Site Links
Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Quick Links
Categories
In this Discussion
Please review the site Rules, Terms of Service, and Privacy Policy at your convenience. Rules, TOS, Privacy
Get familiar with the reaction system: Introducing the Reaction System
Modeling woofer cabinets using DSP . . . .
I am asking how you model a speaker/cabinet for a woofer's T/S parameters for use with DSP while seeking the smallest cabinet volume.
Using WinISD and its Filter functions and cone excursion/power applied gets me part of the way there but I am missing something.
Any guidance?
Thanks.
Comments
What feels wrong about what you're doing right now? Got any screen shots of what you're setting up?
The best thing to do with WinISD is play around with stuff. Create two sealed enclosures. One at a Qts of .6, and another with a Qts of .8. See how each responds to different parametric EQ boosts, with different Q values. Add power until each are at Xmax, then dip on over to the SPL graph and see which one outperforms.
Then, without closing those first two, create a ported enclosure or two for the same driver and see how they respond to different boosts or cuts.
Then try a hybrid enclosure, with a PR tuned really low. Does it respond differently to boost?
The nice thing about DSP is that it is unbelievably easy to build a thing, dial in all your proposed filters, and then on a whim throw them all out the window and try something else. You just reload the old filters.
Hey Dirk,
That's the thing about WinISD and other modeling software is that I don't see how I can add boost or shelving in those models because they assume passive filters/XO's from my experience & limited understanding.
To use a smaller box volume [than recommended by whomever] with any given woofer yields either a higher or lower Q depending upon the driver - but I can do stuff with DSP that I can neither change quickly with passive or box volume to see the acoustic effects I have applied.
Right now I am planning on using the Dayton PA310-8 as a woofer from 50~200Hz in a ported enclosure crossed to a sealed 12" sub for the bottom octave and am looking for clean/articulate slap which is missing from my current setup using the Eminence Beta10CX crossed to a sub @ 95Hz. [crossing the BetaCX10 lower introduces audible distortion/muddiness in its sealed 0.56 cu ft of enclosure].
I just don't want to build a box I cannot lift and manage nor have to build another one.
This is what my WinISD models look like right now using a 12" wide by 1.25" tall by 5.42 " deep slot port:
So what happens to the Q /tuning if I try to push the FR of the woofer lower with DSP in that box or increase the output SPL at lower/higher frequencies while keeping the excursion under Xmax?
At some point [I think] that I need a different box volume than WinISD indicates for best performance of the driver.
Maybe I am overthinking it as usual but thanks for chiming-in [gives me a sounding board for sanity check].
I probably need to just pull these PA310-8's out and break them in at "published" Fs for a week, measure them with DATS and ask this question again with that data in hand [afterthought considering the Eminence Beta10CX's were published as being Fs @ 49Hz and wound up being Fs 95Hz after week of break-in]. Cloth surrounds need a lot of exercise to get softened-up/normalized in my lengthy experience with them].
I don't know if this will help but I'd play with different boxes and tunings and plot excursion. You're going to be up against power handling and excursion limits with what ever you design. DSP is going to allow you to boost the signal by X db. This graph will show you how many watts you're going to add for a given boost. If you can make a plot that uses X watts in without going over the limits you should be OK. Concentrate on the low end and ignore the rest of the response. You can always fix it with DSP. If you find you're running into excursion problems below the port tuning you can use a filter to cut the low frequencies by X db. I have not tried to do this myself but I think it should work. If it doesn't just ignore me.
Ron
I thought I'd add an example. This is the woofer sim for my last project which was an 8" woofer in 1.2 cu. ft. Maximum power for the woofer is 150W. If I look at the db scale and subtract 3db from 150 watts I get 75W and I can plot the graph and compare the excursions. I can keep decreasing the power in until it won't excursion limit and reduce the input X db below 27hz.
Ron
RonE, Thanks for the info/contribution - I will absorb this shortly and see where it leads me.
Another thought [way of describing it] is that there is a limit that the driver performance in both fixed enclosure/port dimensions that can work within a FR range/SPL/power/excursion for a given driver using DSP [or even passives for that matter].
I think this is what I am looking for - determining that range for the suggested PA310-8 Woofer.
Freq range is Fs = 39 Hz +10% = 43 Hz to around 200 Hz low-pass.
I guess what Dirk was saying is to just experiment with that range and get the woofer to peak at a desired "Q" within the Freq range needed/desired and then DSP it to perfection by ear/measurements.
WinISD has a "Maximum SPL" graph. I believe it takes the RMS wattage of the driver and plots it against the excursion graph for the modeled enclosure. Any db below the line is theoretically possible with that driver in the modeled enclosure.
You can do a lot with WinISD too. There are several filters available including a Linkwitz Transform (LT).
Here's an example of using a LT. The Transfer Magnitude plot shows a driver in a sealed box (blue line) and the same driver with a LT (red line). I arbitrarily chose the LT parameters to extend the F3 from 150 Hz to down to 60 Hz. The SPL plot shows their output levels when I set the signal level to 10 watts. The Cone Excursion plot shows the result - the LT with 10 watts exceeds the 5 mm Xmax. And importantly, the Amplifier Apparent Load VA plot shows the added demand on the amplifier by adding the LT.
WinISD is different than other modeling tools like VituixCAD or Xsim, in that it assumes active. More specifically, it assumes you're altering the input signal to the amplifier and gives you a list of filters to choose from as likely modifications to the amplifier's input signal. If you place a 2nd order 80hz highpass on a woofer, you will see that the impedance graph remains unchanged and the transfer function follows the ideal -12dB/oct roll-off. If this were simulating a passive filter, we would not see that perfect -12dB roll-off as the filter interacts with (and is made ineffective by) the woofer's impedance peak(s) near the system Fb.
One thing I don't like about WinISD is that if you add +3dB of boost at a given frequency and then bump up the power on the "Signal" tab to 50W, you will actually be pushing 100W at your boosted frequency. It's not immediately intuitive, and once you remember that's how it behaves, it's fine. And to be honest I don't know how I'd design the interface to behave differently. But you always need to keep that in mind if you're wondering whether your subwoofer plate amp amp has enough guts for what you've simulated.
Unless you're specifically trying to do something crazy like @Ed_Perkins is doing with the Linkwitz Transform, design your box to be "not half bad" without DSP. Then use DSP to bring it the rest of the way home. It's easy to navel gaze and worry about getting a line that's as perfectly flat as possible, but all that is going to go to hell once you put it in a room anyway.
For that functionality it would seem necessary to add an additional graph to plot the power applied across the frequency range. The power number defined in the signal tab would likely become the default baseline on the graph to be altered by the added filters.
They do have an "Amplifier apparent load power (VA)" graph. So the data is there, if you know to look for it.
If I'm trying something weird with WinISD, I usually start with an "unaltered" model as my baseline, driven to system max (whether that's input power or Xmax). Then I copy that model, and start fiddling with EQ and other filters on that copy. If my 2nd design ever crosses the baseline model on the SPL graph, I know I'm doing something illegal.
Guys, THANKS.
The thing I was missing/had blinders-on were all the other filters in the drop-down list [duh!].
I now see that I can use a 1.0 cu ft box with a longer port and reduce group delay and not break Xmax until 25 Hz @ 20W which can be rolled-off easily with PEQ.
Whole new ballgame now and less material/effort required.
Thanks again.
Since you won't be really using it all the way down to the port tune frequency, can probably get away with a smaller dia port too.
I'll play with that next - thanks for the nudge.
I'm settled on 1.3 cu ft now . . .
Dirk, This ^ stuff is golden insight, Sir.
Thanks - my perspective of WinISD has changed significantly - like a light came on in a dark room.
I'm a little late to this thread, but I would like to mention the Parametric EQ filters in WinISD. You can set the center frequency, Q factor, and a positive or negative gain. You can have multiple EQs also. The Transfer function magnitude (EQ/Filter) graph shows the sum of the filter profiles.
This is a very powerful feature that I only recently discovered to help model some "Inner Sanctum" designs for InDIYana 2024.
The Amplifier apparent load power graph is a great suggestion also, I had overlooked that one before today.
Yeah, you have to remember that the entered signal value is the nominal power, i.e. without any gain or boost. To see the actual power required you have select the "Amplifier apparent load power (VA)" plot. You get used to it after awhile.
After more experimentation I've decided to go with a 1 cu ft sealed box since I am crossing to a sub for the very bottom octave. It presents the cleanest roll-off and the lowest Group delay while being able to keep up with the 10" midrange without breaking Xmax with up to 10W applied at an SPL I will never use . . .
Time to make sawdust . . .
.
Amp Power and PEQ settings for those interested:
I do not fear looking stupid by asking stupid questions because I realize just how many people too afraid to ask, benefit from my lack of fear.
The truly experienced/helpful people on this planet will help you if you are sufficiently articulate on any given subject to make it worth their effort to respond.