB
PEQ: 1700hz, Q 2, gain 2.5dB
PEQ: 4900hz, Q 1.9, gain -3.3dB

Still gonna play with those PEQ's. There are a number of ways to get the same response so we'll see if we can get something more elegant. The passive version is coming along, but I don't know if I'll get anything released other than a rough draft until late October. I'll be working just about every day for the next five weeks so I won't have to time to get much farther.
What driver is on the waveguide presently, i.e., for these measurements?
B&C DE160. From the measurements at least it does everything the De250 will do, but cheaper. The only tradeoff is lower sensitivity, which isn't typically an issue anyway. Slightly different diaphragm material so there may be some subjective differences, but that could go either way.

Got a woofer for you try for the E'wave. I know you guys are JBL fans, but in fitting with the budget focus for at least the entry level Ewave you might want to look at the Eminence Deltalite 2512 or 2515. It's obviously part of the new family of Eminence drivers like the Kappalite. I'll post measurements when i get a chance but I'll just say the frequency response is very close to and as usable as the 3012HO, Le(x) is very good, and distortion is solid. And I think it retails for like \$109. Check it when you get a chance.

I may do a quicky design for it with the No Quarter cabinet and the \$68 Celestion 1425 compression driver. Assuming the 1425 can handle going low-ish, I think this could be one those designs where the sum is greater than the parts. Finally I think the diy community has the parts available to build not just very good speakers, but world class - and at a reasonable price.
Edited the horizontal plots since I didn't have them labeled quite right. I'll look for a better way to plot more than five response per graph, so I don't have to split them up an make it difficult to visually analyze.
I was confused about the DE160 because you said you were waiting for it, which apparently means a second one.

Thank you for the tip on Deltalite. We have several members wanting to do high efficiency, but it's not mainstream yet.

We're building the QSC waveguide into the PE KD trapezoidal over there, and have snugged up the C/C distance a tad. By virtue of the location of the stock woofer cutout, we have had to rear mount the woofer, which turns out to be advantageous, perhaps, since the QSC is ~1.25" deeper than eWave.

Looks like we can grab another 1/2" or so out of the waveguide flange, but I believe I'd rather see a roundover to the woofer, instead, which is gonna require getting clever about how to mount it. A 10" will go in nicely on an adapter ring.

1425 moves the HF acoustic center back even more, no...?
Lookin good!!

btw what are the outer dimensions of the QSC horn.
Quote:
Originally Posted by DL86

Lookin good!!

btw what are the outer dimensions of the QSC horn.

The retangle horn?

Exactly 10"x14"
Quote:
Originally Posted by penngray

The retangle horn?

Exactly 10"x14"

Great, just to confirm. Yes the rectangle horn.
Quote:
Originally Posted by DL86

Great, just to confirm. Yes the rectangle horn.

No problem, I just free hand my baffle cut out 3/16" deep for it yesterday so its fresh in my brain

3/16" because its actually just less then 1/4 thick and I have veener going on the baffle so its perfectly flush now.
Quote:
Originally Posted by ZilchLab

Looks like we can grab another 1/2" or so out of the waveguide flange, but I believe I'd rather see a roundover to the woofer, instead, which is gonna require getting clever about how to mount it. A 10" will go in nicely on an adapter ring.

Yeah I'm currently working on the rearmount with roundover, but I really think the better way is with the woofer surface mounted like in my original. The response is noticeably smoother in the crossover region, on and off axis. I mean look at the plots I posted, that is pretty darn good response across a very wide arc by anyone's standard. That leaves moving the horn forward somehow. I don't really like things hanging in front of drivers, so I'm looking at getting that room I need simply by using a spacer/protrusion on the front baffle to mount the horn to. It will make the pretty roundover edges more difficult or impossible though. I could do large chamfers instead. Obviously the overhang of this spacer will be a source of reflection for the woofer, but I don't believe it would be as pronounced as rearmounting would be.

Quote:
Originally Posted by ZilchLab

1425 moves the HF acoustic center back even more, no...?

It might. It looks longer, I'd have to throw it and say the DE160 on a baffle and see what the impulses look like. Obviously if I plan for it BEFORE I build the cabinet, it is much easier to correct
Whats rear mounting the woofer mean?

So its better to have the horn protrude out a little bit further compared to the woofer which is sitting slightly further back?
has geddes offered to test the qsc horn?

lol.
Quote:
Originally Posted by DL86

Whats rear mounting the woofer mean?

So its better to have the horn protrude out a little bit further compared to the woofer which is sitting slightly further back?

big horns place the compression driver pretty far back.

rear mounting the woofer gets you a little closer to matching the accoustic centers with big horns.
Quote:
Originally Posted by LTD02

has geddes offered to test the qsc horn?

lol.

Why should he? The last thread regarding the JBL test turned out to be a nightmare for Geddes (IMO), not because of the results but because of the debates that happened afterwards
that thread probably wasnt his favorite lol
Quote:
Originally Posted by Rightbrained

that thread probably wasnt his favorite lol

Apparently not:

Quote:
Originally Posted by gedlee

I've taken the position not to respond in this thread due to the disrespectful manner in which I was treated. If you want to ask the questions over at one of my other threads then I will respond. The sooner this thread dies the better.

Wow, a Geddes free zone. I wonder how long that'll last...
can't anybody take some off-axis measurements for the qsc horn? if i had the capability, i'd do it. it's just not possible for me right now.

also, let's give dr. g. a break. he has contributed much. right or wrong, agree or disagree, he has some thought provoking points.
Quote:
Originally Posted by LTD02

can't anybody take some off-axis measurements for the qsc horn? if i had the capability, i'd do it. it's just not possible for me right now.
.

I thought we have all the QSC horn measurements?
New Posts  All Forums:Forum Nav:
Return Home
Back to Forum: DIY Speakers and Subs
• Can you smell what Brando's cooking!?

### AVS Top Picks

AVS › AVS Forum › Audio › DIY Speakers and Subs › Can you smell what Brando's cooking!?