The Subwoofer DIY Page v1.1
A Hsu Subwoofer with a RD75 Dipole Baffle
last updated: 16 August 2010
Subwoofer DIY v1.1
- Discussion Forum
- Projects
- Links

Related web sites:
Hsu Research
Bohlender-Graebener

 

Brian asked the Bass List recently for 'True Tales of Woofer Integration'. In March 1998 Rudi Blondia and I measured several woofers and crossovers in conjunction with a pair of RD75 dipole baffles at a very large gymnasium. This was the final event for an off-Bass List group of planar/ribbon enthusiasts who were interested in woofer integration with the RD75 planar driver. The 'whole' story would be very long indeed, and would have to include much crossover theory, so this will be a somewhat abridged version.

One critical matter relating to theory is important. That is, the difference in sound attentuation rates at which a line-source and point-source fall as the distance from them increases. A line-source's fallrate of -3dB per doubling of distance is one-half the attenuation rate of a point-source. Line-sources will stay louder, farther from the source, than does a point-source. This physical reality can't be circumvented.

So what does it mean for woofer integration? There will be only ONE point distant from the line-source main speaker and point-source woofer combination at which their amplitude integration CAN be perfect. The output amplitudes of the two different types of drivers do not 'track' over distance. At distances beyond the choosen integration point the woofer's level will be too low; and, when closer than the integration point the woofer will be too loud. The user must choose the integration point's distance and adjust the amplitude levels and crossovers of the drivers accordingly.

A final note before starting: I'm frustrated with the data here and so is Brian. There's nothing to be done about the matter at this time though. Time at the gym was not sufficent to make the comprehesive measurements that would answer the many questions that arise from the data.

Rudi chose to measure his 1259 woofer system in close proximity to his dipole baffle, but I took a different tack, consistent with how I will use the Hsu/RD75 system in my livingroom. The Hsus will be along the sidewalls and the RD75s centered on the backwall about 2.5 feet out from it and about 7 feet apart. There is about 5 feet lateral separation between each Hsu and it's RD75. It is more difficult to get good looking response plots when the woofer and main speaker are separated, but in many cases we either cannot or choose not to put our subwoofers close to the main speakers.

The subwoofer being placed at a such a distance from the main speakers with a 125Hz crossover implies 'sweet spot' listening. Also, one does not wish to put large objects, like woofer cabinets, close to the main dipole speakers due to diffraction issues. For my situation these setup parameters are givens, and are consistent with two goals: 1. To have the RD75 baffles as unobstructed as possible to maximize the effect of their dipole side nulls and reduce diffraction/reflection problems; and 2. To implement a two-way loudspeaker system.

A Parapix active filter/amplifier combination was used in this study for it's active crossover functionality and also to provide amplification for the Hsu subwoofer. It was extensively modified, 'hacked', to as close an approximation of a 'Parametric active filter' concept as time and the situation permitted. Adjustments that one ususally never sees on commericial active filters, such as high & low pass 'Q' adjustments for altering the 'knee' of the filter's response were built-in.

Prior to the test session the 4th order Linkwitz-Riley low pass (to the Hsu) and the 4th order Butterworth (to the RD75) were set at 125Hz with high precision test equipment. The RD75/baffle combination starts it's low end rolloff at approximately that frequency, and in combination with the 4th order Butterworth filter approximates an 8th order Linkwitz-Riley electro-acoustic filter. The 125 Hz 4th order Butterworth high pass to the RD75 is the minimum frequency and slope required by the BG Corporation to maintain the warrenty.

At the Gym, during the course of preparations for the measurements the following additional crossover adjustments on the Parapix were 'eyeballed' by Rudi and I:

1. The 6th order high Pass boost to the Hsu vented subwoofer, which sets the lower frequency response limit of the subwoofer. It is not clear that a true 6th order high pass response was achieved.

2. The amplitude of the subwoofer bandpass relative to the RD75 output, which was set with the subwoofer 'online' with the RD75 baffle. This amplitude was not varied as the Hsu was moved forward relative to the RD75.

3. The low frequency phase control on the RD75 output. The phase control on the RD75 output turned out to be a complete bust, and we'll pretty much ignore it from this point on. It had an extremely small, if any, effect upon frequency response.

An aside about the Dr. Hsu's 6th order vented alignment. Although Dr. Hsu had previously given me an active filter circuit schematic for use with the system I chose to fabricate a different form of the circuit allowing me to vary both the high pass frequency (18Hz to 48Hz) and the Q (0.5 to ~5.0). I think if you will compare the Hsu's low end to that of Rudi's electronically boosted 1259 you will see that the Hsu can 'get way down'. Rudi's website: Audio-X-Stream

During the course of the adjustment of the low frequency boost/Q on the Hsu another issue arose. The Hsu and it's low pass active filter exhibited a 2nd order or -12dB/octave low pass response starting at 125Hz and for several octaves above that point, until finally reaching a fallrate of -24dB/octave rate or greater above 500Hz or so. The electrical response of the 4th order LR lowpass active filter to the Hsu was -6dB down at 125Hz!

So what does it mean? The Hsu clearly has a signicant 'rise' in response above the 125Hz point, say a +12dB/octave rise relative to the bass pass band (20-125Hz)! That rising frequency response clearly 'compensates' the active filter low pass fallrate back from a 4th order to a 2nd order response for a couple octaves above the crossover. Sadly there are no separate plots of either the Hsu or the RD75 baffle.

These are the problems that black box solutions cannot always ameliorate. Without system measurement the question always remains, 'what are the drivers doing?' Remember when you see the plots that the Hsu is NOT exhibiting a -24dB/octave low pass response at 125Hz. Nor is the RD75 exhibit a 24dB/octave high pass at 125Hz. What ultimately matters is the complete electro-acoustic response of the crossover/driver combination.

In my listening room the Hsus can be moved forward or backward along the sidewall relative to the RD75 baffles to improve time alignment. This is part of what the experiment/demonstration was attempting to illustrate, i.e. the effects of physical positioning between RD75 and subwoofer without extensive crossover tweaking.

The link following shows a schematic view of the test setup:

The Test Setup.

The RD75 baffle used in this experiment was one of my new 'equal area' baffles. No other measurements were made on it due to time limitations. It is a one-sided rectangular baffle, and can be expected to have all the benefits of a one-sided baffle relative to a two-sided baffle. It is exactly 15" in total width.

The Test Results.

The plots show that the range of affected response for the Hsu and RD75 integration extends to 600Hz. These are 'engineering' plots with a low level of smoothing (1/12 octave smoothing), no windowing on the FFTs, and 32 MLS averages.

The 1 foot 'infront' condition is flattest through the 125Hz crossover point, yet has the deepest null at ~325Hz. The majority of baffles tested so far at the gym exhibit a dipole cancellation null in the 300Hz range. We should probably conclude that this is an example of that null. We might additionally conclude that the null gets partically 'filled-in' when the Hsu is in different positions due to it's prodigious output above the crossover frequency.

It is clear that as the Hsu is moved farther from the RD75 baffle, closer to the microphone, that it's relative amplitude is increasing compared to the 'online' with the RD75 condition. This is expected from basic physical principles. The amplitude and other parameters the crossover to the Hsu/RD75 would normally be adjusted to best integrate with the RD75 at the listening position. This study shows what happens when those parameters are not adjusted.

When the Hsu is 4' infront of the RD75 do we judge the response to be -2dB at 20Hz (and NO room gain!!!)?

I was worried that the Hsu would have difficulty 'getting up' to 125Hz. This surely doesn't seem to be a problem given the results. Subjectively, the question of whether there is too much distance separation between the RD75s and the Hsus for a 125Hz crossover is a non-issue. The only time one can 'hear' the Hsus is when their level is too high - even with the 125Hz crossover! This is the first time I've implemented stereo subwoofers at home, and I must admit to a secret smile or two over a couple cuts from 'sonic spectacular' CDs that have wild side-to-side low frequency effects.

[Note: Text below is from my final 'Ave Atque Vale' message to the RD75/Woofer Integration Group.]

I believe the general understanding to be derived from the plots is of the sensitivity of the frequency response at crossover to all the variables at issue, including that of the physical time alignment between the drivers.

What Rudi and I have demonstrated to you is that you can create a TWO-WAY system using the RD75 and a subwoofer. This is an amazing thing when you consider the range of frequencies involved. The 125Hz crossover point possible with the RD75 driver opens a range of potential woofer options to the DIYer.

To pontificate a bit, one of my intentions in hacking the ParaPix amplifiers for use with the RD75 and Hsu subwoofer was to demonstate the possibility for the financially challenged DIYer (been there! still there!) to create a system which might approximate a commericial RD75 system costing $25,000 for somewhat less than 1/10th the cost. In fact, it might even be better than the $25K system with the proper dipole baffle selected for use with the RD75 driver. It is my understanding that the RD75s are still available for the original Bass List DIY price of $396 each through aftermarket vendors. It's a pretty good deal, ask an owner.

Go forth and integrate your woofers.

John Whittaker
May 12, 1998