crossover confusion

[Kittle]

Ive been experimenting with assembling some crossovers for a hypothetical 2 way system and wound up with some puzzling results.
1st off, I used this site to calculate my values from:
http://ccs.exl.info/calc_cr.html

Using a handy breadboard & a trip to radio shack I constructed a 2nd order high pass network for a vifa tweeter I have kicking around. It behaved as expected. -- namely that as my test tone went below the xo point, the audiable volume dropped off drasticly.

So then I try the same exercise for a woofer, but I get either No effect, or the opposite results of what I expect. What I expect is that as my test tone goes up in frequency past the xo point the audible volume drops off drasticly.

My drivers measured out thus:
tweeter: 4.9ohms
woofer: 6.5ohms
xo point: 2000hz

The above site gives me L2 = 0.73mH and C2=8.6uF.

I clipped a 10uF (I didnt have an 8.6 one) cap across the terminals of my woofer, then clipped a 0.7mH coil in-line with the positive lead, and finally clipped the negative lead to the other woofer terminal.

Test tone (generated by WinISD) has no audiable volume difference once it passes 2k hz.

So what am I doing wrong here?
Someone please point me in the right direction..........

[navin]

run the woofer au naturel. then run the woofer with only the coil in series. then add teh cap. let us know what u find.

[Bill Fitzpatrick]

Put a Zobel across the woofer and it will be a whole 'nother story.

[Kittle]

Bill is probably right.. but im trying to keep things simple -- for now.

By "connect the coil in series" I assume you mean one end goes on the woofer terminal & the other and on the wire leading to my amp. -- This I have tried.. with different coil values (1.0mH and 0.3mH) with no change in the output. This was with a 2k test tone.

[navin]

forget the test tone for now....use a CD player and play some music and see if you hear a difference between 0.3mh and 1mh.

my guess is that the output impedance of yout tone generator and the input impeandace of your amp might have something to do with this.

[navin]

Originally posted by Kittle
So then I try the same exercise for a woofer, but I get either No effect, or the opposite results of what I expect. What I expect is that as my test tone goes up in frequency past the xo point the audible volume drops off drasticly....I clipped a 10uF (I didnt have an 8.6 one) cap across the terminals of my woofer, then clipped a 0.7mH coil in-line with the positive lead, and finally clipped the negative lead to the other woofer terminal.

Test tone (generated by WinISD) has no audiable volume difference once it passes 2k hz.

hold it here. do you mean that as the tone increased the volume decreased but there was no further decrease below 2000Hz?

if this is true then Bill RC solution would work.

if however there is no volume change no matter what freq. then you might have an impedance mismatch.

[jomor]

Sorry to spoil the fun but its not gonna work this way. You cant just apply a formula from the web and make a
crossover. These formulas DONT work, since they assume the load ( driver ) is resistive. Which is not. To make
the driver appear as a resistive load to the filter, you must apply a RC zobel filter in parallel with the driver to
equalize the driver's impedance rise. This zobel CAN'T be calculated by the usual formulas that there are on the
web. They usually dont give correct results, sorry. The only way to go, is to measure the total impedance of the
combination driver+zobel and adjust the R and/or C values accordingly to get a resistive load. Then you can use
the filter desing programs. After completing the crossover's implementation according to the software, you must
measure the frequency response to see if the crossover design is successful. An important factor, is that the crossover's output depends on the driver's frequency response graph's shape too when the driver is in the box, not just on the crossover's topology and parts' values. Which can only be verified by spl measurement. After that, you must also do a lot of listening tests with many kinds of music, and fine tune the
crossover to get better results.

Its not a plug and play project, it needs patience, but the result will pay you off.

Good luck and keep us updated

Yannis


PS: the comments that the formulas dont work, are not mine, i ve seen it in practice too, but they re comments of professional experienced speaker designers I had the chance to talk with.

[Kittle]

Originally posted by navin

hold it here. do you mean that as the tone increased the volume decreased but there was no further decrease below 2000Hz?

if this is true then Bill RC solution would work.

if however there is no volume change no matter what freq. then you might have an impedance mismatch.

What seems to be happening is no volume change no matter what the frequency. I tried tones from 50hz up to 9khz. And while the 5k+ tones got softer just due to the woofer not being able to reproduce said tones.. I got no rollof starting at 2k like I expected.

Originally posted by jomor
Sorry to spoil the fun but its not gonna work this way. You

ahh.. but this IS the fun part

cant just apply a formula from the web and make a
crossover. These formulas DONT work, since they assume the load ( driver ) is resistive. Which is not. To make
the driver appear as a resistive load to the filter, you must apply a RC zobel filter in parallel with the driver to
equalize the driver's impedance rise. This zobel CAN'T be calculated by the usual formulas that there are on the
web. They usually dont give correct results, sorry. The only way to go, is to measure the total impedance of the
combination driver+zobel and adjust the R and/or C values accordingly to get a resistive load.
[/B]

okay.. lets start here then. I need a zobel filter.
but since your saying I cant use the values from the web.. What do I use? do I just grab a random resistor & cap and plug them in? or... ?

Then this "measure impedance" ... last time I tried to measure impedance I fried a good meter. I'll need guidance there too.

I begin to see why people get paid big $$ to design crossovers

[Swedish Chef]

This is probably one of the very best introductions to speaker crossover design I have ever read:
http://www.pispeakers.com/Speaker_Crossover.doc

/M

[jomor]

Originally posted by Kittle

Then this "measure impedance" ... last time I tried to measure impedance I fried a good meter.

What meter?

Impedance varies with frequency. Thus you should be able to measure it in a way that you ll get as a result a graph that will be a frequency - impedance plot. This is usually done via special software-hardware combination. You can use speakerworkshop ( www.speakerworkshop.com ) and your sound card to do an accurate impedance plot. No microphone needed for this. Download the program, read the help file carefully, and follow the instructions at Claudio's site STEP BY STEP and you ll do it.
( http://planeta.terra.com.br/educacao...negro/english/ )

Its not a plug and play thing, but its worthed the trouble ( as you said, thats the fun of it). Here, you wanted fun? Eat speaker workshop ) It rules !!

and dont forget to tell us how it goes