No, that's excellent. Work on the sim with the files you've got and then work on it afterwards with my files that will include more information. All the work you put in is a helpful learning experience (even when you get it wrong and sometimes especially when you get it wrong) and the differences between the 2 sims should be instructive as well.
I simply worked on the files you attached. I assumed you didn't take actual measurements, which would automatically add the Z-offset. I therefore added a Z-offset, 25mm I think but don't remember exactly. Since there was no diffraction response, I took a 6.5" baffle width target and created an approximation.
What you would typically want to do in the absence of zero measurements is add the Z-offset when you have the drivers in hand. And once you have the boxes, set a baffle loss target frequency and roll the driver off to match both that frequency and your crossover target in one smooth slope.
I measured the Z offset to approximately 24mm so 25 is in the ballpark.
I am beginning to think I am thick around the ears for asking so many questions.
Also jReave, you did not include your files or will you do that later. No rush mind you, it's just that I did not see them.
Don
From the article in True Audio the baffle loss would be around 650Hz. Does that sound correct? How do I add the Z offset to Xsim?
I am beginning to think I am thick around the ears for asking so many questions.
Also jReave, you did not include your files or will you do that later. No rush mind you, it's just that I did not see them.
Don
I should really avoid posting when I should be sleeping instead.
I had designed on woofer axis, so no Z-offset was added. This way the rising top end of the tweeter would be tamed.
Depending on your placement and listening axis, the Z-offset might be different from the one you measured, which assumes dead flat tweeter listening axis. This is not always practical or possible because the tweeter is just a decimal point at most normal listening distances, so don't worry about it being a bit off.
Also if they're ceiling mounted a lot of all of this is moot anyway. Most of the design best practices are based on guesstimates around typical home hifi listening, which don't apply to a large number of applications (such as yours). Also the axial response doesn't always provide a true picture of the overall performance of the speaker, so don't worry about a few slips here and there.
These will be but, I am thinking ahead. If I can get a good understanding of how everything integrates together and I can manage to do a decent job with what I have, I might consider something a bit bigger in the future for pure music enjoyment. I have a few spare amps and av receivers as well as more speakers that I can count including Altec's A7 The Voice of the Theater, a few B&W's, some old Radio Shack Machs 1 as well as some home made speakers and subs that are someone else's design.
Don
Don, you are a beginner so in my mind every question is valid.
I've gone over your design and here are some comments:
- with 1/2" wood, I get about 4.9L gross for your cabinet. A 1.75" diameter port tuned to about 65Hz needs to be about 6.25"-8" long. Depends on net volume and the amount of stuffing.
- with the drivers centered on the baffle, I get the best on-axis baffle diffraction response with the tweeter at about 9" from the bottom. With the high xo point the mid and tweeter should be pretty close together so I put the mid 5" from the bottom. But maybe the speaker looks better to you with the drivers a little lower, like with the tweeter at about 8.25" from the bottom. Diffraction is a bit worse with a little less output between about 3000Hz -5000Hz. Offsetting the tweeter about 3/4" from the center line is an option too but some people don't like the look of that either.
-if the speakers are going right up against a wall, the baffle diffraction sim and therefore the xo sim should be revisited.
- I've attached a new XSim file below with the new driver files included. The driver data now includes baffle effects as noted, the mid/woofer frd and zma files include the effects of the vented box, phase has been "derived" for all the files (it's worth looking at what appropriate tailing is for the drivers) and I set the mid/woofer delay to .6".
- just to clarify: the acoustic center is what you measure on each driver. The z off-set is the difference between the acoustic centers of the 2 drivers based on how you mount them on the baffle. And the XSim mod delay is just slightly different again as it is actually the difference in pathlengths between a measuring/listening position and the drivers' acoustic centers which means that it will also take into account the difference in the drivers' vertical and horizontal baffle positions (if applicable). It comes down to just some basic trigonometry, but as I said, don't worry too much about that last one for now.
- I've also attached a pic below which shows a rough look at how to deal with the 6dB baffle step loss. Black is what your woofer looks like with the baffle effects included. Blue is driver S2 with a traditional baffle step compensation filter. And green is driver S3 compensating for the baffle loss by just using the single inductor but with a higher value. The curve gets flattened out just like with the S2 filter but it also starts rolling off higher up which is what you want in the xo region anyways. Two birds with 1 stone kind of thing.
- When you open XSim, you'll also find 2 extra drivers labelled as the HP and LP filter targets. These are really useful for beginners as they give you something to try to match the drivers' FR's to instead of just going blind. I chose 4th order (24dB/octave) filters at 3000Hz because that tweeter really shouldn't be played too low. So as low as is we probably can and with steep acoustic slopes.
- Here is a quick guide to starting out in XSim:
- In the Frequency Response graph, turn off system phase
- choose to show the phase for each driver.
- I find it easier to read the FR graphs if the response colors are always the same - black for System, red for the tweeter and blue for the woofer. Just my preferences though.
- set your "minimum frequency" down to 20Hz and set your "vertical center" so that you can see enough of what the drivers are doing 40 or 50dB below the fundamental.
- for the target filters, select S4 and S5 (driver only). I find they work well in grey.
- and here is the key: you want to set the target filter at about the same level as the woofer at about 200Hz. Or in XSim with this method, you are actually setting the woofer level equal to the target SPL at about 200Hz. Do that by increasing the amp output until they match, in this case at about 10W.
- then start work on the woofer and once that matches well, start the work on the tweeter.
- for the purposes of learning, I have set it up with a typical 2nd order electric on the woofer and 3rd order electric on the tweeter with an L-pad to adjust the tweeter level but all the values are simply what the program sets when you add each new component. There is also 1 additional resistor, R1, before the filter which you may or may not want to use. You will find that where a resistor is placed in the circuit will slightly change the effects, so sometimes 1 is a better choice than the other or sometimes both work well together.
- I've also included 3 components that are not connected. With C4, match the woofer FR to the target 1st and then make the connection. It will create a deep notch somewhere. Adjust until it hits that nasty peak up around 10,000Hz.
- L3 and C5 are perhaps optional. They are an impedance compensation filter meant to hit the tweeter response at about 1100Hz, to really kill any chance of its resonance frequency being any kind of a problem. Up to you if you want to use it. Getting these 2 values right I found rather difficult as a beginner so don't worry too much if it proves a tough nut to crack.
Lastly, there are 3 primary and simultaneous goals when you do xo work:
- a decent summed frequency response
- get the phase of each driver to match up as best you can within the region of the xo
- make sure the impedance response doesn't dip too low for your amplifier. Peaks on the other hand are generally acceptable for most of todays solid state amplifiers.
I'm sure I've probably left something out but that should get you going for now.
I've gone over your design and here are some comments:
- with 1/2" wood, I get about 4.9L gross for your cabinet. A 1.75" diameter port tuned to about 65Hz needs to be about 6.25"-8" long. Depends on net volume and the amount of stuffing.
- with the drivers centered on the baffle, I get the best on-axis baffle diffraction response with the tweeter at about 9" from the bottom. With the high xo point the mid and tweeter should be pretty close together so I put the mid 5" from the bottom. But maybe the speaker looks better to you with the drivers a little lower, like with the tweeter at about 8.25" from the bottom. Diffraction is a bit worse with a little less output between about 3000Hz -5000Hz. Offsetting the tweeter about 3/4" from the center line is an option too but some people don't like the look of that either.
-if the speakers are going right up against a wall, the baffle diffraction sim and therefore the xo sim should be revisited.
- I've attached a new XSim file below with the new driver files included. The driver data now includes baffle effects as noted, the mid/woofer frd and zma files include the effects of the vented box, phase has been "derived" for all the files (it's worth looking at what appropriate tailing is for the drivers) and I set the mid/woofer delay to .6".
- just to clarify: the acoustic center is what you measure on each driver. The z off-set is the difference between the acoustic centers of the 2 drivers based on how you mount them on the baffle. And the XSim mod delay is just slightly different again as it is actually the difference in pathlengths between a measuring/listening position and the drivers' acoustic centers which means that it will also take into account the difference in the drivers' vertical and horizontal baffle positions (if applicable). It comes down to just some basic trigonometry, but as I said, don't worry too much about that last one for now.
- I've also attached a pic below which shows a rough look at how to deal with the 6dB baffle step loss. Black is what your woofer looks like with the baffle effects included. Blue is driver S2 with a traditional baffle step compensation filter. And green is driver S3 compensating for the baffle loss by just using the single inductor but with a higher value. The curve gets flattened out just like with the S2 filter but it also starts rolling off higher up which is what you want in the xo region anyways. Two birds with 1 stone kind of thing.
- When you open XSim, you'll also find 2 extra drivers labelled as the HP and LP filter targets. These are really useful for beginners as they give you something to try to match the drivers' FR's to instead of just going blind. I chose 4th order (24dB/octave) filters at 3000Hz because that tweeter really shouldn't be played too low. So as low as is we probably can and with steep acoustic slopes.
- Here is a quick guide to starting out in XSim:
- In the Frequency Response graph, turn off system phase
- choose to show the phase for each driver.
- I find it easier to read the FR graphs if the response colors are always the same - black for System, red for the tweeter and blue for the woofer. Just my preferences though.
- set your "minimum frequency" down to 20Hz and set your "vertical center" so that you can see enough of what the drivers are doing 40 or 50dB below the fundamental.
- for the target filters, select S4 and S5 (driver only). I find they work well in grey.
- and here is the key: you want to set the target filter at about the same level as the woofer at about 200Hz. Or in XSim with this method, you are actually setting the woofer level equal to the target SPL at about 200Hz. Do that by increasing the amp output until they match, in this case at about 10W.
- then start work on the woofer and once that matches well, start the work on the tweeter.
- for the purposes of learning, I have set it up with a typical 2nd order electric on the woofer and 3rd order electric on the tweeter with an L-pad to adjust the tweeter level but all the values are simply what the program sets when you add each new component. There is also 1 additional resistor, R1, before the filter which you may or may not want to use. You will find that where a resistor is placed in the circuit will slightly change the effects, so sometimes 1 is a better choice than the other or sometimes both work well together.
- I've also included 3 components that are not connected. With C4, match the woofer FR to the target 1st and then make the connection. It will create a deep notch somewhere. Adjust until it hits that nasty peak up around 10,000Hz.
- L3 and C5 are perhaps optional. They are an impedance compensation filter meant to hit the tweeter response at about 1100Hz, to really kill any chance of its resonance frequency being any kind of a problem. Up to you if you want to use it. Getting these 2 values right I found rather difficult as a beginner so don't worry too much if it proves a tough nut to crack.
Lastly, there are 3 primary and simultaneous goals when you do xo work:
- a decent summed frequency response
- get the phase of each driver to match up as best you can within the region of the xo
- make sure the impedance response doesn't dip too low for your amplifier. Peaks on the other hand are generally acceptable for most of todays solid state amplifiers.
I'm sure I've probably left something out but that should get you going for now.
Attachments
The point is, as jReave stated, the actual position of the driver wrt the listening position. For example if you mount a regular MT speaker on the ceiling firing downward, the path length to the woofer actually increases.
For any two way not using a coaxial driver, the positions on the baffle in 2D geometry are never usually the correct goalposts. A speaker mounted high on a wall and pointing downwards is pretty much a worst case scenario for any crossover designer, as any speaker designed for that will likely not work well anywhere else, and a speaker designed with a standard listening geometry will never sound good in a situation like that (unless it had a coax driver, in which case it might work).
The reason for this is that in order to get the top-end response linear the speaker has to be angled down, and the woofer now goes much further back than it actually sits on the baffle. This increases Z-offset, and from some of the pictures I've seen of such setups, the amount is usually in hundreds of millimeters. You cannot have a spaced driver setup work properly across such a wide range of listening angles.
I went to bed in the wee hours of the morning finishing the read on Wintermute's thread, yours and jRead's explanation and it all kicked in. I finally understand what all this means and I now get your point about the woofer being further back. Now, I will practice that newfound knowledge on the published specks of other speakers and manufactures till I get my crossovers which, won't be for another couple of weeks. Two more questions before I leave:1- In Wintermutes thread on baffle edge diffraction, the woofer is measure at 3' while the tweeter is measure at 8'. Is this the usual way o do it?
2-Does all what we have been discussing apply to sealed box as well? I know the boxes wont have the same volume, xover point and value but are they treated the same way in the calculations?
And with that in mind, I want to thank everyone that helped an old geezer in understanding some of the complexities in the joy of DIY speaker building.
1 - Don't worry about that. It's about how drivers are measured to produce the spec sheet responses. The distance and timing of the measurements varies with frequency is more or less what it is saying. Many driver spec sheet measurements are using methods that get rid of baffle diffraction effects anyways so subtracting any out would be a mistake. For other drivers with possible baffle effects included, the effects are so small it's debatable if it's worth worrying about them.
2 - "all that we have been discussing" is fairly large but if you're asking what I think you are, the answer is yes, same for sealed boxes.
Two more tips for xo design:
1. Always try to get the first series component to do the most work in the xo filter. In a 3rd order electrical on a tweeter then for example, the 1st series capacitor would always be smaller than the 2nd capacitor. And for series inductors, the 1st would should always be bigger than the 2nd.
2. Also be careful with all your parallel lines to ground, also referred to as shunts. Because... a shunt to ground with no resistance is simply a short which is not exactly a good thing. So for capacitors, you need to be cautious with very large values placed in parallel because they become more and more conductive (so less and less resistive) the larger the value. Conversely, you need to be cautious with very small inductor values in parallel because they become more and more conductive (less and less resistive) as they get smaller. The caveat here is that things are complicated by what the value of the series component(s) is in front of the shunt. Or in other words, it's the combination of the series and the parallel component values that determine the amount of current running down the parallel line to ground. And additionally, it can depend on if there are other components in series in the shunt too and what their values are.
2 - "all that we have been discussing" is fairly large but if you're asking what I think you are, the answer is yes, same for sealed boxes.
Two more tips for xo design:
1. Always try to get the first series component to do the most work in the xo filter. In a 3rd order electrical on a tweeter then for example, the 1st series capacitor would always be smaller than the 2nd capacitor. And for series inductors, the 1st would should always be bigger than the 2nd.
2. Also be careful with all your parallel lines to ground, also referred to as shunts. Because... a shunt to ground with no resistance is simply a short which is not exactly a good thing. So for capacitors, you need to be cautious with very large values placed in parallel because they become more and more conductive (so less and less resistive) the larger the value. Conversely, you need to be cautious with very small inductor values in parallel because they become more and more conductive (less and less resistive) as they get smaller. The caveat here is that things are complicated by what the value of the series component(s) is in front of the shunt. Or in other words, it's the combination of the series and the parallel component values that determine the amount of current running down the parallel line to ground. And additionally, it can depend on if there are other components in series in the shunt too and what their values are.
That's a slip up on my part! Not sure how I managed to swap from 3' to 8' when doing the sims. They should both have been done at the same distance. I suspect that the difference is negligible, but I should correct it.
I should also do the next follow up posts. Ones involving real measurements and a real crossover design! Maybe I should make a new years resolution
Tony.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.