These speakers were my first self-designed units and the original write up is at the bottom of this page. I took the crossover values of an Internet calculator and then adjusted by ear. Adjusting by ear is almost impossible. I knew the tweeters were a little high as that is what everyone tells me and I now know I tend to bring up tweeters to compensate for my hearing loss in the high range.
Now that I have measurement equipment, I decided to revisit this project. I went through three XO designs. The first was me foundering about and I got a reasonable measurement but with a climbing SPL. This bothered me a lot even though they sounded OK. Then I saw an XO on the Internet and tried that. It was worse than mine but with started with a different topology. With tweaking I made an improved over my first try. The main thing this new version added was a Zobel. I Posted on the mad board and John Nail suggested I try the design below. With his design, it was very easy to get a good response with fewer components. This is the final measurement.
Note the gate is low so everything below 200 Hz is not reading.
I did not end up with the dip as he recommends. I landed on a good measurement early on and decided to keep it. But the advice is good. :) This is his post:
In your LSPcad model, try doing something like this... Model the response so that starting at 1K there is the starting of a slight dip that reaches a depth of about 3-4dB at 3-4K. Model the tweeter with a cap value starting at 4MFD and a shunt of .25mH. DO NOT use an LPad on the tweeter. Rather use an inserted resistance value right after the series cap on the tweeter, but BEFORE the coil. Then start tweaking the tweeter resistor so that the tweeter level from 8K to 12K is at the same level as the majority of the woofer. Then start tweaking the cap and shunt coil values of the tweeter so that the combined system response has a -2 - -2.5db point of 5K and a -3 - -4dB point at 3-4K. The overall system response should look like it has a broad dip starting at 1K and center at 3.5K where it reaches a down point of -3.5 - -4dB then start back up and reaches the "0" reference level at 8K or near enough there that it doesn't matter. The woofer should have a simple coil and a modified zobel, not a textbook zobel. It should be somewhere in the neighbor hood of 5-6mfd and 4-5 ohms. Use that modified zobel to adjust the woofer response to get the transition smooth a progressive. The finished XO should be simple to build. Build it a give a listen. I think you might like what you hear. John
Here is the schematic:
This is the original write up:
I had the two spare ACI Emerald boxes hanging around the house so I congured up some deals on the older Vifa D27-45 tweeters and P13 woofers. I used a straight 2nd order xo and padding. I gave them to my daughter. They sound pretty good and measure decently. They have the usual P13 roll off around 70Hz. But it is nice to train my kids ears on decent speakers and not a boom box.
It is true what everyone says about these Vifa drivers. They are well behaved and you can get good results with simple alignments. Note these are not Emerald copies. Different drivers and different xo. And no, they do not sound quite as good as the Emerald. But close.
Not a very good pic. How many 13 year olds have nice Vifas, sand filled stands and 80 watts class A in a mint vintage receiver?
Not bad for throwing a couple of Vifa components into a box. The drop off at 10k is my mic.
Copyright Peter Jay Smith 2005 Return to helarc.com