[B]Part 1. [/B] This is my first post on the Audiosmile Forum, although I’ve been quite active elsewhere. This series of posts was prompted by Simon who suggested that I post here as my experiences might be of interest to others. This is a long post so I’ve split it into several parts. Please feel free to comment. I’ve been a user of active ‘speakers since the mid ‘80s as eliminating the high-level passive crossover and driving each loudspeaker with its own amplifier has always seems a far more sensible approach. I had Meridian ‘speakers since the 80s, firstly the M2s, then DSP5000 supplemented by DSP1500 subwoofers. However, after some 13 years with the DSP5000s, I wanted to try something different, especially a bit of DIY. The loudspeaker I had always admired was the B&W 801. I heard these at various studios, and always thought the best sound I’d ever heard came from a pair of 801s in Danmark Radio’s IEC listening rooms. To cut a long story short, I bought a pair of 801Fs, which were the second series, still with a sealed bass bin rather than the later reflex Matrix series, but with the improved Fibrecrete head rather than wood (hence the 801F). I planned from the start to make them active, given that if they’re that good passive, they should be rather better (or at least no worse) active. Using a DSP crossover with DSP parametric and graphic equalisation, I should be able to set them up to a high level of accuracy. Having had bass-reflex speakers for 20+ years, and Transmission Lines (IMF TLS50II) before that, I was quite keen to try sealed box bass for a change. I got them home and listened to them as they were, driven by an old Yamaha CR1000 receiver, and compared them with the Meridians, but took no measurements yet. They did some things better than the Meridians, like vocal clarity, and the bass was different ( A single large sealed box rather than bass-reflex) but I couldn’t work out if any better, just different. Anyway, I was happy that they were both working correctly, given that they were 30 years old, and had come originally from Decca Studios so would have had some considerable usage. [IMG]http://i1200.photobucket.com/albums/bb328/sergeauckland/801OriginalLores.jpg[/IMG] I had previously decided to use the Behringer DCX2496 as a crossover, supplemented by the DEQ2496 Digital EQ to provide fine equalisation control. I got the DCX on order (I already had the DEQ) and looked for power amps. I needed 6 channels, and decided that to provide the necessary headroom, I would like 100 watts per channel as that would provide some 105dBSPL capability, given that the passive system was rated at 85dBSPL/watt. In fact, with active systems, the midrange and treble are often more efficient that the bass unit, and with music energy being predominately bass-heavy, I could have used smaller amplifiers for the midrange and top, but decided to keep all the amps the same. I looked around at new and second-hand, thinking in terms perhaps of three Quad 405s, but decided I could get three brand-new Behringer A500 power amps for rather less money than three used Quads, and get balanced ins, power metering and a warranty. I ordered the three A500s and started on the project. The Behringer A500 is very cheap, around £150 for a 130+watt stereo amplifier, so it might be thought of as being hopelessly compromised or flawed. Not a bit of it! It’s well made inside, and whilst the performance is hardly state-of-the-art, and its published spec is somewhat optimistic, it is transparent by any objective measure. All three amplifiers measured identically, so consistency was good! I got close to 30 volts out into 8 ohms at the onset of clipping, which is 115watts with <0.06% distortion at both 1kHz and 10kHz. Of greater interest was the distortion at 1Watt, which is the level I’m much more likely to be listening at for most of the time. Here it was 0.07% at both frequencies, albeit with a bit of crossover distortion visible. Whilst I would have preferred no visible crossover distortion, at these levels it’s nothing to worry about and at the price, I couldn’t build myself anything better. As I have often said, adequate i.e. transparent, is all I need and the A500 is perfectly adequate. Due to a minor fault (low batteries), my generator was giving out 0.05% distortion, so I can expect the amplifier distortion to have been lower than noted above. One of these days, I might get round to remeasuring the amplifier distortion. Output impedance was a little higher than spec at 0.125 ohms, but that’s also low enough not to be of concern. All in all for around £150, I can’t fault the A500. Noise was within spec at -100dB A weighted, but only -90dB unweighted, more about this later. The DCX crossover is so simple to use, it’s nice when a product has been well designed. All I needed to do was to dial-in the crossover frequencies, which I kept to B&Ws as I’m sure they got that right, and decide on what sort of crossover slopes and type. There’s no good reason I can think of not to use Linkwitz-Riley (Bessel and Butterworth are the other options) and the greatest slope available which is 48dB/octave. The original passive crossovers were 24dB/octave, and that’s about as steep as can be achieved with passive components, and the B&W crossover is pretty complicated at that. As I had the option of 48dB/octave, I decided to use it as it would give better defined bandwidths to each driver, so reducing the out-of-band signals they were sent and therefore reducing distortion, especially intermodulation distortion. The DCX has both analogue and digital inputs, and 6 analogue outputs. I noticed that if a 0dBFS signal was put in, the analogue output clipped. Reducing the internal gain even by 0.2dB cleared the clipping, so I set the internal gain to -0.5dB. This problem I’ve found remarkably common with DAC products from several manufacturers. I rejected several “pro” ADC-DAC devices as the analogue outputs would clip on 0dBFS. It was irritating, but simple to get round, but I wouldn’t have noticed it if I hadn’t done the measurements. So, with the amplifiers, crossovers and equalisers available, I could start on the project proper. The first thing was to measure what the 801s were doing passive. I had the original manuals with the individual factory-produced frequency response graphs, but whether it was differences in the measuring methods or 30 years of use, my measurements indicated that the frequency response variations were larger than the original graphs, with a pronounced extreme HF droop after 16kHz (not that I can hear that any more) which was also rather worse on one channel than the other. [IMG]http://i1200.photobucket.com/albums/bb328/sergeauckland/801F20Freq20Plot20lores.jpg[/IMG] The interchannel differences across the frequency range were also rather worse than the 0.25dB originally specced. My self-imposed task was to achieve a frequency response to within +-1dB (above 200Hz) and with a 0.5dB channel balance. Using ARTA impulse measuring software I can get pseudo-anechoic frequency response measurements, but only above 200Hz due to the limitations of my room. To get lower, it would need a much larger room (or better still, open air on a high platform) such that the early reflections take longer to reach the measuring microphone. In any event, below 200Hz the effect of the room will swamp whatever errors the loudspeaker has so as long as these errors are sensibly small, it’s not necessary to equalise LF to the same extent as mid and HF. I was also expecting that the distortion would be less due to driving each loudspeaker directly rather than through a passive crossover. More on this later. S.