@donavabdear
This should put your mind at ease about the Bryston Bax-1 external crossover:
Technical Note by: David A. Rich
Unlike the analog active crossover network in the ELAC Navis loudspeakers or Carlo’s old active system that he referenced, Bryston chose to go the DSP route in the BAX-1 crossover. There are several intelligent reasons for this:
-
- 1. One can design a much more complex filter in DSP compared to a few inductors, resistors, and capacitors in a passive or even an analog active system.
- 2. DSP allows filters that cannot be synthesized in the passive domain. Finite impulse response filters are possible. Computation that is not causal is possible – one can look forward and back at a digital signal to determine a sample point to be sent to a driver. Phase correction is possible with non-causal filters although it is unclear if that does anything positive. Bryston is not correcting the phase here.
- 3. Drivers present complex impedances. This interacts with a passive crossover. In an active system, with the amplifier directly connected to the drivers, the speaker sees a flat response at the terminals. The acoustic performance of the driver is isolated. The non-ideal response of the driver, now isolated, can be corrected with the DSP filter. Custom compensation for a specific driver shipped with the speakers is possible although this presents issues if a replacement is needed as the DSP code must be updated.
- 4. In passive speakers, crossover components see very large voltage swings and high current flows in them. This results in non-linearities that can be measured easily and are high enough to be audible, especially in iron core inductors and electrolytic capacitors. Moving to air-core inductors and film capacitors are prohibitively expensive at lower crossover frequencies. Passive speakers also have very complex impedance plots and can be difficult to drive. This is due to the passive crossover network and the overlapping impedances of the drivers. Dealing with an individual driver gives us a much easier impedance to drive. What comes out of the amplifier terminal in an active design winds up at the voice coil exactly.
- 5. Passive components have tolerances. Capacitors and inductors are within 5% or even 10% of the specified value. The speaker’s impedance also varies from sample to sample. Combine all the tolerance changes of each component and the desired frequency response of the speaker has changed by a significant amount. This results in pair matching errors that effect stereo imaging. DSP has no such tolerance errors. Every box produces the same response.
- 6. Bryston’s approach has significant advantages for those who want to keep a speaker for a long time. The amplifiers inside an active speaker have little space and occasional ventilation problems. Parts availability may be 5 years or less if the company still exists. The DSP section of an all in one box speaker may also have parts that cannot be replaced and sometimes the company puts a limit on the time it will repair the electronics.
- 7. To ensure 10 years or more of uptime the DSP filter code should be portable to a newer external DSP box if the old one becomes obsolete or is not reparable.
- 8. Bryston can continue development of the DSP code for the speaker. If they make an improvement, all they have to do is send you the code to download.
(see the measurement graph using this link for the comments addressing it below)
Above is a curve pair of the response of the passive Middle T and active Middle T provided by Bryston. Each curve shows the listening window response and power response. Note that they are not flattening the listening window with DSP completely. You would see the active curve get as flat as ice if they did that.
What Bryston does is use the fact that they are applying DSP correction to each driver, not to the whole speaker.
That gives an extra degree of freedom when applying DSP filters to each driver span around the crossover points. For the woofer – midrange the crossover is 500Hz and they overlap between 300Hz – 1.8kHz. The midrange – tweeter crossover at 2kHz and the overlap runs from about 2.2kHz to 3.5kHz. In these regions, Bryston can adjust the sound power and the listening window in different directions. This is something other speaker designers may take advantage of but they talk of it in very technical terms. Bryston has come up with a brilliant explanation and it never occurred to me before that it could be this easy to explain.
The graph pair above shows the Bryston concept in practice. Both the early reflection curve and the power spectrum have been improved. Outside the crossover area, the changes track each other. In the crossover, the correction becomes almost independent.
