Go Back   Home > Forums > Design & Build > Software Tools

Software Tools SPICE, PCB CAD, speaker design and measurement software, calculators

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 10th March 2013, 12:57 PM   #1
tvrgeek is offline tvrgeek  United States
diyAudio Member
 
Join Date: Dec 2009
Location: Md
Default Verification of a CAD crossover using SPICE

Playing around with several of the crossover design tools, I found it is very easy to come up with many solutions that give a very good response through the crossover region, but it also occurred to me that you know nothing about the filter Q's you are coming up with. when you start just flipping through values. You can get some hint looking at the output phase, but not much.

So, I have taken to modeling thew potential solutions in LTSpice using resistors as loads. Sure enough, many solutions that look great in the sim are very high Q or very low Q. Some are quite reasonable. The other great thing about this is fiddling in Spice, you may see some of the more subtle actions of the filters that were giving you issues in the crossover sim or that may be more transient related when you build it.
  Reply With Quote
Old 10th March 2013, 01:16 PM   #2
diyAudio Member
 
vacuphile's Avatar
 
Join Date: Apr 2011
Location: Right behind you.
You can also model the load to resemble a real speaker to spice things up.
__________________
Everything eats everything else, given appetite and opportunity.
  Reply With Quote
Old 10th March 2013, 02:13 PM   #3
dlr is offline dlr  United States
diyAudio Member
 
Join Date: Mar 2005
Location: Canton, MA
Quote:
Originally Posted by tvrgeek View Post
Playing around with several of the crossover design tools, I found it is very easy to come up with many solutions that give a very good response through the crossover region, but it also occurred to me that you know nothing about the filter Q's you are coming up with. when you start just flipping through values. You can get some hint looking at the output phase, but not much.

So, I have taken to modeling thew potential solutions in LTSpice using resistors as loads. Sure enough, many solutions that look great in the sim are very high Q or very low Q. Some are quite reasonable. The other great thing about this is fiddling in Spice, you may see some of the more subtle actions of the filters that were giving you issues in the crossover sim or that may be more transient related when you build it.
Most design tools (all that I've used over the years) also show you the electrical circuit transfer function. For almost all real-world crossovers, there is no Q. It's an overall transfer function. What matters is not the Q of any particular circuit section, what is important is the acoustic output of the driver. The goal is a specific acoustic output. To achieve that goal, you have to create a crossosver that, when combined with the raw driver output, yields the desired acoustic output. There are different ways to achieve this within reason, the final sensitivity may be different while still having the same acoustic output shape/order in the passband and the final system impedance profile will be different, so you have to decide which way is the best for your needs. Lower/higher impedance output, lower/higher overall sensitivity, etc. But in all cases, the crossover impedance when terminated by a resistive load is unimportant. The electrical transfer function required to meet a target is dictated by the electro-acoustic response of the driver measured when mounted on a baffle, since diffraction will inherently be a factor as well.

With rare exceptions, you won't find any drivers with a nearly flat impedance profile. The load to the crossover is the electrical impedance of the driver. Exchange a resistive load with a real driver and the final system impedance will change, sometimes rather drastically, so it's not of much use to examine the electrical response to a resistive load.

Dave
  Reply With Quote
Old 10th March 2013, 02:21 PM   #4
tvrgeek is offline tvrgeek  United States
diyAudio Member
 
Join Date: Dec 2009
Location: Md
Yes you can, and using WT-II, it even gives it to you. ( a few companies are starting to publish them) Probably a good thing to do to verify any notch is centered correctly because you can't do that with a "nominal" resister load. But that is not the point. The point is to see the filter itself, not the composite. It's to keep clear which direction you are going.

As an example, I would not have guessed the difference in Q between using a series pad in front of the HP filter and an L-Pad at the tweeter. Learning experience.
  Reply With Quote
Old 10th March 2013, 02:31 PM   #5
tvrgeek is offline tvrgeek  United States
diyAudio Member
 
Join Date: Dec 2009
Location: Md
Dave,
Where I agree with all you say here, a couple small points. 1: All tools don't let you isolate the filter from the load. 2: This is not to pick the correct values, but to keep you from going too far astray. Educational. When doing the curve fitting with WinPCD, I found side effects of the changes I was making that were not clear why. Ripple shifting the opposite direction than I expected. Looking at the raw filter in a resistive load can enlighten you on what you are doing wrong. Then you have to go back to the real impedance, phase, and SPL in the model to get the solution. Change the overlap, or change the Q? I think it can help.
  Reply With Quote
Old 10th March 2013, 02:36 PM   #6
dlr is offline dlr  United States
diyAudio Member
 
Join Date: Mar 2005
Location: Canton, MA
As a learning experience, sure. The one aspect where I would concede that it may be useful is in the sensitivity to variation of individual components since component tolerance will have an effect with higher Q sections being more sensitive.

Dave
  Reply With Quote
Old 11th March 2013, 07:54 PM   #7
tvrgeek is offline tvrgeek  United States
diyAudio Member
 
Join Date: Dec 2009
Location: Md
Plop in a few of your crossovers and see what it shows. See if you think it is helpful. I granted that those with many years experience have a better sense in the first place and probably would not get too far out of whack. For folks just starting to use the modeling tools, and those who do not have an electronics background like we do, it could be helpful.
  Reply With Quote
Old 11th March 2013, 10:44 PM   #8
dlr is offline dlr  United States
diyAudio Member
 
Join Date: Mar 2005
Location: Canton, MA
Quote:
Originally Posted by tvrgeek View Post
Plop in a few of your crossovers and see what it shows. See if you think it is helpful. I granted that those with many years experience have a better sense in the first place and probably would not get too far out of whack. For folks just starting to use the modeling tools, and those who do not have an electronics background like we do, it could be helpful.
Jeff provided text files that he called Null Drivers with zero for magnitude and phase, I'm not sure why off-hand. I just edited both to change the frd file to be nominal 90db flat and the zma to be nominal 8 ohms flat. With these files imported, you can then enter only a parallel trap in series or with a series resistor added to provide a voltage divider, you can then enter parallel traps and see the resultant response of those circuits only. This lets you see the nature of the circuit response. There's no need to do anything special.

If you want to see what the primary crossover circuit does alone, just set up a 2-way. In PCD and WinPCD you have the graphs for the transfer functions that show the response of the circuit transfer function alone. With nothing but primary crossover elements, that show precisely the transfer function of those in isolation.

This obviates the need to go for special effort (and at no extra cost) in using something such as Spice.

One note - in WinPCD you'll have to import them into woofer and tweeter before the program will calculate the driver responses. This is a bit quirky, but it's the way it is for now. It simplified some of the internal logic.

Dave
  Reply With Quote
Old 12th March 2013, 12:54 AM   #9
tvrgeek is offline tvrgeek  United States
diyAudio Member
 
Join Date: Dec 2009
Location: Md
LTspice is free. No extra cost. BIG screen display. Plus it gives you a nice schematic to keep in the build file that is easy to read. OK, I am reaching a little there, but it does.

A null driver would do the trick if one remembered to turn off the mic cal. To see the transfer function, you would need to change it to whatever the nominal-at-crossover your driver is. Easier to change one value in Spice.

You could of course run Monte-Carlo in the crossover, but as we know, just the room temp, let alone Re heating is way larger that the variance in the crossover if using film caps. If you are using electro's still, then precision is not your concern so why bother. (hint to those who have not read the data sheets for an electro, that 20% tolerance is with all other parameters constant, which they are not. 40% is a closer range in the wild.)

It may be more fun once to take the driver model and put in the variables for heating and suspension limbering up to see how much effect it has on the crossover. You could also use it to judge different configurations with regard to predicted harmonic distortion. Maybe people shouldn't when the see what a couple of mH coils is doing to them
  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
LT-spice VS Eagle... CAD cooperation... MiiB Solid State 5 15th May 2009 09:52 PM
SPICE modelling for crossover design zebra100 Multi-Way 2 15th May 2009 07:55 AM
Transport Verification ZenCow Digital Source 6 19th September 2008 08:00 PM
linear technology spice cad wonder_b Solid State 20 25th December 2004 12:51 AM


New To Site? Need Help?

All times are GMT. The time now is 07:17 AM.


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