Toccata Grande, Part 2

By Lars Mytting
20 Jan 1998

Losing phase

I spent many hours in front of the Calsod screen while working on the crossover. Getting five drivers to cooperate is about as easy as getting five cats to stay away from a salmon. Our trusty Clio-system was in use elsewhere at that moment, so I did all the tuning by ear. After simulating and listening to many different filters, I ended with something I found quite reasonable. I had offered my speaker kit dealer to listen to the Toccatas, and one late evening I did the final assembly of the crossover. Next day after work, we drove the speakers to the store.

That evening was not one of the best. We hooked the Toccatas to Electrocompaniet amplifiers, and fired them up in the well-damped listening room. The speakers sure had dynamic resources, but something was... very wrong. The midrange was very badly defined. It was all over nosy and dull, but at the same time it sounded undistorted. The tonal balance was quite weird.

Measurements made the next day showed further disgrace. The frequency response was so horribly bad, so disreputable that it looked like a printout from a seismograph after a earthquake. Nearly the whole midrange area was missing in a huge, wide dip.

The speakers travelled home. After a frantic debug of the speaker, I discovered that I -- during the late-night assembly -- had connected one of the midranges out of phase. This effectively cancelled out the whole midrange, and of course, it was this speaker that had been measured.

Correct phase can be checked by connecting a 1.5 volt battery to the drivers. If drivers are in phase, they will move in the same direction. The connection must bypass any series capacitors, as these will block the DC from the battery. Do not use this test on tweeters, as they may get damaged.

Get the Clio, find the facts
Somewhat later (and somewhat wiser), I cranked up the Clio measurement system and started on final tuning. The final crossover looks like this: The final filter is very simple. I had used as many as 16 components for the first filters, but with the aid of Calsod and Clio I managed to reduce the number to ten while maintaining a fairly flat frequency response and a phase response without too many flaws Fig. 9. Crossover schematic. Measured crossover frequencies (driver rolloff included) are ca. 350 Hz and 2000 Hz. All rolloffs are (approximately) 4th order Linkwitz-Riley response Fig. 10. Frequency response and phase Fig. 11. Frequency response in room.

The dual drivers in bass and midrange results in 4 ohm impedance, which reduces inductor values (and their costs) to the half. The high sensitivity is demonstrated by the fact that there is no damping resistor for the tweeter. By calculating from Seas' measured figures of their kits with the T25, I estimate the sensitivity to 94 dB.

Depending on room, amplifier and taste, I recommend using a resistor with a value between 1-3 ohm in series with midrange and tweeter if you find the sound too bright. This will of course reduce the average sensitivity with 2-3 dB.

The impedance reveals some flaws Fig. 12. Impedance and electrical phase. Impedance is generally flat, but also low, averagely 5 ohm. Highest impedance is 7,8 ohm at 43 Hz, and lowest is 3,0 ohm at 480 Hz. Notable are the highly damped peaks in the bass region. The amplifier meets a troublesome area from 90-200 Hz, with 3,2 ohm at 130 Hz. The steepest phase-shift is 38º at 90 Hz. However, a modern amplifier in healthy shape will be able to cope with these demands, especially since the sensitivity is high.

About the
author

Print this
article

Email this
article