Audio - Bi-amping

System hookup

A

challenge I recently took on was to bi-amp my Magnepan IIBs. I managed to get hold of two beautiful Bryston 4BSST amps and a matching Bryston 10B-LR active crossover. The active crossover splits the line level frequencies and feeds the amps, one per side, and bypasses the Magnepan factory crossovers going direct to the high and low panels. I don’t have any issues with the factory passive crossover, I just wanted to be able to dedicate a channel to the highs and probably more importantly, have a dedicated channel for the lows.

Magnepan IIB back panel

The Magnepan MG-IIB model is just begging to be bi-amped. The friendly folks at Magnepan were thoughtful enough to build the IIB with removable jumpers making bi-amping a simple plug-and-press-PLAY►. To use the internal passive crossover, three jumpers are located across the posts as labeled (3-4, 5-6, and 7-8) and the single pair of speaker cables connect across the posts labeled "AMPLIFIER" (1 and 2). In order to bypass the internal crossover, all three jumpers are pulled, the low is connected to posts 3 and 5, and the high is connected to posts 4 and 8. The jumpers can then kept in a location you will soon forget.

Here is an excerpt from page 6 of the Magneplanar MG II B Instruction Manual:

CROSSOVER POINT--DUE TO THE FREQUENCY RESPONSE OF THE INDIVIDUAL DRIVER SECTIONS IN THE MG-IIB, IT IS NECESSARY TO SPREAD THE CROSSOVER POINTS. THE LOW PASS (BASS) CROSSOVER POINT IS SET (-3DB) ELECTRICALLY AT 400HZ. THE HIGH PASS (TWEETER) CROSSOVER POINT IS SET (-3DB) ELECTRICALLY AT 1600HZ. BOTH SLOPES ARE 6DB PER OCTAVE. THE ARRANGEMENT RESULTS IN A FLAT ACOUSTICAL RESPONSE. TO OBTAIN THE IDEAL ACOUSTICAL RESPONSE AS PROVIDED BY THE CROSSOVER BUILT INTO THE SPEAKER, YOUR CROSSOVER MUST BE SET AT:
     LOW PASS--400HZ, -3DB WITH 6DB PER OCTAVE SLOPE
     HIGH PASS--1600HZ, -3DB WITH 6DB PER OCTAVE SLOPE

Apparently in the days of Air Supply and acid-washed jeans, audio manuals were written in all-caps. Perhaps there is a joke in there about audiophiles being deaf and needing to yell - but I'm not going to touch it. Interestingly though, Magnepan chose to "spread" the low and high pass - I'm guessing this is to flatten the lobing error produced by the lower order Butterworth crossover and create what they refer to as a "flat acoustical response".

Magnepan IIB factory crossover

Remembering that the frequency doubles for every octave, there is a gap of two octaves between 400Hz and 1600Hz. The midway point is 800Hz and is reached at -9dB with the 6^dB/_oct. slope.

So to summarize, these are the factory specs:

Low-Pass (LP):	(400Hz, -3dB) at 6dB/oct.
High-Pass (HP):	(1600Hz, -3dB) at 6dB/oct.

This gives an effective crossover point of 800Hz at -9dB.

Without getting into a “what’s-the-optimal-crossover-point” discussion, for now I trust the brains at Magnepan and I will keep the same crossover frequency of 800Hz as designed. But since the 10B-LR’s Linkwitz-Riley filters are formed by cascading two second order filters, it has a much steeper slope of 24^dB/_oct.. The question then became, what should the new low-pass (LP') and high-pass (HP') be?

Using the same low-pass and high-pass as the factory specs would create a huge gap. I knew my new LP' and HP' would need to be much closer. I decided to work under the assumption that Magnepan's primary motivation for the "spread" was compensation for the slow roll-off of the Butterworth filters and avoidance of a lobing error. Fortunately with a Linkwitz-Riley filter I don't need to worry about a spread at all since it produces moderate roll-off and a flat sum, resulting in zero lobing error. Without a spread, both my LP' and HP' can be 800Hz at a slightly louder -6dB.

Proposed crossover

(Un)fortunately I learned all this because of some hindsight. Initially I thought I needed a ½-octave gap between the LP' and HP'. I was thinking the low-pass and high-pass frequencies where defined at 0dB, not the -6dB for Linkwitz-Riley filters. I didn't discover this mistake until I got it all together and heard an audible gap around the crossover point. I reviewed my logic, dug deeper, learned some more and found my error. (Note, the pictures below are for incorrect LP and HP.)

So the following are my desired specs:

Low-Pass (LP'):	(800Hz, -6dB) at 24dB/oct.
High-Pass (HP'):	(800Hz, -6dB) at 24dB/oct.

This gives an effective crossover point of 800Hz at -6dB.

The programming boards


Single channel Linkwitz-Riley board

As you might have noticed, there are no frequency selection knobs on the faceplate of the 10B-LR. Instead of presets, the low and high-pass are set using "programming boards" which plug into each channel’s pull-out circuit board. Each of these programming boards (a total of four per channel) is programmed using a pair of ±1% metal film fixed ¼W resistors. At a price, Bryston will build these little boards for your preferred frequency – but it’s really easy to reuse and modify existing boards with some simple de-soldering.

Doing some rounding and using the BRYSTON 10B-LR Crossover Additional Frequency/Resistor Tables^[PDF] document, the following resistors are used:

LP' at 806Hz			HP' at 806Hz
Ra:	6.98kΩ		R1:	29.4kΩ
Rb:	6.98kΩ		R2:	59.0kΩ

Note that there are multiple tables for different versions of the 10B-LR based on the capacitors used on the channel. In this case the low-pass is C₁ = 40nF and C₂ = 20nF and high-pass is C_a = C_b = 4.7nF.

Not only do Magnepans look unusual, these also run at a nonstandard 5Ω. Each channel of the 4BSST delivers a hearty 300W at 8Ω. So admittedly, it is a bit too much umph for these circa 1981 speakers, however I’ve found it quite safe up to just over half volume on my Bryston BP-25 DA.

The listening room

I’m very pleased with this active crossover configuration. The low end is exceptional – so much so, I now need to do a DIY on sound-deadening wall panels to make my listening room acoustically larger. The highs of course are also noticeably better. When the hungry Maggies demand more power to drive the bass panels, the channels dedicated to the highs remain unfazed. The Linkwitz-Riley filters are amazing - I can't believe it's not Butterworth... Sorry, uncalled for. (Actually I guess it is Butterworth, just two in series)

Besides all that, it looks pretty cool too.