Tweet
This is my method of doing things:
I use speaker workshop, spltrace, and jeff B's response modeler all of which are free to do crossover simulation. Jeff B also has a free crossover simulator called PCD but I'm more experienced with speaker workshop so that's what I use. Maybe someone who uses PCD can post a quick PCD guide if they feel up to it.
Speaker workshop: http://www.speakerworkshop.com/SW/Download.htm
SPL trace: http://www.pvconsultants.com/audio/utility/spl.htm
Response modeler: http://audio.claub.net/software/jbagby.html#FRM (requires excel)
PCD: http://www.pvconsultants.com/audio/crossover/pcd.htm (requires excel)
My process is as such:
Find decent response plots of drivers (if you can get them all from the same site for example zaph's site, your results will be better as testing conditions will not be way off between drivers).
Save the plots to your disk, open them in paint, select all, copy.
Open SPL trace, file>import from clipboard.
Set low amplitude (I do 60dB)
Set High amplitude (I do 90dB)
Move your mouse around and look in the lower right corner to be sure the values it's reading are correct. If not then reset the lines.
Then set the low and high frequencies (i always try to do 20-20000)
Under trace>options, I set the trace limits to 20-20000, 200 samples.
Then trace>start spl trace. Line up the white line on the left with the point in the graph where the response is, then click. Repeat 200 times :P
When you reach the end it will say "trace limit reached."
Then go to trace-stop trace. It will ask if you want to save it. Say yes. I label it the name of the driver plus trace. For example RS180-4trace.frd
I do not trace impedance as I model this in response modeler.
Repeat the SPL trace for any drivers you might be using.
Now onto response modeler.
Fire it up and make sure you enable the macros. You should see a blue trace in the graph like in the screenshot.
Click the blue button "import frd file to modify" and find one of the files you traced earlier. It will load it an it will show up in the graph.
Next I model the impedance. Scroll down to the impedance/box model section and input the T/S parameters of the driver you want to model. Look at the impedance from the measured driver plot and adjust the Le coeff. and Le exp. (red text) until the rising impedance curve in response modeler matches what the actual measured impedance looks like. I model it instead of tracing because there's never enough detail when tracing to get a smooth plot. Once it looks good click "save modified result to zma file" (in red) and save it. I name it driver+impedance to keep it simple. So going by the previous example, RS180-4impedance.zma
I do the same thing for tweeters. Sometimes they don't have T/s parameters listed so you just have to fiddle. Put in the FS at least, then I set both qes and qms to 1.2. vas 1liter. Box type sealed, volume 1 liter. Fiddle with the box size and qes/qms until the "bump" at the resonant frequency is of the correct amplitude. For Le i start with .1mh, then adjust the Le coeff and exponent to match the measured impedance plot. The closer you can get things the better.
Now we get to model the baffle effects. Scroll down the bottom area of response modeler for baffle diffraction response modeling. Input the baffle width and height, where the speaker is located on the baffle, what size roundover if any, and the approximate size of the woofer. Then click "save baffle diffraction curve to BDS register above" (in blue). If you scroll back up to the top you'll see it has been applied to your imported response.
Now you could be done here but usually we don't need a full 6dB of baffle step simulation, so I then do the following:
Near the top where it says "baffle step: simple step or from below," click where it says "off" a few times until it says "inverted." Then set the baffle width. I usually set it to about 45% inverted, which is like accounting for half baffle step which is a good comprimise for placement flexibility. You'll see that the curve is not as pronounced as it was with the full step.
Now on the right, click "save modified result to frd file" (in red). I save it as driver name+baffle dimensions, for example RS180-4_10x15.frd
Now one more step, we need to extract minimum phase. Scroll down a bit more and to the right slightly, you'll see a button labeled "auto-extract phase using hilbert bode transform." Click this and find the file you just saved. If it saved something about fft.dll just click ok and ignore it. Wait a few seconds while it thinks. Then it will prompt you to save the new file. I save it the same name as before but I add "phase" to the end so I know it's the one with phase data in it. So RS180-4_10x15phase.frd
Repeat that whole proccess for the rest of your drivers (actually doesn't take all that long).
Now they are ready for simulation! Fire up speaker workshop and create a new project, give it a name.
Right click in the sidebar to the left and click "import."
Set file type to .frd and find one of the "phase" files you made in response modeler. Import it. Repeat for the rest of the drivers you're using.
Then right click>import, change file type to impedance .zma files, import those.
The right click>new>driver. Name it the driver (RS180-4) .Right click the window that pops up and select "properties" down at the bottom of the list. Click over to the "data" tab.
Click the question mark next to the "impedance" box and load the impedance date for that driver. Repeat for frequency response. Click OK, and you can close the driver window. Repeat for your other drivers.
Now rightclick on sidebar, new>network. Call it crossover or something like that. Double click the icon for the network and it will pop up a mostly white window with a signal generator icon. Right click on the white space in that window and select "properties." Make sure "individual response" and "calculate network impedance" and both checked. Click "OK" to get out of properties.
Rightclick on whitespace again, insert>driver. Click the question mark, select a driver. Make sure "individual response" is checked. Offset should be zero for mids/woofers, 1inch for tweeters. Click ok and poof! Your driver is there. Repeat for the other drivers. Now you can insert capacitors, resistors, inductors, etc to build your crossover. To connect multiple parts to the same point, you need to hold down the shift key while dragging them together. You'll get the hang of it. I usually use an online calculator to get rough values such as lalena.com
Right click on the white space and select "calculate response" to simulate the crossover. It will show up on the lefthand sidebar as crossover.total response. Double click this to open a window where you can see it. Right click on this graph and select "chart properties." Under the data sets tab, i usually add the individual response plots of the drivers and make them different colors, so I can see what each driver is doing. X-axis is usually fine as-is. Y axis I turn off "auto minimax" under the scale section, and set the minimum at 50dB and the maximum at 95dB. Then under grid lines below that, major lines every 5dB. Y2 axis is the phase. I turn of minmax and set the min/max to -180/180, and then turn off grid lines to keep things uncluttered. Press OK and you're in business!
Now just keep tweaking values and right clicking to regenerate the response.
Keep in mind that sine we didn't measure the data ourselves, the final design will likely need tweaking by ear or with the aid of measurement to truly be perfect. Good luck!
I use speaker workshop, spltrace, and jeff B's response modeler all of which are free to do crossover simulation. Jeff B also has a free crossover simulator called PCD but I'm more experienced with speaker workshop so that's what I use. Maybe someone who uses PCD can post a quick PCD guide if they feel up to it.
Speaker workshop: http://www.speakerworkshop.com/SW/Download.htm
SPL trace: http://www.pvconsultants.com/audio/utility/spl.htm
Response modeler: http://audio.claub.net/software/jbagby.html#FRM (requires excel)
PCD: http://www.pvconsultants.com/audio/crossover/pcd.htm (requires excel)
My process is as such:
Find decent response plots of drivers (if you can get them all from the same site for example zaph's site, your results will be better as testing conditions will not be way off between drivers).
Save the plots to your disk, open them in paint, select all, copy.
Open SPL trace, file>import from clipboard.
Set low amplitude (I do 60dB)
Set High amplitude (I do 90dB)
Move your mouse around and look in the lower right corner to be sure the values it's reading are correct. If not then reset the lines.
Then set the low and high frequencies (i always try to do 20-20000)
Under trace>options, I set the trace limits to 20-20000, 200 samples.
Then trace>start spl trace. Line up the white line on the left with the point in the graph where the response is, then click. Repeat 200 times :P
When you reach the end it will say "trace limit reached."
Then go to trace-stop trace. It will ask if you want to save it. Say yes. I label it the name of the driver plus trace. For example RS180-4trace.frd
I do not trace impedance as I model this in response modeler.
Repeat the SPL trace for any drivers you might be using.
Now onto response modeler.
Fire it up and make sure you enable the macros. You should see a blue trace in the graph like in the screenshot.
Click the blue button "import frd file to modify" and find one of the files you traced earlier. It will load it an it will show up in the graph.
Next I model the impedance. Scroll down to the impedance/box model section and input the T/S parameters of the driver you want to model. Look at the impedance from the measured driver plot and adjust the Le coeff. and Le exp. (red text) until the rising impedance curve in response modeler matches what the actual measured impedance looks like. I model it instead of tracing because there's never enough detail when tracing to get a smooth plot. Once it looks good click "save modified result to zma file" (in red) and save it. I name it driver+impedance to keep it simple. So going by the previous example, RS180-4impedance.zma
I do the same thing for tweeters. Sometimes they don't have T/s parameters listed so you just have to fiddle. Put in the FS at least, then I set both qes and qms to 1.2. vas 1liter. Box type sealed, volume 1 liter. Fiddle with the box size and qes/qms until the "bump" at the resonant frequency is of the correct amplitude. For Le i start with .1mh, then adjust the Le coeff and exponent to match the measured impedance plot. The closer you can get things the better.
Now we get to model the baffle effects. Scroll down the bottom area of response modeler for baffle diffraction response modeling. Input the baffle width and height, where the speaker is located on the baffle, what size roundover if any, and the approximate size of the woofer. Then click "save baffle diffraction curve to BDS register above" (in blue). If you scroll back up to the top you'll see it has been applied to your imported response.
Now you could be done here but usually we don't need a full 6dB of baffle step simulation, so I then do the following:
Near the top where it says "baffle step: simple step or from below," click where it says "off" a few times until it says "inverted." Then set the baffle width. I usually set it to about 45% inverted, which is like accounting for half baffle step which is a good comprimise for placement flexibility. You'll see that the curve is not as pronounced as it was with the full step.
Now on the right, click "save modified result to frd file" (in red). I save it as driver name+baffle dimensions, for example RS180-4_10x15.frd
Now one more step, we need to extract minimum phase. Scroll down a bit more and to the right slightly, you'll see a button labeled "auto-extract phase using hilbert bode transform." Click this and find the file you just saved. If it saved something about fft.dll just click ok and ignore it. Wait a few seconds while it thinks. Then it will prompt you to save the new file. I save it the same name as before but I add "phase" to the end so I know it's the one with phase data in it. So RS180-4_10x15phase.frd
Repeat that whole proccess for the rest of your drivers (actually doesn't take all that long).
Now they are ready for simulation! Fire up speaker workshop and create a new project, give it a name.
Right click in the sidebar to the left and click "import."
Set file type to .frd and find one of the "phase" files you made in response modeler. Import it. Repeat for the rest of the drivers you're using.
Then right click>import, change file type to impedance .zma files, import those.
The right click>new>driver. Name it the driver (RS180-4) .Right click the window that pops up and select "properties" down at the bottom of the list. Click over to the "data" tab.
Click the question mark next to the "impedance" box and load the impedance date for that driver. Repeat for frequency response. Click OK, and you can close the driver window. Repeat for your other drivers.
Now rightclick on sidebar, new>network. Call it crossover or something like that. Double click the icon for the network and it will pop up a mostly white window with a signal generator icon. Right click on the white space in that window and select "properties." Make sure "individual response" and "calculate network impedance" and both checked. Click "OK" to get out of properties.
Rightclick on whitespace again, insert>driver. Click the question mark, select a driver. Make sure "individual response" is checked. Offset should be zero for mids/woofers, 1inch for tweeters. Click ok and poof! Your driver is there. Repeat for the other drivers. Now you can insert capacitors, resistors, inductors, etc to build your crossover. To connect multiple parts to the same point, you need to hold down the shift key while dragging them together. You'll get the hang of it. I usually use an online calculator to get rough values such as lalena.com
Right click on the white space and select "calculate response" to simulate the crossover. It will show up on the lefthand sidebar as crossover.total response. Double click this to open a window where you can see it. Right click on this graph and select "chart properties." Under the data sets tab, i usually add the individual response plots of the drivers and make them different colors, so I can see what each driver is doing. X-axis is usually fine as-is. Y axis I turn off "auto minimax" under the scale section, and set the minimum at 50dB and the maximum at 95dB. Then under grid lines below that, major lines every 5dB. Y2 axis is the phase. I turn of minmax and set the min/max to -180/180, and then turn off grid lines to keep things uncluttered. Press OK and you're in business!
Now just keep tweaking values and right clicking to regenerate the response.
Keep in mind that sine we didn't measure the data ourselves, the final design will likely need tweaking by ear or with the aid of measurement to truly be perfect. Good luck!
Comment