How to Properly Design a Subsonic Filter

I’ve got an old XEQ-2 active crossover made by EV that I’ve been using for the crossover on my subwoofer for my home theater. The one thing I always disliked about this EQ was the 30-Hz built-in high pass filter. Worse than that, the options for adjusting this filter don’t offer any solutions. Each click of the switch adds +6 dB boost at varying frequencies from 29Hz to 60Hz. Definitely not optimized for HT and certainly not for a subwoofer that is, by design, flat to 20 Hz. What I really wanted was to do was adjust the filter for a predetermined cutoff frequency that corresponded exactly to the maximum excursion capability of my new Audiopulse vented subwoofer. And this is how I did it.

First off, why put a high pass filter on a subwoofer at all? Commonly referred to as a rumble filter or subsonic filter, the idea is to remove frequencies below that which you can’t hear or those frequencies below where your sub can adequately produce, in an effort to reduce driver excursion and reduce stresses put on your amplifier. Any vented subwoofer will essentially unload below tuning which can cause the driver to exceed its rated xmax (or even xmech) causing potential damage to the driver. However, the problem with most rumble/subsonic filters is the cutoff frequency is arbitrarily chosen and usually does not represent the ideal cutoff frequency for the subwoofer to which it is connected. It is either too high, in which case it negatively impacts the natural frequency response of the driver/enclosure, or it’s too low and still allows the driver to exceed its rated excursion capability. An ideal subsonic filter is achieved by setting the cutoff frequency to be low enough to have a negligible impact on the acoustic response of the driver/enclosure while still maintaining just the right amount of safety margin below tuning for the driver so that it does not exceed xmax.

The EV XEQ-2 is the perfect crossover for achieving this. The first thing I needed to find was the service manual for this unit. I found it in an unlikely place – someone’s eBay auction that had ended months ago. They had a link to both the manual and the service manual. The service manual has the schematic for the unit which is key into figuring out how to adjust the high pass circuitry. This is what I learned while looking over the schematic. The filter for the low pass section is made up of a single-chip dual op-amp. The one op-amp is for the 18 dB/octave low pass section and the other is for the high pass section. This is how the high pass filter and boost work according to the service manual:

“U2 (1) is an active high pass filter. S2, the LF EQ switch, controls the cutoff frequency. In the FLAT position, U2 (1) becomes a second order (12 dB/oct.) high pass filter [which is down -3dB at 30Hz]. At other positions, U2 (1) becomes an underdamped high pass filter. In an underdamped second order filter, a peak occurs in the frequency response just before the curve breaks away from a flat response. Once past the peak frequency, the curve falls at 12 dB/octave. The peak frequency is noted in the swtich markings on the front panel. The peak amplitude is approximately 6 dB.”

This is the basic function for many bass boost circuits found in a lot of plate amps. It’s just an underdamped, i.e., high-Q, high pass filter which creates a peak is the response at a calculated frequency. The calculation for determining the f3 (or -3dB point) of an active 12dB/octave filter such as this is represented by the formula: f3 = 1/((2*pi*(R1*R2*C1*C2)^0.5). Damping is represented by Q and is calculated by the formula: Q = 1/2*(R1/R2)^0.5 where C1=C2. Since am trying to achieve a maximally flat response, I will shoot for a Q = .707, which results in a very simple resistor calculation represented by R2=R1*2.

For my application I wanted to reduce the FLAT response so that the corner of the filter is somewhere below 20Hz, or the same frequency that corresponds to the point where by sub begins to reach xmax. To determine the best -3dB point for my rumble filter, I checked out the excursion plots from Unibox for my sub design. Excursion doesn’t become a problem at 500W input until we go below 18 Hz. So that is about the point that I need the corner of this filter to be set to.

Though the formulas are usually enough for determining the necessary values, I ran across a really cool simulation tool that let’s you build a simple schematic and then run an AC analysis to see just how your circuit will look in the real world. It’s called National Instruments Multisim Analog Devices Edition. I found this tool while checking out some op-amps from Analog’s Devices web site. I won’t go into a lot of detail on the capabilities of the free version of this software, other than I found it works very well for modeling and simulating active filters, among many other things. I drew up the schematic, following the schematic in the service manual, of just the low and high pass portion of the entire EQ. This version is limited to 25 components, so I was unable to add the input unity gain buffer circuitry since that pushed it over the 25 components. For the simulation though it doesn’t really need it since it assumes the AC voltage source is a already a low impedance source. If you were to want to build this circuit, I would recommend adding the input buffer so the high output impedance of various receivers does not inadvertently alter your low pass cutoff frequency. A full schematic is shown here as well. I set up the low pass for around 80Hz with resistor values that were calculated using the formulas as defined in the owner’s manual for the XEQ-2.

UPDATE: Now you can create simple circuit simulations using MultismLive. It’s a web-based interface that requires no downloads, has no part count limitation and works great for simulating a wide variety of active filters. Plus you can export raw .csv data of the response which can be imported into Unibox so you can model the combined driver and excursion responses with your filter. I’ve simulated this circuit here, so go check it out and while you’re there, go and create some cool circuits of your own!

Multisim Live: EV XEQ-2 Filter

I entered the values for FLAT which is a 36k and 75k ohm resistor and ran an AC simulation from 1Hz to 1kHz. The results show up in a graph in a new window. The simulation appeared very accurate as the -3dB point was exactly 30Hz, right where EV says it should be, and right where I’ve measured it before. Also, the 80Hz low pass modeled correctly as well. Satisfied that my schematic is correct, and the simulation is working properly, I began adjusting R23 and R29 according to the results I had calculate above shooting for a Q of .707 and recording where the actual f3 turned up. In most cases everything was dead on. The simulated response was pretty much right where the formula said it would be and had no hump or bump, hence critically damped, and perfect for my subsonic filter.

Now this is all fun and games, but how do I decide what cutoff is right for my sub to prevent overexcursion? And how will it affect the frequency response? This is where a great feature in Unibox 4.08 becomes extremely useful. There is the option to import an .frd file, which is summed with the box response giving you combined response of your enclosure and your active EQ. Using NI Multisim I was able to export each of the simulated FR plots and pull them into Unibox to see just how the different cutoff frequencies affect both the FR and the excursion. My first filter cutoff was set to 15Hz, and although it had very little impact on the FR of the sub, it still allowed the sub to be overdriven below 15Hz at 500W input power. So I adjusted the filter and imported a 19Hz cutoff point. Now the excursion is in check and never exceeds 22mm at 500W input power and the FR is minimally impacted. This is most likely the ideal spot, if I had just one cutoff point to choose, but the great thing about the XEQ-2, is that it offers six different cutoff points which are easily selected between with a switch on the front panel. This is where as massive DP6T (dual-pole-six-throw) switch comes in really handy. I can select resistors values in pairs for all 6 switch positions which correspond to 6 different cutoff frequencies and remove using any boost completely.

So the plan is to replace R23-R36 with resistor values which correspond to cutoffs of 12Hz, 14Hz, 16HZ, 18Hz, and 20Hz with no bass boost. Then the last switch setting of FLAT I will remove the resistors completely which basically removes the filter from the circuit and makes the response flat to 1 Hz. So when it says flat, it will truly mean flat.

Download the Service Manual Here

About Dan

For a complete biography, stop by www.danmarx.org and click around for a while.
This entry was posted in DIY Audio and tagged , , , , , . Bookmark the permalink.

1 Response to How to Properly Design a Subsonic Filter

  1. Josef Szaday says:

    Awesome….
    I just picked one of these XEQ-2 units up for this same purpose and didn’t know what I would need to do to accomplish this…
    Thank you and a GREAT job!

Leave a Reply

Your email address will not be published.