Am I on the right track? Bookshelf speaker Fountek FW146 / Dayton DC28F

[crossbound]

Since it has been a rainy weekend here in Florida, I finally decided to start learning a little more about designing my own speakers and started playing around with PCD. The drivers are the Fountek FW146 woofer and Dayton DC28F
It seems to LOOK good to me but since this is my first, I'm not sure. One thing I do not account for is baffle step diffraction loss. My understanding is that the smaller the baffle the greater the loss!? Jeff Bagby's Frequency Response Modeler would not run on my computer (Win10 64-bit). Can I estimate the loss somehow and bump the response curve accordingly?
Thanks for any input and advise.

[Wolf]

Your schematic shows a short circuit at the input to the xover through the shunt coil for the tweeter, and also above the Fc of cap on the woofer. IE- your xover schematic is laid out incorrectly.

Either way, you'll also need to suppress the breakup of the woofer more to at least -25dB from reference, and the tweeter can play to 1.7k really without issues.

Later,
Wolf

[dcibel]

http://www.tolvan.com/index.php?page=/edge/edge.php

[Chris Roemer]

The width of the baffle slides the "step" curve up or down, but the amount of the step (about -6dB) is basically the same.
"Diffraction" is a different thing, but related. It can be affected by generous chamfers or roundovers, while baffle-step really can't be.

Like Ben said, if you dropped the 1st (parallel "shunt") component from each of your filters, they'd look more topologically correct (not saying that the values would be right, though). Shoot for more like a 2k Fc, probably. I'd be lookin' at about a 0.4cf ported box. Maybe about a 1-3/8"d x 4" long port?

[crossbound]

So after trying to figure out what Wolf was exactly talking about I discovered that my PCD project file is corrupted. Some of the options and values are not visible or accessible. Specifically L2, R2 in the woofer section and C9, R11 in the tweeter section. Time to start over.
I will drop the crossover frequency down to 1.7k-2k and post my new response curve.
Thank you dcible for the link. I will check out that program.

[isaeagle4031]

The values aren't missing, you have PCD set to provide a series crossover where you schematic is a parallel.

[crossbound]

Hmm.. isaeagle, I'm pretty sure I had set it to parallel. Makes sense though. It was a lot easier to get the response curve right when set to parallel

[WernerM]

What Wolf was talking about that first popped out at me also is that the breakup peak of the woofer high up in the frequency range between 5k and 10k. If you notice it is playing only 10dB below the tweeter and will most likely be audible.

Also when you get close to a finished crossover design make sure you look up the DCR value of the inductors you will be using in your build and add those values in the sim. It will change the sim slightly.

[crossbound]

Marc,
If I run a resistor in series with an inductor, do I just add up the resistor and the DCR of the inductor for the simulation? And if needed adjust the resistor value accordingly?

Thx,
Billy

[WernerM]

It has been a while since I simmed, but IIRC there is a spot for series resistance in the crossover data.

[wogg]

PCD should let you add the series resistor in with the inductor leg, though you would get the same result simply adding to the DCR for the inductor. I use WinPCD, not sure about the Excel version. I also use Jeff's response modeler on a Windows 10x64 PC with Office 2013, it works OK for me, not sure what trouble you're running into.

As for the woofer breakup, I've found the easiest way to add attenuation to kill that is by paralleling a cap with the series low pass inductor. The cap creates a notch filter with the inductor. The frequency of the notch is calculated with f = 1/(2*pi*sqrt(L*C)). So you'd use your inductor value for the low pass and do some algebra work to find the C value that would create the notch at the frequency you need.

[crossbound]

I just looked at the screenshots of winPCD and there is a slight difference. There is no separate field for a resistor and the DCR in the excel version.

[WernerM]

IIRC when I added the DCR resistance, the largest effect I saw (which was still minimal) was in the impedance and phase shift at the crossover points. One sim I did after adding the DCR to all inductors I could see the difference in my reverse null, was about 5dB less at the crossover point on one and little effect on the other. It mostly matters with higher DCR inductors, but is always good practice to add it in on your sim.

[Chris Roemer]

The DCR of any (series) inductors on the mid-woof filter SHOULD be accounted for in your BOX MODEL (as it can affect the "Q").

[crossbound]

So I guess that's why one would use 18 or thicker gauge inductors in order to reduce the DCR.. correct?