Ian's Tinkerings and Musings: Adventures in DIY Hi Fi

Introduction

I meant to keep a blog of a couple of little projects I've started (and sometimes finished) over the last couple of years but never quite got around to it until now. As a result, this first post will be a bit long and probably tedious. I may post updates later on some other things which I've started earlier, so this isn't going to be quite as chronological as the term "blog" might suggest.

Bi-Amping Speakers

My most recent bit of tinkering has been to do with bi-amping my Tannoy 638 loudspeakers. I've always liked the idea of active speaker systems, and had some success at DIY bi-amping with my previous speakers, a pair of Mordaunt-Short MS-904s, so when I got the Tannoys I had the idea of bi-amping them as well. Both speakers are pictured right. This turned out to be a little trickier than I'd imagined.

638s are floor-standing 2 1/2 way speakers using an 8" dual concentric driver backed up by an 8" woofer below about 200Hz. Dual concentric drivers are the one genuine reason for buying Tannoy speakers. They're very clever bits of kit, and nothing does stereo imaging quite like them. But they do have flaws, and getting an active crossover that works well with them takes a bit of head-scratching and, unless you're a brilliant electronic engineer (I'm not), some experimentation.

First, I tried the crossover I'd been using with the 904s. That didn't work too badly, actually, but was not quite right. That was a 4th order Linkwitz-Riley, as described here: http://sound.westhost.com/project09.htm but using my own pcb and crossing over at 2.74 kHz - at that frequency, the separation of the driver centres is equal to the wavelength. With that crossover frequency, I'm surprised it sounded any good at all with the Tannoy's, which have a much bigger bass/mid driver, as you can see. Using a lower crossover frequency made the sound utterly dreadful and very "shouty" unless the the high frequency outputs were attenuated a lot - then it sounded muffled. Time to make some measurements.

Below are the frequency responses of the two parts of the dual concentric, 500Hz - 5kHz for the low frequency driver and 1kHz-10kHz for the tweeter. Measurements were made using a signal generator, microphone and oscilloscope and the vertical scale is arbitrary - amplitude of microphone signal relative to signal generator output expressed in dB.

You'll note the massive plateau from 2KHz to about 7kHz - the tweeter output continues to decline, but rather more slowly, up to 20kHz. Then there's the pit of oblivion in the low frequency driver's output at about 1.7 kHz. That would explain the shoutiness of a low crossover frequency and the muffling effect (i.e. no highs at all) of simply turning down the tweeter to try and fix it. A crossover which produces a flat summed electrical output isn't going to fix either of those.

So what to do? Some sort of band-reject filter to selectively bring down the region from 2-7 kHz coupled with an acoustic crossover frequency somewhere just below 2kHz so the tweeter can fill in that big dip and maybe lift the high frequencies above 10kHz a bit seemed like a sensible approach.

A little experimentation with a circuit simulator came up with a crossover whose electrical response looks like this

Which sounded a bit light at the top end. Well, muffled actually. Not surprising as the measured output looks like this

Note how the high frequencies are quieter than the low ones. Still, the great pit has gone although it's still not exactly a flat response even allowing for the attenuated treble. Moving the cut-off frequency of the high pass filter down an octave gave a worthwhile improvement.

That's brought the high frequencies nicely up to the same level as the lows. And that's the state of play as of today. They're sounding pretty good, even despite the ups and downs, notably the peak at 1.25 kHz and the dip at 630Hz. I may do some more tinkering with this, but for now, I'm going to listen to music through them for a while - that is why I did this, after all.

A couple more pictures before I go.

Here's the schematic for the crossover