The sub I've been using for quiet music I got cheap at Goodwill. It had an amp built in, but a proprietary wireless system, and I didn't know how to bypass it and hard wire the input, so I took it out and ended up replacing it with a huge 3rd order filter I calculated online and assembled, before I knew much about doing things the right way, to force a better response curve that extends much lower than the sub was designed for. It doesn't sound great, but it's not bad, it's just quiet, since most of the work it's been doing is below the tuning frequency, but it's sufficient when there's minimal other noise to compete with.
Anyway, this time around I have Vituix, and I was messing around to see if it's even possible to make a passive filter that can modify the curve without being so affected by the wild impedance curve of a ported sub. I arrived at this solution pictured below, which appears to force a steady impedance curve.
My question is: is this just plain stupid, or will it work? I don't yet have enough capacitors to try it. It looks like a big waste of power, so it may not be that good of an option for a portable speaker, but it should allow for much louder spl by compensating the response curve of an compact undersized low tuned sub.
Maybe there are more practical solutions, but with my limited knowledge I can't see why this isn't a reasonable choice among others. It's just a concept now, and I'll probably try it out if you don't think it's not going to work.
I was able to make a variety of curves, but this one looked clean while keeping the impedance form getting too low. Some would go low even when I didn't expect it to, even after putting an 8 ohm resistor after the capacitor (now between the two), which is how I got started on that path.
The original impedance curve came from a simulated enclosure tuned to 28hz
Anyway, this time around I have Vituix, and I was messing around to see if it's even possible to make a passive filter that can modify the curve without being so affected by the wild impedance curve of a ported sub. I arrived at this solution pictured below, which appears to force a steady impedance curve.
My question is: is this just plain stupid, or will it work? I don't yet have enough capacitors to try it. It looks like a big waste of power, so it may not be that good of an option for a portable speaker, but it should allow for much louder spl by compensating the response curve of an compact undersized low tuned sub.
Maybe there are more practical solutions, but with my limited knowledge I can't see why this isn't a reasonable choice among others. It's just a concept now, and I'll probably try it out if you don't think it's not going to work.
I was able to make a variety of curves, but this one looked clean while keeping the impedance form getting too low. Some would go low even when I didn't expect it to, even after putting an 8 ohm resistor after the capacitor (now between the two), which is how I got started on that path.
The original impedance curve came from a simulated enclosure tuned to 28hz
2mF and 5 mF capacitors are in series - they are equivalent of one 1.43 mF capacitor. 1 ohm resistor parallel to driver is mistake - it drops impedance level.
How about line-level low-pass RC filter? Much cheaper, better amplifier load too - just the driver.
How about line-level low-pass RC filter? Much cheaper, better amplifier load too - just the driver.
Thanks, I see that now. The 1 ohm resistor made a significant change in the right direction, but that's a lot more wasted power than I was thinking. I thought the resistor between the 2 capacitors changes how they respond, at least that's how it seemed when experimenting, but yes, removing one and changing the other to 1.43 does the same.
Yes, I've thought about a line level filter. Maybe my reason for shying away from it was how it may respond differently to different sources and amps. Anyway, this was an attempt to find a viable solution after the amp, something to store inside the speaker box.
Yes, I've thought about a line level filter. Maybe my reason for shying away from it was how it may respond differently to different sources and amps. Anyway, this was an attempt to find a viable solution after the amp, something to store inside the speaker box.
The impedance curve now seems of relatively little use with how this circuit seems to force a new impedance curve similar to when using the default flat "curve". but here it is.
You did not describe how the complete system is set up. Since this is a sub, how do you sum the left and right channel? I assume the sub does not have its own amplifier. Is it a mono system?
Anyway, you should filter the signal to the sub in some way. Too high frequencies cause cone break-up or distortion. So that was a good decision.
However, a 1 ohm series resistor and the 20 mH inductor form a terrible damping factor, which ruins any box design effort. If you insist of staying passive, I think you should think of an inductor only to maintain at least some damping factor.
Summing and filtering at line level and feeding the amplifier from its own amplifier is preferable by far. For the price of a single inductor you'd have a class D amplifier which matches your power needs. Class D is fine for a sub, most distortion is imperceptible.
Since the speaker originally was active it already has a power supply. That is one reason more to look for an active sub.
Anyway, you should filter the signal to the sub in some way. Too high frequencies cause cone break-up or distortion. So that was a good decision.
However, a 1 ohm series resistor and the 20 mH inductor form a terrible damping factor, which ruins any box design effort. If you insist of staying passive, I think you should think of an inductor only to maintain at least some damping factor.
Summing and filtering at line level and feeding the amplifier from its own amplifier is preferable by far. For the price of a single inductor you'd have a class D amplifier which matches your power needs. Class D is fine for a sub, most distortion is imperceptible.
Since the speaker originally was active it already has a power supply. That is one reason more to look for an active sub.
Thanks for your input.
So far, the filter is just a proof of concept attempt to see if I could overcome the impedance peaks of a bass reflex to force whatever response curve I may want, because I've had trouble figuring this out in the past, probably because I was using crossover calculators and just typing in the resistance rating of the driver and getting bad results. I haven't quite figured out a high pass filter that works like this yet.
I know nothing about damping factor, so it looks like I have more research to do. The series resistor was my attempt to keep the amp from seeing too low of impedance.
So far, the filter is just a proof of concept attempt to see if I could overcome the impedance peaks of a bass reflex to force whatever response curve I may want, because I've had trouble figuring this out in the past, probably because I was using crossover calculators and just typing in the resistance rating of the driver and getting bad results. I haven't quite figured out a high pass filter that works like this yet.
I know nothing about damping factor, so it looks like I have more research to do. The series resistor was my attempt to keep the amp from seeing too low of impedance.
Last edited:
The .zma data seems to have some issues with it. For example, at 3000.72 Hz the impedance is 1.283 ohms. It continues dropping until at 12kHz it's only 0.001 ohms. Something seems to be quite wrong, as the impedance should be rising due to the voice-coil inductance. Was that a measurement of just the raw driver in a vented-box enclosure?
At low frequencies, between 5Hz and 100Hz, I can identify the twin impedance peaks and the impedance minimum. Their values seem to be quite reasonable.
Output impedance from source always is low, so it will not interfere with the line-level RC filter. Input impedance of the amplifier will have effect on the frequency response of the RC filter, but it is very easy and very cheap to make, say, three different RC filters for three different amp input impedances (the most common are 47K, 22K and 10K), selected by 3 position 2 pole switch.
That impedance curve is very wrong! It is not real!
Nevertheless, here is passive LC filter loaded with that impedance curve:
Schematic:
Frequency response:
Unlike line-level RC filter, frequency passband of this LC filter is not flat, because of driver impedance peaks.
Impedance:
Last edited:
Good idea
I found I had an other trace for a lower tuned sub to experiment with. I'm getting the hang of designing what I need, so it's not nearly the challenge I thought it was, just a simple low pass filter with resistors to help fine tune the slope angle: one parallel to the inductor, and one in series with the capacitor.
It was a short band trace from WinISD, and Vituix estimated the high end wrong.
I found I had an other trace for a lower tuned sub to experiment with. I'm getting the hang of designing what I need, so it's not nearly the challenge I thought it was, just a simple low pass filter with resistors to help fine tune the slope angle: one parallel to the inductor, and one in series with the capacitor.
Last edited:
No. I don't have the tools to take measurements. I do realize there are inaccuracies in simulations vs the real world. I did do a real world test, though, and found a second order filter worked well for what I used to think only a 3rd order filter could do well enough, based on my past experience of some real tests and simulation confirmations.
Yes, I know that, and those peaks are part of what I was trying to overcome. I think in real life the peaks are not as high anyway from what I've read from other's experiences. Anyway, I seem to be managing the peaks better than in the past.
Last edited:
I looked up the datasheet for the Dayton Audio DCS165-4 6-1/2" Classic Subwoofer 4 Ohm, and the free-air impedance curve is shown below. The vented-box impedance curve for this driver would be expected to have a similar impedance curve above 150Hz or so.
The measured data in the .zma file produces the following plot. The impedance curve above 100Hz looks quite wrong.
The measured data in the .zma file produces the following plot. The impedance curve above 100Hz looks quite wrong.
That's what happens when I give Vituix a graph that goes only up to 200hz to trace. It makes up the rest above 200hz, but I didn't care about what happens that high.
That's okay. It's possible to get the free-air impedance .ZMA file for the Dayton Audio DCS165-4 driver from the Dayton Audio web site. VituixCAD can be used to simulate the vented-box enclosure response using the Thiele–Small parameters in the datasheet. For a passive low-pass filter to work properly, you will need to use impedance data that is as representative as possible.
I've taken the .ZMA file for the Dayton Audio DCS165-4 driver and used VituixCAD to compute the "Extended Z model" parameters. These provide a reasonable match to the measured data. If you use these values, then VituixCAD's filter simulations will be more accurate for your purposes. Note that you will have to export the .ZMA file for the vented-box enclosure tuning that you've chosen, which I think you may have previously done.
The VituixCAD "Edit parameters" Thiele–Small and impedance parameters data entry screen that I created is shown below.
The region between 100Hz and 1kHz is quite important when trying to design a low-pass filter for the driver. There can be some interactions between the complex impedance and the filter component values that need to be taken into account.
Peaks may be high - depends on driver parameters. Nothing wrong with that. You don't have to "manage" the peaks - accept them as they are.
It is not clear to me - did you enter Thiele-Small parameter in WinISD (or VituixCad) to use simulated impedance curve, or did you trace graphically some measured driver impedance?
If you didn't enter Le in the Thiele-Small parameters, than you will get wrong .zma impedance which is going to zero in the high frequencies - just as in your post #4.
Frequencies above 200 Hz are important!