Hello, and My Second 3-Way

[Brian B]

Hello all,
I'm currently in the process of building my second 3-way. I did my first 3-way last winter. The only other design I've done was a powered subwoofer (over 10 years ago). I've assembled a couple of other people's 2-ways designs, but these are the first loudspeakers I've designed and built myself.
I find building these larger cabinets time consuming, and I typically get a little frustrated with the crossover and getting them to sound good (to me). But I may continue to do at least one design a year.
I already had most of the woodworking tools needed to build the cabinets, but since my speaker building has been a seldom thing, I have not acquired many helpful “tools” (software or electronic). As a result most of my crossover work (other than a starting point from simple software calculations) has been done by ear with sine waves and music CDs. Obviously not a formula for a speaker that measures well. And perhaps not a formula for any “good” speaker. But I enjoy the results anyway.

[Millstonemike]

You need to post basic data on your build to get a response, and likely, needed help. Which drivers, cabinet dimensions / layout and port dimensions or no port. Crossover numbers, components, etc.

That may allow some of the gurus here to point you in the right direction and build a good speaker,

[Brian B]

^nd order crossovers all around at approximately 300Hz and 3000Hz.

[Brian B]

[westrock2000]

Good start. It took me a while to get the software stuff. I have Dayton Omnimic and DATS combo and it's great. Very simple to use and really adds a whole other dimension to the hobby. I have also tried REW (using the Omnimic mic) and it does work and it's free (plus cost of a USB mic), but I like the Omnimic interface better. But REW does have some cool features, like making a very tight frequency sweep that allowed me to find the exact frequency a panel was vibrating at.

[Brian B]

[Chris Roemer]

Even in only 1.6cf, adding a (3" x 9" long "Precision") port will drop your F3 by 10Hz (50 down to 40) as well as add a dB or 2 of "lift" near 100Hz, which could be used as part of your BSC.

[Brian B]

I'll give some consideration to a port, but I generally prefer sealed designs. I know ported designs have been pretty much standard for many years now, but they can sound somewhat tubby or one-note to me at times. I may try it once I get these speakers making noise. Conversely, I don't like the approximate 3dB loss that comes with a sealed design.

Never been a big fan of BSC either because of the losses. Of course, the use of a port (with the "gain") helps offset the loss in a BSC circuit.

I beginning to sense how important the overall efficiency is to a lively speaker. It seems like the less sensitive a speaker is, the less dynamic it is, no matter how much power is thrown at it.

[bullittstang]

...Never been a big fan of BSC either because of the losses. Of course, the use of a port (with the "gain") helps offset the loss in a BSC circuit

This statement illustrates you really should read some more about the finer points of designing a competent Speaker/XO design, IF you want the best sounding speaker possible. We would all love to ignore BSC too, but it's PHYSICS, you CAN ignore it, but your design will suffer because of it.

Decent choice of components - you really should optimize the XO and get the most out of them you possibly can.

[Brian B]

It's Alive!

The tweeter will need some padding as expected. The mid is considerably hotter than desired above 1-2K. So the WinPCD 1^st pass crossover (based on published graphs) isn't very pleasant sounding (not really a surprise). But it definitely shows promise.

[Chris Roemer]

That F5 has 2 pretty bad breakups, one near 1.5kHz (notch) and again near 5k (don't take it that high).

[dynamo]

[scottvalentin]

Those are great drivers in all respects, with some work and some help here you will have a very capable speaker. As Chris and Dynamo allude to, those breakups will need flattening and single order probably won't cut it on the F5, you will want a steeper slope. This also helps the XT25 as well, although at 3k it should work pretty good.

With FPTrace and PCD you should be able to come up with a pretty good sim that will be able to help you notch out the breakups and get the drivers to work together pretty good. Check out this writeup on how to piece together a sim using traces:
https://sites.google.com/site/undefi...d-measurements

If you follow these steps with the manufacturer data sheets, you should get pretty close and will then be able to tweak it to your liking.

[Millstonemike]

Two out of three ain't bad ...

You need the third leg of XO design: Response Modeler.
Response Modeler will allow you to account for BSC (and diffraction which can cause humps above nominal dB at certain freq.'s) as well as including the lower freq.'s of the box model (and min phase).

Trace the graph, process with Response Modeler, input the Response Modeler output into WinPCD. If the driver spec's are accurate, the model will be good.

Bullittstang's comment on BSC is right on.

[Brian B]

I'll plan to look at those additional tools.

In the mean time, I've worked more with the crossover and have it sounding pretty good (to my ears). It's all simple 2^nd order stuff at this point. WinPCD says it should produce a SPL like this:




And maybe it does. At any rate I'm liking it.

I have tailored the crossover a few different ways that produce a flatter SPL in PCD, but the octave to octave balance of the one above sounds more pleasant and natural to me than the those others. There obviously could be some things going on that I'm ignorant about (and/or that are outside the model), so for now I'll probably stop the crossover tweaking in order to review the other software tools suggested.

As an aside, I had enough scrap MDF to put together an Overnight Sensation over the last couple of days. Impressive sound for such a small speaker. But everyone knows that already.