The fundamental challenge of designing a dipole speaker is that the frequency response rolls off at 6db per octave as the frequency is reduced. This is a far cry from a flat frequency response! To counteract this, equalization is used to boost the power at lower frequencies relative to higher frequencies. This correction is applied by the Orion ASP throughout the passbands of the bass and midrange drivers. The conundrum is that if too much boost is applied, the drivers will run out of excursion capability and bottom out.
The Standard Solution
The drivers themselves must have enough linear excursion capability (Xmax) to be driven as far as the equalization demands in order for the equalization to work. Drive them too far and they will break. The power delivered to the woofers must be limited in some fashion.
If you follow Linkwitz’ recommendations and use 60 watt amps for each channel, the amp’s output voltage is not sufficient to drive the woofers beyond Xmax above 22Hz. (See his discussion and graphs here.) Above 88Hz, the amp’s current limit prevents the driver from reaching Xmax. In either case, using 60 watt amps limits the output capabilities of the woofers to about 72db at 22Hz and 94db at 88Hz. In this region, there is sufficient, unused excursion to allow significantly higher SPL before reaching Xmax if more power was available.
The recommended, smaller amps reduce potential damage to the woofers from exceeding Xmax. This is a simple solution to the problem. Another approach is to use the Orions with subwoofers integrated at 40Hz or even 50Hz, but that will not be discussed here.
Many Orion owners using 60 watt amp report that they don’t perceive any limitations. I noticed many problems with my first set of 60 watt amps when pushed to their limits. The bass distortion, though infrequent and momentary, was simply not acceptable. When auditioning replacement amps, I selected 200 watt amps to drive each woofer.
The extra power and headroom produces higher SPL and cleaner sound. But I was constantly worried about bottoming out the woofers and so I built a subwoofer integrated at 50Hz. This works very well for loud music, and also for restoring a sense of overall spaciousness and envelopment on classical recordings made in concert halls. But the quality of the bass—timbre, subtle and not-so-subtle envelopes on the bass, unusual sounding bass instruments—suffers a little bit with the subwoofer on.
Three years ago, I posted the idea of using a frequency-dependent limiter on the Orion User’s Group forum. Recently, I solved the problem by rigging up a frequency-dependent limiter system using an off-the-shelf stereo compressor and a sidechain filter. This is apparently not a new idea, as it’s mentioned in the manual for the dbx 166A compressor I got, and also discussed elsewhere, for example, in this article.
An old dbx 166A compressor is used as a limiter. It is inserted between the woofer outputs of the ASP and the woofer amps. The compressor ratio is set to hard limit the output voltage (set to 1:∞) when the input voltage exceeds the compression threshold. The signal that drives the limiter is a modified version of the input woofer signal, frequency-weighted so that lower frequency signals cause more limiting than higher frequencies. The limiting reduces the entire bass signal when an excessive peak is detected, not just the particular peak itself. This is the only drawback of the system, but is no worse in this respect than having a 60 watt amp clip and limit.
The control signal is modified by a filter in the compressor’s sidechain. It took several sets of filters and measurements to come up with one that matched a set of frequency/voltage targets. The final filter is a 4th order passive lowpass filter at 10Hz. The filter was designed specifically for this compressor’s sidechain. The sidechain’s output impedance is 2kΩ and its input impedance is 10kΩ. The filter is simply four cascaded 1st order RC filters made of 2kΩ resistors and 3µF capacitors.
This filter response was determined by first measuring the actual amp output voltages that drove a woofer to Xmax, comparing them against the predicted values from Linkwitz’ closed-box1.xls spreadsheet, and then calculating the compressor output voltages that would take the woofer to 90% of Xmax, taking account of the gain of my amps, 27db or 22.4x.
Here are some pictures showing how I measured displacement versus voltage.
This table shows modeled and projected displacements for an input voltage and the voltages needed to reach Xmax.
Once each filter was installed, I adjusted the compressor threshold so it would just start compressing when the input was 0.87v at 20Hz, giving the same output. For the second (and final) calibration of the 4th order filter, I adjusted the threshold for 3.1v/64Hz instead of 0.87v/20Hz to keep the voltage below the maximum at all frequencies.
The following table shows the measured voltages.
The filter slope deviates the most from a straight line as the filter transitions from passband to stopband at the deepest bass notes where the limiting needs to be most accurate. At lower frequencies, less power is needed to hit Xmax. I tried moving the filter knee down in frequency for more linear slope above 10Hz, but the insertion loss in the transition band was very high, approaching 50db at 20Hz for a filter frequency of 5Hz. To get straight line attenuation down to 10Hz without this much insertion loss requires an active filter with some gain set to an extremely low frequency perhaps two octaves below 10Hz at 2.5Hz.
Here are two photos of the compressor and sidechain filter box (unplugged) and the internals of the filter. The caps are surplus 1.5µF film caps, two in parallel per filter section. The box cost more than all the other parts combined!
A Practical Problem
I bought the dbx 166A compressor off craigslist. The seller threw in an extra one that wasn’t working. There was some odd behavior with the system and I wanted to compare the two compressors to eliminate them as the fault so I eventually fixed the bad one. An internal AC transformer fuse had blown, a resistor in one leg of each cross-coupled balanced driver had fried, and both balanced driver opamps had blown.
A problem showed up on about half of the CDs I played. It mostly happened during extremely soft passages with no bass, and on the fadeout of many songs. The compressor’s gate function would kick in and out very briefly, even though the gate threshold was set to OFF. The turn on and turn off was so quick that the gate’s LEDs would flicker almost imperceptibly while the gain reduction LEDs would all light up, indicating maximum compression. The result was a slight chuffing noise during these passages due to the harmonics added to the woofer signal not being masked by the quiet music.
I tried to isolate the problem by
- swapping compressors: both had the same problem
- adding a very low frequency blocking filter to the sidechain filter: no change
- disconnecting the gate circuit output from the rest of the circuitry inside the compressor
Disconnecting the gate circuit fixed the problem. My theory is that some anomaly in the signal (DC offset, low level low frequency bursts from faders, etc) is triggering the gate circuitry in the compressor. It could be a design flaw in the compressors, or it could be that both of them are out of spec.
I can listen to anything now through the Orions at any level without worrying about damaging the woofers. When the bass compression becomes noticeable on loud music, I just turn to the subwoofer. For most listening, the compressor as limiter works really well at keeping the woofers in their safe operating area while still providing more usable power at higher frequencies.
I’m not inclined to find a compressor with a gate that behaves better since things are working as they are now.
In response to Charles Port’s comment and my reply below, I’ve created the following graph showing the different SPL levels with 200w and 60w amps and the regions where the SPL is limited by Xmax, voltage, and current. This shows where the limiter protects the woofers from excess power, even protecting the woofers when very low frequency transients occur with a 60w amp.