Ken Perkins on Baffle
Step Correction in Series Cross-Overs
Hey Andy,
I think I've done it! I
finally figured out an easy to do way to implement BSC for the simple,
classic first order series crossover. It doesn't waste
any more power that a parallel xo w/BSC, doesn't need an eq circuit
before the xo and only uses one more part. It's actually
inspired by the AR sxo and some of the work Tony Gee has done and
should be easy to calculate within your spreadsheet.
First, you determine
the frequency where the sound begins to roll off below the rise
caused by the baffle gain. For a typical "bookshelf"
style stand-mount, this seems to be about 300 to 600 hz. Calculate
a coil value needed for a textbook first order Butterworth parallel
low-pass, using this frequency, and this coil is then wired directly
in series with the woofer.
To the crossover, this
coil will appear to be part of the woofer's natural inductance if
the DCR of the coil is VERY low. Add a slightly over-damped
zobel like Jon Risch describes (textbook R*1.25 and textbook C*1.4)
and place it in the normal position on the woofer. Then
subtract approx. 3 dB from the published SPL/w/m of the woofer and
calculate the l-pad from this new point. For instance,
if the woofer is 89 dB/w/m, it will become 86 and THIS becomes the
reference point for the l-pad, since this is the actual "new"
mid-band response of the woofer after the BSC coil does it business.
I'm also going to try
putting the zobel BEFORE the BSC inductor and woofer,as if they
were a whole unit, to see if I can completely neutralize theTOTAL
inductance of this part of the circuit. This new zobel
will require a resistor about the same value as normal but a cap
of roughly twice the size of normal.
I've included two schematics
for you to see. The first one is my current crossover
for the HiVi BG8N woofer and SEAS 27TBFC/G tweeter and it sounds
excellent! The BSC balance is perfect and the crossover
is occurring at about 2.8k with a zeta of 1.1-ish. The
second schematic is one I'm going to try which will taken into account
and neutralize the total inductance the xo sees.
BTW, I've modelled
this circuit in SpeakerWorkshop and it's doing exactly what I think
it is. The beauty part is that it can be calculated easily
without modelling software and is predictable if the woofer has
no severe break-up to tame and the tweeter can handle first order
series. I suspect the second order type nature of the
woofer section can actually REDUCE a not too severe break-up in
the woofer there is one and if used properly.
Having fun! Ken
