Next generation active speaker electronics
Posted 3rd December 2013 at 05:12 AM by abraxalito
Updated 18th December 2013 at 02:59 PM by abraxalito
Updated 18th December 2013 at 02:59 PM by abraxalito
I'd like to be sure that my amp and the passive XOs in my speakers aren't significantly the bottleneck in the sound I'm getting from my DAC prototypes. To get a second reference point I'm starting to build a passive line-level crossover. Here's a pic of it taking shape - its using AD605s because I have plenty to hand and they're cheap and capable of being subjectively transparent with the right power arrangements. The pot core inductors are implementing LR4 XOs in balanced. Using the AD605s with their low input resistance and noise allows manageable inductor sizes for the XO - the largest being 9mH for a 3.5kHz crossover frequency.
Ah - I forgot to mention how I propose to do the transformation from normal (2V) line levels to the 70ohm impedance/20mV paradigm at the AD605 inputs. I shall be winding a pair of ferrite-cored step down transformers with centre-tapped secondaries
Lots of turns to look forward to with 0.1mm or perhaps finer wire as I'd like to get at least 5H primary inductance 
Update - progress is very gradual - new pic attached. There are new hand-wound inductors for power supply filtering and output load decoupling. Also the HP section needs them as an ultrasonic filter to keep IMD down as low as possible. With judicious tweaking of the Q here this LPF can be made to do a very passable impression of NOS droop correction. The clump of TO-92s bottom centre is an experiment at using multiple TL431s to get lower noise for the PSU...
I spent some time looking for cheap low-noise regs and whilst there are low noise regs, none I'd call cheap. TL431 is noisy but does have low cost firmly on its side. I can buy over 100 for one LTC6655 which would be my next choice. LM329 is good too but too high voltage for this application. And although the DS plot for TL431 shows far from outstanding noise, the waveform shown for 0.1 - 10Hz noise (the bit that LPF filtering can't deal with easily without huge caps) looks very respectable when set alongside other offerings. So TL431 it is for this experiment
Ah - I forgot to mention how I propose to do the transformation from normal (2V) line levels to the 70ohm impedance/20mV paradigm at the AD605 inputs. I shall be winding a pair of ferrite-cored step down transformers with centre-tapped secondaries
Lots of turns to look forward to with 0.1mm or perhaps finer wire as I'd like to get at least 5H primary inductance Update - progress is very gradual - new pic attached. There are new hand-wound inductors for power supply filtering and output load decoupling. Also the HP section needs them as an ultrasonic filter to keep IMD down as low as possible. With judicious tweaking of the Q here this LPF can be made to do a very passable impression of NOS droop correction. The clump of TO-92s bottom centre is an experiment at using multiple TL431s to get lower noise for the PSU...
I spent some time looking for cheap low-noise regs and whilst there are low noise regs, none I'd call cheap. TL431 is noisy but does have low cost firmly on its side. I can buy over 100 for one LTC6655 which would be my next choice. LM329 is good too but too high voltage for this application. And although the DS plot for TL431 shows far from outstanding noise, the waveform shown for 0.1 - 10Hz noise (the bit that LPF filtering can't deal with easily without huge caps) looks very respectable when set alongside other offerings. So TL431 it is for this experiment Total Comments 0
