Saturday, 4 June 2016

Speaker thoughts

As mentioned earlier (http://johnsraspi.blogspot.co.uk/2016/03/thinking-so-far.html) I have also been thinking about loudspeakers, especially active with DSP-implemented crossovers. This approach makes a great deal of sense to me:
  • Realising analogue crossovers is an incredible PitA, and has lots of technical drawbacks
    • Getting the right component values and adjusting them (actually impossible!)
    • The general inefficiency of driving lots of additional analogue crap as well as the speakers
    • Their phase response is horrible - and I'm utterly convinced that accurate phase response across the audio band is critical to good quality! For example, it's a characteristic of the Chord DACs which I've been very impressed with in the past
    • The only slight advantage is they generally only require a single amp per speaker, which is ok, but actually is another drawback - any distortion is reflected in all units, and the different demands of low and high frequencies are conflated
  • Digital crossovers, OTOH, implemented in software, allow incredible flexibility and power, including phase correction, response correction and time alignment
So I've been looking at the various approaches to DSP-implemented active crossovers...
  • Buy one - miniDSP for example
    • + easy peasy!! Relatively anyway. 
    • - cost - £200+ probably
    • - flexibility; not sure how true this is, but it would either be a standalone analogue in/out device, like the miniDSP 2x4, or more priceily, that or another variant with possible I2S or USB in/out; but if I want 2- or 3-way XOver capability, then I need enough digital/analogue outputs to enable that; the 2x4 only has 4 outputs, enough for the LXMini, but nothing else; how open-ended do I need it?
  • Use software on some handy platform...
    • RPi - appears to be powerful enough, certainly cheap!
    • Mac OSX - except I want to keep using my lappie...
    • Linux on a cheap PC - also pretty simple
  • IIR vs. FIR - seems to me that FIR is a Good Thing to aim for, if possible, purely because of the phase and impulse response management possible
So, assuming for the moment I'll be going with software-based, since I can do this without a huge outlay initially, to get some experience, I need 
  1. FIR parameter generating software, preferably easy to use
  2. An FIR implementation toolkit 
that can be run on something I have i.e. Mac, Windows or RPi, or some combination.

Currently, BruteFIR looks like a reasonable candidate for 2), and runs on RPi, admittedly less well than on Intel Pentium with the SSE instructions. RePhase is a possibility for 1), except it runs on Windows which means I have to borrow my wife's PC... Since she's currently away for the weekend, maybe now is a good time!

Then I'll need some speakers to test it out with, and so on. I guess I could try the old Boston A100 speakers hidden somewhere in the bedroom - bypass the passive crossover in a reversible way (since they actually belong to my wife!), strap some amplifiers up somehow, maybe just use the RPi and the Yamaha AV amp I have which will accept HDMI from the RPi, and amplify appropriately. Gotta be worth a try!!


No comments:

Post a Comment