No I don't believe I'll run full sim on SW, I just haven't had the opportunity to go back through and do an impedence trace in SPLTrace So I'm lacking some parameters.
Why are you using SPL trace for the impedance? The reason I ask is that there are measuring programs that will export files to use. Or if you are using manufacturers data, does it include impedance phase?
Modelling the box resonance won't help you either if you can measure the response. It isn't always easy to get the lowest octaves when you measure inside though. What program are you using to measure with?
Modelling the box resonance won't help you either if you can measure the response. It isn't always easy to get the lowest octaves when you measure inside though. What program are you using to measure with?
Measure, do you mean as in using a microphone? None.
From what I can see yes the manfacturer included Impedence graphs joined into their Frequency reponse graph. Unless I am mistaken.
From what I can see yes the manfacturer included Impedence graphs joined into their Frequency reponse graph. Unless I am mistaken.
I'm in a couple of threads like this, Istoc, excuse me if I lose track or repeat myself. 😉 Anyway, you have a good start but there are a few things you'll need to remember if you're using manufacturers data.
Firstly, when you don't have acoustic phase data, don't pay attention to the total response in SW (or any simulator). The best you can do is create a target rolloff for each driver and work on them separately to achieve that goal.
When the only plot you have was taken at 0 degrees, keep away from the response extremes (in particular the upper response of a woofer) as it becomes inaccurate. In other words, cross lower.
When you don't have impedance phase data the crossover will be somewhat inaccurate so if something doesn't sound right you might want to tweak by ear.
You'll need to consider the effect of the baffle yourself. If you apply a simmed correction to your response files before you construct your crossover it can simplify the crossover, but a full simulated baffle might be going too far. There is a way to work this out using some without figuring in your baffle, using a few steps, so you could forget about this for now but just keep it in mind.
Firstly, when you don't have acoustic phase data, don't pay attention to the total response in SW (or any simulator). The best you can do is create a target rolloff for each driver and work on them separately to achieve that goal.
When the only plot you have was taken at 0 degrees, keep away from the response extremes (in particular the upper response of a woofer) as it becomes inaccurate. In other words, cross lower.
When you don't have impedance phase data the crossover will be somewhat inaccurate so if something doesn't sound right you might want to tweak by ear.
You'll need to consider the effect of the baffle yourself. If you apply a simmed correction to your response files before you construct your crossover it can simplify the crossover, but a full simulated baffle might be going too far. There is a way to work this out using some without figuring in your baffle, using a few steps, so you could forget about this for now but just keep it in mind.
No worries Allen, I understand 🙂. Consider the effect of the baffle? I'm sorry but can you elaborate? As in using a Unibox style program to set up a box sim?
Now that I re-read that it does sound unclear. The specific effect I'm talking about is where the lower frequencies wrap around the cabinet as their wavelength is larger (wavelength is the distance sound will travel during one cycle of sound at a given frequency). They therefore enter the room at a wider coverage angle rather than mostly moving forward as the higher frequencies will, meaning the higher frequencies would measure louder and may come across as louder simply because they are concentrated toward the front. Often when measurements are not taken, the mids and treble are reduced in an attempt to tip the balance in favour of the forward response against the total response. Usually referred to as the baffle step, it often occurs between maybe 200Hz and 2kHz depending on baffle conditions (mostly width).
By and large, I consider the outside of the box requires more work and thought than the inside of the box. Anyway, it is possible to model this and related effects using tools such as the 'baffle diffraction simulator' spreadsheet, or programs like 'the edge'. I'd encourage you to take a look if only to get a feel of what is happening to the forward sound.
It is possible to add this simulated effect to your response file and proceed to design a crossover that makes it flat. This is where I was suggesting that doing so may produce an unnaturally bassy sound simply because both the forward sound and the total sound are important. I was also suggesting that you could ignore it for now and find the correct adjustment in time using steps. The first step is to use SW to design a crossover that gives you the highest sensitivity flat response that you can.
By and large, I consider the outside of the box requires more work and thought than the inside of the box. Anyway, it is possible to model this and related effects using tools such as the 'baffle diffraction simulator' spreadsheet, or programs like 'the edge'. I'd encourage you to take a look if only to get a feel of what is happening to the forward sound.
It is possible to add this simulated effect to your response file and proceed to design a crossover that makes it flat. This is where I was suggesting that doing so may produce an unnaturally bassy sound simply because both the forward sound and the total sound are important. I was also suggesting that you could ignore it for now and find the correct adjustment in time using steps. The first step is to use SW to design a crossover that gives you the highest sensitivity flat response that you can.
Alright I managed to log about 20 minutes and get some of the other crossover stuff done. Thanks again for being so helpful Allen this is incredible. Now should I start generating Impedence curves for the ranges specified with Manufacturers data using SPLTrace?
I in essence got the crossover networks mapped out in SW and had to run but progress is progress.
I in essence got the crossover networks mapped out in SW and had to run but progress is progress.
Looks like the only thing stopping you from starting the crossover is the impedance information. The only other things you'll want to do first is choose a level, choose two crossover frequencies and generate target curves for each driver.
Is using the blue line on my Manufacturers Data accurate for the impedance curves in SPLTrace correct? Also if its not too much bother how are those three things accomplished? Is it within SW or external programs?
I am assuming the blue line is the impedance magnitude (if I misunderstand that, show me the curve you're referring to). Just two caveats with this data. Firstly, look at the scale and see if it is log or linear (2,4,8,16 or 4,6,8,10,12) and use SPLtrace appropriately. Secondly, since you don't have phase data make sure SW can handle it, ie look at the plot and also make sure the phase data shows a flat zero as expected. One other thought is that you could use a driver modelling program to produce a simmed impedance curve which might happen to be more accurate. I don't do this myself so I can't recommend a program to do it with, and don't stress about the impedance down near the box resonance if you can't sort that, it's not too close to the crossover.
Choosing crossover frequencies and levels depends on the drivers, their sizes, their dispersion, their capabilities and their failings. If you want me to assist with that can you remind me of the drivers you've settled on, or where you posted that or link me to the specs?
Once you know how loud, where to cross and the slope to use, the target is generated in SW and overlaid onto the crossover response chart so you can tweak the response to match it.
Choosing crossover frequencies and levels depends on the drivers, their sizes, their dispersion, their capabilities and their failings. If you want me to assist with that can you remind me of the drivers you've settled on, or where you posted that or link me to the specs?
Once you know how loud, where to cross and the slope to use, the target is generated in SW and overlaid onto the crossover response chart so you can tweak the response to match it.
http://www.diyaudio.com/forums/multi-way/202082-3-way-crossover-assistance-request.html#post2811425
As it happens I posted the links to the webpages I've been using for the data on my first post 🙂.
As it happens I posted the links to the webpages I've been using for the data on my first post 🙂.
Ok, I recall making a comment on the upper crossover frequency in this post. A quick look at the woofer and mid would suggest somewhere in the 400Hz to 800Hz region.
A few questions. What are the dimensions of your box? What is the size of your midrange enclosure? Have you formed a preference for anything to do with the crossover at this point?
At the moment I'm thinking that 86dB might be a good sensitivity to shoot for. This is the level of the mid, which happens to be the least sensitive. This leaves the woofer a couple of dB louder which will be OK as long as we don't allow it to be that way above, say, 100-200Hz.
We can create targets now, and produce a crossover when you have the impedance data ready.
How do you feel about trialling a crossover and then tweaking it according to your thoughts? This will naturally take time, and you'll probably need a few extra components. My thoughts would be to produce this first run crossover, then to find the best polarity and estimate the phase issues by ear. Then depending on your responses, to try a partial baffle compensation, and to investigate relaxing the treble and the critical 4kHz region.
A few questions. What are the dimensions of your box? What is the size of your midrange enclosure? Have you formed a preference for anything to do with the crossover at this point?
At the moment I'm thinking that 86dB might be a good sensitivity to shoot for. This is the level of the mid, which happens to be the least sensitive. This leaves the woofer a couple of dB louder which will be OK as long as we don't allow it to be that way above, say, 100-200Hz.
We can create targets now, and produce a crossover when you have the impedance data ready.
How do you feel about trialling a crossover and then tweaking it according to your thoughts? This will naturally take time, and you'll probably need a few extra components. My thoughts would be to produce this first run crossover, then to find the best polarity and estimate the phase issues by ear. Then depending on your responses, to try a partial baffle compensation, and to investigate relaxing the treble and the critical 4kHz region.
I believe when I built my current crossovers they were designed for around that crossover range on the low end.
The Midrange external dimensions are 13 1/4" Tall 10 1/2" Wide 8" Deep.
No real preference on what to do to modify the crossover yet. I was also thinking that lowering the other two drivers to pair with the midrange sensitivity might also be the way to go, it was actually the concept I thought paramount when I started this modification venture.
I have some errands today but I will have time and I will map the impedance magnitude from my Manu Data then post again and I guess we can start working things down. 🙂
Allen you sir are awesome.
The Midrange external dimensions are 13 1/4" Tall 10 1/2" Wide 8" Deep.
No real preference on what to do to modify the crossover yet. I was also thinking that lowering the other two drivers to pair with the midrange sensitivity might also be the way to go, it was actually the concept I thought paramount when I started this modification venture.
I have some errands today but I will have time and I will map the impedance magnitude from my Manu Data then post again and I guess we can start working things down. 🙂
Allen you sir are awesome.
Allen, I finally got the chance to sit down and do some work on SPL Trace to attempt an Impedance graph. However it seems a little off to me. Does this look accurate to you?
Its also a pain in the butt atm to get my graphs for Tracing as Madisound(The location of the images) is buggy right now and not displaying the web pages. So I am having to save a PDF Copy of the specs sheets then screenshot it then work the screenshot down to just the Graph then copy the graph to SPL. But more to the point what do you think of this tracing?
Its also a pain in the butt atm to get my graphs for Tracing as Madisound(The location of the images) is buggy right now and not displaying the web pages. So I am having to save a PDF Copy of the specs sheets then screenshot it then work the screenshot down to just the Graph then copy the graph to SPL. But more to the point what do you think of this tracing?
Yeah thats where I'm Copying the PDF from to do the whole process 🙂. Thanks for confirming that it looks ok. I'll do my other two and then I guess its time to start combining and breaking down?
There are bumps in my graphs where apparantly it decided to jump to my mouse for a single plot for no reason but with my SPLTrace destroying my eyes, its acting funny..., I'm going to disregard the little bumps and use them anyway.
My SPLTrace is showing a white background with a slightly off white line for the tracing line. Its near impossible to tell the difference between them on my screen, which causes significant strain however I now have all my data I believe outside Baffle/Enclosure information in SW. What do I start with?
My SPLTrace is showing a white background with a slightly off white line for the tracing line. Its near impossible to tell the difference between them on my screen, which causes significant strain however I now have all my data I believe outside Baffle/Enclosure information in SW. What do I start with?
WRT the screen shot rigmarole, that's the way I used to do it.
WRT the bump, open your zma file in notepad. Find the frequency in the first column and alter the value in the second column.
WRT where to go from here, import your zma files into SW, add them to your driver objects, create networks for your drivers, create target curves for your networks and add them to the network response charts.
WRT the bump, open your zma file in notepad. Find the frequency in the first column and alter the value in the second column.
WRT where to go from here, import your zma files into SW, add them to your driver objects, create networks for your drivers, create target curves for your networks and add them to the network response charts.
Networks I assume you mean the crossover relays? Thanks for the tip about the text editing. All fixed and pretty 🙂. How would I begin to set target curves?
Network is another word for circuit. For now, create one per driver. When you have a source correctly connected to a driver through some components, you'll be able to activate the 'calculate response' option from the network menu which will create a chart for each driver in the network, and one as a total which will be the same if you only have one driver in the network.
In the network menu (which appears when you highlight a network window), choose create goal. This creates a chart with a target slope. Add that chart to the network response plot so you can compare them as you tweak.
In the network menu (which appears when you highlight a network window), choose create goal. This creates a chart with a target slope. Add that chart to the network response plot so you can compare them as you tweak.
- Status
- Not open for further replies.