newbie - confirming a few crossover design points - diyAudio
Go Back   Home > Forums > Loudspeakers > Multi-Way

Multi-Way Conventional loudspeakers with crossovers

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 15th March 2005, 06:39 AM   #1
diyAudio Member
 
tinkerbell's Avatar
 
Join Date: Feb 2005
Location: Sydney
Send a message via AIM to tinkerbell Send a message via MSN to tinkerbell Send a message via Yahoo to tinkerbell
Question newbie - confirming a few crossover design points

i have read that you should use the impedance of the driver at the cross-over point (say 4.5kHz) to calculate the crossover values, as the driver impedance changes with frequency.

i have measured the impedance using the method described here:

http://sound.westhost.com/lr-passive.htm

and the problem is that i get a 18 ohm impedance at cross-over frequency on the woofer (it is a Response 6.5in carbon fibre woofer from Jaycar).

the tweeter reading calculates to be about right at 5.x ohms (it is a SEAS H623 al dome)

the calculated 12db cross-over for the woofer is a 1353mH and 0.92uF

does this sound right?

can anyone provide a different source for a similar test or further reading sources for me to understand the processes better, as the ESP site is the only one i have found that talks about this stuff...

please help, as i am nervous about following through with the cross-over construction without independently confirming the proper values.

thanks,

david..
  Reply With Quote
Old 15th March 2005, 11:26 PM   #2
diyAudio Member
 
tinkerbell's Avatar
 
Join Date: Feb 2005
Location: Sydney
Send a message via AIM to tinkerbell Send a message via MSN to tinkerbell Send a message via Yahoo to tinkerbell
is this the wrong forum for cross-over questions?
  Reply With Quote
Old 16th March 2005, 02:25 AM   #3
jamesj is offline jamesj  New Zealand
diyAudio Member
 
Join Date: Jul 2004
Location: Middle Earth
Those are most definitely the wrong values for an 8-ohm woofer. The rise in impedence for your Jaycar driver is due to self inductance, which must be flattened with a zobel before you arrive at the component values.

James
  Reply With Quote
Old 16th March 2005, 06:45 AM   #4
diyAudio Member
 
tinkerbell's Avatar
 
Join Date: Feb 2005
Location: Sydney
Send a message via AIM to tinkerbell Send a message via MSN to tinkerbell Send a message via Yahoo to tinkerbell
so do i need to implement a zobel network before i measure the impedance at the desired cross-over frequency?
  Reply With Quote
Old 16th March 2005, 06:45 AM   #5
diyAudio Member
 
tinkerbell's Avatar
 
Join Date: Feb 2005
Location: Sydney
Send a message via AIM to tinkerbell Send a message via MSN to tinkerbell Send a message via Yahoo to tinkerbell
Quote:
Originally posted by jamesj
Those are most definitely the wrong values for an 8-ohm woofer. The rise in impedence for your Jaycar driver is due to self inductance, which must be flattened with a zobel before you arrive at the component values.

James
Hmmm, but i cant make a Zobel without measuring the driver can i?

maybe i need some more literature?!?!?



any online sources?!?

searching here only brings up knowledgeable guys discussing too indepth stuff for me to get into...
  Reply With Quote
Old 16th March 2005, 10:37 AM   #6
diyAudio Member
 
Join Date: Feb 2004
Location: Queensland Australia
Dear Tinkerbell, Can't talk much tonight no time. Happy to catch up tommorow/fri', your questions are sensible and not too hard to respond to. It helps with responses if you're in the same time zone too. Often you can post in Oz and by the time the rest of the world is on line your post is off the main board. I've got the Jaycar graphs in front of me. we should be able to generate something useful. Catch you later. Greeting from "The Smart State" as Pete calls it. Jonathan Bright
__________________
"It was the Springtime of the year when aunt is calling to aunt like mastodons bellowing across primeval swamps." P.G. Wodehouse.
  Reply With Quote
Old 16th March 2005, 01:32 PM   #7
rabbitz is offline rabbitz  Australia
diyAudio Member
 
rabbitz's Avatar
 
Join Date: Jul 2003
Location: Newcastle, Australia
18 ohm at 4500Hz sounds reasonable for a 6.5" woofer due to the rising impedance. There are different thoughts on the use of a zobel....... with or without...... do some research on it.

The choice is up to the builder and any circuit you add must be justified. You could always try it both ways as it's the only way to be sure and I can tell you now that normally the calculated crossover values will not be final and is only a starting point. Testing is a must for crossovers IMO....... 5% calculations, 95% testing.

For without a zobel, calculate on your measured impedance.

To calculate the Zobel you can get starting values by:
Resistor (Rz) = 1.25 x driver Re
Cap (Cz) = driver Le (henries)/Rz˛
..... then adjust them for flat impedance.


Cheers
  Reply With Quote
Old 16th March 2005, 08:24 PM   #8
diyAudio Member
 
Join Date: Feb 2004
Location: Queensland Australia
Hi David,
You are wise to use 12db/ove xover. They are less sensitive to the variables involved than higher order networks and they give resonable protection. Can I suggest that you don't stress too much on the accuracy of actual figures? People print data to three decimal places but there is usually a wide range in practise. (One of my wife's favourite stories is the one where the calculator salesman would offered a customer the new "Economist Model Calculator".....which guesses to 15 decimal places!)

I take it the woofer is the one on page 247 of the 2004 Jaycar cat. It is therefore nominally 4ohms.

If that's the case then you could try the following. A Zobel network is resistor in series with a cap and placed across the speaker terminals. For your unit you can't go wrong with a 4.7 ohm and probably a 5.6uF cap. Amps are usually better into a slightly rising impedence (inductive) rather than a capacitive load.
Make the resistor about 5watts rating or higher. The caps should all be bipolar.
Most midbass units enter some sort of "break up" in the region you are looking at. The graph shows a peak between 3 and 4 kHz. That's typical. The geometry of the usual box also contributes to a rise. In that case people often use an overdamped network. Lets take 4k Hz as a starting point. For 4ohms the "text book" values are 0.225 mH and 7uF. Now an overdamped network would have larger inductors and smaller cap. (scaled proportionally) So something like 0.260mH and 6uF (or maybe 0.3 and 5uF) would start to look promising for the lowpass section. I am not familar with the SEAS unit but there is usually less of a problem there. I wasn't quite sure what "5.x ohms" meant in your initial post.
I put Seas H623 into google and got a good data sheet. They call it a 6ohm model and the resonance is fairly high (1.800c/s) In that case stock values are 0.337 mHs and 5.25 uF (for 4kHz for 6ohms.) Now given the resonance frequency and the availability of "prefered value" components I'd do the following. I'd use a 4.7 uF cap and a 0.39 mH inductor. This is about the right LC time constant and gives a better damping of the tweeter.
Now this may seem a bit sloppy but it will work quite well. One of the previous posts was quite right in his approach. He quoted all the equations about how to accurately calculate the values and then suggested that you experiment to get the right result! But that's not bad advice. Much of this stuff is best described as highly technical theory combined with experience, guess work and measurement. It is something of a "black art".
But having done your best from theory you may well get a result where the two boxes are not the same due to legitimate component variations and then after listening for a while you decide to let the midrange out 3 db because you like it that way with the vocals more prominent and by doing that you have then diverged exactly 100% from classical theory becasue 3db is a doubling of power! Few other disciplines will take that sort of 100% error band in their stride but it is not uncommon in this field.

Anyway get back to this stuff later this evening. Looks like we are both going to get some rain in the next day or two. Good. Take care, Jonathan.
__________________
"It was the Springtime of the year when aunt is calling to aunt like mastodons bellowing across primeval swamps." P.G. Wodehouse.
  Reply With Quote
Old 17th March 2005, 08:10 AM   #9
diyAudio Member
 
tinkerbell's Avatar
 
Join Date: Feb 2005
Location: Sydney
Send a message via AIM to tinkerbell Send a message via MSN to tinkerbell Send a message via Yahoo to tinkerbell
thanks so much for you many words of wisdom.

but i must state that the purpose of my post seems to have been misconstrued (no offence!)

i did not wish for anyone to design my crossover, i simply wished to confirm the theory that i was using was correct enough to provide me with a result that i could use to construct a cross-over.

I also understand that the calculations will simply be approximations, and will be affected by soooo many variables, eg box shape/size/material/stuffing along with room size, and driver behaviour, but I like to understand the theory and see it applied in practice (ie empirically testing and then experimenting)

ie my initial questions were intended to ask:

1. is it right to try to measure the impedance of the driver at cross-over point (eg 4.5kHz) AND use this impedance in the calculations for L and C? or should i use nominal impedance to calculate it?

2. is the method described by Rod Elliot at his page the correct one to determine impedance at the desired frequency? http://sound.westhost.com/lr-passive.htm

3. does 18 ohms at 4500Hz sound right for a 4 ohm woofer? (thanks rabbitz for your response!)

4. are there any great online resources for understanding the testing of loud-speakers that will teach me more of the basics? Or should I bite the bullet and buy “the Cookbook”?

Now, since the helpful replies I have a few more questions about theory/practice:

a. should I test the loudspeaker for its impedance in free air OR on the intended baffle OR in the intended box?

b. rabbitz makes the point that calculated values are the starting point, how do I “test” the loudspeaker to determine final values? With a PC program, then with my ears?

c. if I want to run the Zobel network to keep the impedance of the woofer constant (and keep the impedance low at cross-over point) do I implement the Zobel network BEFORE measuring the woofers impedance for the L and C value calculations?

c2. the corollary of using a Zobel network *seems* to be lower L values and higher C values, is this a desirable thing? (It is in terms of expense!) is this what Jonathan is referring to when mentioning the preferability of the inductive load vs capacitive load?

d. the Jaycar drivers I have are the 4 ohm CS-2240 as you note Jon, the graph shows a freq. response up to about 5000Hz, where it begins to roll off, is it OK (or in fact desirable) to have a high 4.5kHz cross-over point in this situation (considering this point was chosen due to the high Fs of the Al dome tweeter at 1800Hz).

d2. is such a high freq. range of the drivers due to the Carbon Fibre cone, and hence 'atypical' of mid-bass drivers?

e. why do people typically use overdamped networks when there is peakiness in the freq. response around cross-over frequency (have I understood you correctly Jonathan?)

f. the 5.x ohms for the tweeter was "5 point something ohms" (could not exactly recall) so it seemed close to the nominal impedance. What is a ‘right’ time constant for a cross-over? Can this be calculated or is it intuitive?

g. are there any quirky things I need to be aware of when using Al dome tweeters?

Thanks for any input in relation to the above!!!

Sorry for the long post.

David..
  Reply With Quote
Old 17th March 2005, 09:01 AM   #10
diyAudio Member
 
Join Date: Feb 2004
Location: Queensland Australia
Hi David, Me again! Me thinks you'll find this a frustrating thread. I take your point about the original question: sorry. Now todays qu's
Q.1 Yes or No.
Yes the measured impedence is the one to use if you don't flattern the impedence curve. That's where people started offering zobel networks as they bring it back to nominal values.
Q.2 haven't read it but his stuff is usually v.good and relaible.
Q3. if it measure 18ohms then that's right. They do vary a bit but that value is not out of the ball park.
Q4 can't help on that one as I'm pre digital and I own books. I think that there is a good one called "Speaker lab" or something.

Q(a) in the box. But at that crossover frequency it won't make much difference. It is more important at the low frequency end ie below 200c/s
Q(b) Measure the response with a mic' or meter. Again there are certainly stuff for PCs.
Q(c) Yes
Q (c2) Yes, lower cost. Also it minimises the length of copper wire in the inductor and most of us want to reduce the wire a signal passes thru' Also a flatter curve is easier for amp' to drive. The bit about inductive or capcitive load refers to this area. Where an impedence curve does diviate from the straight and narrow it is better that it is a little inductive rather than a little capacitive. So if you got the zobel values too low it would be worse than a little high. Hence the values I suggested.
Q(d) I don't think many of us would want to go much higher that 4k for that sort of driver. Those published curves will be a bit "charitable". I like the c/over to have an octave to work on but the difference between 4 and 4.5 will not be great.
Q (d2) Too subjective an area to give a reply on. Kevlar has many fans.
Q(e) an overdamped network rolls off sooner (lower down) and hence tames the peak.
Q(f) you can work in the frequency domain or the time domain. It was probably a confusing term to use. If you are going to alter one of any pair of components (the L or the C) then the product should remain similar. IE if you multiply the L by the C. Thats all I did this morning when a shifted the hi pass network to "preferred" values. You can't go overboard on this but 10-20%
changes in each component will not do any damage. There is no "right" time constant. Each combination of L and C has its own inherant value. Sorry, forget I ever mentioned it!

Q(g) You start asking questions like that and you'll get a very subjective response. Those who love them will see no quirky features. Others will see them "under every spreading tree."
__________________
"It was the Springtime of the year when aunt is calling to aunt like mastodons bellowing across primeval swamps." P.G. Wodehouse.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
crossover points ddoyle Multi-Way 1 16th December 2006 09:33 PM
Anyone willing to help newbie with crossover design? zachary Multi-Way 2 21st November 2004 12:57 AM
Bss Fds340 Crossover Points ! Javier Alvarez Solid State 0 2nd June 2004 02:59 PM
Do crossover points really mess things up? BAM Multi-Way 3 2nd June 2002 05:57 AM
crossover points ogp Multi-Way 1 17th January 2002 10:20 PM


New To Site? Need Help?

All times are GMT. The time now is 04:08 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2